/**
Highlight operators (such as $, ~, %, as well as escaped characters.
*/
static void highlight_param( const wcstring &buffstr, std::vector<int> &colors, int pos, wcstring_list_t *error )
{
const wchar_t * const buff = buffstr.c_str();
enum {e_unquoted, e_single_quoted, e_double_quoted} mode = e_unquoted;
size_t in_pos, len = buffstr.size();
int bracket_count=0;
int normal_status = colors.at(0);
for (in_pos=0; in_pos<len; in_pos++)
{
wchar_t c = buffstr.at(in_pos);
switch( mode )
{
/*
Mode 0 means unquoted string
*/
case e_unquoted:
{
if( c == L'\\' )
{
size_t start_pos = in_pos;
in_pos++;
if( wcschr( L"~%", buff[in_pos] ) )
{
if( in_pos == 1 )
{
colors.at(start_pos) = HIGHLIGHT_ESCAPE;
colors.at(in_pos+1) = normal_status;
}
}
else if( buff[in_pos]==L',' )
{
if( bracket_count )
{
colors.at(start_pos) = HIGHLIGHT_ESCAPE;
colors.at(in_pos+1) = normal_status;
}
}
else if( wcschr( L"abefnrtv*?$(){}[]'\"<>^ \\#;|&", buff[in_pos] ) )
{
colors.at(start_pos)=HIGHLIGHT_ESCAPE;
colors.at(in_pos+1)=normal_status;
}
else if( wcschr( L"c", buff[in_pos] ) )
{
colors.at(start_pos)=HIGHLIGHT_ESCAPE;
if (in_pos+2 < colors.size())
colors.at(in_pos+2)=normal_status;
}
else if( wcschr( L"uUxX01234567", buff[in_pos] ) )
{
int i;
long long res=0;
int chars=2;
int base=16;
wchar_t max_val = ASCII_MAX;
switch( buff[in_pos] )
{
case L'u':
{
chars=4;
max_val = UCS2_MAX;
break;
}
case L'U':
{
chars=8;
max_val = WCHAR_MAX;
break;
}
case L'x':
{
break;
}
case L'X':
{
max_val = BYTE_MAX;
break;
}
default:
{
base=8;
chars=3;
in_pos--;
break;
}
}
for( i=0; i<chars; i++ )
{
int d = convert_digit( buff[++in_pos],base);
if( d < 0 )
{
in_pos--;
break;
}
res=(res*base)|d;
}
if( (res <= max_val) )
{
colors.at(start_pos) = HIGHLIGHT_ESCAPE;
colors.at(in_pos+1) = normal_status;
}
else
{
colors.at(start_pos) = HIGHLIGHT_ERROR;
colors.at(in_pos+1) = normal_status;
}
}
}
else
{
switch( buff[in_pos]){
case L'~':
case L'%':
{
if( in_pos == 0 )
{
colors.at(in_pos) = HIGHLIGHT_OPERATOR;
colors.at(in_pos+1) = normal_status;
}
break;
}
case L'$':
{
wchar_t n = buff[in_pos+1];
colors.at(in_pos) = (n==L'$'||wcsvarchr(n))? HIGHLIGHT_OPERATOR:HIGHLIGHT_ERROR;
colors.at(in_pos+1) = normal_status;
break;
}
case L'*':
case L'?':
case L'(':
case L')':
{
colors.at(in_pos) = HIGHLIGHT_OPERATOR;
colors.at(in_pos+1) = normal_status;
break;
}
case L'{':
{
colors.at(in_pos) = HIGHLIGHT_OPERATOR;
colors.at(in_pos+1) = normal_status;
bracket_count++;
break;
}
case L'}':
{
colors.at(in_pos) = HIGHLIGHT_OPERATOR;
colors.at(in_pos+1) = normal_status;
bracket_count--;
break;
}
case L',':
{
if( bracket_count )
{
colors.at(in_pos) = HIGHLIGHT_OPERATOR;
colors.at(in_pos+1) = normal_status;
}
break;
}
case L'\'':
{
colors.at(in_pos) = HIGHLIGHT_QUOTE;
mode = e_single_quoted;
break;
}
case L'\"':
{
colors.at(in_pos) = HIGHLIGHT_QUOTE;
mode = e_double_quoted;
break;
}
}
}
break;
}
/*
Mode 1 means single quoted string, i.e 'foo'
*/
case e_single_quoted:
{
if( c == L'\\' )
{
int start_pos = in_pos;
switch( buff[++in_pos] )
{
case '\\':
case L'\'':
{
colors.at(start_pos) = HIGHLIGHT_ESCAPE;
colors.at(in_pos+1) = HIGHLIGHT_QUOTE;
break;
}
case 0:
{
return;
}
}
}
if( c == L'\'' )
{
mode = e_unquoted;
colors.at(in_pos+1) = normal_status;
}
break;
}
/*
Mode 2 means double quoted string, i.e. "foo"
*/
case e_double_quoted:
{
switch( c )
{
case '"':
{
mode = e_unquoted;
colors.at(in_pos+1) = normal_status;
break;
}
case '\\':
{
int start_pos = in_pos;
switch( buff[++in_pos] )
{
case L'\0':
{
return;
}
case '\\':
case L'$':
case '"':
{
colors.at(start_pos) = HIGHLIGHT_ESCAPE;
colors.at(in_pos+1) = HIGHLIGHT_QUOTE;
break;
}
}
break;
}
case '$':
{
wchar_t n = buff[in_pos+1];
colors.at(in_pos) = (n==L'$'||wcsvarchr(n))? HIGHLIGHT_OPERATOR:HIGHLIGHT_ERROR;
colors.at(in_pos+1) = HIGHLIGHT_QUOTE;
break;
}
}
break;
}
}
}
}
wchar_t *unescape( const wchar_t * orig, int flags )
{
int mode = 0;
int in_pos, out_pos, len;
int c;
int bracket_count=0;
wchar_t prev=0;
wchar_t *in;
int unescape_special = flags & UNESCAPE_SPECIAL;
int allow_incomplete = flags & UNESCAPE_INCOMPLETE;
CHECK( orig, 0 );
len = wcslen( orig );
in = wcsdup( orig );
if( !in )
DIE_MEM();
for( in_pos=0, out_pos=0;
in_pos<len;
(prev=(out_pos>=0)?in[out_pos]:0), out_pos++, in_pos++ )
{
c = in[in_pos];
switch( mode )
{
/*
Mode 0 means unquoted string
*/
case 0:
{
if( c == L'\\' )
{
switch( in[++in_pos] )
{
/*
A null character after a backslash is an
error, return null
*/
case L'\0':
{
if( !allow_incomplete )
{
free(in);
return 0;
}
}
/*
Numeric escape sequences. No prefix means
octal escape, otherwise hexadecimal.
*/
case L'0':
case L'1':
case L'2':
case L'3':
case L'4':
case L'5':
case L'6':
case L'7':
case L'u':
case L'U':
case L'x':
case L'X':
{
int i;
long long res=0;
int chars=2;
int base=16;
int byte = 0;
wchar_t max_val = ASCII_MAX;
switch( in[in_pos] )
{
case L'u':
{
chars=4;
max_val = UCS2_MAX;
break;
}
case L'U':
{
chars=8;
max_val = WCHAR_MAX;
break;
}
case L'x':
{
break;
}
case L'X':
{
byte=1;
max_val = BYTE_MAX;
break;
}
default:
{
base=8;
chars=3;
in_pos--;
break;
}
}
for( i=0; i<chars; i++ )
{
int d = convert_digit( in[++in_pos],base);
if( d < 0 )
{
in_pos--;
break;
}
res=(res*base)|d;
}
if( (res <= max_val) )
{
in[out_pos] = (byte?ENCODE_DIRECT_BASE:0)+res;
}
else
{
free(in);
return 0;
}
break;
}
/*
\a means bell (alert)
*/
case L'a':
{
in[out_pos]=L'\a';
break;
}
/*
\b means backspace
*/
case L'b':
{
in[out_pos]=L'\b';
break;
}
/*
\cX means control sequence X
*/
case L'c':
{
in_pos++;
if( in[in_pos] >= L'a' &&
in[in_pos] <= (L'a'+32) )
{
in[out_pos]=in[in_pos]-L'a'+1;
}
else if( in[in_pos] >= L'A' &&
in[in_pos] <= (L'A'+32) )
{
in[out_pos]=in[in_pos]-L'A'+1;
}
else
{
free(in);
return 0;
}
break;
}
/*
\x1b means escape
*/
case L'e':
{
in[out_pos]=L'\x1b';
break;
}
/*
\f means form feed
*/
case L'f':
{
in[out_pos]=L'\f';
break;
}
/*
\n means newline
*/
case L'n':
{
in[out_pos]=L'\n';
break;
}
/*
\r means carriage return
*/
case L'r':
{
in[out_pos]=L'\r';
break;
}
/*
\t means tab
*/
case L't':
{
in[out_pos]=L'\t';
break;
}
/*
\v means vertical tab
*/
case L'v':
{
in[out_pos]=L'\v';
break;
}
default:
{
if( unescape_special )
in[out_pos++] = INTERNAL_SEPARATOR;
in[out_pos]=in[in_pos];
break;
}
}
}
else
{
switch( in[in_pos])
{
case L'~':
{
if( unescape_special && (in_pos == 0) )
{
in[out_pos]=HOME_DIRECTORY;
}
else
{
in[out_pos] = L'~';
}
break;
}
case L'%':
{
if( unescape_special && (in_pos == 0) )
{
in[out_pos]=PROCESS_EXPAND;
}
else
{
in[out_pos]=in[in_pos];
}
break;
}
case L'*':
{
if( unescape_special )
{
if( out_pos > 0 && in[out_pos-1]==ANY_STRING )
{
out_pos--;
in[out_pos] = ANY_STRING_RECURSIVE;
}
else
in[out_pos]=ANY_STRING;
}
else
{
in[out_pos]=in[in_pos];
}
break;
}
case L'?':
{
if( unescape_special )
{
in[out_pos]=ANY_CHAR;
}
else
{
in[out_pos]=in[in_pos];
}
break;
}
case L'$':
{
if( unescape_special )
{
in[out_pos]=VARIABLE_EXPAND;
}
else
{
in[out_pos]=in[in_pos];
}
break;
}
case L'{':
{
if( unescape_special )
{
bracket_count++;
in[out_pos]=BRACKET_BEGIN;
}
else
{
in[out_pos]=in[in_pos];
}
break;
}
case L'}':
{
if( unescape_special )
{
bracket_count--;
in[out_pos]=BRACKET_END;
}
else
{
in[out_pos]=in[in_pos];
}
break;
}
case L',':
{
if( unescape_special && bracket_count && prev!=BRACKET_SEP)
{
in[out_pos]=BRACKET_SEP;
}
else
{
in[out_pos]=in[in_pos];
}
break;
}
case L'\'':
{
mode = 1;
if( unescape_special )
in[out_pos] = INTERNAL_SEPARATOR;
else
out_pos--;
break;
}
case L'\"':
{
mode = 2;
if( unescape_special )
in[out_pos] = INTERNAL_SEPARATOR;
else
out_pos--;
break;
}
default:
{
in[out_pos] = in[in_pos];
break;
}
}
}
break;
}
/*
Mode 1 means single quoted string, i.e 'foo'
*/
case 1:
{
if( c == L'\\' )
{
switch( in[++in_pos] )
{
case '\\':
case L'\'':
case L'\n':
{
in[out_pos]=in[in_pos];
break;
}
case 0:
{
if( !allow_incomplete )
{
free(in);
return 0;
}
else
{
//We may ever escape a NULL character, but still appending a \ in case I am wrong.
in[out_pos] = L'\\';
}
}
break;
default:
{
in[out_pos++] = L'\\';
in[out_pos]= in[in_pos];
}
}
}
if( c == L'\'' )
{
if( unescape_special )
in[out_pos] = INTERNAL_SEPARATOR;
else
out_pos--;
mode = 0;
}
else
{
in[out_pos] = in[in_pos];
}
break;
}
/*
Mode 2 means double quoted string, i.e. "foo"
*/
case 2:
{
switch( c )
{
case '"':
{
mode = 0;
if( unescape_special )
in[out_pos] = INTERNAL_SEPARATOR;
else
out_pos--;
break;
}
case '\\':
{
switch( in[++in_pos] )
{
case L'\0':
{
if( !allow_incomplete )
{
free(in);
return 0;
}
else
{
//We probably don't need it since NULL character is always appended before ending this function.
in[out_pos]=in[in_pos];
}
}
break;
case '\\':
case L'$':
case '"':
case '\n':
{
in[out_pos]=in[in_pos];
break;
}
default:
{
in[out_pos++] = L'\\';
in[out_pos] = in[in_pos];
break;
}
}
break;
}
case '$':
{
if( unescape_special )
{
in[out_pos]=VARIABLE_EXPAND_SINGLE;
}
else
{
in[out_pos]=in[in_pos];
}
break;
}
default:
{
in[out_pos] = in[in_pos];
break;
}
}
break;
}
}
}
if( !allow_incomplete && mode )
{
free( in );
return 0;
}
in[out_pos]=L'\0';
return in;
}
