// Process a triple quoted string, the leading """ of which has been seen, but
// not consumed
static token_t* triple_string(lexer_t* lexer)
{
consume_chars(lexer, 3); // Leading """
while(true)
{
if(is_eof(lexer))
return literal_doesnt_terminate(lexer);
char c = look(lexer);
if((c == '\"') && (lookn(lexer, 2) == '\"') && (lookn(lexer, 3) == '\"'))
{
consume_chars(lexer, 3);
// Triple strings can end with 3 or more "s. If there are more than 3
// the extra ones are part of the string contents
while(look(lexer) == '\"')
{
append_to_token(lexer, '\"');
consume_chars(lexer, 1);
}
normalise_string(lexer);
return make_token_with_text(lexer, TK_STRING);
}
consume_chars(lexer, 1);
append_to_token(lexer, c);
}
}
// Append the given value to the current token text, UTF-8 encoded
static void append_utf8(lexer_t* lexer, int value)
{
assert(value >= 0 && value <= 0x10FFFF);
if(value <= 0x7F)
{
append_to_token(lexer, (char)(value & 0x7F));
}
else if(value <= 0x7FF)
{
append_to_token(lexer, (char)(0xC0 | (value >> 6)));
append_to_token(lexer, (char)(0x80 | (value & 0x3F)));
}
// Make a token with the specified ID and current token text
static token_t* make_token_with_text(lexer_t* lexer, token_id id)
{
token_t* t = make_token(lexer, id);
append_to_token(lexer, '\0');
token_set_string(t, stringtab(lexer->buffer));
return t;
}
void nmea_process_character(char c) {
switch (c) {
case '$': /* a new sentence is starting */
sentence_started();
/* reset and enable checksum calculation */
checksum = 0;
checksum_state = CS_CALC;
break;
case ',':
token_finished();
break;
case '*': /* checksum is following */
token_finished();
checksum_state = CS_READ;
break;
case '\r':
/* \n is following soon, we ignore this */
break;
case '\n':
token_finished();
ATOMIC(ATOMIC_FORCEON) {
sentence_finished();
}
checksum_state = CS_UNKNOWN;
break;
default:
append_to_token(c);
}
if (checksum_state == CS_CALC && c != '$') {
add_to_checksum(c);
}
}
// Process a triple quoted string, the leading """ of which has been seen, but
// not consumed
static token_t* triple_string(lexer_t* lexer)
{
consume_chars(lexer, 3); // Leading """
while(true)
{
if(is_eof(lexer))
return literal_doesnt_terminate(lexer);
char c = look(lexer);
if((c == '\"') && (lookn(lexer, 2) == '\"') && (lookn(lexer, 3) == '\"'))
{
consume_chars(lexer, 3);
normalise_string(lexer);
return make_token_with_text(lexer, TK_STRING);
}
consume_chars(lexer, 1);
append_to_token(lexer, c);
}
}
/*
Parse a field into tokens as defined by rfc822.
*/
static TOKEN
parse_field (HDR_LINE hdr)
{
static const char specials[] = "<>@.,;:\\[]\"()";
static const char specials2[] = "<>@.,;:";
static const char tspecials[] = "/?=<>@,;:\\[]\"()";
static const char tspecials2[] = "/?=<>@.,;:"; /* FIXME: really
include '.'?*/
static struct
{
const unsigned char *name;
size_t namelen;
} tspecial_header[] = {
{ "Content-Type", 12},
{ "Content-Transfer-Encoding", 25},
{ "Content-Disposition", 19},
{ NULL, 0}
};
const char *delimiters;
const char *delimiters2;
const unsigned char *line, *s, *s2;
size_t n;
int i, invalid = 0;
TOKEN t, tok, *tok_tail;
errno = 0;
if (!hdr)
return NULL;
tok = NULL;
tok_tail = &tok;
line = hdr->line;
if (!(s = strchr (line, ':')))
return NULL; /* oops */
n = s - line;
if (!n)
return NULL; /* oops: invalid name */
delimiters = specials;
delimiters2 = specials2;
for (i = 0; tspecial_header[i].name; i++)
{
if (n == tspecial_header[i].namelen
&& !memcmp (line, tspecial_header[i].name, n))
{
delimiters = tspecials;
delimiters2 = tspecials2;
break;
}
}
s++; /* Move over the colon. */
for (;;)
{
while (!*s)
{
if (!hdr->next || !hdr->next->cont)
return tok; /* Ready. */
/* Next item is a header continuation line. */
hdr = hdr->next;
s = hdr->line;
}
if (*s == '(')
{
int level = 1;
int in_quote = 0;
invalid = 0;
for (s++;; s++)
{
while (!*s)
{
if (!hdr->next || !hdr->next->cont)
goto oparen_out;
/* Next item is a header continuation line. */
hdr = hdr->next;
s = hdr->line;
}
if (in_quote)
{
if (*s == '\"')
in_quote = 0;
else if (*s == '\\' && s[1]) /* what about continuation? */
s++;
}
else if (*s == ')')
{
if (!--level)
break;
}
else if (*s == '(')
level++;
else if (*s == '\"')
in_quote = 1;
}
oparen_out:
if (!*s)
; /* Actually this is an error, but we don't care about it. */
else
s++;
}
else if (*s == '\"' || *s == '[')
{
/* We do not check for non-allowed nesting of domainliterals */
int term = *s == '\"' ? '\"' : ']';
invalid = 0;
s++;
t = NULL;
for (;;)
{
for (s2 = s; *s2; s2++)
{
if (*s2 == term)
break;
else if (*s2 == '\\' && s2[1]) /* what about continuation? */
s2++;
}
t = (t
? append_to_token (t, s, s2 - s)
: new_token (term == '\"'? tQUOTED : tDOMAINLIT, s, s2 - s));
if (!t)
goto failure;
if (*s2 || !hdr->next || !hdr->next->cont)
break;
/* Next item is a header continuation line. */
hdr = hdr->next;
s = hdr->line;
}
*tok_tail = t;
tok_tail = &t->next;
s = s2;
if (*s)
s++; /* skip the delimiter */
}
else if ((s2 = strchr (delimiters2, *s)))
{ /* Special characters which are not handled above. */
invalid = 0;
t = new_token (tSPECIAL, s, 1);
if (!t)
goto failure;
*tok_tail = t;
tok_tail = &t->next;
s++;
}
else if (*s == ' ' || *s == '\t' || *s == '\r' || *s == '\n')
{
invalid = 0;
s = skip_ws (s + 1);
}
else if (*s > 0x20 && !(*s & 128))
{ /* Atom. */
invalid = 0;
for (s2 = s + 1; *s2 > 0x20
&& !(*s2 & 128) && !strchr (delimiters, *s2); s2++)
;
t = new_token (tATOM, s, s2 - s);
if (!t)
goto failure;
*tok_tail = t;
tok_tail = &t->next;
s = s2;
}
else
{ /* Invalid character. */
if (!invalid)
{ /* For parsing we assume only one space. */
t = new_token (tSPACE, NULL, 0);
if (!t)
goto failure;
*tok_tail = t;
tok_tail = &t->next;
invalid = 1;
}
s++;
}
}
/*NOTREACHED*/
failure:
{
int save = errno;
release_token_list (tok);
errno = save;
}
return NULL;
}
/****************
* Parse a field into tokens as defined by rfc822.
*/
static TOKEN
parse_field( HDR_LINE hdr )
{
static const char specials[] = "<>@.,;:\\[]\"()";
static const char specials2[]= "<>@.,;:";
static const char tspecials[] = "/?=<>@,;:\\[]\"()";
static const char tspecials2[]= "/?=<>@.,;:";
static struct {
const char *name;
int namelen;
} tspecial_header[] = {
{ "Content-Type", 12 },
{ "Content-Transfer-Encoding", 25 },
{ NULL, 0 }
};
const char *delimiters;
const char *delimiters2;
const char *line, *s, *s2;
size_t n;
int i, invalid = 0;
TOKEN t, tok, *tok_head;
if( !hdr )
return NULL;
tok = NULL;
tok_head = &tok;
line = hdr->line;
if( !(s = strchr( line, ':' )) )
return NULL; /* oops */
n = s - line;
if( !n )
return NULL; /* oops: invalid name */
delimiters = specials;
delimiters2 = specials2;
for(i=0; tspecial_header[i].name; i++ ) {
if( n == tspecial_header[i].namelen
&& !memicmp( line, tspecial_header[i].name, n ) )
{
delimiters = tspecials;
delimiters2 = tspecials2;
break;
}
}
/* Add this point we could store the fieldname in the parsing structure.
* If we decide to do this, we should lowercase the name except for the
* first character which should be uppercased. This way we don't
* need to apply the case insensitive compare in the future
*/
s++; /* move over the colon */
for(;;) {
if( !*s ) {
if( !hdr->next || !hdr->next->cont )
break;
hdr = hdr->next;
s = hdr->line;
}
if( *s == '(' ) {
int level = 1;
int in_quote = 0;
invalid = 0;
for(s++ ; ; s++ ) {
if( !*s ) {
if( !hdr->next || !hdr->next->cont )
break;
hdr = hdr->next;
s = hdr->line;
}
if( in_quote ) {
if( *s == '\"' )
in_quote = 0;
else if( *s == '\\' && s[1] ) /* what about continuation?*/
s++;
}
else if( *s == ')' ) {
if( !--level )
break;
}
else if( *s == '(' )
level++;
else if( *s == '\"' )
in_quote = 1;
}
if( !*s )
;/* actually this is an error, but we don't care about it */
else
s++;
}
else if( *s == '\"' || *s == '[' ) {
/* We do not check for non-allowed nesting of domainliterals */
int term = *s == '\"' ? '\"' : ']';
invalid = 0;
s++;
t = NULL;
for(;;) {
for( s2 = s; *s2; s2++ ) {
if( *s2 == term )
break;
else if( *s2 == '\\' && s2[1] ) /* what about continuation?*/
s2++;
}
t = t ? append_to_token( t, s, s2-s)
: new_token( term == '\"'? tQUOTED
: tDOMAINLIT, s, s2-s);
if( *s2 || !hdr->next || !hdr->next->cont )
break;
hdr = hdr->next;
s = hdr->line;
}
*tok_head = t;
tok_head = &t->next;
s = s2;
if( *s )
s++; /* skip the delimiter */
}
else if( (s2 = strchr( delimiters2, *s )) ) {
/* special characters which are not handled above */
invalid = 0;
t = new_token( tSPECIAL, s, 1 );
*tok_head = t;
tok_head = &t->next;
s++;
}
else if( *s == ' ' || *s == '\t' || *s == '\r' || *s == '\n' ) {
invalid = 0;
s = skip_ws(s+1);
}
else if( *s > 0x20 && !(*s & 128) ) { /* atom */
invalid = 0;
for( s2 = s+1; *s2 > 0x20
&& !(*s2 & 128 )
&& !strchr( delimiters, *s2 ); s2++ )
;
t = new_token( tATOM, s, s2-s );
*tok_head = t;
tok_head = &t->next;
s = s2;
}
else { /* invalid character */
if( !invalid ) { /* for parsing we assume only one space */
t = new_token( tSPACE, NULL, 0);
*tok_head = t;
tok_head = &t->next;
invalid = 1;
}
s++;
}
}
return tok;
}
/**
* Removes longest common prefix indentation from every line in a triple
* quoted string. If the string begins with an empty line, that line is removed
* entirely.
*/
static void normalise_string(lexer_t* lexer)
{
if(lexer->buflen == 0)
return;
// Make sure we have a null terminated string.
append_to_token(lexer, '\0');
// If we aren't multiline, do nothing.
if(memchr(lexer->buffer, '\n', lexer->buflen) == NULL)
return;
// Calculate leading whitespace.
char* buf = lexer->buffer;
size_t ws = lexer->buflen;
size_t ws_this_line = 0;
bool in_leading_ws = true;
for(size_t i = 0; i < lexer->buflen; i++)
{
char c = lexer->buffer[i];
if(in_leading_ws)
{
if(c == ' ' || c == '\t')
{
ws_this_line++;
}
else if((c != '\r') && (c != '\n'))
{
if(ws_this_line < ws)
ws = ws_this_line;
in_leading_ws = false;
}
}
if(c == '\n')
{
ws_this_line = 0;
in_leading_ws = true;
}
}
// Trim leading whitespace on each line.
if(ws > 0)
{
char* line_start = lexer->buffer;
char* compacted = lexer->buffer;
size_t rem = lexer->buflen;
while(rem > 0)
{
char* line_end = strchr(line_start, '\n');
size_t line_len =
(line_end == NULL) ? rem : (size_t)(line_end - line_start + 1);
if(line_start != line_end)
{
size_t trim = (line_len < ws) ? line_len : ws;
memmove(compacted, line_start + trim, line_len - trim);
compacted += line_len - trim;
}
else {
memmove(compacted, line_start, line_len);
compacted += line_len;
}
line_start += line_len;
rem -= line_len;
}
}
// Trim a leading newline if there is one.
buf = lexer->buffer;
if((buf[0] == '\r') && (buf[1] == '\n'))
{
lexer->buflen -= 2;
memmove(&buf[0], &buf[2], lexer->buflen);
}
else if(buf[0] == '\n') {
lexer->buflen--;
memmove(&buf[0], &buf[1], lexer->buflen);
}
}