FUNCTION_PTR Parser::Parse_DeclareFunction(int *token_id, const char *fn_name, bool is_local)
{
FUNCTION_PTR ptr = (FUNCTION_PTR)POV_MALLOC(sizeof(FUNCTION), "Function ID");
ExprNode *expression = NULL;
FunctionCode function;
// default type is float function
*token_id = FUNCT_ID_TOKEN;
FNCode f(this, &function, is_local, fn_name);
f.Parameter();
Parse_Begin();
Get_Token();
if(Token.Token_Id == INTERNAL_TOKEN)
{
GET(LEFT_PAREN_TOKEN);
Get_Token();
if(Token.Function_Id != FLOAT_TOKEN)
Expectation_Error("internal function identifier");
expression = FNSyntax_GetTrapExpression((unsigned int)(Token.Token_Float));
function.flags = FN_INLINE_FLAG;
GET(RIGHT_PAREN_TOKEN);
}
else if(Token.Token_Id == TRANSFORM_TOKEN)
{
if(function.parameter_cnt != 0)
Error("Function parameters for transform functions are not allowed.");
expression = FNSyntax_GetTrapExpression(1); // 1 refers to POVFPU_TrapSTable[1] = f_transform [trf]
function.private_copy_method = (FNCODE_PRIVATE_COPY_METHOD)Copy_Transform;
function.private_destroy_method = (FNCODE_PRIVATE_DESTROY_METHOD)Destroy_Transform;
function.private_data = reinterpret_cast<void *>(Parse_Transform_Block());
function.return_size = 3; // returns a 3d vector!!!
// function type is vector function
*token_id = VECTFUNCT_ID_TOKEN;
}
else if(Token.Token_Id == SPLINE_TOKEN)
{
if(function.parameter_cnt != 0)
Error("Function parameters for spline functions are not allowed.");
Experimental_Flag |= EF_SPLINE;
expression = FNSyntax_GetTrapExpression(2); // 2 refers to POVFPU_TrapSTable[2] = f_spline [trf]
function.private_copy_method = (FNCODE_PRIVATE_COPY_METHOD)Copy_Spline;
function.private_destroy_method = (FNCODE_PRIVATE_DESTROY_METHOD)Destroy_Spline;
Parse_Begin();
function.private_data = reinterpret_cast<void *>(Parse_Spline());
Parse_End();
function.return_size = (reinterpret_cast<SPLINE *>(function.private_data))->Terms; // returns a 2d, 3d, 4d or 5d vector!!!
// function type is vector function
*token_id = VECTFUNCT_ID_TOKEN;
}
else if(Token.Token_Id == PIGMENT_TOKEN)
{
if(function.parameter_cnt != 0)
Error("Function parameters for pigment functions are not allowed.");
expression = FNSyntax_GetTrapExpression(0); // 0 refers to POVFPU_TrapSTable[0] = f_pigment [trf]
function.private_copy_method = (FNCODE_PRIVATE_COPY_METHOD)Copy_Pigment;
function.private_destroy_method = (FNCODE_PRIVATE_DESTROY_METHOD)Destroy_Pigment;
Parse_Begin();
function.private_data = reinterpret_cast<void *>(Create_Pigment());
Parse_Pigment(reinterpret_cast<PIGMENT **>(&function.private_data));
Parse_End();
Post_Pigment(reinterpret_cast<PIGMENT *>(function.private_data));
function.return_size = 5; // returns a color!!!
// function type is vector function
*token_id = VECTFUNCT_ID_TOKEN;
}
else if(Token.Token_Id == PATTERN_TOKEN)
{
if(function.parameter_cnt != 0)
Error("Function parameters for pattern functions are not allowed.");
expression = FNSyntax_GetTrapExpression(77); // 77 refers to POVFPU_TrapTable[77] = f_pattern [trf]
function.private_copy_method = (FNCODE_PRIVATE_COPY_METHOD)Copy_Pigment;
function.private_destroy_method = (FNCODE_PRIVATE_DESTROY_METHOD)Destroy_Pigment;
Parse_Begin();
function.private_data = reinterpret_cast<void *>(Create_Pigment()); // Yes, this is a pigment! [trf]
Parse_PatternFunction(reinterpret_cast<TPATTERN *>(function.private_data));
Parse_End();
Post_Pigment(reinterpret_cast<PIGMENT *>(function.private_data));
}
else if(Token.Token_Id == STRING_LITERAL_TOKEN)
{
f.SetFlag(2, Token.Token_String);
Get_Token();
if(Token.Token_Id == COMMA_TOKEN)
{
Get_Token();
if(Token.Token_Id != STRING_LITERAL_TOKEN)
Expectation_Error("valid function expression");
f.SetFlag(1, Token.Token_String);
}
else
{
Unget_Token();
expression = FNSyntax_ParseExpression();
}
}
else
{
Unget_Token();
expression = FNSyntax_ParseExpression();
}
f.Compile(expression);
FNSyntax_DeleteExpression(expression);
Parse_End();
*ptr = sceneData->functionVM->AddFunction(&function);
return ptr;
}
static void suni_hz(unsigned long from_timer)
{
struct suni_priv *walk;
struct atm_dev *dev;
struct k_sonet_stats *stats;
for (walk = sunis; walk; walk = walk->next) {
dev = walk->dev;
stats = &walk->sonet_stats;
PUT(0,MRI); /* latch counters */
udelay(1);
ADD_LIMITED(section_bip,(GET(RSOP_SBL) & 0xff) |
((GET(RSOP_SBM) & 0xff) << 8));
ADD_LIMITED(line_bip,(GET(RLOP_LBL) & 0xff) |
((GET(RLOP_LB) & 0xff) << 8) |
((GET(RLOP_LBM) & 0xf) << 16));
ADD_LIMITED(path_bip,(GET(RPOP_PBL) & 0xff) |
((GET(RPOP_PBM) & 0xff) << 8));
ADD_LIMITED(line_febe,(GET(RLOP_LFL) & 0xff) |
((GET(RLOP_LF) & 0xff) << 8) |
((GET(RLOP_LFM) & 0xf) << 16));
ADD_LIMITED(path_febe,(GET(RPOP_PFL) & 0xff) |
((GET(RPOP_PFM) & 0xff) << 8));
ADD_LIMITED(corr_hcs,GET(RACP_CHEC) & 0xff);
ADD_LIMITED(uncorr_hcs,GET(RACP_UHEC) & 0xff);
ADD_LIMITED(rx_cells,(GET(RACP_RCCL) & 0xff) |
((GET(RACP_RCC) & 0xff) << 8) |
((GET(RACP_RCCM) & 7) << 16));
ADD_LIMITED(tx_cells,(GET(TACP_TCCL) & 0xff) |
((GET(TACP_TCC) & 0xff) << 8) |
((GET(TACP_TCCM) & 7) << 16));
}
if (from_timer) mod_timer(&poll_timer,jiffies+HZ);
}
static int HalfSize(GLint components, GLint width, GLint height, const unsigned char *data, unsigned char *d, int filter) {
int x, y, c;
int line = width*components;
if (width > 1 && height > 1) {
if (filter)
for (y = 0; y < height; y += 2) {
for (x = 0; x < width; x += 2) {
for (c = 0; c < components; c++) {
*d++ = (GET(0)+GET(components)+GET(line)+GET(line+components)) / 4;
data++;
}
data += components;
}
data += line;
}
else
for (y = 0; y < height; y += 2) {
for (x = 0; x < width; x += 2) {
for (c = 0; c < components; c++) {
*d++ = GET(0);
data++;
}
data += components;
}
data += line;
}
}
else if (width > 1 && height == 1) {
if (filter)
for (y = 0; y < height; y += 1) {
for (x = 0; x < width; x += 2) {
for (c = 0; c < components; c++) {
*d++ = (GET(0)+GET(components)) / 2;
data++;
}
data += components;
}
}
else
for (y = 0; y < height; y += 1) {
for (x = 0; x < width; x += 2) {
for (c = 0; c < components; c++) {
*d++ = GET(0);
data++;
}
data += components;
}
}
}
else if (width == 1 && height > 1) {
if (filter)
for (y = 0; y < height; y += 2) {
for (x = 0; x < width; x += 1) {
for (c = 0; c < components; c++) {
*d++ = (GET(0)+GET(line)) / 2;
data++;
}
}
data += line;
}
else
for (y = 0; y < height; y += 2) {
for (x = 0; x < width; x += 1) {
for (c = 0; c < components; c++) {
*d++ = GET(0);
data++;
}
}
data += line;
}
}
else {
return 0;
}
return 1;
}
size_t
do_scrub_chars (size_t (*get) (char *, size_t), char *tostart, size_t tolen)
{
char *to = tostart;
char *toend = tostart + tolen;
char *from;
char *fromend;
size_t fromlen;
register int ch, ch2 = 0;
/* Character that started the string we're working on. */
static char quotechar;
/*State 0: beginning of normal line
1: After first whitespace on line (flush more white)
2: After first non-white (opcode) on line (keep 1white)
3: after second white on line (into operands) (flush white)
4: after putting out a .linefile, put out digits
5: parsing a string, then go to old-state
6: putting out \ escape in a "d string.
7: no longer used
8: no longer used
9: After seeing symbol char in state 3 (keep 1white after symchar)
10: After seeing whitespace in state 9 (keep white before symchar)
11: After seeing a symbol character in state 0 (eg a label definition)
-1: output string in out_string and go to the state in old_state
-2: flush text until a '*' '/' is seen, then go to state old_state
#ifdef TC_V850
12: After seeing a dash, looking for a second dash as a start
of comment.
#endif
#ifdef DOUBLEBAR_PARALLEL
13: After seeing a vertical bar, looking for a second
vertical bar as a parallel expression separator.
#endif
#ifdef TC_PREDICATE_START_CHAR
14: After seeing a predicate start character at state 0, looking
for a predicate end character as predicate.
15: After seeing a predicate start character at state 1, looking
for a predicate end character as predicate.
#endif
#ifdef TC_Z80
16: After seeing an 'a' or an 'A' at the start of a symbol
17: After seeing an 'f' or an 'F' in state 16
#endif
*/
/* I added states 9 and 10 because the MIPS ECOFF assembler uses
constructs like ``.loc 1 20''. This was turning into ``.loc
120''. States 9 and 10 ensure that a space is never dropped in
between characters which could appear in an identifier. Ian
Taylor, [email protected].
I added state 11 so that something like "Lfoo add %r25,%r26,%r27" works
correctly on the PA (and any other target where colons are optional).
Jeff Law, [email protected].
I added state 13 so that something like "cmp r1, r2 || trap #1" does not
get squashed into "cmp r1,r2||trap#1", with the all important space
between the 'trap' and the '#1' being eliminated. [email protected] */
/* This macro gets the next input character. */
#define GET() \
(from < fromend \
? * (unsigned char *) (from++) \
: (saved_input = NULL, \
fromlen = (*get) (input_buffer, sizeof input_buffer), \
from = input_buffer, \
fromend = from + fromlen, \
(fromlen == 0 \
? EOF \
: * (unsigned char *) (from++))))
/* This macro pushes a character back on the input stream. */
#define UNGET(uch) (*--from = (uch))
/* This macro puts a character into the output buffer. If this
character fills the output buffer, this macro jumps to the label
TOFULL. We use this rather ugly approach because we need to
handle two different termination conditions: EOF on the input
stream, and a full output buffer. It would be simpler if we
always read in the entire input stream before processing it, but
I don't want to make such a significant change to the assembler's
memory usage. */
#define PUT(pch) \
do \
{ \
*to++ = (pch); \
if (to >= toend) \
goto tofull; \
} \
while (0)
if (saved_input != NULL)
{
from = saved_input;
fromend = from + saved_input_len;
}
else
{
fromlen = (*get) (input_buffer, sizeof input_buffer);
if (fromlen == 0)
return 0;
from = input_buffer;
fromend = from + fromlen;
}
while (1)
{
/* The cases in this switch end with continue, in order to
branch back to the top of this while loop and generate the
next output character in the appropriate state. */
switch (state)
{
case -1:
ch = *out_string++;
if (*out_string == '\0')
{
state = old_state;
old_state = 3;
}
PUT (ch);
continue;
case -2:
for (;;)
{
do
{
ch = GET ();
if (ch == EOF)
{
as_warn (_("end of file in comment"));
goto fromeof;
}
if (ch == '\n')
PUT ('\n');
}
while (ch != '*');
while ((ch = GET ()) == '*')
;
if (ch == EOF)
{
as_warn (_("end of file in comment"));
goto fromeof;
}
if (ch == '/')
break;
UNGET (ch);
}
state = old_state;
UNGET (' ');
continue;
case 4:
ch = GET ();
if (ch == EOF)
goto fromeof;
else if (ch >= '0' && ch <= '9')
PUT (ch);
else
{
while (ch != EOF && IS_WHITESPACE (ch))
ch = GET ();
if (ch == '"')
{
quotechar = ch;
state = 5;
old_state = 3;
PUT (ch);
}
else
{
while (ch != EOF && ch != '\n')
ch = GET ();
state = 0;
PUT (ch);
}
}
continue;
case 5:
/* We are going to copy everything up to a quote character,
with special handling for a backslash. We try to
optimize the copying in the simple case without using the
GET and PUT macros. */
{
char *s;
ptrdiff_t len;
for (s = from; s < fromend; s++)
{
ch = *s;
if (ch == '\\'
|| ch == quotechar
|| ch == '\n')
break;
}
len = s - from;
if (len > toend - to)
len = toend - to;
if (len > 0)
{
memcpy (to, from, len);
to += len;
from += len;
if (to >= toend)
goto tofull;
}
}
ch = GET ();
if (ch == EOF)
{
/* This buffer is here specifically so
that the UNGET below will work. */
static char one_char_buf[1];
as_warn (_("end of file in string; '%c' inserted"), quotechar);
state = old_state;
from = fromend = one_char_buf + 1;
fromlen = 1;
UNGET ('\n');
PUT (quotechar);
}
else if (ch == quotechar)
{
state = old_state;
PUT (ch);
}
#ifndef NO_STRING_ESCAPES
else if (ch == '\\')
{
state = 6;
PUT (ch);
}
#endif
else if (scrub_m68k_mri && ch == '\n')
{
/* Just quietly terminate the string. This permits lines like
bne label loop if we haven't reach end yet. */
state = old_state;
UNGET (ch);
PUT ('\'');
}
else
{
PUT (ch);
}
continue;
case 6:
state = 5;
ch = GET ();
switch (ch)
{
/* Handle strings broken across lines, by turning '\n' into
'\\' and 'n'. */
case '\n':
UNGET ('n');
add_newlines++;
PUT ('\\');
continue;
case EOF:
as_warn (_("end of file in string; '%c' inserted"), quotechar);
PUT (quotechar);
continue;
case '"':
case '\\':
case 'b':
case 'f':
case 'n':
case 'r':
case 't':
case 'v':
case 'x':
case 'X':
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
break;
default:
#ifdef ONLY_STANDARD_ESCAPES
as_warn (_("unknown escape '\\%c' in string; ignored"), ch);
#endif
break;
}
PUT (ch);
continue;
#ifdef DOUBLEBAR_PARALLEL
case 13:
ch = GET ();
if (ch != '|')
abort ();
/* Reset back to state 1 and pretend that we are parsing a
line from just after the first white space. */
state = 1;
PUT ('|');
#ifdef TC_TIC6X
/* "||^" is used for SPMASKed instructions. */
ch = GET ();
if (ch == EOF)
goto fromeof;
else if (ch == '^')
PUT ('^');
else
UNGET (ch);
#endif
continue;
#endif
#ifdef TC_Z80
case 16:
/* We have seen an 'a' at the start of a symbol, look for an 'f'. */
ch = GET ();
if (ch == 'f' || ch == 'F')
{
state = 17;
PUT (ch);
}
else
{
state = 9;
break;
}
case 17:
/* We have seen "af" at the start of a symbol,
a ' here is a part of that symbol. */
ch = GET ();
state = 9;
if (ch == '\'')
/* Change to avoid warning about unclosed string. */
PUT ('`');
else if (ch != EOF)
UNGET (ch);
break;
#endif
}
/* OK, we are somewhere in states 0 through 4 or 9 through 11. */
/* flushchar: */
ch = GET ();
#ifdef TC_PREDICATE_START_CHAR
if (ch == TC_PREDICATE_START_CHAR && (state == 0 || state == 1))
{
state += 14;
PUT (ch);
continue;
}
else if (state == 14 || state == 15)
{
if (ch == TC_PREDICATE_END_CHAR)
{
state -= 14;
PUT (ch);
ch = GET ();
}
else
{
PUT (ch);
continue;
}
}
#endif
recycle:
#if defined TC_ARM && defined OBJ_ELF
/* We need to watch out for .symver directives. See the comment later
in this function. */
if (symver_state == NULL)
{
if ((state == 0 || state == 1) && ch == symver_pseudo[0])
symver_state = symver_pseudo + 1;
}
else
{
/* We advance to the next state if we find the right
character. */
if (ch != '\0' && (*symver_state == ch))
++symver_state;
else if (*symver_state != '\0')
/* We did not get the expected character, or we didn't
get a valid terminating character after seeing the
entire pseudo-op, so we must go back to the beginning. */
symver_state = NULL;
else
{
/* We've read the entire pseudo-op. If this is the end
of the line, go back to the beginning. */
if (IS_NEWLINE (ch))
symver_state = NULL;
}
}
#endif /* TC_ARM && OBJ_ELF */
#ifdef TC_M68K
/* We want to have pseudo-ops which control whether we are in
MRI mode or not. Unfortunately, since m68k MRI mode affects
the scrubber, that means that we need a special purpose
recognizer here. */
if (mri_state == NULL)
{
if ((state == 0 || state == 1)
&& ch == mri_pseudo[0])
mri_state = mri_pseudo + 1;
}
else
{
/* We advance to the next state if we find the right
character, or if we need a space character and we get any
whitespace character, or if we need a '0' and we get a
'1' (this is so that we only need one state to handle
``.mri 0'' and ``.mri 1''). */
if (ch != '\0'
&& (*mri_state == ch
|| (*mri_state == ' '
&& lex[ch] == LEX_IS_WHITESPACE)
|| (*mri_state == '0'
&& ch == '1')))
{
mri_last_ch = ch;
++mri_state;
}
else if (*mri_state != '\0'
|| (lex[ch] != LEX_IS_WHITESPACE
&& lex[ch] != LEX_IS_NEWLINE))
{
/* We did not get the expected character, or we didn't
get a valid terminating character after seeing the
entire pseudo-op, so we must go back to the
beginning. */
mri_state = NULL;
}
else
{
/* We've read the entire pseudo-op. mips_last_ch is
either '0' or '1' indicating whether to enter or
leave MRI mode. */
do_scrub_begin (mri_last_ch == '1');
mri_state = NULL;
/* We continue handling the character as usual. The
main gas reader must also handle the .mri pseudo-op
to control expression parsing and the like. */
}
}
#endif
if (ch == EOF)
{
if (state != 0)
{
as_warn (_("end of file not at end of a line; newline inserted"));
state = 0;
PUT ('\n');
}
goto fromeof;
}
switch (lex[ch])
{
case LEX_IS_WHITESPACE:
do
{
ch = GET ();
}
while (ch != EOF && IS_WHITESPACE (ch));
if (ch == EOF)
goto fromeof;
if (state == 0)
{
/* Preserve a single whitespace character at the
beginning of a line. */
state = 1;
UNGET (ch);
PUT (' ');
break;
}
#ifdef KEEP_WHITE_AROUND_COLON
if (lex[ch] == LEX_IS_COLON)
{
/* Only keep this white if there's no white *after* the
colon. */
ch2 = GET ();
if (ch2 != EOF)
UNGET (ch2);
if (!IS_WHITESPACE (ch2))
{
state = 9;
UNGET (ch);
PUT (' ');
break;
}
}
#endif
if (IS_COMMENT (ch)
|| ch == '/'
|| IS_LINE_SEPARATOR (ch)
|| IS_PARALLEL_SEPARATOR (ch))
{
if (scrub_m68k_mri)
{
/* In MRI mode, we keep these spaces. */
UNGET (ch);
PUT (' ');
break;
}
goto recycle;
}
/* If we're in state 2 or 11, we've seen a non-white
character followed by whitespace. If the next character
is ':', this is whitespace after a label name which we
normally must ignore. In MRI mode, though, spaces are
not permitted between the label and the colon. */
if ((state == 2 || state == 11)
&& lex[ch] == LEX_IS_COLON
&& ! scrub_m68k_mri)
{
state = 1;
PUT (ch);
break;
}
switch (state)
{
case 1:
/* We can arrive here if we leave a leading whitespace
character at the beginning of a line. */
goto recycle;
case 2:
state = 3;
if (to + 1 < toend)
{
/* Optimize common case by skipping UNGET/GET. */
PUT (' '); /* Sp after opco */
goto recycle;
}
UNGET (ch);
PUT (' ');
break;
case 3:
#ifndef TC_KEEP_OPERAND_SPACES
/* For TI C6X, we keep these spaces as they may separate
functional unit specifiers from operands. */
if (scrub_m68k_mri)
#endif
{
/* In MRI mode, we keep these spaces. */
UNGET (ch);
PUT (' ');
break;
}
goto recycle; /* Sp in operands */
case 9:
case 10:
#ifndef TC_KEEP_OPERAND_SPACES
if (scrub_m68k_mri)
#endif
{
/* In MRI mode, we keep these spaces. */
state = 3;
UNGET (ch);
PUT (' ');
break;
}
state = 10; /* Sp after symbol char */
goto recycle;
case 11:
if (LABELS_WITHOUT_COLONS || flag_m68k_mri)
state = 1;
else
{
/* We know that ch is not ':', since we tested that
case above. Therefore this is not a label, so it
must be the opcode, and we've just seen the
whitespace after it. */
state = 3;
}
UNGET (ch);
PUT (' '); /* Sp after label definition. */
break;
default:
BAD_CASE (state);
}
break;
case LEX_IS_TWOCHAR_COMMENT_1ST:
ch2 = GET ();
if (ch2 == '*')
{
for (;;)
{
do
{
ch2 = GET ();
if (ch2 != EOF && IS_NEWLINE (ch2))
add_newlines++;
}
while (ch2 != EOF && ch2 != '*');
while (ch2 == '*')
ch2 = GET ();
if (ch2 == EOF || ch2 == '/')
break;
/* This UNGET will ensure that we count newlines
correctly. */
UNGET (ch2);
}
if (ch2 == EOF)
as_warn (_("end of file in multiline comment"));
ch = ' ';
goto recycle;
}
#ifdef DOUBLESLASH_LINE_COMMENTS
else if (ch2 == '/')
{
do
{
ch = GET ();
}
while (ch != EOF && !IS_NEWLINE (ch));
if (ch == EOF)
as_warn ("end of file in comment; newline inserted");
state = 0;
PUT ('\n');
break;
}
#endif
else
{
if (ch2 != EOF)
UNGET (ch2);
if (state == 9 || state == 10)
state = 3;
PUT (ch);
}
break;
case LEX_IS_STRINGQUOTE:
quotechar = ch;
if (state == 10)
{
/* Preserve the whitespace in foo "bar". */
UNGET (ch);
state = 3;
PUT (' ');
/* PUT didn't jump out. We could just break, but we
know what will happen, so optimize a bit. */
ch = GET ();
old_state = 3;
}
else if (state == 9)
old_state = 3;
else
old_state = state;
state = 5;
PUT (ch);
break;
#ifndef IEEE_STYLE
case LEX_IS_ONECHAR_QUOTE:
#ifdef H_TICK_HEX
if (state == 9 && enable_h_tick_hex)
{
char c;
c = GET ();
as_warn ("'%c found after symbol", c);
UNGET (c);
}
#endif
if (state == 10)
{
/* Preserve the whitespace in foo 'b'. */
UNGET (ch);
state = 3;
PUT (' ');
break;
}
ch = GET ();
if (ch == EOF)
{
as_warn (_("end of file after a one-character quote; \\0 inserted"));
ch = 0;
}
if (ch == '\\')
{
ch = GET ();
if (ch == EOF)
{
as_warn (_("end of file in escape character"));
ch = '\\';
}
else
ch = process_escape (ch);
}
sprintf (out_buf, "%d", (int) (unsigned char) ch);
/* None of these 'x constants for us. We want 'x'. */
if ((ch = GET ()) != '\'')
{
#ifdef REQUIRE_CHAR_CLOSE_QUOTE
as_warn (_("missing close quote; (assumed)"));
#else
if (ch != EOF)
UNGET (ch);
#endif
}
if (strlen (out_buf) == 1)
{
PUT (out_buf[0]);
break;
}
if (state == 9)
old_state = 3;
else
old_state = state;
state = -1;
out_string = out_buf;
PUT (*out_string++);
break;
#endif
case LEX_IS_COLON:
#ifdef KEEP_WHITE_AROUND_COLON
state = 9;
#else
if (state == 9 || state == 10)
state = 3;
else if (state != 3)
state = 1;
#endif
PUT (ch);
break;
case LEX_IS_NEWLINE:
/* Roll out a bunch of newlines from inside comments, etc. */
if (add_newlines)
{
--add_newlines;
UNGET (ch);
}
/* Fall through. */
case LEX_IS_LINE_SEPARATOR:
state = 0;
PUT (ch);
break;
case LEX_IS_PARALLEL_SEPARATOR:
state = 1;
PUT (ch);
break;
#ifdef TC_V850
case LEX_IS_DOUBLEDASH_1ST:
ch2 = GET ();
if (ch2 != '-')
{
if (ch2 != EOF)
UNGET (ch2);
goto de_fault;
}
/* Read and skip to end of line. */
do
{
ch = GET ();
}
while (ch != EOF && ch != '\n');
if (ch == EOF)
as_warn (_("end of file in comment; newline inserted"));
state = 0;
PUT ('\n');
break;
#endif
#ifdef DOUBLEBAR_PARALLEL
case LEX_IS_DOUBLEBAR_1ST:
ch2 = GET ();
if (ch2 != EOF)
UNGET (ch2);
if (ch2 != '|')
goto de_fault;
/* Handle '||' in two states as invoking PUT twice might
result in the first one jumping out of this loop. We'd
then lose track of the state and one '|' char. */
state = 13;
PUT ('|');
break;
#endif
case LEX_IS_LINE_COMMENT_START:
/* FIXME-someday: The two character comment stuff was badly
thought out. On i386, we want '/' as line comment start
AND we want C style comments. hence this hack. The
whole lexical process should be reworked. xoxorich. */
if (ch == '/')
{
ch2 = GET ();
if (ch2 == '*')
{
old_state = 3;
state = -2;
break;
}
else
{
UNGET (ch2);
}
}
if (state == 0 || state == 1) /* Only comment at start of line. */
{
int startch;
startch = ch;
do
{
ch = GET ();
}
while (ch != EOF && IS_WHITESPACE (ch));
if (ch == EOF)
{
as_warn (_("end of file in comment; newline inserted"));
PUT ('\n');
break;
}
if (ch < '0' || ch > '9' || state != 0 || startch != '#')
{
/* Not a cpp line. */
while (ch != EOF && !IS_NEWLINE (ch))
ch = GET ();
if (ch == EOF)
{
as_warn (_("end of file in comment; newline inserted"));
PUT ('\n');
}
else /* IS_NEWLINE (ch) */
{
/* To process non-zero add_newlines. */
UNGET (ch);
}
state = 0;
break;
}
/* Looks like `# 123 "filename"' from cpp. */
UNGET (ch);
old_state = 4;
state = -1;
if (scrub_m68k_mri)
out_string = "\tlinefile ";
else
out_string = "\t.linefile ";
PUT (*out_string++);
break;
}
#ifdef TC_D10V
/* All insns end in a char for which LEX_IS_SYMBOL_COMPONENT is true.
Trap is the only short insn that has a first operand that is
neither register nor label.
We must prevent exef0f ||trap #1 to degenerate to exef0f ||trap#1 .
We can't make '#' LEX_IS_SYMBOL_COMPONENT because it is
already LEX_IS_LINE_COMMENT_START. However, it is the
only character in line_comment_chars for d10v, hence we
can recognize it as such. */
/* An alternative approach would be to reset the state to 1 when
we see '||', '<'- or '->', but that seems to be overkill. */
if (state == 10)
PUT (' ');
#endif
/* We have a line comment character which is not at the
start of a line. If this is also a normal comment
character, fall through. Otherwise treat it as a default
character. */
if (strchr (tc_comment_chars, ch) == NULL
&& (! scrub_m68k_mri
|| (ch != '!' && ch != '*')))
goto de_fault;
if (scrub_m68k_mri
&& (ch == '!' || ch == '*' || ch == '#')
&& state != 1
&& state != 10)
goto de_fault;
/* Fall through. */
case LEX_IS_COMMENT_START:
#if defined TC_ARM && defined OBJ_ELF
/* On the ARM, `@' is the comment character.
Unfortunately this is also a special character in ELF .symver
directives (and .type, though we deal with those another way).
So we check if this line is such a directive, and treat
the character as default if so. This is a hack. */
if ((symver_state != NULL) && (*symver_state == 0))
goto de_fault;
#endif
#ifdef TC_ARM
/* For the ARM, care is needed not to damage occurrences of \@
by stripping the @ onwards. Yuck. */
if (to > tostart && *(to - 1) == '\\')
/* Do not treat the @ as a start-of-comment. */
goto de_fault;
#endif
#ifdef WARN_COMMENTS
if (!found_comment)
as_where (&found_comment_file, &found_comment);
#endif
do
{
ch = GET ();
}
while (ch != EOF && !IS_NEWLINE (ch));
if (ch == EOF)
as_warn (_("end of file in comment; newline inserted"));
state = 0;
PUT ('\n');
break;
#ifdef H_TICK_HEX
case LEX_IS_H:
/* Look for strings like H'[0-9A-Fa-f] and if found, replace
the H' with 0x to make them gas-style hex characters. */
if (enable_h_tick_hex)
{
char quot;
quot = GET ();
if (quot == '\'')
{
UNGET ('x');
ch = '0';
}
else
UNGET (quot);
}
/* FALL THROUGH */
#endif
case LEX_IS_SYMBOL_COMPONENT:
if (state == 10)
{
/* This is a symbol character following another symbol
character, with whitespace in between. We skipped
the whitespace earlier, so output it now. */
UNGET (ch);
state = 3;
PUT (' ');
break;
}
#ifdef TC_Z80
/* "af'" is a symbol containing '\''. */
if (state == 3 && (ch == 'a' || ch == 'A'))
{
state = 16;
PUT (ch);
ch = GET ();
if (ch == 'f' || ch == 'F')
{
state = 17;
PUT (ch);
break;
}
else
{
state = 9;
if (ch == EOF || !IS_SYMBOL_COMPONENT (ch))
{
if (ch != EOF)
UNGET (ch);
break;
}
}
}
#endif
if (state == 3)
state = 9;
/* This is a common case. Quickly copy CH and all the
following symbol component or normal characters. */
if (to + 1 < toend
&& mri_state == NULL
#if defined TC_ARM && defined OBJ_ELF
&& symver_state == NULL
#endif
)
{
char *s;
ptrdiff_t len;
for (s = from; s < fromend; s++)
{
int type;
ch2 = *(unsigned char *) s;
type = lex[ch2];
if (type != 0
&& type != LEX_IS_SYMBOL_COMPONENT)
break;
}
if (s > from)
/* Handle the last character normally, for
simplicity. */
--s;
len = s - from;
if (len > (toend - to) - 1)
len = (toend - to) - 1;
if (len > 0)
{
PUT (ch);
memcpy (to, from, len);
to += len;
from += len;
if (to >= toend)
goto tofull;
ch = GET ();
}
}
/* Fall through. */
default:
de_fault:
/* Some relatively `normal' character. */
if (state == 0)
{
state = 11; /* Now seeing label definition. */
}
else if (state == 1)
{
state = 2; /* Ditto. */
}
else if (state == 9)
{
if (!IS_SYMBOL_COMPONENT (ch))
state = 3;
}
else if (state == 10)
{
if (ch == '\\')
{
/* Special handling for backslash: a backslash may
be the beginning of a formal parameter (of a
macro) following another symbol character, with
whitespace in between. If that is the case, we
output a space before the parameter. Strictly
speaking, correct handling depends upon what the
macro parameter expands into; if the parameter
expands into something which does not start with
an operand character, then we don't want to keep
the space. We don't have enough information to
make the right choice, so here we are making the
choice which is more likely to be correct. */
if (to + 1 >= toend)
{
/* If we're near the end of the buffer, save the
character for the next time round. Otherwise
we'll lose our state. */
UNGET (ch);
goto tofull;
}
*to++ = ' ';
}
state = 3;
}
PUT (ch);
break;
}
}
/*NOTREACHED*/
fromeof:
/* We have reached the end of the input. */
return to - tostart;
tofull:
/* The output buffer is full. Save any input we have not yet
processed. */
if (fromend > from)
{
saved_input = from;
saved_input_len = fromend - from;
}
else
saved_input = NULL;
return to - tostart;
}
/*
* Actual conversion; called from iconv()
*/
size_t
_icv_iconv(struct _icv_state *st, char **inbuf, size_t *inbytesleft,
char **outbuf, size_t *outbytesleft)
{
int cset, stat;
unsigned char *op, ic, offset;
char *ip;
size_t ileft, oleft;
size_t retval;
cset = st->_st_cset;
stat = st->_st_stat;
if ((inbuf == 0) || (*inbuf == 0)) {
cset = CS_0;
stat = ST_INIT;
op = (unsigned char *)*outbuf;
oleft = *outbytesleft;
retval = 0;
goto ret2;
}
ip = *inbuf;
op = (unsigned char *)*outbuf;
ileft = *inbytesleft;
oleft = *outbytesleft;
offset = 0;
/* Everything down to here was taken unchanged from @(#)ISO-2022-JP%SJIS.
=======================================================================
*
* Main loop; basically 1 loop per 1 input byte
*/
while (ileft > 0)
{
GET(ic);
/*
If the char is one of the following [ / ] { | } then convert
it to its corresponding value. In all other cases if the char
is greater than octal \178 ( ie a high bit char) convert it
to an underscore (_), as it has no mapping to 7 bit ASCII.
Otrherwise the char is the same in both cose sets.
*/
ic=__ib2_to_wi2[ic];
PUT(ic);
/*
Put the converted character into the output buffer, and decrement
the count of chars left in both the in and out buffers.
If we have no space left in the out buffer, but we have no reached
the end of the input buffer. We return what we have, and set the
errno (Error) to E2BIG.
*/
if ((oleft < 1) && (ileft > 0))
{
errno = E2BIG;
retval = ERR_RETURN;
goto ret;
}
}
/*
We only get here if the end of the in buffer has been reached, we therefore return the
value 0 to denote that we have sucesfully converted the inbuffer.
*/
retval = ileft;
/* Taken unchanged from @(#)ISO-2022-JP%SJIS. */
ret:
st->_st_cset = cset;
st->_st_stat = stat;
*inbuf = ip;
*inbytesleft = ileft;
ret2:
*outbuf = (char *)op;
*outbytesleft = oleft;
return (retval);
}
static int
set_start_bits(const uschar *code, uschar *start_bits, BOOL caseless,
BOOL utf8, compile_data *cd)
{
register int c;
int yield = SSB_DONE;
int table_limit = utf8? 16:32;
#if 0
/* ========================================================================= */
/* The following comment and code was inserted in January 1999. In May 2006,
when it was observed to cause compiler warnings about unused values, I took it
out again. If anybody is still using OS/2, they will have to put it back
manually. */
/* This next statement and the later reference to dummy are here in order to
trick the optimizer of the IBM C compiler for OS/2 into generating correct
code. Apparently IBM isn't going to fix the problem, and we would rather not
disable optimization (in this module it actually makes a big difference, and
the pcre module can use all the optimization it can get). */
volatile int dummy;
/* ========================================================================= */
#endif
do
{
const uschar *tcode = code + (((int)*code == OP_CBRA)? 3:1) + LINK_SIZE;
BOOL try_next = TRUE;
while (try_next) /* Loop for items in this branch */
{
int rc;
switch(*tcode)
{
/* Fail if we reach something we don't understand */
default:
return SSB_FAIL;
/* If we hit a bracket or a positive lookahead assertion, recurse to set
bits from within the subpattern. If it can't find anything, we have to
give up. If it finds some mandatory character(s), we are done for this
branch. Otherwise, carry on scanning after the subpattern. */
case OP_BRA:
case OP_SBRA:
case OP_CBRA:
case OP_SCBRA:
case OP_ONCE:
case OP_ASSERT:
rc = set_start_bits(tcode, start_bits, caseless, utf8, cd);
if (rc == SSB_FAIL) return SSB_FAIL;
if (rc == SSB_DONE) try_next = FALSE; else
{
do tcode += GET(tcode, 1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
}
break;
/* If we hit ALT or KET, it means we haven't found anything mandatory in
this branch, though we might have found something optional. For ALT, we
continue with the next alternative, but we have to arrange that the final
result from subpattern is SSB_CONTINUE rather than SSB_DONE. For KET,
return SSB_CONTINUE: if this is the top level, that indicates failure,
but after a nested subpattern, it causes scanning to continue. */
case OP_ALT:
yield = SSB_CONTINUE;
try_next = FALSE;
break;
case OP_KET:
case OP_KETRMAX:
case OP_KETRMIN:
return SSB_CONTINUE;
/* Skip over callout */
case OP_CALLOUT:
tcode += 2 + 2*LINK_SIZE;
break;
/* Skip over lookbehind and negative lookahead assertions */
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
do tcode += GET(tcode, 1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
break;
/* Skip over an option setting, changing the caseless flag */
case OP_OPT:
caseless = (tcode[1] & PCRE_CASELESS) != 0;
tcode += 2;
break;
/* BRAZERO does the bracket, but carries on. */
case OP_BRAZERO:
case OP_BRAMINZERO:
if (set_start_bits(++tcode, start_bits, caseless, utf8, cd) == SSB_FAIL)
return SSB_FAIL;
/* =========================================================================
See the comment at the head of this function concerning the next line,
which was an old fudge for the benefit of OS/2.
dummy = 1;
========================================================================= */
do tcode += GET(tcode,1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
break;
/* SKIPZERO skips the bracket. */
case OP_SKIPZERO:
tcode++;
do tcode += GET(tcode,1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
break;
/* Single-char * or ? sets the bit and tries the next item */
case OP_STAR:
case OP_MINSTAR:
case OP_POSSTAR:
case OP_QUERY:
case OP_MINQUERY:
case OP_POSQUERY:
tcode = set_table_bit(start_bits, tcode + 1, caseless, cd, utf8);
break;
/* Single-char upto sets the bit and tries the next */
case OP_UPTO:
case OP_MINUPTO:
case OP_POSUPTO:
tcode = set_table_bit(start_bits, tcode + 3, caseless, cd, utf8);
break;
/* At least one single char sets the bit and stops */
case OP_EXACT: /* Fall through */
tcode += 2;
case OP_CHAR:
case OP_CHARNC:
case OP_PLUS:
case OP_MINPLUS:
case OP_POSPLUS:
(void)set_table_bit(start_bits, tcode + 1, caseless, cd, utf8);
try_next = FALSE;
break;
/* Special spacing and line-terminating items. These recognize specific
lists of characters. The difference between VSPACE and ANYNL is that the
latter can match the two-character CRLF sequence, but that is not
relevant for finding the first character, so their code here is
identical. */
case OP_HSPACE:
SET_BIT(0x09);
SET_BIT(0x20);
if (utf8)
{
SET_BIT(0xC2); /* For U+00A0 */
SET_BIT(0xE1); /* For U+1680, U+180E */
SET_BIT(0xE2); /* For U+2000 - U+200A, U+202F, U+205F */
SET_BIT(0xE3); /* For U+3000 */
}
else SET_BIT(0xA0);
try_next = FALSE;
break;
case OP_ANYNL:
case OP_VSPACE:
SET_BIT(0x0A);
SET_BIT(0x0B);
SET_BIT(0x0C);
SET_BIT(0x0D);
if (utf8)
{
SET_BIT(0xC2); /* For U+0085 */
SET_BIT(0xE2); /* For U+2028, U+2029 */
}
else SET_BIT(0x85);
try_next = FALSE;
break;
/* Single character types set the bits and stop. Note that if PCRE_UCP
is set, we do not see these op codes because \d etc are converted to
properties. Therefore, these apply in the case when only characters less
than 256 are recognized to match the types. */
case OP_NOT_DIGIT:
set_nottype_bits(start_bits, cbit_digit, table_limit, cd);
try_next = FALSE;
break;
case OP_DIGIT:
set_type_bits(start_bits, cbit_digit, table_limit, cd);
try_next = FALSE;
break;
/* The cbit_space table has vertical tab as whitespace; we have to
ensure it is set as not whitespace. */
case OP_NOT_WHITESPACE:
set_nottype_bits(start_bits, cbit_space, table_limit, cd);
start_bits[1] |= 0x08;
try_next = FALSE;
break;
/* The cbit_space table has vertical tab as whitespace; we have to
not set it from the table. */
case OP_WHITESPACE:
c = start_bits[1]; /* Save in case it was already set */
set_type_bits(start_bits, cbit_space, table_limit, cd);
start_bits[1] = (start_bits[1] & ~0x08) | c;
try_next = FALSE;
break;
case OP_NOT_WORDCHAR:
set_nottype_bits(start_bits, cbit_word, table_limit, cd);
try_next = FALSE;
break;
case OP_WORDCHAR:
set_type_bits(start_bits, cbit_word, table_limit, cd);
try_next = FALSE;
break;
/* One or more character type fudges the pointer and restarts, knowing
it will hit a single character type and stop there. */
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEPOSPLUS:
tcode++;
break;
case OP_TYPEEXACT:
tcode += 3;
break;
/* Zero or more repeats of character types set the bits and then
try again. */
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEPOSUPTO:
tcode += 2; /* Fall through */
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEPOSSTAR:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEPOSQUERY:
switch(tcode[1])
{
default:
case OP_ANY:
case OP_ALLANY:
return SSB_FAIL;
case OP_HSPACE:
SET_BIT(0x09);
SET_BIT(0x20);
if (utf8)
{
SET_BIT(0xC2); /* For U+00A0 */
SET_BIT(0xE1); /* For U+1680, U+180E */
SET_BIT(0xE2); /* For U+2000 - U+200A, U+202F, U+205F */
SET_BIT(0xE3); /* For U+3000 */
}
else SET_BIT(0xA0);
break;
case OP_ANYNL:
case OP_VSPACE:
SET_BIT(0x0A);
SET_BIT(0x0B);
SET_BIT(0x0C);
SET_BIT(0x0D);
if (utf8)
{
SET_BIT(0xC2); /* For U+0085 */
SET_BIT(0xE2); /* For U+2028, U+2029 */
}
else SET_BIT(0x85);
break;
case OP_NOT_DIGIT:
set_nottype_bits(start_bits, cbit_digit, table_limit, cd);
break;
case OP_DIGIT:
set_type_bits(start_bits, cbit_digit, table_limit, cd);
break;
/* The cbit_space table has vertical tab as whitespace; we have to
ensure it gets set as not whitespace. */
case OP_NOT_WHITESPACE:
set_nottype_bits(start_bits, cbit_space, table_limit, cd);
start_bits[1] |= 0x08;
break;
/* The cbit_space table has vertical tab as whitespace; we have to
avoid setting it. */
case OP_WHITESPACE:
c = start_bits[1]; /* Save in case it was already set */
set_type_bits(start_bits, cbit_space, table_limit, cd);
start_bits[1] = (start_bits[1] & ~0x08) | c;
break;
case OP_NOT_WORDCHAR:
set_nottype_bits(start_bits, cbit_word, table_limit, cd);
break;
case OP_WORDCHAR:
set_type_bits(start_bits, cbit_word, table_limit, cd);
break;
}
tcode += 2;
break;
/* Character class where all the information is in a bit map: set the
bits and either carry on or not, according to the repeat count. If it was
a negative class, and we are operating with UTF-8 characters, any byte
with a value >= 0xc4 is a potentially valid starter because it starts a
character with a value > 255. */
case OP_NCLASS:
#ifdef SUPPORT_UTF8
if (utf8)
{
start_bits[24] |= 0xf0; /* Bits for 0xc4 - 0xc8 */
memset(start_bits+25, 0xff, 7); /* Bits for 0xc9 - 0xff */
}
#endif
/* Fall through */
case OP_CLASS:
{
tcode++;
/* In UTF-8 mode, the bits in a bit map correspond to character
values, not to byte values. However, the bit map we are constructing is
for byte values. So we have to do a conversion for characters whose
value is > 127. In fact, there are only two possible starting bytes for
characters in the range 128 - 255. */
#ifdef SUPPORT_UTF8
if (utf8)
{
for (c = 0; c < 16; c++) start_bits[c] |= tcode[c];
for (c = 128; c < 256; c++)
{
if ((tcode[c/8] && (1 << (c&7))) != 0)
{
int d = (c >> 6) | 0xc0; /* Set bit for this starter */
start_bits[d/8] |= (1 << (d&7)); /* and then skip on to the */
c = (c & 0xc0) + 0x40 - 1; /* next relevant character. */
}
}
}
/* In non-UTF-8 mode, the two bit maps are completely compatible. */
else
#endif
{
for (c = 0; c < 32; c++) start_bits[c] |= tcode[c];
}
/* Advance past the bit map, and act on what follows */
tcode += 32;
switch (*tcode)
{
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
tcode++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
if (((tcode[1] << 8) + tcode[2]) == 0) tcode += 5;
else try_next = FALSE;
break;
default:
try_next = FALSE;
break;
}
}
break; /* End of bitmap class handling */
} /* End of switch */
} /* End of try_next loop */
static int
set_start_bits(const pcre_uchar *code, pcre_uint8 *start_bits, BOOL utf,
compile_data *cd)
{
register pcre_uint32 c;
int yield = SSB_DONE;
#if defined SUPPORT_UTF && defined COMPILE_PCRE8
int table_limit = utf? 16:32;
#else
int table_limit = 32;
#endif
#if 0
/* ========================================================================= */
/* The following comment and code was inserted in January 1999. In May 2006,
when it was observed to cause compiler warnings about unused values, I took it
out again. If anybody is still using OS/2, they will have to put it back
manually. */
/* This next statement and the later reference to dummy are here in order to
trick the optimizer of the IBM C compiler for OS/2 into generating correct
code. Apparently IBM isn't going to fix the problem, and we would rather not
disable optimization (in this module it actually makes a big difference, and
the pcre module can use all the optimization it can get). */
volatile int dummy;
/* ========================================================================= */
#endif
do
{
BOOL try_next = TRUE;
const pcre_uchar *tcode = code + 1 + LINK_SIZE;
if (*code == OP_CBRA || *code == OP_SCBRA ||
*code == OP_CBRAPOS || *code == OP_SCBRAPOS) tcode += IMM2_SIZE;
while (try_next) /* Loop for items in this branch */
{
int rc;
switch(*tcode)
{
/* If we reach something we don't understand, it means a new opcode has
been created that hasn't been added to this code. Hopefully this problem
will be discovered during testing. */
default:
return SSB_UNKNOWN;
/* Fail for a valid opcode that implies no starting bits. */
case OP_ACCEPT:
case OP_ASSERT_ACCEPT:
case OP_ALLANY:
case OP_ANY:
case OP_ANYBYTE:
case OP_CIRC:
case OP_CIRCM:
case OP_CLOSE:
case OP_COMMIT:
case OP_COND:
case OP_CREF:
case OP_DEF:
case OP_DNCREF:
case OP_DNREF:
case OP_DNREFI:
case OP_DNRREF:
case OP_DOLL:
case OP_DOLLM:
case OP_END:
case OP_EOD:
case OP_EODN:
case OP_EXTUNI:
case OP_FAIL:
case OP_MARK:
case OP_NOT:
case OP_NOTEXACT:
case OP_NOTEXACTI:
case OP_NOTI:
case OP_NOTMINPLUS:
case OP_NOTMINPLUSI:
case OP_NOTMINQUERY:
case OP_NOTMINQUERYI:
case OP_NOTMINSTAR:
case OP_NOTMINSTARI:
case OP_NOTMINUPTO:
case OP_NOTMINUPTOI:
case OP_NOTPLUS:
case OP_NOTPLUSI:
case OP_NOTPOSPLUS:
case OP_NOTPOSPLUSI:
case OP_NOTPOSQUERY:
case OP_NOTPOSQUERYI:
case OP_NOTPOSSTAR:
case OP_NOTPOSSTARI:
case OP_NOTPOSUPTO:
case OP_NOTPOSUPTOI:
case OP_NOTPROP:
case OP_NOTQUERY:
case OP_NOTQUERYI:
case OP_NOTSTAR:
case OP_NOTSTARI:
case OP_NOTUPTO:
case OP_NOTUPTOI:
case OP_NOT_HSPACE:
case OP_NOT_VSPACE:
case OP_PRUNE:
case OP_PRUNE_ARG:
case OP_RECURSE:
case OP_REF:
case OP_REFI:
case OP_REVERSE:
case OP_RREF:
case OP_SCOND:
case OP_SET_SOM:
case OP_SKIP:
case OP_SKIP_ARG:
case OP_SOD:
case OP_SOM:
case OP_THEN:
case OP_THEN_ARG:
return SSB_FAIL;
/* A "real" property test implies no starting bits, but the fake property
PT_CLIST identifies a list of characters. These lists are short, as they
are used for characters with more than one "other case", so there is no
point in recognizing them for OP_NOTPROP. */
case OP_PROP:
if (tcode[1] != PT_CLIST) return SSB_FAIL;
{
const pcre_uint32 *p = PRIV(ucd_caseless_sets) + tcode[2];
while ((c = *p++) < NOTACHAR)
{
#if defined SUPPORT_UTF && defined COMPILE_PCRE8
if (utf)
{
pcre_uchar buff[6];
(void)PRIV(ord2utf)(c, buff);
c = buff[0];
}
#endif
if (c > 0xff) SET_BIT(0xff); else SET_BIT(c);
}
}
try_next = FALSE;
break;
/* We can ignore word boundary tests. */
case OP_WORD_BOUNDARY:
case OP_NOT_WORD_BOUNDARY:
tcode++;
break;
/* If we hit a bracket or a positive lookahead assertion, recurse to set
bits from within the subpattern. If it can't find anything, we have to
give up. If it finds some mandatory character(s), we are done for this
branch. Otherwise, carry on scanning after the subpattern. */
case OP_BRA:
case OP_SBRA:
case OP_CBRA:
case OP_SCBRA:
case OP_BRAPOS:
case OP_SBRAPOS:
case OP_CBRAPOS:
case OP_SCBRAPOS:
case OP_ONCE:
case OP_ONCE_NC:
case OP_ASSERT:
rc = set_start_bits(tcode, start_bits, utf, cd);
if (rc == SSB_FAIL || rc == SSB_UNKNOWN) return rc;
if (rc == SSB_DONE) try_next = FALSE; else
{
do tcode += GET(tcode, 1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
}
break;
/* If we hit ALT or KET, it means we haven't found anything mandatory in
this branch, though we might have found something optional. For ALT, we
continue with the next alternative, but we have to arrange that the final
result from subpattern is SSB_CONTINUE rather than SSB_DONE. For KET,
return SSB_CONTINUE: if this is the top level, that indicates failure,
but after a nested subpattern, it causes scanning to continue. */
case OP_ALT:
yield = SSB_CONTINUE;
try_next = FALSE;
break;
case OP_KET:
case OP_KETRMAX:
case OP_KETRMIN:
case OP_KETRPOS:
return SSB_CONTINUE;
/* Skip over callout */
case OP_CALLOUT:
tcode += 2 + 2*LINK_SIZE;
break;
/* Skip over lookbehind and negative lookahead assertions */
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
do tcode += GET(tcode, 1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
break;
/* BRAZERO does the bracket, but carries on. */
case OP_BRAZERO:
case OP_BRAMINZERO:
case OP_BRAPOSZERO:
rc = set_start_bits(++tcode, start_bits, utf, cd);
if (rc == SSB_FAIL || rc == SSB_UNKNOWN) return rc;
/* =========================================================================
See the comment at the head of this function concerning the next line,
which was an old fudge for the benefit of OS/2.
dummy = 1;
========================================================================= */
do tcode += GET(tcode,1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
break;
/* SKIPZERO skips the bracket. */
case OP_SKIPZERO:
tcode++;
do tcode += GET(tcode,1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
break;
/* Single-char * or ? sets the bit and tries the next item */
case OP_STAR:
case OP_MINSTAR:
case OP_POSSTAR:
case OP_QUERY:
case OP_MINQUERY:
case OP_POSQUERY:
tcode = set_table_bit(start_bits, tcode + 1, FALSE, cd, utf);
break;
case OP_STARI:
case OP_MINSTARI:
case OP_POSSTARI:
case OP_QUERYI:
case OP_MINQUERYI:
case OP_POSQUERYI:
tcode = set_table_bit(start_bits, tcode + 1, TRUE, cd, utf);
break;
/* Single-char upto sets the bit and tries the next */
case OP_UPTO:
case OP_MINUPTO:
case OP_POSUPTO:
tcode = set_table_bit(start_bits, tcode + 1 + IMM2_SIZE, FALSE, cd, utf);
break;
case OP_UPTOI:
case OP_MINUPTOI:
case OP_POSUPTOI:
tcode = set_table_bit(start_bits, tcode + 1 + IMM2_SIZE, TRUE, cd, utf);
break;
/* At least one single char sets the bit and stops */
case OP_EXACT:
tcode += IMM2_SIZE;
/* Fall through */
case OP_CHAR:
case OP_PLUS:
case OP_MINPLUS:
case OP_POSPLUS:
(void)set_table_bit(start_bits, tcode + 1, FALSE, cd, utf);
try_next = FALSE;
break;
case OP_EXACTI:
tcode += IMM2_SIZE;
/* Fall through */
case OP_CHARI:
case OP_PLUSI:
case OP_MINPLUSI:
case OP_POSPLUSI:
(void)set_table_bit(start_bits, tcode + 1, TRUE, cd, utf);
try_next = FALSE;
break;
/* Special spacing and line-terminating items. These recognize specific
lists of characters. The difference between VSPACE and ANYNL is that the
latter can match the two-character CRLF sequence, but that is not
relevant for finding the first character, so their code here is
identical. */
case OP_HSPACE:
SET_BIT(CHAR_HT);
SET_BIT(CHAR_SPACE);
#ifdef SUPPORT_UTF
if (utf)
{
#ifdef COMPILE_PCRE8
SET_BIT(0xC2); /* For U+00A0 */
SET_BIT(0xE1); /* For U+1680, U+180E */
SET_BIT(0xE2); /* For U+2000 - U+200A, U+202F, U+205F */
SET_BIT(0xE3); /* For U+3000 */
#elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32
SET_BIT(0xA0);
SET_BIT(0xFF); /* For characters > 255 */
#endif /* COMPILE_PCRE[8|16|32] */
}
else
#endif /* SUPPORT_UTF */
{
#ifndef EBCDIC
SET_BIT(0xA0);
#endif /* Not EBCDIC */
#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
SET_BIT(0xFF); /* For characters > 255 */
#endif /* COMPILE_PCRE[16|32] */
}
try_next = FALSE;
break;
case OP_ANYNL:
case OP_VSPACE:
SET_BIT(CHAR_LF);
SET_BIT(CHAR_VT);
SET_BIT(CHAR_FF);
SET_BIT(CHAR_CR);
#ifdef SUPPORT_UTF
if (utf)
{
#ifdef COMPILE_PCRE8
SET_BIT(0xC2); /* For U+0085 */
SET_BIT(0xE2); /* For U+2028, U+2029 */
#elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32
SET_BIT(CHAR_NEL);
SET_BIT(0xFF); /* For characters > 255 */
#endif /* COMPILE_PCRE[8|16|32] */
}
else
#endif /* SUPPORT_UTF */
{
SET_BIT(CHAR_NEL);
#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
SET_BIT(0xFF); /* For characters > 255 */
#endif
}
try_next = FALSE;
break;
/* Single character types set the bits and stop. Note that if PCRE_UCP
is set, we do not see these op codes because \d etc are converted to
properties. Therefore, these apply in the case when only characters less
than 256 are recognized to match the types. */
case OP_NOT_DIGIT:
set_nottype_bits(start_bits, cbit_digit, table_limit, cd);
try_next = FALSE;
break;
case OP_DIGIT:
set_type_bits(start_bits, cbit_digit, table_limit, cd);
try_next = FALSE;
break;
/* The cbit_space table has vertical tab as whitespace; we no longer
have to play fancy tricks because Perl added VT to its whitespace at
release 5.18. PCRE added it at release 8.34. */
case OP_NOT_WHITESPACE:
set_nottype_bits(start_bits, cbit_space, table_limit, cd);
try_next = FALSE;
break;
case OP_WHITESPACE:
set_type_bits(start_bits, cbit_space, table_limit, cd);
try_next = FALSE;
break;
case OP_NOT_WORDCHAR:
set_nottype_bits(start_bits, cbit_word, table_limit, cd);
try_next = FALSE;
break;
case OP_WORDCHAR:
set_type_bits(start_bits, cbit_word, table_limit, cd);
try_next = FALSE;
break;
/* One or more character type fudges the pointer and restarts, knowing
it will hit a single character type and stop there. */
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEPOSPLUS:
tcode++;
break;
case OP_TYPEEXACT:
tcode += 1 + IMM2_SIZE;
break;
/* Zero or more repeats of character types set the bits and then
try again. */
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEPOSUPTO:
tcode += IMM2_SIZE; /* Fall through */
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEPOSSTAR:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEPOSQUERY:
switch(tcode[1])
{
default:
case OP_ANY:
case OP_ALLANY:
return SSB_FAIL;
case OP_HSPACE:
SET_BIT(CHAR_HT);
SET_BIT(CHAR_SPACE);
#ifdef SUPPORT_UTF
if (utf)
{
#ifdef COMPILE_PCRE8
SET_BIT(0xC2); /* For U+00A0 */
SET_BIT(0xE1); /* For U+1680, U+180E */
SET_BIT(0xE2); /* For U+2000 - U+200A, U+202F, U+205F */
SET_BIT(0xE3); /* For U+3000 */
#elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32
SET_BIT(0xA0);
SET_BIT(0xFF); /* For characters > 255 */
#endif /* COMPILE_PCRE[8|16|32] */
}
else
#endif /* SUPPORT_UTF */
#ifndef EBCDIC
SET_BIT(0xA0);
#endif /* Not EBCDIC */
break;
case OP_ANYNL:
case OP_VSPACE:
SET_BIT(CHAR_LF);
SET_BIT(CHAR_VT);
SET_BIT(CHAR_FF);
SET_BIT(CHAR_CR);
#ifdef SUPPORT_UTF
if (utf)
{
#ifdef COMPILE_PCRE8
SET_BIT(0xC2); /* For U+0085 */
SET_BIT(0xE2); /* For U+2028, U+2029 */
#elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32
SET_BIT(CHAR_NEL);
SET_BIT(0xFF); /* For characters > 255 */
#endif /* COMPILE_PCRE16 */
}
else
#endif /* SUPPORT_UTF */
SET_BIT(CHAR_NEL);
break;
case OP_NOT_DIGIT:
set_nottype_bits(start_bits, cbit_digit, table_limit, cd);
break;
case OP_DIGIT:
set_type_bits(start_bits, cbit_digit, table_limit, cd);
break;
/* The cbit_space table has vertical tab as whitespace; we no longer
have to play fancy tricks because Perl added VT to its whitespace at
release 5.18. PCRE added it at release 8.34. */
case OP_NOT_WHITESPACE:
set_nottype_bits(start_bits, cbit_space, table_limit, cd);
break;
case OP_WHITESPACE:
set_type_bits(start_bits, cbit_space, table_limit, cd);
break;
case OP_NOT_WORDCHAR:
set_nottype_bits(start_bits, cbit_word, table_limit, cd);
break;
case OP_WORDCHAR:
set_type_bits(start_bits, cbit_word, table_limit, cd);
break;
}
tcode += 2;
break;
/* Character class where all the information is in a bit map: set the
bits and either carry on or not, according to the repeat count. If it was
a negative class, and we are operating with UTF-8 characters, any byte
with a value >= 0xc4 is a potentially valid starter because it starts a
character with a value > 255. */
#if defined SUPPORT_UTF || !defined COMPILE_PCRE8
case OP_XCLASS:
if ((tcode[1 + LINK_SIZE] & XCL_HASPROP) != 0)
return SSB_FAIL;
/* All bits are set. */
if ((tcode[1 + LINK_SIZE] & XCL_MAP) == 0 && (tcode[1 + LINK_SIZE] & XCL_NOT) != 0)
return SSB_FAIL;
#endif
/* Fall through */
case OP_NCLASS:
#if defined SUPPORT_UTF && defined COMPILE_PCRE8
if (utf)
{
start_bits[24] |= 0xf0; /* Bits for 0xc4 - 0xc8 */
memset(start_bits+25, 0xff, 7); /* Bits for 0xc9 - 0xff */
}
#endif
#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
SET_BIT(0xFF); /* For characters > 255 */
#endif
/* Fall through */
case OP_CLASS:
{
pcre_uint8 *map;
#if defined SUPPORT_UTF || !defined COMPILE_PCRE8
map = NULL;
if (*tcode == OP_XCLASS)
{
if ((tcode[1 + LINK_SIZE] & XCL_MAP) != 0)
map = (pcre_uint8 *)(tcode + 1 + LINK_SIZE + 1);
tcode += GET(tcode, 1);
}
else
#endif
{
tcode++;
map = (pcre_uint8 *)tcode;
tcode += 32 / sizeof(pcre_uchar);
}
/* In UTF-8 mode, the bits in a bit map correspond to character
values, not to byte values. However, the bit map we are constructing is
for byte values. So we have to do a conversion for characters whose
value is > 127. In fact, there are only two possible starting bytes for
characters in the range 128 - 255. */
#if defined SUPPORT_UTF || !defined COMPILE_PCRE8
if (map != NULL)
#endif
{
#if defined SUPPORT_UTF && defined COMPILE_PCRE8
if (utf)
{
for (c = 0; c < 16; c++) start_bits[c] |= map[c];
for (c = 128; c < 256; c++)
{
if ((map[c/8] && (1 << (c&7))) != 0)
{
int d = (c >> 6) | 0xc0; /* Set bit for this starter */
start_bits[d/8] |= (1 << (d&7)); /* and then skip on to the */
c = (c & 0xc0) + 0x40 - 1; /* next relevant character. */
}
}
}
else
#endif
{
/* In non-UTF-8 mode, the two bit maps are completely compatible. */
for (c = 0; c < 32; c++) start_bits[c] |= map[c];
}
}
/* Advance past the bit map, and act on what follows. For a zero
minimum repeat, continue; otherwise stop processing. */
switch (*tcode)
{
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSSTAR:
case OP_CRPOSQUERY:
tcode++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
if (GET2(tcode, 1) == 0) tcode += 1 + 2 * IMM2_SIZE;
else try_next = FALSE;
break;
default:
try_next = FALSE;
break;
}
}
break; /* End of bitmap class handling */
} /* End of switch */
} /* End of try_next loop */
static inline int GET_ALLOC( void *p ) {
return GET(p) & 0x1;
}
static int
find_minlength(const pcre2_real_code *re, PCRE2_SPTR code,
PCRE2_SPTR startcode, BOOL utf, recurse_check *recurses, int *countptr,
int *backref_cache)
{
int length = -1;
int prev_cap_recno = -1;
int prev_cap_d = 0;
int prev_recurse_recno = -1;
int prev_recurse_d = 0;
uint32_t once_fudge = 0;
BOOL had_recurse = FALSE;
BOOL dupcapused = (re->flags & PCRE2_DUPCAPUSED) != 0;
recurse_check this_recurse;
int branchlength = 0;
PCRE2_UCHAR *cc = (PCRE2_UCHAR *)code + 1 + LINK_SIZE;
/* If this is a "could be empty" group, its minimum length is 0. */
if (*code >= OP_SBRA && *code <= OP_SCOND) return 0;
/* Skip over capturing bracket number */
if (*code == OP_CBRA || *code == OP_CBRAPOS) cc += IMM2_SIZE;
/* A large and/or complex regex can take too long to process. */
if ((*countptr)++ > 1000) return -1;
/* Scan along the opcodes for this branch. If we get to the end of the branch,
check the length against that of the other branches. If the accumulated length
passes 16-bits, stop. */
for (;;)
{
int d, min, recno;
PCRE2_UCHAR *cs, *ce;
PCRE2_UCHAR op = *cc;
if (branchlength >= UINT16_MAX) return UINT16_MAX;
switch (op)
{
case OP_COND:
case OP_SCOND:
/* If there is only one branch in a condition, the implied branch has zero
length, so we don't add anything. This covers the DEFINE "condition"
automatically. If there are two branches we can treat it the same as any
other non-capturing subpattern. */
cs = cc + GET(cc, 1);
if (*cs != OP_ALT)
{
cc = cs + 1 + LINK_SIZE;
break;
}
goto PROCESS_NON_CAPTURE;
/* There's a special case of OP_ONCE, when it is wrapped round an
OP_RECURSE. We'd like to process the latter at this level so that
remembering the value works for repeated cases. So we do nothing, but
set a fudge value to skip over the OP_KET after the recurse. */
case OP_ONCE:
if (cc[1+LINK_SIZE] == OP_RECURSE && cc[2*(1+LINK_SIZE)] == OP_KET)
{
once_fudge = 1 + LINK_SIZE;
cc += 1 + LINK_SIZE;
break;
}
/* Fall through */
case OP_ONCE_NC:
case OP_BRA:
case OP_SBRA:
case OP_BRAPOS:
case OP_SBRAPOS:
PROCESS_NON_CAPTURE:
d = find_minlength(re, cc, startcode, utf, recurses, countptr,
backref_cache);
if (d < 0) return d;
branchlength += d;
do cc += GET(cc, 1); while (*cc == OP_ALT);
cc += 1 + LINK_SIZE;
break;
/* To save time for repeated capturing subpatterns, we remember the
length of the previous one. Unfortunately we can't do the same for
the unnumbered ones above. Nor can we do this if (?| is present in the
pattern because captures with the same number are not then identical. */
case OP_CBRA:
case OP_SCBRA:
case OP_CBRAPOS:
case OP_SCBRAPOS:
recno = (int)GET2(cc, 1+LINK_SIZE);
if (dupcapused || recno != prev_cap_recno)
{
prev_cap_recno = recno;
prev_cap_d = find_minlength(re, cc, startcode, utf, recurses, countptr,
backref_cache);
if (prev_cap_d < 0) return prev_cap_d;
}
branchlength += prev_cap_d;
do cc += GET(cc, 1); while (*cc == OP_ALT);
cc += 1 + LINK_SIZE;
break;
/* ACCEPT makes things far too complicated; we have to give up. */
case OP_ACCEPT:
case OP_ASSERT_ACCEPT:
return -1;
/* Reached end of a branch; if it's a ket it is the end of a nested
call. If it's ALT it is an alternation in a nested call. If it is END it's
the end of the outer call. All can be handled by the same code. If an
ACCEPT was previously encountered, use the length that was in force at that
time, and pass back the shortest ACCEPT length. */
case OP_ALT:
case OP_KET:
case OP_KETRMAX:
case OP_KETRMIN:
case OP_KETRPOS:
case OP_END:
if (length < 0 || (!had_recurse && branchlength < length))
length = branchlength;
if (op != OP_ALT) return length;
cc += 1 + LINK_SIZE;
branchlength = 0;
had_recurse = FALSE;
break;
/* Skip over assertive subpatterns */
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
do cc += GET(cc, 1); while (*cc == OP_ALT);
/* Fall through */
/* Skip over things that don't match chars */
case OP_REVERSE:
case OP_CREF:
case OP_DNCREF:
case OP_RREF:
case OP_DNRREF:
case OP_FALSE:
case OP_TRUE:
case OP_CALLOUT:
case OP_SOD:
case OP_SOM:
case OP_EOD:
case OP_EODN:
case OP_CIRC:
case OP_CIRCM:
case OP_DOLL:
case OP_DOLLM:
case OP_NOT_WORD_BOUNDARY:
case OP_WORD_BOUNDARY:
cc += PRIV(OP_lengths)[*cc];
break;
case OP_CALLOUT_STR:
cc += GET(cc, 1 + 2*LINK_SIZE);
break;
/* Skip over a subpattern that has a {0} or {0,x} quantifier */
case OP_BRAZERO:
case OP_BRAMINZERO:
case OP_BRAPOSZERO:
case OP_SKIPZERO:
cc += PRIV(OP_lengths)[*cc];
do cc += GET(cc, 1); while (*cc == OP_ALT);
cc += 1 + LINK_SIZE;
break;
/* Handle literal characters and + repetitions */
case OP_CHAR:
case OP_CHARI:
case OP_NOT:
case OP_NOTI:
case OP_PLUS:
case OP_PLUSI:
case OP_MINPLUS:
case OP_MINPLUSI:
case OP_POSPLUS:
case OP_POSPLUSI:
case OP_NOTPLUS:
case OP_NOTPLUSI:
case OP_NOTMINPLUS:
case OP_NOTMINPLUSI:
case OP_NOTPOSPLUS:
case OP_NOTPOSPLUSI:
branchlength++;
cc += 2;
#ifdef SUPPORT_UNICODE
if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEPOSPLUS:
branchlength++;
cc += (cc[1] == OP_PROP || cc[1] == OP_NOTPROP)? 4 : 2;
break;
/* Handle exact repetitions. The count is already in characters, but we
may need to skip over a multibyte character in UTF mode. */
case OP_EXACT:
case OP_EXACTI:
case OP_NOTEXACT:
case OP_NOTEXACTI:
branchlength += GET2(cc,1);
cc += 2 + IMM2_SIZE;
#ifdef SUPPORT_UNICODE
if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
case OP_TYPEEXACT:
branchlength += GET2(cc,1);
cc += 2 + IMM2_SIZE + ((cc[1 + IMM2_SIZE] == OP_PROP
|| cc[1 + IMM2_SIZE] == OP_NOTPROP)? 2 : 0);
break;
/* Handle single-char non-literal matchers */
case OP_PROP:
case OP_NOTPROP:
cc += 2;
/* Fall through */
case OP_NOT_DIGIT:
case OP_DIGIT:
case OP_NOT_WHITESPACE:
case OP_WHITESPACE:
case OP_NOT_WORDCHAR:
case OP_WORDCHAR:
case OP_ANY:
case OP_ALLANY:
case OP_EXTUNI:
case OP_HSPACE:
case OP_NOT_HSPACE:
case OP_VSPACE:
case OP_NOT_VSPACE:
branchlength++;
cc++;
break;
/* "Any newline" might match two characters, but it also might match just
one. */
case OP_ANYNL:
branchlength += 1;
cc++;
break;
/* The single-byte matcher means we can't proceed in UTF mode. (In
non-UTF mode \C will actually be turned into OP_ALLANY, so won't ever
appear, but leave the code, just in case.) */
case OP_ANYBYTE:
#ifdef SUPPORT_UNICODE
if (utf) return -1;
#endif
branchlength++;
cc++;
break;
/* For repeated character types, we have to test for \p and \P, which have
an extra two bytes of parameters. */
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEPOSSTAR:
case OP_TYPEPOSQUERY:
if (cc[1] == OP_PROP || cc[1] == OP_NOTPROP) cc += 2;
cc += PRIV(OP_lengths)[op];
break;
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEPOSUPTO:
if (cc[1 + IMM2_SIZE] == OP_PROP
|| cc[1 + IMM2_SIZE] == OP_NOTPROP) cc += 2;
cc += PRIV(OP_lengths)[op];
break;
/* Check a class for variable quantification */
case OP_CLASS:
case OP_NCLASS:
#ifdef SUPPORT_WIDE_CHARS
case OP_XCLASS:
/* The original code caused an unsigned overflow in 64 bit systems,
so now we use a conditional statement. */
if (op == OP_XCLASS)
cc += GET(cc, 1);
else
cc += PRIV(OP_lengths)[OP_CLASS];
#else
cc += PRIV(OP_lengths)[OP_CLASS];
#endif
switch (*cc)
{
case OP_CRPLUS:
case OP_CRMINPLUS:
case OP_CRPOSPLUS:
branchlength++;
/* Fall through */
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSSTAR:
case OP_CRPOSQUERY:
cc++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
branchlength += GET2(cc,1);
cc += 1 + 2 * IMM2_SIZE;
break;
default:
branchlength++;
break;
}
break;
/* Backreferences and subroutine calls (OP_RECURSE) are treated in the same
way: we find the minimum length for the subpattern. A recursion
(backreference or subroutine) causes an a flag to be set that causes the
length of this branch to be ignored. The logic is that a recursion can only
make sense if there is another alternative that stops the recursing. That
will provide the minimum length (when no recursion happens).
If PCRE2_MATCH_UNSET_BACKREF is set, a backreference to an unset bracket
matches an empty string (by default it causes a matching failure), so in
that case we must set the minimum length to zero. */
/* Duplicate named pattern back reference. We cannot reliably find a length
for this if duplicate numbers are present in the pattern. */
case OP_DNREF:
case OP_DNREFI:
if (dupcapused) return -1;
if ((re->overall_options & PCRE2_MATCH_UNSET_BACKREF) == 0)
{
int count = GET2(cc, 1+IMM2_SIZE);
PCRE2_UCHAR *slot =
(PCRE2_UCHAR *)((uint8_t *)re + sizeof(pcre2_real_code)) +
GET2(cc, 1) * re->name_entry_size;
d = INT_MAX;
/* Scan all groups with the same name; find the shortest. */
while (count-- > 0)
{
int dd, i;
recno = GET2(slot, 0);
if (recno <= backref_cache[0] && backref_cache[recno] >= 0)
dd = backref_cache[recno];
else
{
ce = cs = (PCRE2_UCHAR *)PRIV(find_bracket)(startcode, utf, recno);
if (cs == NULL) return -2;
do ce += GET(ce, 1); while (*ce == OP_ALT);
if (cc > cs && cc < ce) /* Simple recursion */
{
dd = 0;
had_recurse = TRUE;
}
else
{
recurse_check *r = recurses;
for (r = recurses; r != NULL; r = r->prev)
if (r->group == cs) break;
if (r != NULL) /* Mutual recursion */
{
dd = 0;
had_recurse = TRUE;
}
else
{
this_recurse.prev = recurses;
this_recurse.group = cs;
dd = find_minlength(re, cs, startcode, utf, &this_recurse,
countptr, backref_cache);
if (dd < 0) return dd;
}
}
backref_cache[recno] = dd;
for (i = backref_cache[0] + 1; i < recno; i++) backref_cache[i] = -1;
backref_cache[0] = recno;
}
if (dd < d) d = dd;
if (d <= 0) break; /* No point looking at any more */
slot += re->name_entry_size;
}
}
else d = 0;
cc += 1 + 2*IMM2_SIZE;
goto REPEAT_BACK_REFERENCE;
/* Single back reference. We cannot find a length for this if duplicate
numbers are present in the pattern. */
case OP_REF:
case OP_REFI:
if (dupcapused) return -1;
recno = GET2(cc, 1);
if (recno <= backref_cache[0] && backref_cache[recno] >= 0)
d = backref_cache[recno];
else
{
int i;
if ((re->overall_options & PCRE2_MATCH_UNSET_BACKREF) == 0)
{
ce = cs = (PCRE2_UCHAR *)PRIV(find_bracket)(startcode, utf, recno);
if (cs == NULL) return -2;
do ce += GET(ce, 1); while (*ce == OP_ALT);
if (cc > cs && cc < ce) /* Simple recursion */
{
d = 0;
had_recurse = TRUE;
}
else
{
recurse_check *r = recurses;
for (r = recurses; r != NULL; r = r->prev) if (r->group == cs) break;
if (r != NULL) /* Mutual recursion */
{
d = 0;
had_recurse = TRUE;
}
else
{
this_recurse.prev = recurses;
this_recurse.group = cs;
d = find_minlength(re, cs, startcode, utf, &this_recurse, countptr,
backref_cache);
if (d < 0) return d;
}
}
}
else d = 0;
backref_cache[recno] = d;
for (i = backref_cache[0] + 1; i < recno; i++) backref_cache[i] = -1;
backref_cache[0] = recno;
}
cc += 1 + IMM2_SIZE;
/* Handle repeated back references */
REPEAT_BACK_REFERENCE:
switch (*cc)
{
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSSTAR:
case OP_CRPOSQUERY:
min = 0;
cc++;
break;
case OP_CRPLUS:
case OP_CRMINPLUS:
case OP_CRPOSPLUS:
min = 1;
cc++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
min = GET2(cc, 1);
cc += 1 + 2 * IMM2_SIZE;
break;
default:
min = 1;
break;
}
/* Take care not to overflow: (1) min and d are ints, so check that their
product is not greater than INT_MAX. (2) branchlength is limited to
UINT16_MAX (checked at the top of the loop). */
if ((d > 0 && (INT_MAX/d) < min) || UINT16_MAX - branchlength < min*d)
branchlength = UINT16_MAX;
else branchlength += min * d;
break;
/* Recursion always refers to the first occurrence of a subpattern with a
given number. Therefore, we can always make use of caching, even when the
pattern contains multiple subpatterns with the same number. */
case OP_RECURSE:
cs = ce = (PCRE2_UCHAR *)startcode + GET(cc, 1);
recno = GET2(cs, 1+LINK_SIZE);
if (recno == prev_recurse_recno)
{
branchlength += prev_recurse_d;
}
else
{
do ce += GET(ce, 1); while (*ce == OP_ALT);
if (cc > cs && cc < ce) /* Simple recursion */
had_recurse = TRUE;
else
{
recurse_check *r = recurses;
for (r = recurses; r != NULL; r = r->prev) if (r->group == cs) break;
if (r != NULL) /* Mutual recursion */
had_recurse = TRUE;
else
{
this_recurse.prev = recurses;
this_recurse.group = cs;
prev_recurse_d = find_minlength(re, cs, startcode, utf, &this_recurse,
countptr, backref_cache);
if (prev_recurse_d < 0) return prev_recurse_d;
prev_recurse_recno = recno;
branchlength += prev_recurse_d;
}
}
}
cc += 1 + LINK_SIZE + once_fudge;
once_fudge = 0;
break;
/* Anything else does not or need not match a character. We can get the
item's length from the table, but for those that can match zero occurrences
of a character, we must take special action for UTF-8 characters. As it
happens, the "NOT" versions of these opcodes are used at present only for
ASCII characters, so they could be omitted from this list. However, in
future that may change, so we include them here so as not to leave a
gotcha for a future maintainer. */
case OP_UPTO:
case OP_UPTOI:
case OP_NOTUPTO:
case OP_NOTUPTOI:
case OP_MINUPTO:
case OP_MINUPTOI:
case OP_NOTMINUPTO:
case OP_NOTMINUPTOI:
case OP_POSUPTO:
case OP_POSUPTOI:
case OP_NOTPOSUPTO:
case OP_NOTPOSUPTOI:
case OP_STAR:
case OP_STARI:
case OP_NOTSTAR:
case OP_NOTSTARI:
case OP_MINSTAR:
case OP_MINSTARI:
case OP_NOTMINSTAR:
case OP_NOTMINSTARI:
case OP_POSSTAR:
case OP_POSSTARI:
case OP_NOTPOSSTAR:
case OP_NOTPOSSTARI:
case OP_QUERY:
case OP_QUERYI:
case OP_NOTQUERY:
case OP_NOTQUERYI:
case OP_MINQUERY:
case OP_MINQUERYI:
case OP_NOTMINQUERY:
case OP_NOTMINQUERYI:
case OP_POSQUERY:
case OP_POSQUERYI:
case OP_NOTPOSQUERY:
case OP_NOTPOSQUERYI:
cc += PRIV(OP_lengths)[op];
#ifdef SUPPORT_UNICODE
if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
/* Skip these, but we need to add in the name length. */
case OP_MARK:
case OP_PRUNE_ARG:
case OP_SKIP_ARG:
case OP_THEN_ARG:
cc += PRIV(OP_lengths)[op] + cc[1];
break;
/* The remaining opcodes are just skipped over. */
case OP_CLOSE:
case OP_COMMIT:
case OP_FAIL:
case OP_PRUNE:
case OP_SET_SOM:
case OP_SKIP:
case OP_THEN:
cc += PRIV(OP_lengths)[op];
break;
/* This should not occur: we list all opcodes explicitly so that when
new ones get added they are properly considered. */
default:
return -3;
}
}
/* Control never gets here */
}
/**********************************************************
* add_to_free_list
* adds the free block to the free list
**********************************************************/
void add_to_free_list(void *bp)
{
// printf("IN ADD_TO_FREE_LIST\n");
// printf("adding free block %p\n",bp);
//get the size of the free block
size_t size = GET(HDRP(bp));
int i = get_segregated_index(size);
//print_ptr(bp);
if(segregated_list[i] == NULL)
{
//add the free block to the free list which is NULL
segregated_list[i] = bp;
//set the previous as 0 or (null/nothing)
PUT(LOCATION_PREV_FREE_BLKP(bp),0);
//set the next free block as 0 or (null/nothing)
PUT(LOCATION_NEXT_FREE_BLKP(bp),0);
}
else
{
if(GET_SIZE(HDRP(segregated_list[i]))>GET_SIZE(HDRP(bp))){
if(GET_NEXT_FREE_BLK(segregated_list[i])!=0)
{
// print_seg(1);
// void* sec_ptr = GET_NEXT_FREE_BLK(segregated_list[i]);
PUT(LOCATION_NEXT_FREE_BLKP(bp),GET_NEXT_FREE_BLK(segregated_list[i]));
PUT(GET_PREV_FREE_BLK(LOCATION_NEXT_FREE_BLKP(segregated_list[i])),bp);
PUT(LOCATION_NEXT_FREE_BLKP(segregated_list[i]),bp);
PUT(LOCATION_PREV_FREE_BLKP(bp),segregated_list[i]);
//printf("test esfewfe\n");
// print_ptr(bp);
// print_ptr(segregated_list[i]);
//print_seg(1);
}else{
//printf("test2\n");
//we only have 1 block in the list
PUT(LOCATION_NEXT_FREE_BLKP(segregated_list[i]),bp);
PUT(LOCATION_PREV_FREE_BLKP(bp),segregated_list[i]);
PUT(LOCATION_PREV_FREE_BLKP(segregated_list[i]),0);
PUT(LOCATION_NEXT_FREE_BLKP(bp),0);
}
}else
{
//Set the next block of the new head as the previous head
PUT(LOCATION_NEXT_FREE_BLKP(bp),segregated_list[i]);
//Set the previous block of the new head as NULL
PUT(LOCATION_PREV_FREE_BLKP(bp), 0);
//Set the previous block of the previous head to the new head
PUT(LOCATION_PREV_FREE_BLKP(segregated_list[i]),bp);
segregated_list[i] = bp;
}
}
}
//
// Read the size and allocated fields from address p
//
static inline size_t GET_SIZE( void *p ) {
return GET(p) & ~0x7;
}
/* test routine */
int test_main(int args, char ** argv) {
int i,j;
waypoint_vec_t wmap;
region_map_t regs;
INIT(wmap);
INIT(regs);
bool ret;
for (i=0; i<7; i++) {
PUSH_BACK(regs, regions[i]);
}
int self_refs = 0;
bool r = makeWaypointTable(&wmap, ®s, NR_POINTS, points);
iprintf ("Waypoint Table: \n");
iprintf (" \t ");
for (i=0; i<SIZE(wmap); ++i) {
iprintf("%6d ", i);
}
iprintf("\n");
for (i=0; i<SIZE(wmap); ++i) {
iprintf ("%d: %5s\t (%7.1f,%7.1f) ", i, points[i].comment,
GET(wmap, i).pos.x, GET(wmap, i).pos.y);
for (j=0; j<SIZE(wmap); j++) {
double dist = GET(wmap, i).distances[j];
char sep = ((j+1)%5) == 0? '|':' ';
if (dist == INFINITY) {
iprintf (" %c", sep);
} else if (dist < 0.01) {
if (i == j) {
iprintf (" * %c", sep);
self_refs++;
} else {
iprintf (" X %c", sep);
}
} else {
iprintf ("%6.1f%c", GET(wmap, i).distances[j], sep);
}
}
iprintf ("\n");
}
iprintf("\nSelf References: %d\n", self_refs);
iprintf ("Starting waypoint-waypoint tests.\n");
for (i=0; i<SIZE(wmap); ++i) {
int_vec_t path;
INIT(path);
ret = pathFind(&path, GET(wmap,0).pos, GET(wmap, i).pos,
&wmap, ®s);
iprintf ("RESULT: %d to %d: ", 0, i);
for (j=0; j<SIZE(path); ++j) {
iprintf ("%s (%d) ", points[GET(path,j)].comment,
GET(path,j));
}
iprintf ("\n");
}
char startbuffer[128];
char endbuffer[128];
iprintf ("Interactive mode. Hit Ctrl-C to exit.\n");
while (true) {
iprintf ("\nStart: ");
fgets(startbuffer, 127, stdin);
startbuffer[127] = 0;
iprintf ("End: ");
fgets(endbuffer, 127, stdin);
endbuffer[127] = 0;
char *s, *e;
if (strchr(startbuffer, '\n')) {
(*strchr(startbuffer, '\n')) = 0;
}
if (strchr(endbuffer, '\n')) {
(*strchr(endbuffer, '\n')) = 0;
}
s = startbuffer;
while (isspace(*s))
s++;
e = endbuffer;
while (isspace(*e))
e++;
int start=-1, end=-1;
for (i=0; i<SIZE(wmap) && start < 0; ++i) {
if (!strcasecmp(s, points[i].comment)) {
start = i;
}
}
for (i=0; i<SIZE(wmap) && end < 0; ++i) {
if (!strcasecmp(e, points[i].comment)) {
end = i;
}
}
if (start < 0) {
iprintf("%s not found.\n", s);
continue;
}
if (end < 0) {
iprintf("%s not found.\n", e);
continue;
}
iprintf ("Routing from %s to %s...\n", points[start].comment,
points[end].comment);
int_vec_t path;
INIT(path);
ret = pathFind(&path, GET(wmap, start).pos, GET(wmap, end).pos,
&wmap, ®s);
for (j=0; j<SIZE(path); ++j) {
iprintf ("%s (%d) ", points[GET(path,j)].comment,
GET(path,j));
}
}
}
static double pathFindWork(int_vec_t *destpath,
int pos, int dest,
path_bitmap_t seen,
const waypoint_vec_t* wmap,
const region_map_t* regs) {
BEGIN_DEPTH;
int_vec_t wpoints;
INIT(wpoints);
assert(pos >=0);
assert(dest >=0);
// already there.
if (pos == dest) {
END_DEPTH;
return 0.0;
}
if (full_set(&seen)) {
END_DEPTH;
return INFINITY;
}
if (GET(*wmap,pos).distances[dest] != INFINITY) {
PUSH_BACK(*destpath, dest);
END_DEPTH;
return GET(*wmap,pos).distances[dest];
}
sortedWaypoints(&wpoints, pos, dest, wmap, regs);
// remove those we've already seen.
int j = 0;
while (j < SIZE(wpoints)) {
if (is_set(&seen, GET(wpoints,j))) {
ERASE(wpoints, j);
}
else {
++j;
}
}
if (SIZE(wpoints) == 0) {
END_DEPTH;
return INFINITY;
}
const int offset = SIZE(*destpath);
int lowest_idx = -1;
double lowest_value = INFINITY;
int_vec_t tmp;
INIT(tmp);
COPY(tmp, *destpath);
int i;
for (i=0; i<SIZE(wpoints); ++i) {
if (!is_set(&seen, GET(wpoints,i))) {
double val;
PUSH_BACK(tmp, GET(wpoints,i));
path_bitmap_t local_seen = seen;
set(&local_seen, GET(wpoints,i));
val = pathFindWork(&tmp, GET(wpoints,i), dest, local_seen, wmap, regs)
+ GET(*wmap,pos).distances[GET(wpoints,i)];
if (lowest_value == INFINITY
|| (val != INFINITY && val < lowest_value)) {
lowest_value = val;
lowest_idx = GET(wpoints,i);
COPY(*destpath, tmp);
END_DEPTH;
// with this return, it's an A* search. without, it's
// a DFS.
return val;
}
RESIZE(tmp, offset);
}
}
// at the end, destpath has the minimum path we've seen so far,
// and now we're returning our distance to the path we've appended on.
END_DEPTH;
return lowest_value; // + distance(wmap[pos].pos, wmap[lowest_idx].pos, regs);
}
static void
pcre2_printint(pcre2_code *re, FILE *f, BOOL print_lengths)
{
PCRE2_SPTR codestart, nametable, code;
uint32_t nesize = re->name_entry_size;
BOOL utf = (re->overall_options & PCRE2_UTF) != 0;
nametable = (PCRE2_SPTR)((uint8_t *)re + sizeof(pcre2_real_code));
code = codestart = nametable + re->name_count * re->name_entry_size;
for(;;)
{
PCRE2_SPTR ccode;
uint32_t c;
int i;
const char *flag = " ";
unsigned int extra = 0;
if (print_lengths)
fprintf(f, "%3d ", (int)(code - codestart));
else
fprintf(f, " ");
switch(*code)
{
/* ========================================================================== */
/* These cases are never obeyed. This is a fudge that causes a compile-
time error if the vectors OP_names or OP_lengths, which are indexed
by opcode, are not the correct length. It seems to be the only way to do
such a check at compile time, as the sizeof() operator does not work in
the C preprocessor. */
case OP_TABLE_LENGTH:
case OP_TABLE_LENGTH +
((sizeof(OP_names)/sizeof(const char *) == OP_TABLE_LENGTH) &&
(sizeof(OP_lengths) == OP_TABLE_LENGTH)):
break;
/* ========================================================================== */
case OP_END:
fprintf(f, " %s\n", OP_names[*code]);
fprintf(f, "------------------------------------------------------------------\n");
return;
case OP_CHAR:
fprintf(f, " ");
do
{
code++;
code += 1 + print_char(f, code, utf);
}
while (*code == OP_CHAR);
fprintf(f, "\n");
continue;
case OP_CHARI:
fprintf(f, " /i ");
do
{
code++;
code += 1 + print_char(f, code, utf);
}
while (*code == OP_CHARI);
fprintf(f, "\n");
continue;
case OP_CBRA:
case OP_CBRAPOS:
case OP_SCBRA:
case OP_SCBRAPOS:
if (print_lengths) fprintf(f, "%3d ", GET(code, 1));
else fprintf(f, " ");
fprintf(f, "%s %d", OP_names[*code], GET2(code, 1+LINK_SIZE));
break;
case OP_BRA:
case OP_BRAPOS:
case OP_SBRA:
case OP_SBRAPOS:
case OP_KETRMAX:
case OP_KETRMIN:
case OP_KETRPOS:
case OP_ALT:
case OP_KET:
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
case OP_ONCE:
case OP_ONCE_NC:
case OP_COND:
case OP_SCOND:
case OP_REVERSE:
if (print_lengths) fprintf(f, "%3d ", GET(code, 1));
else fprintf(f, " ");
fprintf(f, "%s", OP_names[*code]);
break;
case OP_CLOSE:
fprintf(f, " %s %d", OP_names[*code], GET2(code, 1));
break;
case OP_CREF:
fprintf(f, "%3d %s", GET2(code,1), OP_names[*code]);
break;
case OP_DNCREF:
{
PCRE2_SPTR entry = nametable + (GET2(code, 1) * nesize) + IMM2_SIZE;
fprintf(f, " %s Cond ref <", flag);
print_custring(f, entry);
fprintf(f, ">%d", GET2(code, 1 + IMM2_SIZE));
}
break;
case OP_RREF:
c = GET2(code, 1);
if (c == RREF_ANY)
fprintf(f, " Cond recurse any");
else
fprintf(f, " Cond recurse %d", c);
break;
case OP_DNRREF:
{
PCRE2_SPTR entry = nametable + (GET2(code, 1) * nesize) + IMM2_SIZE;
fprintf(f, " %s Cond recurse <", flag);
print_custring(f, entry);
fprintf(f, ">%d", GET2(code, 1 + IMM2_SIZE));
}
break;
case OP_FALSE:
fprintf(f, " Cond false");
break;
case OP_TRUE:
fprintf(f, " Cond true");
break;
case OP_STARI:
case OP_MINSTARI:
case OP_POSSTARI:
case OP_PLUSI:
case OP_MINPLUSI:
case OP_POSPLUSI:
case OP_QUERYI:
case OP_MINQUERYI:
case OP_POSQUERYI:
flag = "/i";
/* Fall through */
case OP_STAR:
case OP_MINSTAR:
case OP_POSSTAR:
case OP_PLUS:
case OP_MINPLUS:
case OP_POSPLUS:
case OP_QUERY:
case OP_MINQUERY:
case OP_POSQUERY:
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEPOSSTAR:
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEPOSPLUS:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEPOSQUERY:
fprintf(f, " %s ", flag);
if (*code >= OP_TYPESTAR)
{
if (code[1] == OP_PROP || code[1] == OP_NOTPROP)
{
print_prop(f, code + 1, "", " ");
extra = 2;
}
else fprintf(f, "%s", OP_names[code[1]]);
}
else extra = print_char(f, code+1, utf);
fprintf(f, "%s", OP_names[*code]);
break;
case OP_EXACTI:
case OP_UPTOI:
case OP_MINUPTOI:
case OP_POSUPTOI:
flag = "/i";
/* Fall through */
case OP_EXACT:
case OP_UPTO:
case OP_MINUPTO:
case OP_POSUPTO:
fprintf(f, " %s ", flag);
extra = print_char(f, code + 1 + IMM2_SIZE, utf);
fprintf(f, "{");
if (*code != OP_EXACT && *code != OP_EXACTI) fprintf(f, "0,");
fprintf(f, "%d}", GET2(code,1));
if (*code == OP_MINUPTO || *code == OP_MINUPTOI) fprintf(f, "?");
else if (*code == OP_POSUPTO || *code == OP_POSUPTOI) fprintf(f, "+");
break;
case OP_TYPEEXACT:
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEPOSUPTO:
if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
{
print_prop(f, code + IMM2_SIZE + 1, " ", " ");
extra = 2;
}
else fprintf(f, " %s", OP_names[code[1 + IMM2_SIZE]]);
fprintf(f, "{");
if (*code != OP_TYPEEXACT) fprintf(f, "0,");
fprintf(f, "%d}", GET2(code,1));
if (*code == OP_TYPEMINUPTO) fprintf(f, "?");
else if (*code == OP_TYPEPOSUPTO) fprintf(f, "+");
break;
case OP_NOTI:
flag = "/i";
/* Fall through */
case OP_NOT:
fprintf(f, " %s [^", flag);
extra = print_char(f, code + 1, utf);
fprintf(f, "]");
break;
case OP_NOTSTARI:
case OP_NOTMINSTARI:
case OP_NOTPOSSTARI:
case OP_NOTPLUSI:
case OP_NOTMINPLUSI:
case OP_NOTPOSPLUSI:
case OP_NOTQUERYI:
case OP_NOTMINQUERYI:
case OP_NOTPOSQUERYI:
flag = "/i";
/* Fall through */
case OP_NOTSTAR:
case OP_NOTMINSTAR:
case OP_NOTPOSSTAR:
case OP_NOTPLUS:
case OP_NOTMINPLUS:
case OP_NOTPOSPLUS:
case OP_NOTQUERY:
case OP_NOTMINQUERY:
case OP_NOTPOSQUERY:
fprintf(f, " %s [^", flag);
extra = print_char(f, code + 1, utf);
fprintf(f, "]%s", OP_names[*code]);
break;
case OP_NOTEXACTI:
case OP_NOTUPTOI:
case OP_NOTMINUPTOI:
case OP_NOTPOSUPTOI:
flag = "/i";
/* Fall through */
case OP_NOTEXACT:
case OP_NOTUPTO:
case OP_NOTMINUPTO:
case OP_NOTPOSUPTO:
fprintf(f, " %s [^", flag);
extra = print_char(f, code + 1 + IMM2_SIZE, utf);
fprintf(f, "]{");
if (*code != OP_NOTEXACT && *code != OP_NOTEXACTI) fprintf(f, "0,");
fprintf(f, "%d}", GET2(code,1));
if (*code == OP_NOTMINUPTO || *code == OP_NOTMINUPTOI) fprintf(f, "?");
else
if (*code == OP_NOTPOSUPTO || *code == OP_NOTPOSUPTOI) fprintf(f, "+");
break;
case OP_RECURSE:
if (print_lengths) fprintf(f, "%3d ", GET(code, 1));
else fprintf(f, " ");
fprintf(f, "%s", OP_names[*code]);
break;
case OP_REFI:
flag = "/i";
/* Fall through */
case OP_REF:
fprintf(f, " %s \\%d", flag, GET2(code,1));
ccode = code + OP_lengths[*code];
goto CLASS_REF_REPEAT;
case OP_DNREFI:
flag = "/i";
/* Fall through */
case OP_DNREF:
{
PCRE2_SPTR entry = nametable + (GET2(code, 1) * nesize) + IMM2_SIZE;
fprintf(f, " %s \\k<", flag);
print_custring(f, entry);
fprintf(f, ">%d", GET2(code, 1 + IMM2_SIZE));
}
ccode = code + OP_lengths[*code];
goto CLASS_REF_REPEAT;
case OP_CALLOUT:
fprintf(f, " %s %d %d %d", OP_names[*code], code[1 + 2*LINK_SIZE],
GET(code, 1), GET(code, 1 + LINK_SIZE));
break;
case OP_CALLOUT_STR:
c = code[1 + 4*LINK_SIZE];
fprintf(f, " %s %c", OP_names[*code], c);
extra = GET(code, 1 + 2*LINK_SIZE);
print_custring_bylen(f, code + 2 + 4*LINK_SIZE, extra - 3 - 4*LINK_SIZE);
for (i = 0; PRIV(callout_start_delims)[i] != 0; i++)
if (c == PRIV(callout_start_delims)[i])
{
c = PRIV(callout_end_delims)[i];
break;
}
fprintf(f, "%c %d %d %d", c, GET(code, 1 + 3*LINK_SIZE), GET(code, 1),
GET(code, 1 + LINK_SIZE));
break;
case OP_PROP:
case OP_NOTPROP:
print_prop(f, code, " ", "");
break;
/* OP_XCLASS cannot occur in 8-bit, non-UTF mode. However, there's no harm
in having this code always here, and it makes it less messy without all
those #ifdefs. */
case OP_CLASS:
case OP_NCLASS:
case OP_XCLASS:
{
unsigned int min, max;
BOOL printmap;
BOOL invertmap = FALSE;
uint8_t *map;
uint8_t inverted_map[32];
fprintf(f, " [");
if (*code == OP_XCLASS)
{
extra = GET(code, 1);
ccode = code + LINK_SIZE + 1;
printmap = (*ccode & XCL_MAP) != 0;
if ((*ccode & XCL_NOT) != 0)
{
invertmap = (*ccode & XCL_HASPROP) == 0;
fprintf(f, "^");
}
ccode++;
}
else
{
printmap = TRUE;
ccode = code + 1;
}
/* Print a bit map */
if (printmap)
{
map = (uint8_t *)ccode;
if (invertmap)
{
for (i = 0; i < 32; i++) inverted_map[i] = ~map[i];
map = inverted_map;
}
for (i = 0; i < 256; i++)
{
if ((map[i/8] & (1 << (i&7))) != 0)
{
int j;
for (j = i+1; j < 256; j++)
if ((map[j/8] & (1 << (j&7))) == 0) break;
if (i == '-' || i == ']') fprintf(f, "\\");
if (PRINTABLE(i)) fprintf(f, "%c", i);
else fprintf(f, "\\x%02x", i);
if (--j > i)
{
if (j != i + 1) fprintf(f, "-");
if (j == '-' || j == ']') fprintf(f, "\\");
if (PRINTABLE(j)) fprintf(f, "%c", j);
else fprintf(f, "\\x%02x", j);
}
i = j;
}
}
ccode += 32 / sizeof(PCRE2_UCHAR);
}
/* For an XCLASS there is always some additional data */
if (*code == OP_XCLASS)
{
PCRE2_UCHAR ch;
while ((ch = *ccode++) != XCL_END)
{
BOOL not = FALSE;
const char *notch = "";
switch(ch)
{
case XCL_NOTPROP:
not = TRUE;
notch = "^";
/* Fall through */
case XCL_PROP:
{
unsigned int ptype = *ccode++;
unsigned int pvalue = *ccode++;
switch(ptype)
{
case PT_PXGRAPH:
fprintf(f, "[:%sgraph:]", notch);
break;
case PT_PXPRINT:
fprintf(f, "[:%sprint:]", notch);
break;
case PT_PXPUNCT:
fprintf(f, "[:%spunct:]", notch);
break;
default:
fprintf(f, "\\%c{%s}", (not? 'P':'p'),
get_ucpname(ptype, pvalue));
break;
}
}
break;
default:
ccode += 1 + print_char(f, ccode, utf);
if (ch == XCL_RANGE)
{
fprintf(f, "-");
ccode += 1 + print_char(f, ccode, utf);
}
break;
}
}
}
/* Indicate a non-UTF class which was created by negation */
fprintf(f, "]%s", (*code == OP_NCLASS)? " (neg)" : "");
/* Handle repeats after a class or a back reference */
CLASS_REF_REPEAT:
switch(*ccode)
{
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRPLUS:
case OP_CRMINPLUS:
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSSTAR:
case OP_CRPOSPLUS:
case OP_CRPOSQUERY:
fprintf(f, "%s", OP_names[*ccode]);
extra += OP_lengths[*ccode];
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
min = GET2(ccode,1);
max = GET2(ccode,1 + IMM2_SIZE);
if (max == 0) fprintf(f, "{%u,}", min);
else fprintf(f, "{%u,%u}", min, max);
if (*ccode == OP_CRMINRANGE) fprintf(f, "?");
else if (*ccode == OP_CRPOSRANGE) fprintf(f, "+");
extra += OP_lengths[*ccode];
break;
/* Do nothing if it's not a repeat; this code stops picky compilers
warning about the lack of a default code path. */
default:
break;
}
}
break;
case OP_MARK:
case OP_PRUNE_ARG:
case OP_SKIP_ARG:
case OP_THEN_ARG:
fprintf(f, " %s ", OP_names[*code]);
print_custring_bylen(f, code + 2, code[1]);
extra += code[1];
break;
case OP_THEN:
fprintf(f, " %s", OP_names[*code]);
break;
case OP_CIRCM:
case OP_DOLLM:
flag = "/m";
/* Fall through */
/* Anything else is just an item with no data, but possibly a flag. */
default:
fprintf(f, " %s %s", flag, OP_names[*code]);
break;
}
code += OP_lengths[*code] + extra;
fprintf(f, "\n");
}
}
/* MDELEM: Process ELEMENT declaration.
*/
VOID mdelem(UNCH *tbuf) /* Work area for tokenization (tbuf). */
{
UNCH *ranksuff = lbuf; /* Rank suffix. */
UNS dctype = 0; /* Declared content type (from dctab). */
UNCH fmin = 0; /* Minimization bit flags. */
int i; /* Loop counter. */
UNS u; /* Temporary variable. */
struct etd **mexgrp, **pexgrp; /* Ptr to model exceptions array. */
struct thdr *cmod, *cmodsv; /* Ptr to content model. */
UNCH *etdgi; /* GI of current etd (when going through group).*/
int minomitted = 0; /* Tag minimization parameters omitted. */
mdname = key[KELEMENT]; /* Identify declaration for messages. */
subdcl = NULL; /* No subject as yet. */
parmno = 0; /* No parameters as yet. */
mdessv = es; /* Save es level for entity nesting check. */
ranksuff[0] = 0;
mexgrp = pexgrp = 0;
/* PARAMETER 1: Element name or a group of them.
*/
parsemd(tbuf, NAMECASE, &pcblitp, NAMELEN);
TRACEMD("1: element name or grp");
switch (pcbmd.action) {
case NAS:
nmgrp[0] = etddef(tbuf);
nmgrp[1] = 0;
break;
case GRPS:
parsegrp(nmgrp, &pcbgrnm, tbuf);
break;
default:
mderr(121, (UNCH *)0, (UNCH *)0);
return;
}
/* Save first GI for trace and error messages. */
if (nmgrp[0])
subdcl = nmgrp[0]->etdgi+1;
/* PARAMETER 1A: Rank suffix (optional).
*/
parsemd(tbuf, NAMECASE, &pcblitp, NAMELEN);
TRACEMD("1A: rank suffix");
switch (pcbmd.action) {
case NUM:
ustrcpy(ranksuff, tbuf);
parsemd(tbuf, NAMECASE, &pcblitp, NAMELEN);
default:
break;
}
/* PARAMETER 2A: Start-tag minimization.
*/
TRACEMD("2A: start min");
switch (pcbmd.action) {
case CDR:
break;
case NAS:
if (!ustrcmp(tbuf+1, key[KO])) {
if (OMITTAG==YES) SET(fmin, SMO);
break;
}
/* fall through */
default:
if (OMITTAG==NO) {minomitted=1; break;}
mderr(129, tbuf+1, (UNCH *)0);
return;
}
/* Must omit omitted end-tag minimization, if omitted
start-tag minimization was omitted (because OMITTAG == NO). */
if (!minomitted) {
/* PARAMETER 2B: End-tag minimization.
*/
parsemd(tbuf, NAMECASE, &pcblitp, NAMELEN);
TRACEMD("2B: end min");
switch (pcbmd.action) {
case NAS:
if (ustrcmp(tbuf+1, key[KO])) {mderr(129, tbuf+1, (UNCH *)0); return;}
if (OMITTAG==YES) SET(fmin, EMO);
break;
case CDR:
SET(fmin, EMM);
break;
default:
mderr(129, tbuf+1, (UNCH *)0);
return;
}
/* PARAMETER 3: Declared content.
*/
parsemd(tbuf, NAMECASE, &pcblitp, NAMELEN);
}
TRACEMD("3: declared content");
switch (pcbmd.action) {
case NAS:
dctype = mapsrch(dctab, tbuf+1);
if (!dctype) {mderr(24, tbuf+1, (UNCH *)0); return;}
/* Eliminate incompatibilities among parameters. */
if (GET(fmin, SMO) && GET(dctype, MNONE+MCDATA+MRCDATA)) {
mderr(58, (UNCH *)0, (UNCH *)0);
RESET(fmin, SMO);
}
if (GET(dctype, MNONE) && BITON(fmin, EMM)) {
mderr(87, (UNCH *)0, (UNCH *)0);
SET(fmin, EMO);
}
/* If valid, process like a content model. */
case GRPS:
cmodsv = parsemod((int)(pcbmd.action==GRPS ? 0 : dctype));
if (cmodsv==0) return;
u = (dctype ? 1 : cmodsv->tu.tnum+2) * THSZ;
cmod = (struct thdr *)rmalloc(u);
memcpy((UNIV)cmod , (UNIV)cmodsv, u );
ds.modcnt += cmod->tu.tnum;
TRACEMOD(cmod);
break;
default:
mderr(130, (UNCH *)0, (UNCH *)0);
return;
}
/* PARAMETERS 3A, 3B: Exceptions or end.
*/
parsemd(tbuf, NAMECASE, &pcblitp, NAMELEN);
if (BITOFF(cmod->ttype, MCDATA+MRCDATA+MNONE)) {
/* PARAMETER 3A: Minus exceptions.
*/
TRACEMD("3A: -grp");
switch (pcbmd.action) {
case MGRP:
/* We cheat and use nnmgrp for this. */
mexgrp = copygrp((PETD *)nnmgrp,
u = parsegrp((PETD *)nnmgrp, &pcbgrnm, tbuf));
++ds.pmexgcnt; ds.pmexcnt += u-1;
TRACEGRP(mexgrp);
parsemd(tbuf, NAMECASE, &pcblitp, NAMELEN);
default:
break;
}
/* PARAMETER 3B: Plus exceptions.
*/
TRACEMD("3B: +grp");
switch (pcbmd.action) {
case PGRP:
pexgrp = copygrp((PETD *)nnmgrp,
u = parsegrp((PETD *)nnmgrp, &pcbgrnm, tbuf));
++ds.pmexgcnt; ds.pmexcnt += u-1;
TRACEGRP(pexgrp);
parsemd(tbuf, NAMECASE, &pcblitp, NAMELEN);
default:
break;
}
}
/* PARAMETER 4: End of declaration.
*/
TRACEMD(emd);
if (pcbmd.action!=EMD) mderr(126, (UNCH *)0, (UNCH *)0);
if (es!=mdessv) synerr(37, &pcbmd);
/* EXECUTE: Store the definition for each element name specified.
*/
TRACEGRP(nmgrp);
for (i = -1; nmgrp[++i];) {
etdgi = nmgrp[i]->etdgi;
if (*ranksuff) {
if ((tbuf[0] = *etdgi + ustrlen(ranksuff)) - 2 > NAMELEN) {
mderr(131, etdgi+1, ranksuff);
continue;
}
memcpy(tbuf+1, etdgi+1, *etdgi-1);
ustrcpy(tbuf+*etdgi-1, ranksuff);
etdcan(etdgi);
nmgrp[i] = etddef(tbuf);
}
if (nmgrp[i]->etdmod) {mderr(56, etdgi+1, (UNCH *)0); continue;}
etdset(nmgrp[i], fmin+ETDDCL, cmod, mexgrp, pexgrp, nmgrp[i]->etdsrm);
++ds.etdcnt;
if (nmgrp[i]->adl) etdadl(nmgrp[i]); /* Check ETD conflicts. */
TRACEETD(nmgrp[i]);
}
}
static int
set_start_bits(pcre2_real_code *re, PCRE2_SPTR code, BOOL utf)
{
uint32_t c;
int yield = SSB_DONE;
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
int table_limit = utf? 16:32;
#else
int table_limit = 32;
#endif
do
{
BOOL try_next = TRUE;
PCRE2_SPTR tcode = code + 1 + LINK_SIZE;
if (*code == OP_CBRA || *code == OP_SCBRA ||
*code == OP_CBRAPOS || *code == OP_SCBRAPOS) tcode += IMM2_SIZE;
while (try_next) /* Loop for items in this branch */
{
int rc;
uint8_t *classmap = NULL;
switch(*tcode)
{
/* If we reach something we don't understand, it means a new opcode has
been created that hasn't been added to this function. Hopefully this
problem will be discovered during testing. */
default:
return SSB_UNKNOWN;
/* Fail for a valid opcode that implies no starting bits. */
case OP_ACCEPT:
case OP_ASSERT_ACCEPT:
case OP_ALLANY:
case OP_ANY:
case OP_ANYBYTE:
case OP_CIRC:
case OP_CIRCM:
case OP_CLOSE:
case OP_COMMIT:
case OP_COND:
case OP_CREF:
case OP_FALSE:
case OP_TRUE:
case OP_DNCREF:
case OP_DNREF:
case OP_DNREFI:
case OP_DNRREF:
case OP_DOLL:
case OP_DOLLM:
case OP_END:
case OP_EOD:
case OP_EODN:
case OP_EXTUNI:
case OP_FAIL:
case OP_MARK:
case OP_NOT:
case OP_NOTEXACT:
case OP_NOTEXACTI:
case OP_NOTI:
case OP_NOTMINPLUS:
case OP_NOTMINPLUSI:
case OP_NOTMINQUERY:
case OP_NOTMINQUERYI:
case OP_NOTMINSTAR:
case OP_NOTMINSTARI:
case OP_NOTMINUPTO:
case OP_NOTMINUPTOI:
case OP_NOTPLUS:
case OP_NOTPLUSI:
case OP_NOTPOSPLUS:
case OP_NOTPOSPLUSI:
case OP_NOTPOSQUERY:
case OP_NOTPOSQUERYI:
case OP_NOTPOSSTAR:
case OP_NOTPOSSTARI:
case OP_NOTPOSUPTO:
case OP_NOTPOSUPTOI:
case OP_NOTPROP:
case OP_NOTQUERY:
case OP_NOTQUERYI:
case OP_NOTSTAR:
case OP_NOTSTARI:
case OP_NOTUPTO:
case OP_NOTUPTOI:
case OP_NOT_HSPACE:
case OP_NOT_VSPACE:
case OP_PRUNE:
case OP_PRUNE_ARG:
case OP_RECURSE:
case OP_REF:
case OP_REFI:
case OP_REVERSE:
case OP_RREF:
case OP_SCOND:
case OP_SET_SOM:
case OP_SKIP:
case OP_SKIP_ARG:
case OP_SOD:
case OP_SOM:
case OP_THEN:
case OP_THEN_ARG:
return SSB_FAIL;
/* A "real" property test implies no starting bits, but the fake property
PT_CLIST identifies a list of characters. These lists are short, as they
are used for characters with more than one "other case", so there is no
point in recognizing them for OP_NOTPROP. */
case OP_PROP:
if (tcode[1] != PT_CLIST) return SSB_FAIL;
{
const uint32_t *p = PRIV(ucd_caseless_sets) + tcode[2];
while ((c = *p++) < NOTACHAR)
{
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
if (utf)
{
PCRE2_UCHAR buff[6];
(void)PRIV(ord2utf)(c, buff);
c = buff[0];
}
#endif
if (c > 0xff) SET_BIT(0xff); else SET_BIT(c);
}
}
try_next = FALSE;
break;
/* We can ignore word boundary tests. */
case OP_WORD_BOUNDARY:
case OP_NOT_WORD_BOUNDARY:
tcode++;
break;
/* If we hit a bracket or a positive lookahead assertion, recurse to set
bits from within the subpattern. If it can't find anything, we have to
give up. If it finds some mandatory character(s), we are done for this
branch. Otherwise, carry on scanning after the subpattern. */
case OP_BRA:
case OP_SBRA:
case OP_CBRA:
case OP_SCBRA:
case OP_BRAPOS:
case OP_SBRAPOS:
case OP_CBRAPOS:
case OP_SCBRAPOS:
case OP_ONCE:
case OP_ONCE_NC:
case OP_ASSERT:
rc = set_start_bits(re, tcode, utf);
if (rc == SSB_FAIL || rc == SSB_UNKNOWN) return rc;
if (rc == SSB_DONE) try_next = FALSE; else
{
do tcode += GET(tcode, 1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
}
break;
/* If we hit ALT or KET, it means we haven't found anything mandatory in
this branch, though we might have found something optional. For ALT, we
continue with the next alternative, but we have to arrange that the final
result from subpattern is SSB_CONTINUE rather than SSB_DONE. For KET,
return SSB_CONTINUE: if this is the top level, that indicates failure,
but after a nested subpattern, it causes scanning to continue. */
case OP_ALT:
yield = SSB_CONTINUE;
try_next = FALSE;
break;
case OP_KET:
case OP_KETRMAX:
case OP_KETRMIN:
case OP_KETRPOS:
return SSB_CONTINUE;
/* Skip over callout */
case OP_CALLOUT:
tcode += PRIV(OP_lengths)[OP_CALLOUT];
break;
case OP_CALLOUT_STR:
tcode += GET(tcode, 1 + 2*LINK_SIZE);
break;
/* Skip over lookbehind and negative lookahead assertions */
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
do tcode += GET(tcode, 1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
break;
/* BRAZERO does the bracket, but carries on. */
case OP_BRAZERO:
case OP_BRAMINZERO:
case OP_BRAPOSZERO:
rc = set_start_bits(re, ++tcode, utf);
if (rc == SSB_FAIL || rc == SSB_UNKNOWN) return rc;
do tcode += GET(tcode,1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
break;
/* SKIPZERO skips the bracket. */
case OP_SKIPZERO:
tcode++;
do tcode += GET(tcode,1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
break;
/* Single-char * or ? sets the bit and tries the next item */
case OP_STAR:
case OP_MINSTAR:
case OP_POSSTAR:
case OP_QUERY:
case OP_MINQUERY:
case OP_POSQUERY:
tcode = set_table_bit(re, tcode + 1, FALSE, utf);
break;
case OP_STARI:
case OP_MINSTARI:
case OP_POSSTARI:
case OP_QUERYI:
case OP_MINQUERYI:
case OP_POSQUERYI:
tcode = set_table_bit(re, tcode + 1, TRUE, utf);
break;
/* Single-char upto sets the bit and tries the next */
case OP_UPTO:
case OP_MINUPTO:
case OP_POSUPTO:
tcode = set_table_bit(re, tcode + 1 + IMM2_SIZE, FALSE, utf);
break;
case OP_UPTOI:
case OP_MINUPTOI:
case OP_POSUPTOI:
tcode = set_table_bit(re, tcode + 1 + IMM2_SIZE, TRUE, utf);
break;
/* At least one single char sets the bit and stops */
case OP_EXACT:
tcode += IMM2_SIZE;
/* Fall through */
case OP_CHAR:
case OP_PLUS:
case OP_MINPLUS:
case OP_POSPLUS:
(void)set_table_bit(re, tcode + 1, FALSE, utf);
try_next = FALSE;
break;
case OP_EXACTI:
tcode += IMM2_SIZE;
/* Fall through */
case OP_CHARI:
case OP_PLUSI:
case OP_MINPLUSI:
case OP_POSPLUSI:
(void)set_table_bit(re, tcode + 1, TRUE, utf);
try_next = FALSE;
break;
/* Special spacing and line-terminating items. These recognize specific
lists of characters. The difference between VSPACE and ANYNL is that the
latter can match the two-character CRLF sequence, but that is not
relevant for finding the first character, so their code here is
identical. */
case OP_HSPACE:
SET_BIT(CHAR_HT);
SET_BIT(CHAR_SPACE);
/* For the 16-bit and 32-bit libraries (which can never be EBCDIC), set
the bits for 0xA0 and for code units >= 255, independently of UTF. */
#if PCRE2_CODE_UNIT_WIDTH != 8
SET_BIT(0xA0);
SET_BIT(0xFF);
#else
/* For the 8-bit library in UTF-8 mode, set the bits for the first code
units of horizontal space characters. */
#ifdef SUPPORT_UNICODE
if (utf)
{
SET_BIT(0xC2); /* For U+00A0 */
SET_BIT(0xE1); /* For U+1680, U+180E */
SET_BIT(0xE2); /* For U+2000 - U+200A, U+202F, U+205F */
SET_BIT(0xE3); /* For U+3000 */
}
else
#endif
/* For the 8-bit library not in UTF-8 mode, set the bit for 0xA0, unless
the code is EBCDIC. */
{
#ifndef EBCDIC
SET_BIT(0xA0);
#endif /* Not EBCDIC */
}
#endif /* 8-bit support */
try_next = FALSE;
break;
case OP_ANYNL:
case OP_VSPACE:
SET_BIT(CHAR_LF);
SET_BIT(CHAR_VT);
SET_BIT(CHAR_FF);
SET_BIT(CHAR_CR);
/* For the 16-bit and 32-bit libraries (which can never be EBCDIC), set
the bits for NEL and for code units >= 255, independently of UTF. */
#if PCRE2_CODE_UNIT_WIDTH != 8
SET_BIT(CHAR_NEL);
SET_BIT(0xFF);
#else
/* For the 8-bit library in UTF-8 mode, set the bits for the first code
units of vertical space characters. */
#ifdef SUPPORT_UNICODE
if (utf)
{
SET_BIT(0xC2); /* For U+0085 (NEL) */
SET_BIT(0xE2); /* For U+2028, U+2029 */
}
else
#endif
/* For the 8-bit library not in UTF-8 mode, set the bit for NEL. */
{
SET_BIT(CHAR_NEL);
}
#endif /* 8-bit support */
try_next = FALSE;
break;
/* Single character types set the bits and stop. Note that if PCRE2_UCP
is set, we do not see these op codes because \d etc are converted to
properties. Therefore, these apply in the case when only characters less
than 256 are recognized to match the types. */
case OP_NOT_DIGIT:
set_nottype_bits(re, cbit_digit, table_limit);
try_next = FALSE;
break;
case OP_DIGIT:
set_type_bits(re, cbit_digit, table_limit);
try_next = FALSE;
break;
case OP_NOT_WHITESPACE:
set_nottype_bits(re, cbit_space, table_limit);
try_next = FALSE;
break;
case OP_WHITESPACE:
set_type_bits(re, cbit_space, table_limit);
try_next = FALSE;
break;
case OP_NOT_WORDCHAR:
set_nottype_bits(re, cbit_word, table_limit);
try_next = FALSE;
break;
case OP_WORDCHAR:
set_type_bits(re, cbit_word, table_limit);
try_next = FALSE;
break;
/* One or more character type fudges the pointer and restarts, knowing
it will hit a single character type and stop there. */
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEPOSPLUS:
tcode++;
break;
case OP_TYPEEXACT:
tcode += 1 + IMM2_SIZE;
break;
/* Zero or more repeats of character types set the bits and then
try again. */
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEPOSUPTO:
tcode += IMM2_SIZE; /* Fall through */
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEPOSSTAR:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEPOSQUERY:
switch(tcode[1])
{
default:
case OP_ANY:
case OP_ALLANY:
return SSB_FAIL;
case OP_HSPACE:
SET_BIT(CHAR_HT);
SET_BIT(CHAR_SPACE);
/* For the 16-bit and 32-bit libraries (which can never be EBCDIC), set
the bits for 0xA0 and for code units >= 255, independently of UTF. */
#if PCRE2_CODE_UNIT_WIDTH != 8
SET_BIT(0xA0);
SET_BIT(0xFF);
#else
/* For the 8-bit library in UTF-8 mode, set the bits for the first code
units of horizontal space characters. */
#ifdef SUPPORT_UNICODE
if (utf)
{
SET_BIT(0xC2); /* For U+00A0 */
SET_BIT(0xE1); /* For U+1680, U+180E */
SET_BIT(0xE2); /* For U+2000 - U+200A, U+202F, U+205F */
SET_BIT(0xE3); /* For U+3000 */
}
else
#endif
/* For the 8-bit library not in UTF-8 mode, set the bit for 0xA0, unless
the code is EBCDIC. */
{
#ifndef EBCDIC
SET_BIT(0xA0);
#endif /* Not EBCDIC */
}
#endif /* 8-bit support */
break;
case OP_ANYNL:
case OP_VSPACE:
SET_BIT(CHAR_LF);
SET_BIT(CHAR_VT);
SET_BIT(CHAR_FF);
SET_BIT(CHAR_CR);
/* For the 16-bit and 32-bit libraries (which can never be EBCDIC), set
the bits for NEL and for code units >= 255, independently of UTF. */
#if PCRE2_CODE_UNIT_WIDTH != 8
SET_BIT(CHAR_NEL);
SET_BIT(0xFF);
#else
/* For the 8-bit library in UTF-8 mode, set the bits for the first code
units of vertical space characters. */
#ifdef SUPPORT_UNICODE
if (utf)
{
SET_BIT(0xC2); /* For U+0085 (NEL) */
SET_BIT(0xE2); /* For U+2028, U+2029 */
}
else
#endif
/* For the 8-bit library not in UTF-8 mode, set the bit for NEL. */
{
SET_BIT(CHAR_NEL);
}
#endif /* 8-bit support */
break;
case OP_NOT_DIGIT:
set_nottype_bits(re, cbit_digit, table_limit);
break;
case OP_DIGIT:
set_type_bits(re, cbit_digit, table_limit);
break;
case OP_NOT_WHITESPACE:
set_nottype_bits(re, cbit_space, table_limit);
break;
case OP_WHITESPACE:
set_type_bits(re, cbit_space, table_limit);
break;
case OP_NOT_WORDCHAR:
set_nottype_bits(re, cbit_word, table_limit);
break;
case OP_WORDCHAR:
set_type_bits(re, cbit_word, table_limit);
break;
}
tcode += 2;
break;
/* Extended class: if there are any property checks, or if this is a
negative XCLASS without a map, give up. If there are no property checks,
there must be wide characters on the XCLASS list, because otherwise an
XCLASS would not have been created. This means that code points >= 255
are always potential starters. */
#ifdef SUPPORT_WIDE_CHARS
case OP_XCLASS:
if ((tcode[1 + LINK_SIZE] & XCL_HASPROP) != 0 ||
(tcode[1 + LINK_SIZE] & (XCL_MAP|XCL_NOT)) == XCL_NOT)
return SSB_FAIL;
/* We have a positive XCLASS or a negative one without a map. Set up the
map pointer if there is one, and fall through. */
classmap = ((tcode[1 + LINK_SIZE] & XCL_MAP) == 0)? NULL :
(uint8_t *)(tcode + 1 + LINK_SIZE + 1);
#endif
/* Enter here for a negative non-XCLASS. In the 8-bit library, if we are
in UTF mode, any byte with a value >= 0xc4 is a potentially valid starter
because it starts a character with a value > 255. In 8-bit non-UTF mode,
there is no difference between CLASS and NCLASS. In all other wide
character modes, set the 0xFF bit to indicate code units >= 255. */
case OP_NCLASS:
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
if (utf)
{
re->start_bitmap[24] |= 0xf0; /* Bits for 0xc4 - 0xc8 */
memset(re->start_bitmap+25, 0xff, 7); /* Bits for 0xc9 - 0xff */
}
#elif PCRE2_CODE_UNIT_WIDTH != 8
SET_BIT(0xFF); /* For characters >= 255 */
#endif
/* Fall through */
/* Enter here for a positive non-XCLASS. If we have fallen through from
an XCLASS, classmap will already be set; just advance the code pointer.
Otherwise, set up classmap for a a non-XCLASS and advance past it. */
case OP_CLASS:
if (*tcode == OP_XCLASS) tcode += GET(tcode, 1); else
{
classmap = (uint8_t *)(++tcode);
tcode += 32 / sizeof(PCRE2_UCHAR);
}
/* When wide characters are supported, classmap may be NULL. In UTF-8
(sic) mode, the bits in a class bit map correspond to character values,
not to byte values. However, the bit map we are constructing is for byte
values. So we have to do a conversion for characters whose code point is
greater than 127. In fact, there are only two possible starting bytes for
characters in the range 128 - 255. */
if (classmap != NULL)
{
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
if (utf)
{
for (c = 0; c < 16; c++) re->start_bitmap[c] |= classmap[c];
for (c = 128; c < 256; c++)
{
if ((classmap[c/8] & (1 << (c&7))) != 0)
{
int d = (c >> 6) | 0xc0; /* Set bit for this starter */
re->start_bitmap[d/8] |= (1 << (d&7)); /* and then skip on to the */
c = (c & 0xc0) + 0x40 - 1; /* next relevant character. */
}
}
}
else
#endif
/* In all modes except UTF-8, the two bit maps are compatible. */
{
for (c = 0; c < 32; c++) re->start_bitmap[c] |= classmap[c];
}
}
/* Act on what follows the class. For a zero minimum repeat, continue;
otherwise stop processing. */
switch (*tcode)
{
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSSTAR:
case OP_CRPOSQUERY:
tcode++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
if (GET2(tcode, 1) == 0) tcode += 1 + 2 * IMM2_SIZE;
else try_next = FALSE;
break;
default:
try_next = FALSE;
break;
}
break; /* End of class handling case */
} /* End of switch for opcodes */
} /* End of try_next loop */
static void pnl_basis_eval_test ()
{
PnlMat *X;
PnlVect *V, *x, *t, *D, *alpha, *lower, *upper;
PnlRng *rng;
PnlBasis *basis;
int j, deg, n;
double t0, x0, tol;
tol = 1E-8;
deg=5; //total degree
n=50;
D=pnl_vect_create(5);
x=pnl_vect_create(n);
t=pnl_vect_create(n);
t0=0.5;
x0=2.5;
rng = pnl_rng_create (PNL_RNG_MERSENNE);
pnl_rng_sseed (rng, 0);
/*
* Random points where the function will be evaluted
*/
pnl_vect_rng_uni(x,n,-5,4,rng);
pnl_vect_rng_uni(t,n,0,1,rng);
basis = pnl_basis_create_from_degree (PNL_BASIS_HERMITIAN, deg, 2);
alpha = pnl_vect_create (basis->nb_func);
X = pnl_mat_create (n, 2);
for(j=0;j<n;j++)
{
MLET (X, j, 0) = GET(t,j);
MLET (X, j, 1) = GET(x,j);
}
V=pnl_vect_create(n);
/*
* Vector of values for the function to recover
*/
for(j=0;j<n;j++)
{
LET(V,j)=fonction_a_retrouver(GET(t,j),GET(x,j));
}
pnl_basis_fit_ls (basis, alpha, X, V);
/*
* compute approximations of the derivatives (first order in time and
* second order in space )
*/
derive_approx_fonction(basis, D, alpha,t0,x0);
pnl_test_eq_abs (pnl_vect_get(D,0), fonction_a_retrouver(t0,x0), tol,
"deriv_approx_fonction", "derivative 0");
pnl_test_eq_abs (derive_x_approx_fonction(basis, alpha, t0, x0),
derive_x_fonction_a_retrouver(t0,x0), tol,
"deriv_approx_fonction", "derivative %% x");
pnl_test_eq_abs (pnl_vect_get(D,2), derive_xx_fonction_a_retrouver(t0,x0),
tol, "deriv_approx_fonction", "derivative %% xx");
pnl_test_eq_abs (pnl_vect_get(D,3), derive_t_fonction_a_retrouver(t0,x0),
tol, "deriv_approx_fonction", "derivative %% t");
pnl_test_eq_abs (pnl_vect_get(D,4), derive_xt_fonction_a_retrouver(t0,x0),
tol, "deriv_approx_fonction", "derivative %% tx");
pnl_basis_free (&basis);
/* reduced basis */
basis = pnl_basis_create_from_degree (PNL_BASIS_HERMITIAN, deg, 2);
lower = pnl_vect_create_from_list (2, 0., -5.);
upper = pnl_vect_create_from_list (2, 1., 4.);
pnl_basis_set_domain (basis, lower, upper);
pnl_basis_fit_ls (basis, alpha, X, V);
derive_approx_fonction(basis, D, alpha,t0,x0);
pnl_test_eq_abs (pnl_vect_get(D,0), fonction_a_retrouver(t0,x0), tol,
"deriv_approx_fonction (reduced)", "derivative 0");
pnl_test_eq_abs (derive_x_approx_fonction(basis, alpha, t0, x0),
derive_x_fonction_a_retrouver(t0,x0), tol,
"deriv_approx_fonction (reduced)", "derivative %% x");
pnl_test_eq_abs (pnl_vect_get(D,2), derive_xx_fonction_a_retrouver(t0,x0),
tol, "deriv_approx_fonction (reduced)", "derivative %% xx");
pnl_test_eq_abs (pnl_vect_get(D,3), derive_t_fonction_a_retrouver(t0,x0),
tol, "deriv_approx_fonction (reduced)", "derivative %% t");
pnl_test_eq_abs (pnl_vect_get(D,4), derive_xt_fonction_a_retrouver(t0,x0),
tol, "deriv_approx_fonction (reduced)", "derivative %% tx");
pnl_basis_free (&basis);
pnl_rng_free (&rng);
pnl_vect_free(&alpha);
pnl_vect_free(&x);
pnl_vect_free(&t);
pnl_vect_free(&V);
pnl_vect_free(&D);
pnl_vect_free(&lower);
pnl_vect_free(&upper);
pnl_mat_free(&X);
}
static BOOL
set_start_bits(const uschar *code, uschar *start_bits, BOOL caseless,
BOOL utf8, compile_data *cd)
{
register int c;
/* This next statement and the later reference to dummy are here in order to
trick the optimizer of the IBM C compiler for OS/2 into generating correct
code. Apparently IBM isn't going to fix the problem, and we would rather not
disable optimization (in this module it actually makes a big difference, and
the pcre module can use all the optimization it can get). */
volatile int dummy;
do
{
const uschar *tcode = code + 1 + LINK_SIZE;
BOOL try_next = TRUE;
while (try_next)
{
/* If a branch starts with a bracket or a positive lookahead assertion,
recurse to set bits from within them. That's all for this branch. */
if ((int)*tcode >= OP_BRA || *tcode == OP_ASSERT)
{
if (!set_start_bits(tcode, start_bits, caseless, utf8, cd))
return FALSE;
try_next = FALSE;
}
else switch(*tcode)
{
default:
return FALSE;
/* Skip over callout */
case OP_CALLOUT:
tcode += 2;
break;
/* Skip over extended extraction bracket number */
case OP_BRANUMBER:
tcode += 3;
break;
/* Skip over lookbehind and negative lookahead assertions */
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
do tcode += GET(tcode, 1); while (*tcode == OP_ALT);
tcode += 1+LINK_SIZE;
break;
/* Skip over an option setting, changing the caseless flag */
case OP_OPT:
caseless = (tcode[1] & PCRE_CASELESS) != 0;
tcode += 2;
break;
/* BRAZERO does the bracket, but carries on. */
case OP_BRAZERO:
case OP_BRAMINZERO:
if (!set_start_bits(++tcode, start_bits, caseless, utf8, cd))
return FALSE;
dummy = 1;
do tcode += GET(tcode,1); while (*tcode == OP_ALT);
tcode += 1+LINK_SIZE;
break;
/* Single-char * or ? sets the bit and tries the next item */
case OP_STAR:
case OP_MINSTAR:
case OP_QUERY:
case OP_MINQUERY:
set_bit(start_bits, tcode[1], caseless, cd);
tcode += 2;
#ifdef SUPPORT_UTF8
if (utf8) while ((*tcode & 0xc0) == 0x80) tcode++;
#endif
break;
/* Single-char upto sets the bit and tries the next */
case OP_UPTO:
case OP_MINUPTO:
set_bit(start_bits, tcode[3], caseless, cd);
tcode += 4;
#ifdef SUPPORT_UTF8
if (utf8) while ((*tcode & 0xc0) == 0x80) tcode++;
#endif
break;
/* At least one single char sets the bit and stops */
case OP_EXACT: /* Fall through */
tcode++;
case OP_CHARS: /* Fall through */
tcode++;
case OP_PLUS:
case OP_MINPLUS:
set_bit(start_bits, tcode[1], caseless, cd);
try_next = FALSE;
break;
/* Single character type sets the bits and stops */
case OP_NOT_DIGIT:
for (c = 0; c < 32; c++)
start_bits[c] |= ~cd->cbits[c+cbit_digit];
try_next = FALSE;
break;
case OP_DIGIT:
for (c = 0; c < 32; c++)
start_bits[c] |= cd->cbits[c+cbit_digit];
try_next = FALSE;
break;
case OP_NOT_WHITESPACE:
for (c = 0; c < 32; c++)
start_bits[c] |= ~cd->cbits[c+cbit_space];
try_next = FALSE;
break;
case OP_WHITESPACE:
for (c = 0; c < 32; c++)
start_bits[c] |= cd->cbits[c+cbit_space];
try_next = FALSE;
break;
case OP_NOT_WORDCHAR:
for (c = 0; c < 32; c++)
start_bits[c] |= ~cd->cbits[c+cbit_word];
try_next = FALSE;
break;
case OP_WORDCHAR:
for (c = 0; c < 32; c++)
start_bits[c] |= cd->cbits[c+cbit_word];
try_next = FALSE;
break;
/* One or more character type fudges the pointer and restarts, knowing
it will hit a single character type and stop there. */
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
tcode++;
break;
case OP_TYPEEXACT:
tcode += 3;
break;
/* Zero or more repeats of character types set the bits and then
try again. */
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
tcode += 2; /* Fall through */
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
switch(tcode[1])
{
case OP_ANY:
return FALSE;
case OP_NOT_DIGIT:
for (c = 0; c < 32; c++)
start_bits[c] |= ~cd->cbits[c+cbit_digit];
break;
case OP_DIGIT:
for (c = 0; c < 32; c++)
start_bits[c] |= cd->cbits[c+cbit_digit];
break;
case OP_NOT_WHITESPACE:
for (c = 0; c < 32; c++)
start_bits[c] |= ~cd->cbits[c+cbit_space];
break;
case OP_WHITESPACE:
for (c = 0; c < 32; c++)
start_bits[c] |= cd->cbits[c+cbit_space];
break;
case OP_NOT_WORDCHAR:
for (c = 0; c < 32; c++)
start_bits[c] |= ~cd->cbits[c+cbit_word];
break;
case OP_WORDCHAR:
for (c = 0; c < 32; c++)
start_bits[c] |= cd->cbits[c+cbit_word];
break;
}
tcode += 2;
break;
/* Character class where all the information is in a bit map: set the
bits and either carry on or not, according to the repeat count. If it was
a negative class, and we are operating with UTF-8 characters, any byte
with a value >= 0xc4 is a potentially valid starter because it starts a
character with a value > 255. */
case OP_NCLASS:
if (utf8)
{
start_bits[24] |= 0xf0; /* Bits for 0xc4 - 0xc8 */
memset(start_bits+25, 0xff, 7); /* Bits for 0xc9 - 0xff */
}
/* Fall through */
case OP_CLASS:
{
tcode++;
/* In UTF-8 mode, the bits in a bit map correspond to character
values, not to byte values. However, the bit map we are constructing is
for byte values. So we have to do a conversion for characters whose
value is > 127. In fact, there are only two possible starting bytes for
characters in the range 128 - 255. */
if (utf8)
{
for (c = 0; c < 16; c++) start_bits[c] |= tcode[c];
for (c = 128; c < 256; c++)
{
if ((tcode[c/8] && (1 << (c&7))) != 0)
{
int d = (c >> 6) | 0xc0; /* Set bit for this starter */
start_bits[d/8] |= (1 << (d&7)); /* and then skip on to the */
c = (c & 0xc0) + 0x40 - 1; /* next relevant character. */
}
}
}
/* In non-UTF-8 mode, the two bit maps are completely compatible. */
else
{
for (c = 0; c < 32; c++) start_bits[c] |= tcode[c];
}
/* Advance past the bit map, and act on what follows */
tcode += 32;
switch (*tcode)
{
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
tcode++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
if (((tcode[1] << 8) + tcode[2]) == 0) tcode += 5;
else try_next = FALSE;
break;
default:
try_next = FALSE;
break;
}
}
break; /* End of bitmap class handling */
} /* End of switch */
} /* End of try_next loop */
static int
find_minlength(const uschar *code, const uschar *startcode, int options)
{
int length = -1;
BOOL utf8 = (options & PCRE_UTF8) != 0;
BOOL had_recurse = FALSE;
register int branchlength = 0;
register uschar *cc = (uschar *)code + 1 + LINK_SIZE;
if (*code == OP_CBRA || *code == OP_SCBRA) cc += 2;
/* Scan along the opcodes for this branch. If we get to the end of the
branch, check the length against that of the other branches. */
for (;;)
{
int d, min;
uschar *cs, *ce;
register int op = *cc;
switch (op)
{
case OP_COND:
case OP_SCOND:
/* If there is only one branch in a condition, the implied branch has zero
length, so we don't add anything. This covers the DEFINE "condition"
automatically. */
cs = cc + GET(cc, 1);
if (*cs != OP_ALT)
{
cc = cs + 1 + LINK_SIZE;
break;
}
/* Otherwise we can fall through and treat it the same as any other
subpattern. */
case OP_CBRA:
case OP_SCBRA:
case OP_BRA:
case OP_SBRA:
case OP_ONCE:
d = find_minlength(cc, startcode, options);
if (d < 0) return d;
branchlength += d;
do cc += GET(cc, 1); while (*cc == OP_ALT);
cc += 1 + LINK_SIZE;
break;
/* Reached end of a branch; if it's a ket it is the end of a nested
call. If it's ALT it is an alternation in a nested call. If it is
END it's the end of the outer call. All can be handled by the same code. */
case OP_ALT:
case OP_KET:
case OP_KETRMAX:
case OP_KETRMIN:
case OP_END:
if (length < 0 || (!had_recurse && branchlength < length))
length = branchlength;
if (*cc != OP_ALT) return length;
cc += 1 + LINK_SIZE;
branchlength = 0;
had_recurse = FALSE;
break;
/* Skip over assertive subpatterns */
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
do cc += GET(cc, 1); while (*cc == OP_ALT);
/* Fall through */
/* Skip over things that don't match chars */
case OP_REVERSE:
case OP_CREF:
case OP_NCREF:
case OP_RREF:
case OP_NRREF:
case OP_DEF:
case OP_OPT:
case OP_CALLOUT:
case OP_SOD:
case OP_SOM:
case OP_EOD:
case OP_EODN:
case OP_CIRC:
case OP_DOLL:
case OP_NOT_WORD_BOUNDARY:
case OP_WORD_BOUNDARY:
cc += _pcre_OP_lengths[*cc];
break;
/* Skip over a subpattern that has a {0} or {0,x} quantifier */
case OP_BRAZERO:
case OP_BRAMINZERO:
case OP_SKIPZERO:
cc += _pcre_OP_lengths[*cc];
do cc += GET(cc, 1); while (*cc == OP_ALT);
cc += 1 + LINK_SIZE;
break;
/* Handle literal characters and + repetitions */
case OP_CHAR:
case OP_CHARNC:
case OP_NOT:
case OP_PLUS:
case OP_MINPLUS:
case OP_POSPLUS:
case OP_NOTPLUS:
case OP_NOTMINPLUS:
case OP_NOTPOSPLUS:
branchlength++;
cc += 2;
#ifdef SUPPORT_UTF8
if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];
#endif
break;
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEPOSPLUS:
branchlength++;
cc += (cc[1] == OP_PROP || cc[1] == OP_NOTPROP)? 4 : 2;
break;
/* Handle exact repetitions. The count is already in characters, but we
need to skip over a multibyte character in UTF8 mode. */
case OP_EXACT:
case OP_NOTEXACT:
branchlength += GET2(cc,1);
cc += 4;
#ifdef SUPPORT_UTF8
if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];
#endif
break;
case OP_TYPEEXACT:
branchlength += GET2(cc,1);
cc += (cc[3] == OP_PROP || cc[3] == OP_NOTPROP)? 6 : 4;
break;
/* Handle single-char non-literal matchers */
case OP_PROP:
case OP_NOTPROP:
cc += 2;
/* Fall through */
case OP_NOT_DIGIT:
case OP_DIGIT:
case OP_NOT_WHITESPACE:
case OP_WHITESPACE:
case OP_NOT_WORDCHAR:
case OP_WORDCHAR:
case OP_ANY:
case OP_ALLANY:
case OP_EXTUNI:
case OP_HSPACE:
case OP_NOT_HSPACE:
case OP_VSPACE:
case OP_NOT_VSPACE:
branchlength++;
cc++;
break;
/* "Any newline" might match two characters */
case OP_ANYNL:
branchlength += 2;
cc++;
break;
/* The single-byte matcher means we can't proceed in UTF-8 mode */
case OP_ANYBYTE:
#ifdef SUPPORT_UTF8
if (utf8) return -1;
#endif
branchlength++;
cc++;
break;
/* For repeated character types, we have to test for \p and \P, which have
an extra two bytes of parameters. */
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEPOSSTAR:
case OP_TYPEPOSQUERY:
if (cc[1] == OP_PROP || cc[1] == OP_NOTPROP) cc += 2;
cc += _pcre_OP_lengths[op];
break;
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEPOSUPTO:
if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;
cc += _pcre_OP_lengths[op];
break;
/* Check a class for variable quantification */
#ifdef SUPPORT_UTF8
case OP_XCLASS:
cc += GET(cc, 1) - 33;
/* Fall through */
#endif
case OP_CLASS:
case OP_NCLASS:
cc += 33;
switch (*cc)
{
case OP_CRPLUS:
case OP_CRMINPLUS:
branchlength++;
/* Fall through */
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
cc++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
branchlength += GET2(cc,1);
cc += 5;
break;
default:
branchlength++;
break;
}
break;
/* Backreferences and subroutine calls are treated in the same way: we find
the minimum length for the subpattern. A recursion, however, causes an
a flag to be set that causes the length of this branch to be ignored. The
logic is that a recursion can only make sense if there is another
alternation that stops the recursing. That will provide the minimum length
(when no recursion happens). A backreference within the group that it is
referencing behaves in the same way.
If PCRE_JAVASCRIPT_COMPAT is set, a backreference to an unset bracket
matches an empty string (by default it causes a matching failure), so in
that case we must set the minimum length to zero. */
case OP_REF:
if ((options & PCRE_JAVASCRIPT_COMPAT) == 0)
{
ce = cs = (uschar *)_pcre_find_bracket(startcode, utf8, GET2(cc, 1));
if (cs == NULL) return -2;
do ce += GET(ce, 1); while (*ce == OP_ALT);
if (cc > cs && cc < ce)
{
d = 0;
had_recurse = TRUE;
}
else d = find_minlength(cs, startcode, options);
}
else d = 0;
cc += 3;
/* Handle repeated back references */
switch (*cc)
{
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
min = 0;
cc++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
min = GET2(cc, 1);
cc += 5;
break;
default:
min = 1;
break;
}
branchlength += min * d;
break;
case OP_RECURSE:
cs = ce = (uschar *)startcode + GET(cc, 1);
if (cs == NULL) return -2;
do ce += GET(ce, 1); while (*ce == OP_ALT);
if (cc > cs && cc < ce)
had_recurse = TRUE;
else
branchlength += find_minlength(cs, startcode, options);
cc += 1 + LINK_SIZE;
break;
/* Anything else does not or need not match a character. We can get the
item's length from the table, but for those that can match zero occurrences
of a character, we must take special action for UTF-8 characters. */
case OP_UPTO:
case OP_NOTUPTO:
case OP_MINUPTO:
case OP_NOTMINUPTO:
case OP_POSUPTO:
case OP_STAR:
case OP_MINSTAR:
case OP_NOTMINSTAR:
case OP_POSSTAR:
case OP_NOTPOSSTAR:
case OP_QUERY:
case OP_MINQUERY:
case OP_NOTMINQUERY:
case OP_POSQUERY:
case OP_NOTPOSQUERY:
cc += _pcre_OP_lengths[op];
#ifdef SUPPORT_UTF8
if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];
#endif
break;
/* Skip these, but we need to add in the name length. */
case OP_MARK:
case OP_PRUNE_ARG:
case OP_SKIP_ARG:
case OP_THEN_ARG:
cc += _pcre_OP_lengths[op] + cc[1];
break;
/* For the record, these are the opcodes that are matched by "default":
OP_ACCEPT, OP_CLOSE, OP_COMMIT, OP_FAIL, OP_PRUNE, OP_SET_SOM, OP_SKIP,
OP_THEN. */
default:
cc += _pcre_OP_lengths[op];
break;
}
}
/* Control never gets here */
}
vidTDeint::vidTDeint (AVDMGenericVideoStream * in, CONFcouple * couples)
{
_in = in;
memcpy (&_info, _in->getInfo (), sizeof (_info));
_info.encoding = 1;
vidCache = new VideoCache (10, in);
_uncompressed=new ADMImage(_info.width,_info.height);
scratch=new ADMImage(_info.width,_info.height);
scratch2=new ADMImage(_info.width,_info.height);
input = cArray = NULL;
_param=new TDEINT_PARAM;
if(!couples)
{
_param->mode=0;
_param->order=-1;
_param->field=-1;
_param->mthreshL=6;
_param->mthreshC=6;
_param->map=0;
_param->type=2;
_param->debug=0;
_param->mtnmode=1;
_param->sharp=1;
_param->full=1;
_param->cthresh=6;
_param->blockx=16;
_param->blocky=16;
_param->chroma=0;
_param->MI=64;
_param->tryWeave=false;
_param->link=2;
_param->denoise=true;
_param->AP=254;
_param->APType=1;
}
else
{
GET(mode);
GET(order);
GET(field);
GET(mthreshL);
GET(mthreshC);
GET(map);
GET(type);
GET(debug);
GET(mtnmode);
GET(sharp);
GET(full);
GET(cthresh);
GET(blockx);
GET(blocky);
GET(chroma);
GET(MI);
GET(tryWeave);
GET(link);
GET(denoise);
GET(AP);
GET(APType);
}
order=1;
orderS=1;
mode=0;
field=0;
fieldS=0;
mthreshL=6;
mthreshLS=6;
mthreshC=6;
mthreshCS=6;
map=0;
ovrDefault=0;
type=2; // kernel deint
debug=true;
mntmode=1;
mtnmode=1;
sharp=true;
hints=false;
full=false;
cthresh=12;
ovr=NULL;
input=NULL;
blocky=16;
blockx=16;
chroma=false;
MI=64;
tryWeave=false;
link=2;
AP=254;
APType=254;
reset();
}
static int
find_minlength(const REAL_PCRE *re, const pcre_uchar *code,
const pcre_uchar *startcode, int options, int recurse_depth)
{
int length = -1;
/* PCRE_UTF16 has the same value as PCRE_UTF8. */
BOOL utf = (options & PCRE_UTF8) != 0;
BOOL had_recurse = FALSE;
register int branchlength = 0;
register pcre_uchar *cc = (pcre_uchar *)code + 1 + LINK_SIZE;
if (*code == OP_CBRA || *code == OP_SCBRA ||
*code == OP_CBRAPOS || *code == OP_SCBRAPOS) cc += IMM2_SIZE;
/* Scan along the opcodes for this branch. If we get to the end of the
branch, check the length against that of the other branches. */
for (;;)
{
int d, min;
pcre_uchar *cs, *ce;
register pcre_uchar op = *cc;
switch (op)
{
case OP_COND:
case OP_SCOND:
/* If there is only one branch in a condition, the implied branch has zero
length, so we don't add anything. This covers the DEFINE "condition"
automatically. */
cs = cc + GET(cc, 1);
if (*cs != OP_ALT)
{
cc = cs + 1 + LINK_SIZE;
break;
}
/* Otherwise we can fall through and treat it the same as any other
subpattern. */
case OP_CBRA:
case OP_SCBRA:
case OP_BRA:
case OP_SBRA:
case OP_CBRAPOS:
case OP_SCBRAPOS:
case OP_BRAPOS:
case OP_SBRAPOS:
case OP_ONCE:
case OP_ONCE_NC:
d = find_minlength(re, cc, startcode, options, recurse_depth);
if (d < 0) return d;
branchlength += d;
do cc += GET(cc, 1); while (*cc == OP_ALT);
cc += 1 + LINK_SIZE;
break;
/* ACCEPT makes things far too complicated; we have to give up. */
case OP_ACCEPT:
case OP_ASSERT_ACCEPT:
return -1;
/* Reached end of a branch; if it's a ket it is the end of a nested
call. If it's ALT it is an alternation in a nested call. If it is END it's
the end of the outer call. All can be handled by the same code. If an
ACCEPT was previously encountered, use the length that was in force at that
time, and pass back the shortest ACCEPT length. */
case OP_ALT:
case OP_KET:
case OP_KETRMAX:
case OP_KETRMIN:
case OP_KETRPOS:
case OP_END:
if (length < 0 || (!had_recurse && branchlength < length))
length = branchlength;
if (op != OP_ALT) return length;
cc += 1 + LINK_SIZE;
branchlength = 0;
had_recurse = FALSE;
break;
/* Skip over assertive subpatterns */
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
do cc += GET(cc, 1); while (*cc == OP_ALT);
/* Fall through */
/* Skip over things that don't match chars */
case OP_REVERSE:
case OP_CREF:
case OP_DNCREF:
case OP_RREF:
case OP_DNRREF:
case OP_DEF:
case OP_CALLOUT:
case OP_SOD:
case OP_SOM:
case OP_EOD:
case OP_EODN:
case OP_CIRC:
case OP_CIRCM:
case OP_DOLL:
case OP_DOLLM:
case OP_NOT_WORD_BOUNDARY:
case OP_WORD_BOUNDARY:
cc += PRIV(OP_lengths)[*cc];
break;
/* Skip over a subpattern that has a {0} or {0,x} quantifier */
case OP_BRAZERO:
case OP_BRAMINZERO:
case OP_BRAPOSZERO:
case OP_SKIPZERO:
cc += PRIV(OP_lengths)[*cc];
do cc += GET(cc, 1); while (*cc == OP_ALT);
cc += 1 + LINK_SIZE;
break;
/* Handle literal characters and + repetitions */
case OP_CHAR:
case OP_CHARI:
case OP_NOT:
case OP_NOTI:
case OP_PLUS:
case OP_PLUSI:
case OP_MINPLUS:
case OP_MINPLUSI:
case OP_POSPLUS:
case OP_POSPLUSI:
case OP_NOTPLUS:
case OP_NOTPLUSI:
case OP_NOTMINPLUS:
case OP_NOTMINPLUSI:
case OP_NOTPOSPLUS:
case OP_NOTPOSPLUSI:
branchlength++;
cc += 2;
#ifdef SUPPORT_UTF
if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEPOSPLUS:
branchlength++;
cc += (cc[1] == OP_PROP || cc[1] == OP_NOTPROP)? 4 : 2;
break;
/* Handle exact repetitions. The count is already in characters, but we
need to skip over a multibyte character in UTF8 mode. */
case OP_EXACT:
case OP_EXACTI:
case OP_NOTEXACT:
case OP_NOTEXACTI:
branchlength += GET2(cc,1);
cc += 2 + IMM2_SIZE;
#ifdef SUPPORT_UTF
if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
case OP_TYPEEXACT:
branchlength += GET2(cc,1);
cc += 2 + IMM2_SIZE + ((cc[1 + IMM2_SIZE] == OP_PROP
|| cc[1 + IMM2_SIZE] == OP_NOTPROP)? 2 : 0);
break;
/* Handle single-char non-literal matchers */
case OP_PROP:
case OP_NOTPROP:
cc += 2;
/* Fall through */
case OP_NOT_DIGIT:
case OP_DIGIT:
case OP_NOT_WHITESPACE:
case OP_WHITESPACE:
case OP_NOT_WORDCHAR:
case OP_WORDCHAR:
case OP_ANY:
case OP_ALLANY:
case OP_EXTUNI:
case OP_HSPACE:
case OP_NOT_HSPACE:
case OP_VSPACE:
case OP_NOT_VSPACE:
branchlength++;
cc++;
break;
/* "Any newline" might match two characters, but it also might match just
one. */
case OP_ANYNL:
branchlength += 1;
cc++;
break;
/* The single-byte matcher means we can't proceed in UTF-8 mode. (In
non-UTF-8 mode \C will actually be turned into OP_ALLANY, so won't ever
appear, but leave the code, just in case.) */
case OP_ANYBYTE:
#ifdef SUPPORT_UTF
if (utf) return -1;
#endif
branchlength++;
cc++;
break;
/* For repeated character types, we have to test for \p and \P, which have
an extra two bytes of parameters. */
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEPOSSTAR:
case OP_TYPEPOSQUERY:
if (cc[1] == OP_PROP || cc[1] == OP_NOTPROP) cc += 2;
cc += PRIV(OP_lengths)[op];
break;
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEPOSUPTO:
if (cc[1 + IMM2_SIZE] == OP_PROP
|| cc[1 + IMM2_SIZE] == OP_NOTPROP) cc += 2;
cc += PRIV(OP_lengths)[op];
break;
/* Check a class for variable quantification */
case OP_CLASS:
case OP_NCLASS:
#if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
case OP_XCLASS:
/* The original code caused an unsigned overflow in 64 bit systems,
so now we use a conditional statement. */
if (op == OP_XCLASS)
cc += GET(cc, 1);
else
cc += PRIV(OP_lengths)[OP_CLASS];
#else
cc += PRIV(OP_lengths)[OP_CLASS];
#endif
switch (*cc)
{
case OP_CRPLUS:
case OP_CRMINPLUS:
case OP_CRPOSPLUS:
branchlength++;
/* Fall through */
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSSTAR:
case OP_CRPOSQUERY:
cc++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
branchlength += GET2(cc,1);
cc += 1 + 2 * IMM2_SIZE;
break;
default:
branchlength++;
break;
}
break;
/* Backreferences and subroutine calls are treated in the same way: we find
the minimum length for the subpattern. A recursion, however, causes an
a flag to be set that causes the length of this branch to be ignored. The
logic is that a recursion can only make sense if there is another
alternation that stops the recursing. That will provide the minimum length
(when no recursion happens). A backreference within the group that it is
referencing behaves in the same way.
If PCRE_JAVASCRIPT_COMPAT is set, a backreference to an unset bracket
matches an empty string (by default it causes a matching failure), so in
that case we must set the minimum length to zero. */
case OP_DNREF: /* Duplicate named pattern back reference */
case OP_DNREFI:
if ((options & PCRE_JAVASCRIPT_COMPAT) == 0)
{
int count = GET2(cc, 1+IMM2_SIZE);
pcre_uchar *slot = (pcre_uchar *)re +
re->name_table_offset + GET2(cc, 1) * re->name_entry_size;
d = INT_MAX;
while (count-- > 0)
{
ce = cs = (pcre_uchar *)PRIV(find_bracket)(startcode, utf, GET2(slot, 0));
if (cs == NULL) return -2;
do ce += GET(ce, 1); while (*ce == OP_ALT);
if (cc > cs && cc < ce)
{
d = 0;
had_recurse = TRUE;
break;
}
else
{
int dd = find_minlength(re, cs, startcode, options, recurse_depth);
if (dd < d) d = dd;
}
slot += re->name_entry_size;
}
}
else d = 0;
cc += 1 + 2*IMM2_SIZE;
goto REPEAT_BACK_REFERENCE;
case OP_REF: /* Single back reference */
case OP_REFI:
if ((options & PCRE_JAVASCRIPT_COMPAT) == 0)
{
ce = cs = (pcre_uchar *)PRIV(find_bracket)(startcode, utf, GET2(cc, 1));
if (cs == NULL) return -2;
do ce += GET(ce, 1); while (*ce == OP_ALT);
if (cc > cs && cc < ce)
{
d = 0;
had_recurse = TRUE;
}
else
{
d = find_minlength(re, cs, startcode, options, recurse_depth);
}
}
else d = 0;
cc += 1 + IMM2_SIZE;
/* Handle repeated back references */
REPEAT_BACK_REFERENCE:
switch (*cc)
{
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSSTAR:
case OP_CRPOSQUERY:
min = 0;
cc++;
break;
case OP_CRPLUS:
case OP_CRMINPLUS:
case OP_CRPOSPLUS:
min = 1;
cc++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
min = GET2(cc, 1);
cc += 1 + 2 * IMM2_SIZE;
break;
default:
min = 1;
break;
}
branchlength += min * d;
break;
/* We can easily detect direct recursion, but not mutual recursion. This is
caught by a recursion depth count. */
case OP_RECURSE:
cs = ce = (pcre_uchar *)startcode + GET(cc, 1);
do ce += GET(ce, 1); while (*ce == OP_ALT);
if ((cc > cs && cc < ce) || recurse_depth > 10)
had_recurse = TRUE;
else
{
branchlength += find_minlength(re, cs, startcode, options,
recurse_depth + 1);
}
cc += 1 + LINK_SIZE;
break;
/* Anything else does not or need not match a character. We can get the
item's length from the table, but for those that can match zero occurrences
of a character, we must take special action for UTF-8 characters. As it
happens, the "NOT" versions of these opcodes are used at present only for
ASCII characters, so they could be omitted from this list. However, in
future that may change, so we include them here so as not to leave a
gotcha for a future maintainer. */
case OP_UPTO:
case OP_UPTOI:
case OP_NOTUPTO:
case OP_NOTUPTOI:
case OP_MINUPTO:
case OP_MINUPTOI:
case OP_NOTMINUPTO:
case OP_NOTMINUPTOI:
case OP_POSUPTO:
case OP_POSUPTOI:
case OP_NOTPOSUPTO:
case OP_NOTPOSUPTOI:
case OP_STAR:
case OP_STARI:
case OP_NOTSTAR:
case OP_NOTSTARI:
case OP_MINSTAR:
case OP_MINSTARI:
case OP_NOTMINSTAR:
case OP_NOTMINSTARI:
case OP_POSSTAR:
case OP_POSSTARI:
case OP_NOTPOSSTAR:
case OP_NOTPOSSTARI:
case OP_QUERY:
case OP_QUERYI:
case OP_NOTQUERY:
case OP_NOTQUERYI:
case OP_MINQUERY:
case OP_MINQUERYI:
case OP_NOTMINQUERY:
case OP_NOTMINQUERYI:
case OP_POSQUERY:
case OP_POSQUERYI:
case OP_NOTPOSQUERY:
case OP_NOTPOSQUERYI:
cc += PRIV(OP_lengths)[op];
#ifdef SUPPORT_UTF
if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
/* Skip these, but we need to add in the name length. */
case OP_MARK:
case OP_PRUNE_ARG:
case OP_SKIP_ARG:
case OP_THEN_ARG:
cc += PRIV(OP_lengths)[op] + cc[1];
break;
/* The remaining opcodes are just skipped over. */
case OP_CLOSE:
case OP_COMMIT:
case OP_FAIL:
case OP_PRUNE:
case OP_SET_SOM:
case OP_SKIP:
case OP_THEN:
cc += PRIV(OP_lengths)[op];
break;
/* This should not occur: we list all opcodes explicitly so that when
new ones get added they are properly considered. */
default:
return -3;
}
}
/* Control never gets here */
}
static double tr_lrs1d_capfloor(TreeLRS1D* Meth, ModelLRS1D* ModelParam, ZCMarketData* ZCMarket, int NumberOfTimeStep, NumFunc_1 *p, double s, double r, double periodicity,double first_reset_date,double contract_maturity, double CapFloorFixedRate)
{
double lambda;
double delta_y; // delta_x1 = space step of the process x at time i ; delta_x2 same at time i+1.
double delta_t, sqrt_delta_t; // time step
double OptionPrice, OptionPrice1, OptionPrice2;
int i, i_s, h_r;
double theta;
double y_r, y_ih, y_00, r_00;
double Ti2, Ti1;
int i_Ti2, i_Ti1, n;
PnlVect* proba_from_ih;
PnlVect* OptionPriceVect1; // Matrix of prices of the option at i
PnlVect* OptionPriceVect2; // Matrix of prices of the option at i+1
proba_from_ih = pnl_vect_create(3);
OptionPriceVect1 = pnl_vect_create(1);
OptionPriceVect2 = pnl_vect_create(1);
///********* Model parameters *********///
lambda = (ModelParam->Lambda);
///**************** PAYOFF at the MATURITY of the OPTION : T(n-1)****************///
Ti2 = contract_maturity;
Ti1 = Ti2 - periodicity;
CapFloor_InitialPayoffLRS1D(Meth, ModelParam, ZCMarket, OptionPriceVect2, p, Ti1, Ti2, CapFloorFixedRate);
///**************** Backward computation of the option price ****************///
n = (int) ((contract_maturity-first_reset_date)/periodicity + 0.1);
if(n>1)
{
for(i = n-2; i>=0; i--)
{
Ti1 = first_reset_date + i * periodicity;
Ti2 = Ti1 + periodicity;
i_Ti2 = indiceTimeLRS1D(Meth, Ti2);
i_Ti1 = indiceTimeLRS1D(Meth, Ti1);
CapFloor_BackwardIterationLRS1D(Meth, ModelParam, ZCMarket, OptionPriceVect1, OptionPriceVect2, i_Ti2, i_Ti1);
CapFloor_InitialPayoffLRS1D(Meth, ModelParam, ZCMarket, OptionPriceVect1, p, Ti1, Ti2, CapFloorFixedRate);
pnl_vect_plus_vect(OptionPriceVect2, OptionPriceVect1);
}
}
///****************** Price of the option at initial time s *******************///
i_s = indiceTimeLRS1D(Meth, s); // Localisation of s on the tree
delta_t = GET(Meth->t, 1) - GET(Meth->t,0);
sqrt_delta_t = sqrt(delta_t);
r_00 = -log(BondPrice(GET(Meth->t, 1), ZCMarket))/delta_t;
y_00 = r_to_y(ModelParam, r_00);
Ti1 = first_reset_date;
i_Ti1 = indiceTimeLRS1D(Meth, Ti1);
if(i_s==0) // If s=0
{
CapFloor_BackwardIterationLRS1D(Meth, ModelParam, ZCMarket, OptionPriceVect1, OptionPriceVect2, i_Ti1, 1);
probabilities(GET(Meth->t,0), y_00, 0, lambda, sqrt_delta_t, ModelParam, ZCMarket, proba_from_ih);
OptionPrice = exp(-r_00*delta_t) * ( GET(proba_from_ih,0) * GET(OptionPriceVect1, 0) + GET(proba_from_ih,1) * GET(OptionPriceVect1,1) + GET(proba_from_ih,2) * GET(OptionPriceVect1, 2));
}
else
{ // We compute the price of the option as a linear interpolation of the prices at the nodes r(i_s,j_r) and r(i_s,j_r+1)
delta_t = GET(Meth->t, i_s+1) - GET(Meth->t,i_s);
sqrt_delta_t = sqrt(delta_t);
delta_y = lambda * sqrt_delta_t;
y_r = r_to_y(ModelParam, r);
h_r = (int) floor(i_s - (y_r-y_00)/delta_y); // y_r between y(h_r) et y(h_r+1) : y(h_r+1) < y_r <= y(h_r)
y_ih = y_00 + (i_s-h_r) * delta_y;
if(h_r < 0 || h_r > 2*i_s)
{
printf("WARNING : Instantaneous futur spot rate is out of tree\n");
exit(EXIT_FAILURE);
}
CapFloor_BackwardIterationLRS1D(Meth, ModelParam, ZCMarket, OptionPriceVect1, OptionPriceVect2, i_Ti1, i_s);
theta = (y_ih - y_r)/delta_y;
OptionPrice1 = MeanPrice(Meth, i_s, h_r, OptionPriceVect2); //Interpolation(Meth, i_s, h_r , OptionPriceVect2, phi0);
OptionPrice2 = MeanPrice(Meth, i_s, h_r+1, OptionPriceVect2); // Interpolation(Meth, i_s, h_r+1 , OptionPriceVect2, phi0);
OptionPrice = (1-theta) * OptionPrice1 + theta * OptionPrice2 ;
}
pnl_vect_free(& OptionPriceVect1);
pnl_vect_free(& OptionPriceVect2);
pnl_vect_free(&proba_from_ih);
return OptionPrice;
}
Decimate::Decimate(AVDMGenericVideoStream *in,CONFcouple *couples)
{
{
int count = 0;
char buf[80];
unsigned int *p;
_in=in;
memcpy(&_info,_in->getInfo(),sizeof(_info));
_info.encoding=1;
_uncompressed=NULL;
_info.encoding=1;
//
// Init here
debug=0;
show=0;
#ifdef USE_SSE
if(CpuCaps::hasSSE())
{
printf("Decimate:SSE enabled\n");
}
#endif
//
_param=new DECIMATE_PARAM;
if(couples)
{
GET(cycle);
GET(mode);
GET(quality);
GET(threshold);
GET(threshold2);
}
else // Default
{
_param->cycle=5;
_param->mode=0;
_param->quality=2;
_param->threshold=0;
_param->threshold2=3.0;
}
ADM_assert(_param->cycle);
vidCache=new VideoCache(_param->cycle*2+1,in);
if (_param->mode == 0 || _param->mode == 2 || _param->mode == 3)
{
num_frames_hi = _info.nb_frames;
_info.nb_frames = _info.nb_frames * (_param->cycle - 1) / _param->cycle;
_info.fps1000=_info.fps1000*(_param->cycle-1);
_info.fps1000=(uint32_t)(_info.fps1000/_param->cycle);
}
last_request = -1;
firsttime = true;
sum = (unsigned int *) ADM_alloc(MAX_BLOCKS * MAX_BLOCKS * sizeof(unsigned int));
ADM_assert(sum);
all_video_cycle = true;
if (debug)
{
char b[80];
sprintf(b, "Decimate %s by Donald Graft, Copyright 2003\n", VERSION);
OutputDebugString(b);
}
}
}
/// Backward computation of the price of a Zero Coupon Bond
static void CapFloor_BackwardIterationLRS1D(TreeLRS1D* Meth, ModelLRS1D* ModelParam, ZCMarketData* ZCMarket, PnlVect* OptionPriceVect1, PnlVect* OptionPriceVect2, int index_last, int index_first)
{
double sigma, rho, kappa, lambda;
int i, j, h;
double delta_y, delta_t, sqrt_delta_t;
double price_up, price_middle, price_down;
double y_00, y_ih, r_ih, phi_ihj, phi_next;
PnlVect* proba_from_ij;
proba_from_ij = pnl_vect_create(3);
///********* Model parameters *********///
kappa = (ModelParam->Kappa);
sigma = (ModelParam->Sigma);
rho = (ModelParam->Rho);
lambda = (ModelParam->Lambda);
delta_t = GET(Meth->t, 1) - GET(Meth->t,0);
y_00 = r_to_y(ModelParam, -log(BondPrice(GET(Meth->t, 1), ZCMarket))/delta_t);
for(i = index_last-1; i>=index_first; i--)
{
pnl_vect_resize(OptionPriceVect1, 6*i-3); // OptionPriceVect1 := Price of the bond in the tree at time t(i)
delta_t = GET(Meth->t, i+1) - GET(Meth->t,i);
sqrt_delta_t = sqrt(delta_t);
delta_y = lambda * sqrt_delta_t;
for( h=0; h<=2*i; h++) /// h : numero de la box
{
y_ih = y_00 + (i-h) * delta_y;
r_ih = y_to_r(ModelParam, y_ih);
for(j=0;j<number_phi_in_box(i, h);j++) /// Boucle sur les valeurs de phi à (i,h)
{
phi_ihj = phi_value(Meth, i, h, j);
phi_next = phi_ihj * (1-2*kappa*delta_t) + SQR(sigma) * pow(y_to_r(ModelParam, y_ih), (2*rho)) * delta_t;
price_up = Interpolation(Meth, i+1, h , OptionPriceVect2, phi_next);
price_middle = Interpolation(Meth, i+1, h+1, OptionPriceVect2, phi_next);
price_down = Interpolation(Meth, i+1, h+2, OptionPriceVect2, phi_next);
probabilities(GET(Meth->t,i), y_ih, phi_ihj, lambda, sqrt_delta_t, ModelParam, ZCMarket, proba_from_ij);
LET(OptionPriceVect1, index_tree(i,h,j)) = exp(-r_ih*delta_t) * (GET(proba_from_ij,0) * price_up + GET(proba_from_ij,1) * price_middle + GET(proba_from_ij,2) * price_down );
}
}
pnl_vect_clone(OptionPriceVect2, OptionPriceVect1); // Copy OptionPriceVect1 in OptionPriceVect2
} // END of the loop on i (time)
pnl_vect_free(&proba_from_ij);
}
static void poll_los(struct atm_dev *dev)
{
dev->signal = GET(RSOP_SIS) & SUNI_RSOP_SIS_LOSV ? ATM_PHY_SIG_LOST :
ATM_PHY_SIG_FOUND;
}
static void
fill_tracks_from_medium (Mb4Medium medium, AlbumDetails *album)
{
Mb4TrackList track_list;
GList *tracks;
unsigned int i;
char buffer[512]; /* for the GET() macro */
track_list = mb4_medium_get_tracklist (medium);
if (!track_list)
return;
album->number = mb4_track_list_size (track_list);
tracks = NULL;
for (i = 0; i < mb4_track_list_size (track_list); i++) {
Mb4Track mbt;
Mb4ArtistCredit credit;
Mb4Recording recording;
TrackDetails *track;
mbt = mb4_track_list_item (track_list, i);
if (!mbt)
continue;
track = g_new0 (TrackDetails, 1);
track->album = album;
track->number = mb4_track_get_position (mbt);
recording = mb4_track_get_recording (mbt);
if (recording != NULL) {
GET (track->title, mb4_recording_get_title, recording);
GET (track->track_id, mb4_recording_get_id, recording);
track->duration = mb4_recording_get_length (recording) / 1000;
credit = mb4_recording_get_artistcredit (recording);
} else {
GET (track->title, mb4_track_get_title, mbt);
track->duration = mb4_track_get_length (mbt) / 1000;
credit = mb4_track_get_artistcredit (mbt);
}
if (credit) {
GList *artists;
artists = get_artist_list (credit);
if (artists) {
get_artist_info (artists, &track->artist,
&track->artist_sortname,
&track->artist_id);
}
track->artists = artists;
}
if (track->artist == NULL)
track->artist = g_strdup (album->artist);
if (track->artist_sortname == NULL)
track->artist_sortname = g_strdup (album->artist_sortname);
if (track->artist_id == NULL)
track->artist_id = g_strdup (album->artist_id);
tracks = g_list_prepend (tracks, track);
}
album->tracks = g_list_reverse (tracks);
}
int do_frd (cmd_tbl_t * cmdtp, int flag, int argc, char *argv[])
{
unsigned int fslnum;
unsigned int num;
unsigned int blocking;
if (argc < 2) {
cmd_usage(cmdtp);
return 1;
}
fslnum = (unsigned int)simple_strtoul (argv[1], NULL, 16);
blocking = (unsigned int)simple_strtoul (argv[2], NULL, 16);
if (fslnum < 0 || fslnum >= XILINX_FSL_NUMBER) {
puts ("Bad number of FSL\n");
cmd_usage(cmdtp);
return 1;
}
switch (fslnum) {
#if (XILINX_FSL_NUMBER > 0)
case 0:
switch (blocking) {
case 0: NGET (num, 0);
break;
case 1: NCGET (num, 0);
break;
case 2: GET (num, 0);
break;
case 3: CGET (num, 0);
break;
default:
return 2;
}
break;
#endif
#if (XILINX_FSL_NUMBER > 1)
case 1:
switch (blocking) {
case 0: NGET (num, 1);
break;
case 1: NCGET (num, 1);
break;
case 2: GET (num, 1);
break;
case 3: CGET (num, 1);
break;
default:
return 2;
}
break;
#endif
#if (XILINX_FSL_NUMBER > 2)
case 2:
switch (blocking) {
case 0: NGET (num, 2);
break;
case 1: NCGET (num, 2);
break;
case 2: GET (num, 2);
break;
case 3: CGET (num, 2);
break;
default:
return 2;
}
break;
#endif
#if (XILINX_FSL_NUMBER > 3)
case 3:
switch (blocking) {
case 0: NGET (num, 3);
break;
case 1: NCGET (num, 3);
break;
case 2: GET (num, 3);
break;
case 3: CGET (num, 3);
break;
default:
return 2;
}
break;
#endif
#if (XILINX_FSL_NUMBER > 4)
case 4:
switch (blocking) {
case 0: NGET (num, 4);
break;
case 1: NCGET (num, 4);
break;
case 2: GET (num, 4);
break;
case 3: CGET (num, 4);
break;
default:
return 2;
}
break;
#endif
#if (XILINX_FSL_NUMBER > 5)
case 5:
switch (blocking) {
case 0: NGET (num, 5);
break;
case 1: NCGET (num, 5);
break;
case 2: GET (num, 5);
break;
case 3: CGET (num, 5);
break;
default:
return 2;
}
break;
#endif
#if (XILINX_FSL_NUMBER > 6)
case 6:
switch (blocking) {
case 0: NGET (num, 6);
break;
case 1: NCGET (num, 6);
break;
case 2: GET (num, 6);
break;
case 3: CGET (num, 6);
break;
default:
return 2;
}
break;
#endif
#if (XILINX_FSL_NUMBER > 7)
case 7:
switch (blocking) {
case 0: NGET (num, 7);
break;
case 1: NCGET (num, 7);
break;
case 2: GET (num, 7);
break;
case 3: CGET (num, 7);
break;
default:
return 2;
}
break;
#endif
default:
return 1;
}
printf ("%01x: 0x%08x - %s %s read\n", fslnum, num,
blocking < 2 ? "non blocking" : "blocking",
((blocking == 1) || (blocking == 3)) ? "control" : "data" );
return 0;
}
/*
* Virtual methods
*/
static GList *
mb4_list_albums (SjMetadata *metadata, char **url, GError **error)
{
SjMetadataMusicbrainz4Private *priv;
GList *albums = NULL;
Mb4ReleaseList releases;
Mb4Release release;
const char *discid = NULL;
char buffer[1024];
int i;
g_return_val_if_fail (SJ_IS_METADATA_MUSICBRAINZ4 (metadata), NULL);
priv = GET_PRIVATE (metadata);
if (sj_metadata_helper_check_media (priv->cdrom, error) == FALSE) {
return NULL;
}
priv->disc = discid_new ();
if (priv->disc == NULL)
return NULL;
if (discid_read (priv->disc, priv->cdrom) == 0)
return NULL;
if (url != NULL)
*url = g_strdup (discid_get_submission_url (priv->disc));
if (g_getenv("MUSICBRAINZ_FORCE_DISC_ID")) {
discid = g_getenv("MUSICBRAINZ_FORCE_DISC_ID");
} else {
discid = discid_get_id (priv->disc);
}
releases = mb4_query_lookup_discid(priv->mb, discid);
if (releases == NULL) {
return NULL;
}
if (mb4_release_list_size (releases) == 0) {
return NULL;
}
for (i = 0; i < mb4_release_list_size (releases); i++) {
AlbumDetails *album;
release = mb4_release_list_item (releases, i);
if (release) {
char *releaseid = NULL;
Mb4Release full_release;
releaseid = NULL;
GET(releaseid, mb4_release_get_id, release);
full_release = mb4_query_lookup_release (priv->mb, releaseid);
g_free (releaseid);
if (full_release) {
Mb4MediumList media;
Mb4Metadata metadata = NULL;
Mb4ReleaseGroup group;
unsigned int j;
group = mb4_release_get_releasegroup (full_release);
if (group) {
/* The release-group information we can extract from the
* lookup_release query doesn't have the url relations for the
* release-group, so run a separate query to get these urls
*/
char *releasegroupid = NULL;
char *params_names[] = { "inc" };
char *params_values[] = { "artists url-rels" };
GET (releasegroupid, mb4_releasegroup_get_id, group);
metadata = mb4_query_query (priv->mb, "release-group", releasegroupid, "",
1, params_names, params_values);
g_free (releasegroupid);
}
if (metadata && mb4_metadata_get_releasegroup (metadata))
group = mb4_metadata_get_releasegroup (metadata);
media = mb4_release_media_matching_discid (full_release, discid);
for (j = 0; j < mb4_medium_list_size (media); j++) {
Mb4Medium medium;
medium = mb4_medium_list_item (media, j);
if (medium) {
album = make_album_from_release (group, full_release, medium);
album->metadata_source = SOURCE_MUSICBRAINZ;
albums = g_list_append (albums, album);
}
}
mb4_metadata_delete (metadata);
mb4_medium_list_delete (media);
mb4_release_delete (full_release);
}
}
}
mb4_release_list_delete (releases);
return albums;
}
static int is_epilogue(void *bp)
{
/* return GET_SIZE(HDRP(bp)) == 0 && GET_ALLOC(HDRP(bp)); */
return 1 == GET(HDRP(bp));
}
void handle_motors() {
if (step_state == 1)
return;
last_active = seconds();
cli();
uint8_t state = step_state;
uint8_t cf = current_fragment;
uint8_t cs = current_sample;
sei();
// Check probe.
bool probed;
if (settings[cf].flags & Settings::PROBING && probe_pin < NUM_DIGITAL_PINS) {
if (state == 0) {
if (GET(probe_pin) ^ bool(pin_flags & 2))
stopping = active_motors;
probed = true;
}
else
probed = false;
}
else
probed = true; // If we didn't need to probe; don't block later on.
if (stopping < 0) {
// Check sensors.
for (uint8_t m = 0; m < active_motors; ++m) {
if (!(motor[m].intflags & Motor::ACTIVE))
continue;
//debug("check %d", m);
// Check sense pins.
if (motor[m].sense_pin < NUM_DIGITAL_PINS) {
if (GET(motor[m].sense_pin) ^ bool(motor[m].flags & Motor::SENSE_STATE)) {
//debug("sense %d %x", m, motor[m].flags);
motor[m].flags ^= Motor::SENSE_STATE;
motor[m].flags |= (motor[m].flags & Motor::SENSE_STATE ? Motor::SENSE1 : Motor::SENSE0);
uint8_t sense_state = motor[m].flags & Motor::SENSE_STATE ? 1 : 0;
cli();
for (int mi = 0; mi < active_motors; ++mi)
motor[mi].sense_pos[sense_state] = motor[mi].current_pos;
sei();
}
}
// Check limit switches.
if (stopping < 0) {
int8_t value = buffer[cf][m][cs];
if (value == 0)
continue;
uint8_t limit_pin = value < 0 ? motor[m].limit_min_pin : motor[m].limit_max_pin;
if (limit_pin < NUM_DIGITAL_PINS) {
bool inverted = motor[m].flags & (value < 0 ? Motor::INVERT_LIMIT_MIN : Motor::INVERT_LIMIT_MAX);
if (GET(limit_pin) ^ inverted) {
debug("hit %d pos %d state %d sample %d", m, motor[m].current_pos, state, buffer[cf][m][cs]);
stopping = m;
motor[m].flags |= Motor::LIMIT;
break;
}
}
}
}
}
if (stopping >= 0) {
// Hit endstop or probe; disable timer interrupt.
step_state = 1;
//debug("hit limit %d curpos %ld cf %d ncf %d lf %d cfp %d", m, F(motor[m].current_pos), cf, notified_current_fragment, last_fragment, cs);
// Notify host.
limit_fragment_pos = cs;
arch_set_speed(0);
return;
}
if (homers > 0) {
// Homing.
if (state == 0) {
probed = true;
for (uint8_t m = 0; m < active_motors; ++m) {
if (!(motor[m].intflags & Motor::ACTIVE))
continue;
// Get twe "wrong" limit pin for the given direction.
uint8_t limit_pin = (buffer[cf][m][cs] < 0 ? motor[m].limit_max_pin : motor[m].limit_min_pin);
bool inverted = motor[m].flags & (buffer[cf][m][cs] < 0 ? Motor::INVERT_LIMIT_MAX : Motor::INVERT_LIMIT_MIN);
if (limit_pin >= NUM_DIGITAL_PINS || GET(limit_pin) ^ inverted) {
// Limit pin still triggered; continue moving.
continue;
}
// Limit pin no longer triggered. Stop moving and possibly notify host.
motor[m].intflags &= ~Motor::ACTIVE;
if (!--homers) {
arch_set_speed(0);
return;
}
}
}
else
probed = false;
}
if (state == 0 && probed) {
if (homers > 0)
current_sample = 0; // Use only the first sample for homing.
step_state = homers > 0 || (settings[cf].flags & Settings::PROBING) ? 2 : 3;
}
}