tidy-ext 0.1.7

data/ext/tidy/streamio.h

@@ -0,0 +1,222 @@
+#ifndef __STREAMIO_H__
+#define __STREAMIO_H__
+
+/* streamio.h -- handles character stream I/O
+
+  (c) 1998-2007 (W3C) MIT, ERCIM, Keio University
+  See tidy.h for the copyright notice.
+
+  CVS Info :
+
+    $Author: arnaud02 $ 
+    $Date: 2007/07/22 09:33:26 $ 
+    $Revision: 1.21 $ 
+
+  Wrapper around Tidy input source and output sink
+  that calls appropriate interfaces, and applies 
+  necessary char encoding transformations: to/from
+  ISO-10646 and/or UTF-8.
+
+*/
+
+#include "forward.h"
+#include "buffio.h"
+#include "fileio.h"
+
+#ifdef __cplusplus
+extern "C"
+{
+#endif
+typedef enum
+{
+  FileIO,
+  BufferIO,
+  UserIO
+} IOType;
+
+/* states for ISO 2022
+
+ A document in ISO-2022 based encoding uses some ESC sequences called
+ "designator" to switch character sets. The designators defined and
+ used in ISO-2022-JP are:
+
+    "ESC" + "(" + ?     for ISO646 variants
+
+    "ESC" + "$" + ?     and
+    "ESC" + "$" + "(" + ?   for multibyte character sets
+*/
+typedef enum
+{
+  FSM_ASCII,
+  FSM_ESC,
+  FSM_ESCD,
+  FSM_ESCDP,
+  FSM_ESCP,
+  FSM_NONASCII
+} ISO2022State;
+
+/************************
+** Source
+************************/
+
+enum
+{
+    CHARBUF_SIZE=5,
+    LASTPOS_SIZE=64
+};
+
+/* non-raw input is cleaned up*/
+struct _StreamIn
+{
+    ISO2022State    state;     /* FSM for ISO2022 */
+    Bool   pushed;
+    TidyAllocator *allocator;
+    tchar* charbuf;
+    uint   bufpos;
+    uint   bufsize;
+    int    tabs;
+    int    lastcols[LASTPOS_SIZE];
+    unsigned short curlastpos; /* current last position in lastcols */ 
+    unsigned short firstlastpos; /* first valid last position in lastcols */ 
+    int    curcol;
+    int    curline;
+    int    encoding;
+    IOType iotype;
+
+    TidyInputSource source;
+
+#ifdef TIDY_WIN32_MLANG_SUPPORT
+    void* mlang;
+#endif
+
+#ifdef TIDY_STORE_ORIGINAL_TEXT
+    tmbstr otextbuf;
+    size_t otextsize;
+    uint   otextlen;
+#endif
+
+    /* Pointer back to document for error reporting */
+    TidyDocImpl* doc;
+};
+
+StreamIn* TY_(initStreamIn)( TidyDocImpl* doc, int encoding );
+void TY_(freeStreamIn)(StreamIn* in);
+
+StreamIn* TY_(FileInput)( TidyDocImpl* doc, FILE* fp, int encoding );
+StreamIn* TY_(BufferInput)( TidyDocImpl* doc, TidyBuffer* content, int encoding );
+StreamIn* TY_(UserInput)( TidyDocImpl* doc, TidyInputSource* source, int encoding );
+
+int       TY_(ReadBOMEncoding)(StreamIn *in);
+uint      TY_(ReadChar)( StreamIn* in );
+void      TY_(UngetChar)( uint c, StreamIn* in );
+Bool      TY_(IsEOF)( StreamIn* in );
+
+
+/************************
+** Sink
+************************/
+
+struct _StreamOut
+{
+    int   encoding;
+    ISO2022State   state;     /* for ISO 2022 */
+    uint  nl;
+
+#ifdef TIDY_WIN32_MLANG_SUPPORT
+    void* mlang;
+#endif
+
+    IOType iotype;
+    TidyOutputSink sink;
+};
+
+StreamOut* TY_(FileOutput)( TidyDocImpl *doc, FILE* fp, int encoding, uint newln );
+StreamOut* TY_(BufferOutput)( TidyDocImpl *doc, TidyBuffer* buf, int encoding, uint newln );
+StreamOut* TY_(UserOutput)( TidyDocImpl *doc, TidyOutputSink* sink, int encoding, uint newln );
+
+StreamOut* TY_(StdErrOutput)(void);
+/* StreamOut* StdOutOutput(void); */
+void       TY_(ReleaseStreamOut)( TidyDocImpl *doc, StreamOut* out );
+
+void TY_(WriteChar)( uint c, StreamOut* out );
+void TY_(outBOM)( StreamOut *out );
+
+ctmbstr TY_(GetEncodingNameFromTidyId)(uint id);
+ctmbstr TY_(GetEncodingOptNameFromTidyId)(uint id);
+int TY_(GetCharEncodingFromOptName)(ctmbstr charenc);
+
+/************************
+** Misc
+************************/
+
+/* character encodings
+*/
+#define RAW         0
+#define ASCII       1
+#define LATIN0      2
+#define LATIN1      3
+#define UTF8        4
+#define ISO2022     5
+#define MACROMAN    6
+#define WIN1252     7
+#define IBM858      8
+
+#if SUPPORT_UTF16_ENCODINGS
+#define UTF16LE     9
+#define UTF16BE     10
+#define UTF16       11
+#endif
+
+/* Note that Big5 and SHIFTJIS are not converted to ISO 10646 codepoints
+** (i.e., to Unicode) before being recoded into UTF-8. This may be
+** confusing: usually UTF-8 implies ISO10646 codepoints.
+*/
+#if SUPPORT_ASIAN_ENCODINGS
+#if SUPPORT_UTF16_ENCODINGS
+#define BIG5        12
+#define SHIFTJIS    13
+#else
+#define BIG5        9
+#define SHIFTJIS    10
+#endif
+#endif
+
+#ifdef TIDY_WIN32_MLANG_SUPPORT
+/* hack: windows code page numbers start at 37 */
+#define WIN32MLANG  36
+#endif
+
+
+/* char encoding used when replacing illegal SGML chars,
+** regardless of specified encoding.  Set at compile time
+** to either Windows or Mac.
+*/
+extern const int TY_(ReplacementCharEncoding);
+
+/* Function for conversion from Windows-1252 to Unicode */
+uint TY_(DecodeWin1252)(uint c);
+
+/* Function to convert from MacRoman to Unicode */
+uint TY_(DecodeMacRoman)(uint c);
+
+#ifdef __cplusplus
+}
+#endif
+
+
+/* Use numeric constants as opposed to escape chars (\r, \n)
+** to avoid conflict Mac compilers that may re-define these.
+*/
+#define CR    0xD
+#define LF    0xA
+
+#if   defined(MAC_OS_CLASSIC)
+#define DEFAULT_NL_CONFIG TidyCR
+#elif defined(_WIN32) || defined(OS2_OS)
+#define DEFAULT_NL_CONFIG TidyCRLF
+#else
+#define DEFAULT_NL_CONFIG TidyLF
+#endif
+
+
+#endif /* __STREAMIO_H__ */

data/ext/tidy/tagask.c

@@ -0,0 +1,286 @@
+/* tagask.c -- Interrogate node type
+
+  (c) 1998-2006 (W3C) MIT, ERCIM, Keio University
+  See tidy.h for the copyright notice.
+
+  CVS Info :
+
+    $Author: arnaud02 $ 
+    $Date: 2006/09/12 15:14:44 $ 
+    $Revision: 1.6 $ 
+
+*/
+
+#include "tidy-int.h"
+#include "tags.h"
+#include "tidy.h"
+
+Bool TIDY_CALL tidyNodeIsText( TidyNode tnod )
+{ return TY_(nodeIsText)( tidyNodeToImpl(tnod) );
+}
+Bool tidyNodeCMIsBlock( TidyNode tnod ); /* not exported yet */
+Bool tidyNodeCMIsBlock( TidyNode tnod )
+{ return TY_(nodeCMIsBlock)( tidyNodeToImpl(tnod) );
+}
+Bool tidyNodeCMIsInline( TidyNode tnod ); /* not exported yet */
+Bool tidyNodeCMIsInline( TidyNode tnod )
+{ return TY_(nodeCMIsInline)( tidyNodeToImpl(tnod) );
+}
+Bool tidyNodeCMIsEmpty( TidyNode tnod ); /* not exported yet */
+Bool tidyNodeCMIsEmpty( TidyNode tnod )
+{ return TY_(nodeCMIsEmpty)( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsHeader( TidyNode tnod )
+{ return TY_(nodeIsHeader)( tidyNodeToImpl(tnod) );
+}
+
+Bool TIDY_CALL tidyNodeIsHTML( TidyNode tnod )
+{ return nodeIsHTML( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsHEAD( TidyNode tnod )
+{ return nodeIsHEAD( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsTITLE( TidyNode tnod )
+{ return nodeIsTITLE( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsBASE( TidyNode tnod )
+{ return nodeIsBASE( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsMETA( TidyNode tnod )
+{ return nodeIsMETA( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsBODY( TidyNode tnod )
+{ return nodeIsBODY( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsFRAMESET( TidyNode tnod )
+{ return nodeIsFRAMESET( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsFRAME( TidyNode tnod )
+{ return nodeIsFRAME( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsIFRAME( TidyNode tnod )
+{ return nodeIsIFRAME( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsNOFRAMES( TidyNode tnod )
+{ return nodeIsNOFRAMES( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsHR( TidyNode tnod )
+{ return nodeIsHR( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsH1( TidyNode tnod )
+{ return nodeIsH1( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsH2( TidyNode tnod )
+{ return nodeIsH2( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsPRE( TidyNode tnod )
+{ return nodeIsPRE( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsLISTING( TidyNode tnod )
+{ return nodeIsLISTING( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsP( TidyNode tnod )
+{ return nodeIsP( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsUL( TidyNode tnod )
+{ return nodeIsUL( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsOL( TidyNode tnod )
+{ return nodeIsOL( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsDL( TidyNode tnod )
+{ return nodeIsDL( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsDIR( TidyNode tnod )
+{ return nodeIsDIR( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsLI( TidyNode tnod )
+{ return nodeIsLI( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsDT( TidyNode tnod )
+{ return nodeIsDT( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsDD( TidyNode tnod )
+{ return nodeIsDD( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsTABLE( TidyNode tnod )
+{ return nodeIsTABLE( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsCAPTION( TidyNode tnod )
+{ return nodeIsCAPTION( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsTD( TidyNode tnod )
+{ return nodeIsTD( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsTH( TidyNode tnod )
+{ return nodeIsTH( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsTR( TidyNode tnod )
+{ return nodeIsTR( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsCOL( TidyNode tnod )
+{ return nodeIsCOL( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsCOLGROUP( TidyNode tnod )
+{ return nodeIsCOLGROUP( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsBR( TidyNode tnod )
+{ return nodeIsBR( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsA( TidyNode tnod )
+{ return nodeIsA( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsLINK( TidyNode tnod )
+{ return nodeIsLINK( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsB( TidyNode tnod )
+{ return nodeIsB( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsI( TidyNode tnod )
+{ return nodeIsI( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsSTRONG( TidyNode tnod )
+{ return nodeIsSTRONG( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsEM( TidyNode tnod )
+{ return nodeIsEM( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsBIG( TidyNode tnod )
+{ return nodeIsBIG( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsSMALL( TidyNode tnod )
+{ return nodeIsSMALL( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsPARAM( TidyNode tnod )
+{ return nodeIsPARAM( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsOPTION( TidyNode tnod )
+{ return nodeIsOPTION( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsOPTGROUP( TidyNode tnod )
+{ return nodeIsOPTGROUP( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsIMG( TidyNode tnod )
+{ return nodeIsIMG( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsMAP( TidyNode tnod )
+{ return nodeIsMAP( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsAREA( TidyNode tnod )
+{ return nodeIsAREA( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsNOBR( TidyNode tnod )
+{ return nodeIsNOBR( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsWBR( TidyNode tnod )
+{ return nodeIsWBR( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsFONT( TidyNode tnod )
+{ return nodeIsFONT( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsLAYER( TidyNode tnod )
+{ return nodeIsLAYER( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsSPACER( TidyNode tnod )
+{ return nodeIsSPACER( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsCENTER( TidyNode tnod )
+{ return nodeIsCENTER( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsSTYLE( TidyNode tnod )
+{ return nodeIsSTYLE( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsSCRIPT( TidyNode tnod )
+{ return nodeIsSCRIPT( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsNOSCRIPT( TidyNode tnod )
+{ return nodeIsNOSCRIPT( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsFORM( TidyNode tnod )
+{ return nodeIsFORM( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsTEXTAREA( TidyNode tnod )
+{ return nodeIsTEXTAREA( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsBLOCKQUOTE( TidyNode tnod )
+{ return nodeIsBLOCKQUOTE( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsAPPLET( TidyNode tnod )
+{ return nodeIsAPPLET( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsOBJECT( TidyNode tnod )
+{ return nodeIsOBJECT( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsDIV( TidyNode tnod )
+{ return nodeIsDIV( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsSPAN( TidyNode tnod )
+{ return nodeIsSPAN( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsINPUT( TidyNode tnod )
+{ return nodeIsINPUT( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsQ( TidyNode tnod )
+{ return nodeIsQ( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsLABEL( TidyNode tnod )
+{ return nodeIsLABEL( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsH3( TidyNode tnod )
+{ return nodeIsH3( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsH4( TidyNode tnod )
+{ return nodeIsH4( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsH5( TidyNode tnod )
+{ return nodeIsH5( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsH6( TidyNode tnod )
+{ return nodeIsH6( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsADDRESS( TidyNode tnod )
+{ return nodeIsADDRESS( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsXMP( TidyNode tnod )
+{ return nodeIsXMP( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsSELECT( TidyNode tnod )
+{ return nodeIsSELECT( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsBLINK( TidyNode tnod )
+{ return nodeIsBLINK( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsMARQUEE( TidyNode tnod )
+{ return nodeIsMARQUEE( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsEMBED( TidyNode tnod )
+{ return nodeIsEMBED( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsBASEFONT( TidyNode tnod )
+{ return nodeIsBASEFONT( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsISINDEX( TidyNode tnod )
+{ return nodeIsISINDEX( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsS( TidyNode tnod )
+{ return nodeIsS( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsSTRIKE( TidyNode tnod )
+{ return nodeIsSTRIKE( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsU( TidyNode tnod )
+{ return nodeIsU( tidyNodeToImpl(tnod) );
+}
+Bool TIDY_CALL tidyNodeIsMENU( TidyNode tnod )
+{ return nodeIsMENU( tidyNodeToImpl(tnod) );
+}
+
+
+/*
+ * local variables:
+ * mode: c
+ * indent-tabs-mode: nil
+ * c-basic-offset: 4
+ * eval: (c-set-offset 'substatement-open 0)
+ * end:
+ */