unicode: Make Unicode functions private

doc/apibuild.py

@@ -32,6 +32,7 @@ ignored_files = {
   "timsort.h": "Internal header only for xpath.c 2.9.0",
   "nanoftp.h": "empty",
   "SAX.h": "empty",
+  "xmlunicode.h": "empty",
 }
 
 ignored_words = {

include/libxml/xmlunicode.h

@@ -2,365 +2,14 @@
  * Summary: Unicode character APIs
  * Description: API for the Unicode character APIs
  *
- * This file is automatically generated from the
+ * Deprecated, don't use.
- * UCS description files of the Unicode Character Database
- * http://www.unicode.org/Public/4.0-Update1/UCD-4.0.1.html
- * using the genUnicode.py Python script.
- *
- * Generation date: Tue Apr 30 17:30:38 2024
- * Sources: Blocks-4.0.1.txt UnicodeData-4.0.1.txt
- * Author: Daniel Veillard
  */
 
 #ifndef __XML_UNICODE_H__
 #define __XML_UNICODE_H__
 
-#include <libxml/xmlversion.h>
+#ifdef __GNUC__
-
+  #warning "libxml/xmlunicode.h is deprecated"
-#ifdef LIBXML_UNICODE_ENABLED
-
-#ifdef __cplusplus
-extern "C" {
 #endif
 
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsAegeanNumbers	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsAlphabeticPresentationForms	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsArabic	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsArabicPresentationFormsA	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsArabicPresentationFormsB	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsArmenian	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsArrows	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsBasicLatin	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsBengali	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsBlockElements	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsBopomofo	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsBopomofoExtended	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsBoxDrawing	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsBraillePatterns	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsBuhid	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsByzantineMusicalSymbols	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCJKCompatibility	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCJKCompatibilityForms	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCJKCompatibilityIdeographs	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCJKCompatibilityIdeographsSupplement	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCJKRadicalsSupplement	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCJKSymbolsandPunctuation	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCJKUnifiedIdeographs	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCJKUnifiedIdeographsExtensionA	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCJKUnifiedIdeographsExtensionB	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCherokee	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCombiningDiacriticalMarks	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCombiningDiacriticalMarksforSymbols	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCombiningHalfMarks	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCombiningMarksforSymbols	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsControlPictures	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCurrencySymbols	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCypriotSyllabary	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCyrillic	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCyrillicSupplement	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsDeseret	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsDevanagari	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsDingbats	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsEnclosedAlphanumerics	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsEnclosedCJKLettersandMonths	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsEthiopic	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsGeneralPunctuation	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsGeometricShapes	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsGeorgian	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsGothic	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsGreek	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsGreekExtended	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsGreekandCoptic	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsGujarati	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsGurmukhi	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsHalfwidthandFullwidthForms	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsHangulCompatibilityJamo	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsHangulJamo	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsHangulSyllables	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsHanunoo	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsHebrew	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsHighPrivateUseSurrogates	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsHighSurrogates	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsHiragana	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsIPAExtensions	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsIdeographicDescriptionCharacters	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsKanbun	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsKangxiRadicals	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsKannada	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsKatakana	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsKatakanaPhoneticExtensions	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsKhmer	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsKhmerSymbols	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsLao	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsLatin1Supplement	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsLatinExtendedA	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsLatinExtendedB	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsLatinExtendedAdditional	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsLetterlikeSymbols	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsLimbu	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsLinearBIdeograms	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsLinearBSyllabary	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsLowSurrogates	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsMalayalam	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsMathematicalAlphanumericSymbols	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsMathematicalOperators	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsMiscellaneousMathematicalSymbolsA	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsMiscellaneousMathematicalSymbolsB	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsMiscellaneousSymbols	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsMiscellaneousSymbolsandArrows	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsMiscellaneousTechnical	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsMongolian	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsMusicalSymbols	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsMyanmar	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsNumberForms	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsOgham	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsOldItalic	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsOpticalCharacterRecognition	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsOriya	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsOsmanya	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsPhoneticExtensions	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsPrivateUse	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsPrivateUseArea	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsRunic	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsShavian	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsSinhala	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsSmallFormVariants	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsSpacingModifierLetters	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsSpecials	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsSuperscriptsandSubscripts	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsSupplementalArrowsA	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsSupplementalArrowsB	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsSupplementalMathematicalOperators	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsSupplementaryPrivateUseAreaA	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsSupplementaryPrivateUseAreaB	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsSyriac	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsTagalog	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsTagbanwa	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsTags	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsTaiLe	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsTaiXuanJingSymbols	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsTamil	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsTelugu	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsThaana	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsThai	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsTibetan	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsUgaritic	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsUnifiedCanadianAboriginalSyllabics	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsVariationSelectors	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsVariationSelectorsSupplement	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsYiRadicals	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsYiSyllables	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsYijingHexagramSymbols	(int code);
-
-XMLPUBFUN int xmlUCSIsBlock	(int code, const char *block);
-
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatC	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatCc	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatCf	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatCo	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatCs	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatL	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatLl	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatLm	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatLo	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatLt	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatLu	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatM	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatMc	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatMe	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatMn	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatN	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatNd	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatNl	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatNo	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatP	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatPc	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatPd	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatPe	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatPf	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatPi	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatPo	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatPs	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatS	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatSc	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatSk	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatSm	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatSo	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatZ	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatZl	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatZp	(int code);
-XML_DEPRECATED
-XMLPUBFUN int xmlUCSIsCatZs	(int code);
-
-XMLPUBFUN int xmlUCSIsCat	(int code, const char *cat);
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif /* LIBXML_UNICODE_ENABLED */
-
 #endif /* __XML_UNICODE_H__ */

include/libxml/xmlversion.h.in

@@ -243,11 +243,9 @@
 /**
  * LIBXML_UNICODE_ENABLED:
  *
- * Whether the Unicode related interfaces are compiled in
+ * Removed in 2.14
  */
-#if @WITH_REGEXPS@
+#undef LIBXML_UNICODE_ENABLED
-#define LIBXML_UNICODE_ENABLED
-#endif
 
 /**
  * LIBXML_REGEXP_ENABLED:

include/private/Makefile.am

@@ -16,6 +16,7 @@ EXTRA_DIST = \
 	string.h \
 	threads.h \
 	tree.h \
+	unicode.h \
 	xinclude.h \
 	xpath.h \
 	xzlib.h

include/private/unicode.h

@@ -0,0 +1,44 @@
+#ifndef XML_UNICODE_H_PRIVATE__
+#define XML_UNICODE_H_PRIVATE__
+
+XML_HIDDEN int xmlUCSIsBlock	(int code, const char *block);
+XML_HIDDEN int xmlUCSIsCat	(int code, const char *cat);
+
+XML_HIDDEN int xmlUCSIsCatC	(int code);
+XML_HIDDEN int xmlUCSIsCatCc	(int code);
+XML_HIDDEN int xmlUCSIsCatCf	(int code);
+XML_HIDDEN int xmlUCSIsCatCo	(int code);
+XML_HIDDEN int xmlUCSIsCatCs	(int code);
+XML_HIDDEN int xmlUCSIsCatL	(int code);
+XML_HIDDEN int xmlUCSIsCatLl	(int code);
+XML_HIDDEN int xmlUCSIsCatLm	(int code);
+XML_HIDDEN int xmlUCSIsCatLo	(int code);
+XML_HIDDEN int xmlUCSIsCatLt	(int code);
+XML_HIDDEN int xmlUCSIsCatLu	(int code);
+XML_HIDDEN int xmlUCSIsCatM	(int code);
+XML_HIDDEN int xmlUCSIsCatMc	(int code);
+XML_HIDDEN int xmlUCSIsCatMe	(int code);
+XML_HIDDEN int xmlUCSIsCatMn	(int code);
+XML_HIDDEN int xmlUCSIsCatN	(int code);
+XML_HIDDEN int xmlUCSIsCatNd	(int code);
+XML_HIDDEN int xmlUCSIsCatNl	(int code);
+XML_HIDDEN int xmlUCSIsCatNo	(int code);
+XML_HIDDEN int xmlUCSIsCatP	(int code);
+XML_HIDDEN int xmlUCSIsCatPc	(int code);
+XML_HIDDEN int xmlUCSIsCatPd	(int code);
+XML_HIDDEN int xmlUCSIsCatPe	(int code);
+XML_HIDDEN int xmlUCSIsCatPf	(int code);
+XML_HIDDEN int xmlUCSIsCatPi	(int code);
+XML_HIDDEN int xmlUCSIsCatPo	(int code);
+XML_HIDDEN int xmlUCSIsCatPs	(int code);
+XML_HIDDEN int xmlUCSIsCatS	(int code);
+XML_HIDDEN int xmlUCSIsCatSc	(int code);
+XML_HIDDEN int xmlUCSIsCatSk	(int code);
+XML_HIDDEN int xmlUCSIsCatSm	(int code);
+XML_HIDDEN int xmlUCSIsCatSo	(int code);
+XML_HIDDEN int xmlUCSIsCatZ	(int code);
+XML_HIDDEN int xmlUCSIsCatZl	(int code);
+XML_HIDDEN int xmlUCSIsCatZp	(int code);
+XML_HIDDEN int xmlUCSIsCatZs	(int code);
+
+#endif /* XML_UNICODE_H_PRIVATE__ */

parser.c

@@ -631,11 +631,7 @@ xmlHasFeature(xmlFeature feature)
             return(0);
 #endif
         case XML_WITH_UNICODE:
-#ifdef LIBXML_UNICODE_ENABLED
-            return(1);
-#else
             return(0);
-#endif
         case XML_WITH_REGEXP:
 #ifdef LIBXML_REGEXP_ENABLED
             return(1);

python/libxml_wrap.h

@@ -13,7 +13,6 @@
 #include <libxml/HTMLtree.h>
 #include <libxml/xinclude.h>
 #include <libxml/xpointer.h>
-#include <libxml/xmlunicode.h>
 #include <libxml/xmlregexp.h>
 #include <libxml/xmlautomata.h>
 #include <libxml/xmlreader.h>

tools/genUnicode.py

@@ -191,12 +191,6 @@ ckeys = sorted(Categories.keys())
 #
 # Generate the resulting files
 #
-try:
-    header = open("include/libxml/xmlunicode.h", "w")
-except:
-    print("Failed to open include/libxml/xmlunicode.h")
-    sys.exit(1)
-
 try:
     output = open("xmlunicode.c", "w")
 except:
@@ -205,34 +199,6 @@ except:
 
 date = time.asctime(time.localtime(time.time()))
 
-header.write(
-"""/*
- * Summary: Unicode character APIs
- * Description: API for the Unicode character APIs
- *
- * This file is automatically generated from the
- * UCS description files of the Unicode Character Database
- * %s
- * using the genUnicode.py Python script.
- *
- * Generation date: %s
- * Sources: %s
- * Author: Daniel Veillard
- */
-
-#ifndef __XML_UNICODE_H__
-#define __XML_UNICODE_H__
-
-#include <libxml/xmlversion.h>
-
-#ifdef LIBXML_UNICODE_ENABLED
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-""" % (webpage, date, sources));
-
 output.write(
 """/*
  * xmlunicode.c: this module implements the Unicode character APIs
@@ -250,13 +216,14 @@ output.write(
 #define IN_LIBXML
 #include "libxml.h"
 
-#ifdef LIBXML_UNICODE_ENABLED
+#ifdef LIBXML_REGEXP_ENABLED
 
 #include <string.h>
 #include <libxml/xmlversion.h>
-#include <libxml/xmlunicode.h>
 #include <libxml/chvalid.h>
 
+#include "private/unicode.h"
+
 typedef int (xmlIntFunc)(int);	/* just to keep one's mind untwisted */
 
 typedef struct {
@@ -272,29 +239,8 @@ typedef struct {
 
 static xmlIntFunc *xmlUnicodeLookup(const xmlUnicodeNameTable *tptr, const char *tname);
 
-static const xmlUnicodeRange xmlUnicodeBlocks[] = {
 """ % (webpage, date, sources));
 
-flag = 0
-for block in bkeys:
-    name = block.replace('-', '')
-    if flag:
-        output.write(',\n')
-    else:
-        flag = 1
-    output.write('  {"%s", xmlUCSIs%s}' % (block, name))
-output.write('};\n\n')
-
-output.write('static const xmlUnicodeRange xmlUnicodeCats[] = {\n')
-flag = 0;
-for name in ckeys:
-    if flag:
-        output.write(',\n')
-    else:
-        flag = 1
-    output.write('  {"%s", xmlUCSIsCat%s}' % (name, name))
-output.write('};\n\n')
-
 #
 # For any categories with more than minTableSize ranges we generate
 # a range table suitable for xmlCharInRange
@@ -335,10 +281,7 @@ for name in ckeys:
 
 
 output.write(
-"""static const xmlUnicodeNameTable xmlUnicodeBlockTbl = {xmlUnicodeBlocks, %s};
+"""/**
-static const xmlUnicodeNameTable xmlUnicodeCatTbl = {xmlUnicodeCats, %s};
-
-/**
  * xmlUnicodeLookup:
  * @tptr: pointer to the name table
  * @name: name to be found
@@ -359,7 +302,8 @@ static xmlIntFunc
     sptr = tptr->table;
     while (low <= high) {
 	mid = (low + high) / 2;
-	if ((cmp=strcmp(tname, sptr[mid].rangename)) == 0)
+	cmp = strcmp(tname, sptr[mid].rangename);
+	if (cmp == 0)
 	    return (sptr[mid].func);
 	if (cmp < 0)
 	    high = mid - 1;
@@ -369,16 +313,15 @@ static xmlIntFunc
     return (NULL);
 }
 
-""" % (len(BlockNames), len(Categories)) )
+""")
 
 for block in bkeys:
     name = block.replace('-', '')
-    header.write("XML_DEPRECATED\nXMLPUBFUN int xmlUCSIs%s\t(int code);\n" % name)
     output.write("/**\n * xmlUCSIs%s:\n * @code: UCS code point\n" % (name))
     output.write(" *\n * Check whether the character is part of %s UCS Block\n"%
                  (block))
     output.write(" *\n * Returns 1 if true 0 otherwise\n */\n");
-    output.write("int\nxmlUCSIs%s(int code) {\n    return(" % name)
+    output.write("static int\nxmlUCSIs%s(int code) {\n    return(" % name)
     flag = 0
     for (start, end) in BlockNames[block]:
         if flag:
@@ -388,32 +331,8 @@ for block in bkeys:
         output.write("((code >= %s) && (code <= %s))" % (start, end))
     output.write(");\n}\n\n")
 
-header.write("\nXMLPUBFUN int xmlUCSIsBlock\t(int code, const char *block);\n\n")
-output.write(
-"""/**
- * xmlUCSIsBlock:
- * @code: UCS code point
- * @block: UCS block name
- *
- * Check whether the character is part of the UCS Block
- *
- * Returns 1 if true, 0 if false and -1 on unknown block
- */
-int
-xmlUCSIsBlock(int code, const char *block) {
-    xmlIntFunc *func;
-
-    func = xmlUnicodeLookup(&xmlUnicodeBlockTbl, block);
-    if (func == NULL)
-	return (-1);
-    return (func(code));
-}
-
-""")
-
 for name in ckeys:
     ranges = Categories[name]
-    header.write("XML_DEPRECATED\nXMLPUBFUN int xmlUCSIsCat%s\t(int code);\n" % name)
     output.write("/**\n * xmlUCSIsCat%s:\n * @code: UCS code point\n" % (name))
     output.write(" *\n * Check whether the character is part of %s UCS Category\n"%
                  (name))
@@ -438,9 +357,54 @@ for name in ckeys:
                          hex(begin), hex(end)))
     output.write(");\n}\n\n")
 
-header.write("\nXMLPUBFUN int xmlUCSIsCat\t(int code, const char *cat);\n")
 output.write(
-"""/**
+"""static const xmlUnicodeRange xmlUnicodeBlocks[] = {""")
+
+flag = 0
+for block in bkeys:
+    name = block.replace('-', '')
+    if flag:
+        output.write(',\n')
+    else:
+        flag = 1
+    output.write('  {"%s", xmlUCSIs%s}' % (block, name))
+output.write('};\n\n')
+
+output.write('static const xmlUnicodeRange xmlUnicodeCats[] = {\n')
+flag = 0;
+for name in ckeys:
+    if flag:
+        output.write(',\n')
+    else:
+        flag = 1
+    output.write('  {"%s", xmlUCSIsCat%s}' % (name, name))
+
+output.write(
+"""};
+
+static const xmlUnicodeNameTable xmlUnicodeBlockTbl = {xmlUnicodeBlocks, %s};
+static const xmlUnicodeNameTable xmlUnicodeCatTbl = {xmlUnicodeCats, %s};
+
+/**
+ * xmlUCSIsBlock:
+ * @code: UCS code point
+ * @block: UCS block name
+ *
+ * Check whether the character is part of the UCS Block
+ *
+ * Returns 1 if true, 0 if false and -1 on unknown block
+ */
+int
+xmlUCSIsBlock(int code, const char *block) {
+    xmlIntFunc *func;
+
+    func = xmlUnicodeLookup(&xmlUnicodeBlockTbl, block);
+    if (func == NULL)
+	return (-1);
+    return (func(code));
+}
+
+/**
  * xmlUCSIsCat:
  * @code: UCS code point
  * @cat: UCS Category name
@@ -459,18 +423,7 @@ xmlUCSIsCat(int code, const char *cat) {
     return (func(code));
 }
 
-#endif /* LIBXML_UNICODE_ENABLED */
+#endif /* LIBXML_REGEXP_ENABLED */
-""")
+""" % (len(BlockNames), len(Categories)))
 
-header.write("""
-#ifdef __cplusplus
-}
-#endif
-
-#endif /* LIBXML_UNICODE_ENABLED */
-
-#endif /* __XML_UNICODE_H__ */
-""");
-
-header.close()
 output.close()

xmlregexp.c

@@ -27,11 +27,11 @@
 #include <libxml/parserInternals.h>
 #include <libxml/xmlregexp.h>
 #include <libxml/xmlautomata.h>
-#include <libxml/xmlunicode.h>
 
 #include "private/error.h"
 #include "private/memory.h"
 #include "private/regexp.h"
+#include "private/unicode.h"
 
 #ifndef SIZE_MAX
 #define SIZE_MAX ((size_t) -1)