bool CPLJSonStreamingParser::StartNewToken(const char*& pStr, size_t& nLength)
{
char ch = *pStr;
if( ch == '{' )
{
if( m_aState.size() == m_nMaxDepth )
{
return EmitException("Too many nested objects and/or arrays");
}
StartObject();
m_aeObjectState.push_back(WAITING_KEY);
m_aState.push_back(OBJECT);
AdvanceChar(pStr, nLength);
}
else if( ch == '"' )
{
m_aState.push_back(STRING);
AdvanceChar(pStr, nLength);
}
else if( ch == '[' )
{
if( m_aState.size() == m_nMaxDepth )
{
return EmitException("Too many nested objects and/or arrays");
}
StartArray();
m_abFirstElement.push_back(true);
m_aState.push_back(ARRAY);
AdvanceChar(pStr, nLength);
}
else if( ch == '-' || ch == '.' || isdigit(ch) )
{
m_aState.push_back(NUMBER);
}
else if( ch == 't' )
{
m_aState.push_back(STATE_TRUE);
}
else if( ch == 'f' )
{
m_aState.push_back(STATE_FALSE);
}
else if( ch == 'n' )
{
m_aState.push_back(STATE_NULL);
}
else
{
assert( false );
}
return true;
}
Bool ParseRepeatAttr( TidyDocImpl* doc, const TidyOptionImpl* option )
{
Bool status = yes;
tmbchar buf[64] = {0};
uint i = 0;
TidyConfigImpl* cfg = &doc->config;
tchar c = SkipWhite( cfg );
while (i < sizeof(buf)-1 && c != EndOfStream && !IsWhite(c))
{
buf[i++] = (tmbchar) c;
c = AdvanceChar( cfg );
}
buf[i] = '\0';
if ( tmbstrcasecmp(buf, "keep-first") == 0 )
cfg->value[ TidyDuplicateAttrs ].v = TidyKeepFirst;
else if ( tmbstrcasecmp(buf, "keep-last") == 0 )
cfg->value[ TidyDuplicateAttrs ].v = TidyKeepLast;
else
{
ReportBadArgument( doc, option->name );
status = no;
}
return status;
}
void CPLJSonStreamingParser::SkipSpace(const char*& pStr, size_t& nLength)
{
while( nLength > 0 && isspace(*pStr) )
{
AdvanceChar(pStr, nLength);
}
}
Bool ParseCharEnc( TidyDocImpl* doc, const TidyOptionImpl* option )
{
tmbchar buf[64] = {0};
uint i = 0;
int enc = ASCII;
Bool validEncoding = yes;
tchar c = SkipWhite( &doc->config );
while ( i < sizeof(buf)-2 && c != EndOfStream && !IsWhite(c) )
{
buf[i++] = (tmbchar) ToLower( c );
c = AdvanceChar( &doc->config );
}
buf[i] = 0;
enc = CharEncodingId( buf );
#ifdef TIDY_WIN32_MLANG_SUPPORT
/* limit support to --input-encoding */
if (option->id != TidyInCharEncoding && enc > WIN32MLANG)
enc = -1;
#endif
if ( enc < 0 )
{
validEncoding = no;
ReportBadArgument( doc, option->name );
}
else
SetOptionInt( doc, option->id, enc );
if ( validEncoding && option->id == TidyCharEncoding )
AdjustCharEncoding( doc, enc );
return validEncoding;
}
/* #508936 - CSS class naming for -clean option */
Bool ParseCSS1Selector( TidyDocImpl* doc, const TidyOptionImpl* option )
{
char buf[256] = {0};
uint i = 0;
uint c = SkipWhite( &doc->config );
while ( i < sizeof(buf)-2 && c != EndOfStream && !IsWhite(c) )
{
buf[i++] = (tmbchar) c;
c = AdvanceChar( &doc->config );
}
buf[i] = '\0';
if ( i == 0 || !IsCSS1Selector(buf) ) {
ReportBadArgument( doc, option->name );
return no;
}
buf[i++] = '-'; /* Make sure any escaped Unicode is terminated */
buf[i] = 0; /* so valid class names are generated after */
/* Tidy appends last digits. */
SetOptionValue( doc, option->id, buf );
return yes;
}
/* cr, lf or crlf */
Bool ParseNewline( TidyDocImpl* doc, const TidyOptionImpl* entry )
{
int nl = -1;
tmbchar work[ 16 ] = {0};
tmbstr cp = work, end = work + sizeof(work);
TidyConfigImpl* cfg = &doc->config;
tchar c = SkipWhite( cfg );
while ( c!=EndOfStream && cp < end && !IsWhite(c) && c != '\r' && c != '\n' )
{
*cp++ = (tmbchar) c;
c = AdvanceChar( cfg );
}
*cp = 0;
if ( tmbstrcasecmp(work, "lf") == 0 )
nl = TidyLF;
else if ( tmbstrcasecmp(work, "crlf") == 0 )
nl = TidyCRLF;
else if ( tmbstrcasecmp(work, "cr") == 0 )
nl = TidyCR;
if ( nl < TidyLF || nl > TidyCR )
ReportBadArgument( doc, entry->name );
else
SetOptionInt( doc, entry->id, nl );
return ( nl >= TidyLF && nl <= TidyCR );
}
Bool ParseString( TidyDocImpl* doc, const TidyOptionImpl* option )
{
TidyConfigImpl* cfg = &doc->config;
tmbchar buf[8192];
uint i = 0;
tchar delim = 0;
Bool waswhite = yes;
tchar c = SkipWhite( cfg );
if ( c == '"' || c == '\'' )
{
delim = c;
c = AdvanceChar( cfg );
}
while ( i < sizeof(buf)-2 && c != EndOfStream && c != '\r' && c != '\n' )
{
if ( delim && c == delim )
break;
if ( IsWhite(c) )
{
if ( waswhite )
{
c = AdvanceChar( cfg );
continue;
}
c = ' ';
}
else
waswhite = no;
buf[i++] = (tmbchar) c;
c = AdvanceChar( cfg );
}
buf[i] = '\0';
SetOptionValue( doc, option->id, buf );
return yes;
}
/*
doctype: omit | auto | strict | loose | <fpi>
where the fpi is a string similar to
"-//ACME//DTD HTML 3.14159//EN"
*/
Bool ParseDocType( TidyDocImpl* doc, const TidyOptionImpl* option )
{
tmbchar buf[ 32 ] = {0};
uint i = 0;
Bool status = yes;
TidyDoctypeModes dtmode = TidyDoctypeAuto;
TidyConfigImpl* cfg = &doc->config;
tchar c = SkipWhite( cfg );
/* "-//ACME//DTD HTML 3.14159//EN" or similar */
if ( c == '"' || c == '\'' )
{
status = ParseString(doc, option);
if (status)
SetOptionInt( doc, TidyDoctypeMode, TidyDoctypeUser );
return status;
}
/* read first word */
while ( i < sizeof(buf)-1 && c != EndOfStream && !IsWhite(c) )
{
buf[i++] = (tmbchar) c;
c = AdvanceChar( cfg );
}
buf[i] = '\0';
if ( tmbstrcasecmp(buf, "auto") == 0 )
dtmode = TidyDoctypeAuto;
else if ( tmbstrcasecmp(buf, "omit") == 0 )
dtmode = TidyDoctypeOmit;
else if ( tmbstrcasecmp(buf, "strict") == 0 )
dtmode = TidyDoctypeStrict;
else if ( tmbstrcasecmp(buf, "loose") == 0 ||
tmbstrcasecmp(buf, "transitional") == 0 )
dtmode = TidyDoctypeLoose;
else
{
ReportBadArgument( doc, option->name );
status = no;
}
if ( status )
SetOptionInt( doc, TidyDoctypeMode, dtmode );
return status;
}
/* a string excluding whitespace */
Bool ParseName( TidyDocImpl* doc, const TidyOptionImpl* option )
{
tmbchar buf[ 1024 ] = {0};
uint i = 0;
uint c = SkipWhite( &doc->config );
while ( i < sizeof(buf)-2 && c != EndOfStream && !IsWhite(c) )
{
buf[i++] = (tmbchar) c;
c = AdvanceChar( &doc->config );
}
buf[i] = 0;
if ( i == 0 )
ReportBadArgument( doc, option->name );
else
SetOptionValue( doc, option->id, buf );
return ( i > 0 );
}
/* unsigned integers */
Bool ParseInt( TidyDocImpl* doc, const TidyOptionImpl* entry )
{
ulong number = 0;
Bool digits = no;
TidyConfigImpl* cfg = &doc->config;
tchar c = SkipWhite( cfg );
while ( IsDigit(c) )
{
number = c - '0' + (10 * number);
digits = yes;
c = AdvanceChar( cfg );
}
if ( !digits )
ReportBadArgument( doc, entry->name );
else
SetOptionInt( doc, entry->id, number );
return digits;
}
/* open the file and parse its contents
*/
int ParseConfigFileEnc( TidyDocImpl* doc, ctmbstr file, ctmbstr charenc )
{
uint opterrs = doc->optionErrors;
tmbstr fname = (tmbstr) ExpandTilde( file );
TidyConfigImpl* cfg = &doc->config;
FILE* fin = fopen( fname, "r" );
int enc = CharEncodingId( charenc );
if ( fin == NULL || enc < 0 )
{
FileError( doc, fname, TidyConfig );
return -1;
}
else
{
tchar c;
cfg->cfgIn = FileInput( doc, fin, enc );
c = FirstChar( cfg );
for ( c = SkipWhite(cfg); c != EndOfStream; c = NextProperty(cfg) )
{
uint ix = 0;
tmbchar name[ TIDY_MAX_NAME ] = {0};
/* // or # start a comment */
if ( c == '/' || c == '#' )
continue;
while ( ix < sizeof(name)-1 && c != '\n' && c != EndOfStream && c != ':' )
{
name[ ix++ ] = (tmbchar) c; /* Option names all ASCII */
c = AdvanceChar( cfg );
}
if ( c == ':' )
{
const TidyOptionImpl* option = lookupOption( name );
c = AdvanceChar( cfg );
if ( option )
option->parser( doc, option );
else
{
if (NULL != doc->pOptCallback)
{
TidyConfigImpl* cfg = &doc->config;
tmbchar buf[8192];
uint i = 0;
tchar delim = 0;
Bool waswhite = yes;
tchar c = SkipWhite( cfg );
if ( c == '"' || c == '\'' )
{
delim = c;
c = AdvanceChar( cfg );
}
while ( i < sizeof(buf)-2 && c != EndOfStream && c != '\r' && c != '\n' )
{
if ( delim && c == delim )
break;
if ( IsWhite(c) )
{
if ( waswhite )
{
c = AdvanceChar( cfg );
continue;
}
c = ' ';
}
else
waswhite = no;
buf[i++] = (tmbchar) c;
c = AdvanceChar( cfg );
}
buf[i] = '\0';
if (no == (*doc->pOptCallback)( name, buf ))
ReportUnknownOption( doc, name );
}
else
ReportUnknownOption( doc, name );
}
}
}
fclose( fin );
MemFree( (void *)cfg->cfgIn->source.sourceData ); /* fix for bug #810259 */
freeStreamIn( cfg->cfgIn );
cfg->cfgIn = NULL;
}
if ( fname != (tmbstr) file )
MemFree( fname );
AdjustConfig( doc );
/* any new config errors? If so, return warning status. */
return (doc->optionErrors > opterrs ? 1 : 0);
}
/* a space or comma separated list of tag names */
Bool ParseTagNames( TidyDocImpl* doc, const TidyOptionImpl* option )
{
TidyConfigImpl* cfg = &doc->config;
tmbchar buf[1024];
uint i = 0, nTags = 0;
uint c = SkipWhite( cfg );
UserTagType ttyp = tagtype_null;
switch ( option->id )
{
case TidyInlineTags: ttyp = tagtype_inline; break;
case TidyBlockTags: ttyp = tagtype_block; break;
case TidyEmptyTags: ttyp = tagtype_empty; break;
case TidyPreTags: ttyp = tagtype_pre; break;
default:
ReportUnknownOption( doc, option->name );
return no;
}
SetOptionValue( doc, option->id, NULL );
FreeDeclaredTags( doc, ttyp );
cfg->defined_tags |= ttyp;
do
{
if (c == ' ' || c == '\t' || c == ',')
{
c = AdvanceChar( cfg );
continue;
}
if ( c == '\r' || c == '\n' )
{
uint c2 = AdvanceChar( cfg );
if ( c == '\r' && c2 == '\n' )
c = AdvanceChar( cfg );
else
c = c2;
if ( !IsWhite(c) )
{
buf[i] = 0;
UngetChar( c, cfg->cfgIn );
UngetChar( '\n', cfg->cfgIn );
break;
}
}
/*
if ( c == '\n' )
{
c = AdvanceChar( cfg );
if ( !IsWhite(c) )
{
buf[i] = 0;
UngetChar( c, cfg->cfgIn );
UngetChar( '\n', cfg->cfgIn );
break;
}
}
*/
while ( i < sizeof(buf)-2 && c != EndOfStream && !IsWhite(c) && c != ',' )
{
buf[i++] = (tmbchar) c;
c = AdvanceChar( cfg );
}
buf[i] = '\0';
if (i == 0) /* Skip empty tag definition. Possible when */
continue; /* there is a trailing space on the line. */
/* add tag to dictionary */
DeclareUserTag( doc, option->id, ttyp, buf );
i = 0;
++nTags;
}
while ( c != EndOfStream );
if ( i > 0 )
DeclareUserTag( doc, option->id, ttyp, buf );
return ( nTags > 0 );
}