diff --git a/ChangeLog b/ChangeLog index df34aa9d..01e06eb4 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,3 +1,11 @@ +Thu May 10 16:14:36 CEST 2001 Daniel Veillard + + * Makefile.am xmlversion.h.in configure.in include/Makefile.am: + integrating catalogs + * catalog.[ch] testCatalog.c: adding a small catalo API + (only SGML catalog support). + * parser.c: restaured xmlKeepBlanksDefault(0) API + Wed May 9 12:50:15 CEST 2001 Daniel Veillard * tree.c: zb@bisp.com reported an error in xmlNodeGetLang() diff --git a/Makefile.am b/Makefile.am index 7dc70174..67dceba3 100644 --- a/Makefile.am +++ b/Makefile.am @@ -4,7 +4,7 @@ SUBDIRS = . include doc example INCLUDES = -I@srcdir@/include -I./include @Z_CFLAGS@ @CORBA_CFLAGS@ -noinst_PROGRAMS=testSAX testHTML testXPath testURI testDocbook +noinst_PROGRAMS=testSAX testHTML testXPath testURI testDocbook testCatalog bin_PROGRAMS = xmllint @@ -20,13 +20,14 @@ libxml2_la_SOURCES = SAX.c entities.c encoding.c error.c parserInternals.c \ parser.c tree.c hash.c list.c xmlIO.c xmlmemory.c uri.c \ valid.c xlink.c HTMLparser.c HTMLtree.c debugXML.c xpath.c \ xpointer.c xinclude.c nanohttp.c nanoftp.c DOCBparser.c \ - strio.c trio.c + catalog.c strio.c trio.c else libxml2_la_SOURCES = SAX.c entities.c encoding.c error.c parserInternals.c \ parser.c tree.c hash.c list.c xmlIO.c xmlmemory.c uri.c \ valid.c xlink.c HTMLparser.c HTMLtree.c debugXML.c xpath.c \ - xpointer.c xinclude.c nanohttp.c nanoftp.c DOCBparser.c + xpointer.c xinclude.c nanohttp.c nanoftp.c DOCBparser.c \ + catalog.c endif @@ -58,6 +59,11 @@ testDocbook_LDFLAGS = testDocbook_DEPENDENCIES = $(DEPS) testDocbook_LDADD= $(LDADDS) +testCatalog_SOURCES=testCatalog.c +testCatalog_LDFLAGS = +testCatalog_DEPENDENCIES = $(DEPS) +testCatalog_LDADD= $(LDADDS) + testXPath_SOURCES=testXPath.c testXPath_LDFLAGS = testXPath_DEPENDENCIES = $(DEPS) diff --git a/catalog.c b/catalog.c new file mode 100644 index 00000000..437474a9 --- /dev/null +++ b/catalog.c @@ -0,0 +1,531 @@ +/** + * catalog.c: set of generic Catalog related routines + * + * Reference: SGML Open Technical Resolution TR9401:1997. + * http://www.jclark.com/sp/catalog.htm + * + * See Copyright for the status of this software. + * + * Daniel.Veillard@imag.fr + */ + +#include "libxml.h" + +#ifdef LIBXML_CATALOG_ENABLED +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#include +#include +#include +#include +#include +#include +#include + +/************************************************************************ + * * + * Types, all private * + * * + ************************************************************************/ + +typedef enum { + XML_CATA_NONE = 0, + XML_CATA_SYSTEM, + XML_CATA_PUBLIC, + XML_CATA_ENTITY, + XML_CATA_PENTITY, + XML_CATA_DOCTYPE, + XML_CATA_LINKTYPE, + XML_CATA_NOTATION, + XML_CATA_DELEGATE, + XML_CATA_BASE, + XML_CATA_CATALOG, + XML_CATA_DOCUMENT, + XML_CATA_SGMLDECL +} xmlCatalogEntryType; + +typedef struct _xmlCatalogEntry xmlCatalogEntry; +typedef xmlCatalogEntry *xmlCatalogEntryPtr; +struct _xmlCatalogEntry { + xmlCatalogEntryType type; + xmlChar *name; + xmlChar *value; +}; + +static xmlHashTablePtr xmlDefaultCatalog; + +/************************************************************************ + * * + * alloc or dealloc * + * * + ************************************************************************/ + +static xmlCatalogEntryPtr +xmlNewCatalogEntry(int type, xmlChar *name, xmlChar *value) { + xmlCatalogEntryPtr ret; + + ret = (xmlCatalogEntryPtr) xmlMalloc(sizeof(xmlCatalogEntry)); + if (ret == NULL) { + xmlGenericError(xmlGenericErrorContext, + "malloc of %d byte failed\n", sizeof(xmlCatalogEntry)); + return(NULL); + } + ret->type = type; + ret->name = name; + ret->value = value; + return(ret); +} + +static void +xmlFreeCatalogEntry(xmlCatalogEntryPtr ret) { + if (ret == NULL) + return; + if (ret->name != NULL) + xmlFree(ret->name); + if (ret->value != NULL) + xmlFree(ret->value); + xmlFree(ret); +} + +/************************************************************************ + * * + * The parser * + * * + ************************************************************************/ + + +#define RAW *cur +#define NEXT cur++; +#define SKIP(x) cur += x; + +#define SKIP_BLANKS while (IS_BLANK(*cur)) NEXT; + +static const xmlChar * +xmlParseCatalogComment(const xmlChar *cur) { + if ((cur[0] != '-') || (cur[1] != '-')) + return(cur); + SKIP(2); + while ((cur[0] != 0) && ((cur[0] != '-') || ((cur[1] != '-')))) + NEXT; + if (cur[0] == 0) { + return(NULL); + } + return(cur); +} + +static const xmlChar * +xmlParseCatalogPubid(const xmlChar *cur, xmlChar **id) { + xmlChar *buf = NULL; + int len = 0; + int size = 50; + xmlChar stop; + int count = 0; + + *id = NULL; + + if (RAW == '"') { + NEXT; + stop = '"'; + } else if (RAW == '\'') { + NEXT; + stop = '\''; + } else { + stop = ' '; + } + buf = (xmlChar *) xmlMalloc(size * sizeof(xmlChar)); + if (buf == NULL) { + xmlGenericError(xmlGenericErrorContext, + "malloc of %d byte failed\n", size); + return(NULL); + } + while (xmlIsPubidChar(*cur)) { + if ((*cur == stop) && (stop != ' ')) + break; + if ((stop == ' ') && (IS_BLANK(*cur))) + break; + if (len + 1 >= size) { + size *= 2; + buf = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); + if (buf == NULL) { + xmlGenericError(xmlGenericErrorContext, + "realloc of %d byte failed\n", size); + return(NULL); + } + } + buf[len++] = *cur; + count++; + NEXT; + } + buf[len] = 0; + if (stop == ' ') { + if (!IS_BLANK(*cur)) { + xmlFree(buf); + return(NULL); + } + } else { + if (*cur != stop) { + xmlFree(buf); + return(NULL); + } + NEXT; + } + *id = buf; + return(cur); +} + +static const xmlChar * +xmlParseCatalogName(const xmlChar *cur, xmlChar **name) { + xmlChar buf[XML_MAX_NAMELEN + 5]; + int len = 0; + int c; + + *name = NULL; + + /* + * Handler for more complex cases + */ + c = *cur; + if ((!IS_LETTER(c) && (c != '_') && (c != ':'))) { + return(NULL); + } + + while (((IS_LETTER(c)) || (IS_DIGIT(c)) || + (c == '.') || (c == '-') || + (c == '_') || (c == ':'))) { + buf[len++] = c; + cur++; + c = *cur; + if (len >= XML_MAX_NAMELEN) + return(NULL); + } + *name = xmlStrndup(buf, len); + return(cur); +} + +static int +xmlParseCatalog(const xmlChar *value, const char *file) { + const xmlChar *cur = value; + xmlChar *base = NULL; + + if ((cur == NULL) || (file == NULL)) + return(-1); + base = xmlStrdup((const xmlChar *) file); + + while ((cur != NULL) && (cur[0] != '0')) { + SKIP_BLANKS; + if ((cur[0] == '-') && (cur[1] == '-')) { + cur = xmlParseCatalogComment(cur); + if (cur == NULL) { + /* error */ + break; + } + } else { + xmlChar *sysid = NULL; + xmlChar *name = NULL; + xmlCatalogEntryType type = XML_CATA_NONE; + + cur = xmlParseCatalogName(cur, &name); + if (name == NULL) { + /* error */ + break; + } + if (!IS_BLANK(*cur)) { + /* error */ + break; + } + SKIP_BLANKS; + if (xmlStrEqual(name, (const xmlChar *) "SYSTEM")) + type = XML_CATA_SYSTEM; + else if (xmlStrEqual(name, (const xmlChar *) "PUBLIC")) + type = XML_CATA_PUBLIC; + else if (xmlStrEqual(name, (const xmlChar *) "DELEGATE")) + type = XML_CATA_DELEGATE; + else if (xmlStrEqual(name, (const xmlChar *) "ENTITY")) + type = XML_CATA_ENTITY; + else if (xmlStrEqual(name, (const xmlChar *) "DOCTYPE")) + type = XML_CATA_DOCTYPE; + else if (xmlStrEqual(name, (const xmlChar *) "LINKTYPE")) + type = XML_CATA_LINKTYPE; + else if (xmlStrEqual(name, (const xmlChar *) "NOTATION")) + type = XML_CATA_NOTATION; + else if (xmlStrEqual(name, (const xmlChar *) "SGMLDECL")) + type = XML_CATA_SGMLDECL; + else if (xmlStrEqual(name, (const xmlChar *) "DOCUMENT")) + type = XML_CATA_DOCUMENT; + else if (xmlStrEqual(name, (const xmlChar *) "CATALOG")) + type = XML_CATA_CATALOG; + else if (xmlStrEqual(name, (const xmlChar *) "BASE")) + type = XML_CATA_BASE; + else if (xmlStrEqual(name, (const xmlChar *) "DELEGATE")) + type = XML_CATA_DELEGATE; + else if (xmlStrEqual(name, (const xmlChar *) "OVERRIDE")) { + xmlFree(name); + cur = xmlParseCatalogName(cur, &name); + if (name == NULL) { + /* error */ + break; + } + continue; + } + xmlFree(name); + + switch(type) { + case XML_CATA_ENTITY: + if (*cur == '%') + type = XML_CATA_PENTITY; + case XML_CATA_PENTITY: + case XML_CATA_DOCTYPE: + case XML_CATA_LINKTYPE: + case XML_CATA_NOTATION: + cur = xmlParseCatalogName(cur, &name); + if (cur == NULL) { + /* error */ + break; + } + if (!IS_BLANK(*cur)) { + /* error */ + break; + } + SKIP_BLANKS; + cur = xmlParseCatalogPubid(cur, &sysid); + if (cur == NULL) { + /* error */ + break; + } + break; + case XML_CATA_PUBLIC: + case XML_CATA_SYSTEM: + case XML_CATA_DELEGATE: + cur = xmlParseCatalogPubid(cur, &name); + if (cur == NULL) { + /* error */ + break; + } + if (!IS_BLANK(*cur)) { + /* error */ + break; + } + SKIP_BLANKS; + cur = xmlParseCatalogPubid(cur, &sysid); + if (cur == NULL) { + /* error */ + break; + } + break; + case XML_CATA_BASE: + case XML_CATA_CATALOG: + case XML_CATA_DOCUMENT: + case XML_CATA_SGMLDECL: + cur = xmlParseCatalogPubid(cur, &sysid); + if (cur == NULL) { + /* error */ + break; + } + break; + default: + break; + } + if (cur == NULL) { + if (name != NULL) + xmlFree(name); + if (sysid != NULL) + xmlFree(sysid); + break; + } else if (type == XML_CATA_BASE) { + if (base != NULL) + xmlFree(base); + base = sysid; + } else if ((type == XML_CATA_PUBLIC) || + (type == XML_CATA_SYSTEM)) { + xmlChar *filename; + + filename = xmlBuildURI(sysid, base); + if (filename != NULL) { + + xmlHashAddEntry(xmlDefaultCatalog, name, + xmlNewCatalogEntry(type, name, filename)); + } + if (sysid != NULL) + xmlFree(sysid); + } else { + /* + * drop anything else we won't handle it + */ + if (name != NULL) + xmlFree(name); + if (sysid != NULL) + xmlFree(sysid); + } + } + } + if (base != NULL) + xmlFree(base); + if (cur == NULL) + return(-1); + return(0); +} + +/************************************************************************ + * * + * Public interfaces * + * * + ************************************************************************/ + +/* + * xmlLoadCatalog: + * @filename: a file path + * + * Load the catalog and makes its definition effective for the default + * external entity loader. + * + * Returns 0 in case of success -1 in case of error + */ +int +xmlLoadCatalog(const char *filename) { + int fd, len, ret; + struct stat info; + xmlChar *content; + + if (filename == NULL) + return(-1); + if (xmlDefaultCatalog == NULL) + xmlDefaultCatalog = xmlHashCreate(20); + if (xmlDefaultCatalog == NULL) + return(-1); + + if (stat(filename, &info) < 0) + return(-1); + + if ((fd = open(filename, O_RDONLY)) < 0) + return(-1); + + content = xmlMalloc(info.st_size + 10); + if (content == NULL) { + xmlGenericError(xmlGenericErrorContext, + "realloc of %d byte failed\n", info.st_size + 10); + } + len = read(fd, content, info.st_size); + if (len < 0) { + xmlFree(content); + return(-1); + } + content[len] = 0; + close(fd); + + ret = xmlParseCatalog(content, filename); + xmlFree(content); + return(ret); +} + +/** + * xmlCatalogCleanup: + * + * Free up all the memory associated with catalogs + */ +void +xmlCatalogCleanup(void) { + if (xmlDefaultCatalog != NULL) + xmlHashFree(xmlDefaultCatalog, + (xmlHashDeallocator) xmlFreeCatalogEntry); + xmlDefaultCatalog = NULL; +} + +/** + * xmlCatalogDumpEntry: + * @entry: the + * @out: the file. + * + * Free up all the memory associated with catalogs + */ +static void +xmlCatalogDumpEntry(xmlCatalogEntryPtr entry, FILE *out) { + if ((entry == NULL) || (out == NULL)) + return; + switch (entry->type) { + case XML_CATA_ENTITY: + fprintf(out, "ENTITY "); break; + case XML_CATA_PENTITY: + fprintf(out, "ENTITY %%"); break; + case XML_CATA_DOCTYPE: + fprintf(out, "DOCTYPE "); break; + case XML_CATA_LINKTYPE: + fprintf(out, "LINKTYPE "); break; + case XML_CATA_NOTATION: + fprintf(out, "NOTATION "); break; + case XML_CATA_PUBLIC: + fprintf(out, "PUBLIC "); break; + case XML_CATA_SYSTEM: + fprintf(out, "SYSTEM "); break; + case XML_CATA_DELEGATE: + fprintf(out, "DELEGATE "); break; + case XML_CATA_BASE: + fprintf(out, "BASE "); break; + case XML_CATA_CATALOG: + fprintf(out, "CATALOG "); break; + case XML_CATA_DOCUMENT: + fprintf(out, "DOCUMENT "); break; + case XML_CATA_SGMLDECL: + fprintf(out, "SGMLDECL "); break; + default: + return; + } + switch (entry->type) { + case XML_CATA_ENTITY: + case XML_CATA_PENTITY: + case XML_CATA_DOCTYPE: + case XML_CATA_LINKTYPE: + case XML_CATA_NOTATION: + fprintf(out, "%s", entry->name); break; + case XML_CATA_PUBLIC: + case XML_CATA_SYSTEM: + case XML_CATA_SGMLDECL: + case XML_CATA_DOCUMENT: + case XML_CATA_CATALOG: + case XML_CATA_BASE: + case XML_CATA_DELEGATE: + fprintf(out, "\"%s\"", entry->name); break; + default: + break; + } + switch (entry->type) { + case XML_CATA_ENTITY: + case XML_CATA_PENTITY: + case XML_CATA_DOCTYPE: + case XML_CATA_LINKTYPE: + case XML_CATA_NOTATION: + case XML_CATA_PUBLIC: + case XML_CATA_SYSTEM: + case XML_CATA_DELEGATE: + fprintf(out, " \"%s\"", entry->value); break; + default: + break; + } + fprintf(out, "\n"); +} + +/** + * xmlCatalogDump: + * @out: the file. + * + * Free up all the memory associated with catalogs + */ +void +xmlCatalogDump(FILE *out) { + if (out == NULL) + return; + if (xmlDefaultCatalog != NULL) { + xmlHashScan(xmlDefaultCatalog, + (xmlHashScanner) xmlCatalogDumpEntry, out); + } +} +#endif /* LIBXML_CATALOG_ENABLED */ diff --git a/catalog.h b/catalog.h new file mode 100644 index 00000000..f91ee94e --- /dev/null +++ b/catalog.h @@ -0,0 +1,32 @@ +/** + * uri.c: interfaces of the Catalog handling system + * + * Reference: SGML Open Technical Resolution TR9401:1997. + * http://www.jclark.com/sp/catalog.htm + * + * See Copyright for the status of this software. + * + * Daniel.Veillard@w3.org + */ + +#ifndef __XML_CATALOG_H__ +#define __XML_CATALOG_H__ + +#include + +#include +#ifdef LIBXML_CATALOG_ENABLED + +#ifdef __cplusplus +extern "C" { +#endif + +int xmlLoadCatalog (const char *URL); +void xmlCatalogCleanup (void); +void xmlCatalogDump (FILE *out); + +#ifdef __cplusplus +} +#endif +#endif /* LIBXML_CATALOG_ENABLED */ +#endif /* __XML_CATALOG_H__ */ diff --git a/configure.in b/configure.in index 31499449..7f5dda94 100644 --- a/configure.in +++ b/configure.in @@ -301,6 +301,18 @@ fi AC_SUBST(WITH_HTML) AC_SUBST(HTML_OBJ) +AC_ARG_WITH(catalog, [ --with-catalog Add the Catalog support (on)]) +if test "$with_catalog" = "no" ; then + echo Disabling Catalog support + WITH_CATALOG=0 + CATALOG_OBJ= +else + WITH_CATALOG=1 + CATALOG_OBJ="catalog.o" +fi +AC_SUBST(WITH_CATALOG) +AC_SUBST(CATALOG_OBJ) + AC_ARG_WITH(docbook, [ --with-docbook Add Docbook SGML support (on)]) if test "$with_docbook" = "no" ; then echo Disabling Docbook support diff --git a/include/Makefile.am b/include/Makefile.am index 5baa7fe1..12d2b1dc 100644 --- a/include/Makefile.am +++ b/include/Makefile.am @@ -27,7 +27,8 @@ xmlinc_HEADERS = \ libxml/valid.h \ libxml/xlink.h \ libxml/xmlversion.h \ - libxml/DOCBparser.h + libxml/DOCBparser.h \ + libxml/catalog.h install-exec-hook: $(mkinstalldirs) $(DESTDIR)$(xmlincdir) $(DESTDIR)$(xmlincdir)/libxml diff --git a/include/libxml/catalog.h b/include/libxml/catalog.h new file mode 100644 index 00000000..f91ee94e --- /dev/null +++ b/include/libxml/catalog.h @@ -0,0 +1,32 @@ +/** + * uri.c: interfaces of the Catalog handling system + * + * Reference: SGML Open Technical Resolution TR9401:1997. + * http://www.jclark.com/sp/catalog.htm + * + * See Copyright for the status of this software. + * + * Daniel.Veillard@w3.org + */ + +#ifndef __XML_CATALOG_H__ +#define __XML_CATALOG_H__ + +#include + +#include +#ifdef LIBXML_CATALOG_ENABLED + +#ifdef __cplusplus +extern "C" { +#endif + +int xmlLoadCatalog (const char *URL); +void xmlCatalogCleanup (void); +void xmlCatalogDump (FILE *out); + +#ifdef __cplusplus +} +#endif +#endif /* LIBXML_CATALOG_ENABLED */ +#endif /* __XML_CATALOG_H__ */ diff --git a/include/libxml/xmlversion.h.in b/include/libxml/xmlversion.h.in index 6725b846..76b7ae08 100644 --- a/include/libxml/xmlversion.h.in +++ b/include/libxml/xmlversion.h.in @@ -61,6 +61,15 @@ extern void xmlCheckVersion(int version); #define LIBXML_HTML_DISABLED #endif +/* + * Whether the Catalog support is configured in + */ +#if @WITH_CATALOG@ +#define LIBXML_CATALOG_ENABLED +#else +#define LIBXML_CATALOG_DISABLED +#endif + /* * Whether the SGML Docbook support is configured in */ diff --git a/parser.c b/parser.c index 95e7c0a6..626461a4 100644 --- a/parser.c +++ b/parser.c @@ -2495,11 +2495,18 @@ get_more: } nbchar = in - ctxt->input->cur; if (nbchar > 0) { - if (IS_BLANK(*ctxt->input->cur) && - areBlanks(ctxt, ctxt->input->cur, nbchar)) { - if (ctxt->sax->ignorableWhitespace != NULL) - ctxt->sax->ignorableWhitespace(ctxt->userData, - ctxt->input->cur, nbchar); + if (IS_BLANK(*ctxt->input->cur)) { + const xmlChar *tmp = ctxt->input->cur; + ctxt->input->cur = in; + if (areBlanks(ctxt, tmp, nbchar)) { + if (ctxt->sax->ignorableWhitespace != NULL) + ctxt->sax->ignorableWhitespace(ctxt->userData, + tmp, nbchar); + } else { + if (ctxt->sax->characters != NULL) + ctxt->sax->characters(ctxt->userData, + tmp, nbchar); + } } else { if (ctxt->sax->characters != NULL) ctxt->sax->characters(ctxt->userData, diff --git a/testCatalog.c b/testCatalog.c new file mode 100644 index 00000000..99f6c66c --- /dev/null +++ b/testCatalog.c @@ -0,0 +1,32 @@ +/* + * testCatalog.c : a small tester program for Catalog loading + * + * See Copyright for the status of this software. + * + * Daniel.Veillard@w3.org + */ + +#include "libxml.h" + +#include +#include +#include + +#include +#include +#include +#include + +int main(int argc, char **argv) { +#ifdef LIBXML_CATALOG_ENABLED + int i; + + for (i = 1; i < argc; i++) + xmlLoadCatalog(argv[i]); + + xmlCatalogDump(stdout); + xmlCatalogCleanup(); + xmlMemoryDump(); +#endif + return(0); +} diff --git a/xmlversion.h.in b/xmlversion.h.in index 6725b846..76b7ae08 100644 --- a/xmlversion.h.in +++ b/xmlversion.h.in @@ -61,6 +61,15 @@ extern void xmlCheckVersion(int version); #define LIBXML_HTML_DISABLED #endif +/* + * Whether the Catalog support is configured in + */ +#if @WITH_CATALOG@ +#define LIBXML_CATALOG_ENABLED +#else +#define LIBXML_CATALOG_DISABLED +#endif + /* * Whether the SGML Docbook support is configured in */