--- parser3/src/main/pa_globals.C 2001/12/29 08:39:05 1.99 +++ parser3/src/main/pa_globals.C 2002/03/05 07:48:06 1.118 @@ -1,34 +1,33 @@ /** @file Parser: globals. - Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - Author: Alexander Petrosyan (http://paf.design.ru) + Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) - $Id: pa_globals.C,v 1.99 2001/12/29 08:39:05 paf Exp $ + $Id: pa_globals.C,v 1.118 2002/03/05 07:48:06 paf Exp $ */ +#include "pa_config_includes.h" + +#ifdef XML +#include "libxslt/extensions.h" +#include "libxslt/xsltutils.h" +extern "C" { +#include "libexslt/exslt.h" +}; +#endif + #include "pa_globals.h" #include "pa_string.h" #include "pa_hash.h" #include "pa_sql_driver_manager.h" #include "pa_dictionary.h" -//#include "pa_stylesheet_manager.h" +#include "pa_stylesheet_manager.h" #include "pa_sapi.h" #include "pa_cache_managers.h" #include "pa_charsets.h" #include "pa_charset.h" - -#ifdef DB2 -#include "pa_db_manager.h" -#endif - -#ifdef XML -//#include "libxml/parser.h" -//#include "libxslt/xslt.h" -//#include "libxslt/libxslt.h" -#include "libxslt/extensions.h" -#include "libexslt/exslt.h" -#endif +#include "pa_threads.h" String *content_type_name; String *charset_name; @@ -79,12 +78,14 @@ String *hash_default_element_name; Hash *untaint_lang_name2enum; -GdomeDOMImplementation *domimpl; - Charset *utf8_charset; short hex_value[0x100]; +#ifdef XML +GdomeDOMImplementation *domimpl; +#endif + static void setup_hex_value() { memset(hex_value, 0, sizeof(hex_value)); hex_value['0'] = 0; @@ -111,19 +112,170 @@ static void setup_hex_value() { hex_value['f'] = 15; } -/* #ifdef XML + +const int MAX_CONCURRENT_XML_GENERIC_ERROR_THREADS=10; + +struct XML_Generic_error_info { + pa_thread_t thread_id; + char *message; +} xml_generic_error_infos[MAX_CONCURRENT_XML_GENERIC_ERROR_THREADS]; + +XML_Generic_error_info *xml_generic_error_info(pa_thread_t thread_id) { + for(int i=0; ithread_id==thread_id) + return p; + } + return 0; +} + static void -xmlGenericErrorDefaultFunc2(void *ctx ATTRIBUTE_UNUSED, const char *msg, ...) { - va_list args; - va_start(args, msg); - throw Exception(0, 0, - 0, - msg, args); - va_end(args); +xmlParserGenericErrorFunc(void *ctx, const char *msg, ...) { + pa_thread_t thread_id=pa_get_thread_id(); + + // infinitely looking for free slot to fill it + while(true) { + SYNCHRONIZED; // find+fill blocked + + // first try to get existing for this thread_id + XML_Generic_error_info *p=xml_generic_error_info(thread_id); + if(!p) { // occupy empty one + p=xml_generic_error_info(0); + if(!p) // wait for empty for it to appear + continue; + } + + p->thread_id=thread_id; + size_t offset=p->message?strlen(p->message):0; + p->message=(char *)realloc(p->message, offset+MAX_STRING); + if(!p->message) + SAPI::die( + "out of memory in 'xmlParserGenericErrorFunc', failed to reallocate to %u bytes", + offset+MAX_STRING); + + va_list args; + va_start(args, msg); + vsnprintf(p->message+offset, MAX_STRING, msg, args); + va_end(args); + + break; + } } + +bool xmlHaveGenericErrors() { + pa_thread_t thread_id=pa_get_thread_id(); + + SYNCHRONIZED; // find blocked + + return xml_generic_error_info(thread_id)!=0; +} + +const char *xmlGenericErrors() { + pa_thread_t thread_id=pa_get_thread_id(); + + SYNCHRONIZED; // find+free blocked + + XML_Generic_error_info *p=xml_generic_error_info(thread_id); + if(!p) // no errors for our thread_id registered + return 0; + + const char *result=p->message; + + // free slot up + memset(p, 0, sizeof(*p)); + + // it is up to caller to free it + return result; +} + +/** + * xmlFileMatchWithLocalhostEqDocumentRoot: + * filename: the URI for matching + * + * check if the URI matches an HTTP one + * + * Returns 1 if matches, 0 otherwise + */ +static int +xmlFileMatchLocalhost(const char *filename) { + if (!strncmp(filename, "http://localhost", 16)) + return(1); + return(0); +} + + +/** + * xmlFileOpenHttpLocalhost : + * filename: the URI for matching + * + * http://localhost/abc -> $ENV{DOCUMENT_ROOT}/abc | ./abc + * + * input from FILE *, supports compressed input + * if filename is " " then the standard input is used + * + * Returns an I/O context or NULL in case of error + */ +static void * +xmlFileOpenLocalhost (const char *filename) { + FILE *fd; + const char* documentRoot; + char path[1000]; + + path[0]=0; + strcat(path, (documentRoot=getenv("DOCUMENT_ROOT"))?documentRoot:"."); + strcat(path, &filename[16]); + +#ifdef WIN32 + fd = fopen(path, "rb"); +#else + fd = fopen(path, "r"); +#endif /* WIN32 */ + return((void *) fd); +} + +/** + * xmlFileRead: + * @context: the I/O context + * @buffer: where to drop data + * @len: number of bytes to write + * + * Read @len bytes to @buffer from the I/O channel. + * + * Returns the number of bytes written + */ +static int +xmlFileRead (void * context, char * buffer, int len) { + return(fread(&buffer[0], 1, len, (FILE *) context)); +} + +/** + * xmlFileWrite: + * @context: the I/O context + * @buffer: where to drop data + * @len: number of bytes to write + * + * Write @len bytes from @buffer to the I/O channel. + * + * Returns the number of bytes written + */ +static int +xmlFileWrite (void * context, const char * buffer, int len) { + return(fwrite(&buffer[0], 1, len, (FILE *) context)); +} + +/** + * xmlFileClose: + * @context: the I/O context + * + * Close an I/O channel + */ +static int +xmlFileClose (void * context) { + return ( ( fclose((FILE *) context) == EOF ) ? -1 : 0 ); +} + #endif -*/ void pa_globals_destroy(void *) { try { @@ -141,6 +293,7 @@ void pa_globals_destroy(void *) { } } +/// @test hint on one should call this for each thread xmlSubstituteEntitiesDefault(1); void pa_globals_init(Pool& pool) { pool.register_cleanup(pa_globals_destroy, 0); @@ -225,6 +378,7 @@ void pa_globals_init(Pool& pool) { charsets->put(*charset_UTF8_name, utf8_charset=NEW Charset(pool, *charset_UTF8_name, 0/*no file=system*/)); + // Status registration, must be initialized before all registrants cache_managers=NEW Cache_managers(pool); @@ -232,12 +386,6 @@ void pa_globals_init(Pool& pool) { cache_managers->put(*NEW String(pool, "sql"), SQL_driver_manager=NEW SQL_Driver_manager(pool)); -#ifdef DB2 - // DB driver manager - cache_managers->put(*NEW String(pool, "db"), - DB_manager=NEW DB_Manager(pool)); -#endif - #ifdef XML // initializing xml libs @@ -252,7 +400,7 @@ void pa_globals_init(Pool& pool) { /* * disable CDATA from being built in the document tree */ - xmlDefaultSAXHandler.cdataBlock = NULL; + // never added yet xmlDefaultSAXHandler.cdataBlock = NULL; /* * Initialization function for the XML parser. @@ -261,35 +409,53 @@ void pa_globals_init(Pool& pool) { */ xmlInitParser(); -/// xmlSetGenericErrorFunc(0, xmlGenericErrorDefaultFunc2); + // 1. this is needed for proper parsing of stylesheets + // there were a situation where honest entity ruined innocent xpath compilation + // doc says "you sould turn it on on stylesheet load" without deepening into details + // 2. when dom tree with entites goes under transform text nodes + // got [erroreosly] cut on first entity occurance + // -- + // that is why this is: + xmlSubstituteEntitiesDefault(1); + + // Bit in the loadsubset context field to tell to do ID/REFs lookups + xmlLoadExtDtdDefaultValue |= XML_DETECT_IDS; + // Bit in the loadsubset context field to tell to do complete the elements attributes lists + // with the ones defaulted from the DTDs + //never added yet xmlLoadExtDtdDefaultValue |= XML_COMPLETE_ATTRS; + +//regretfully this not only replaces entities on parse, but also on generate xmlSubstituteEntitiesDefault(1); + // never switched this on xmlIndentTreeOutput=1; + + memset(xml_generic_error_infos, 0, sizeof(xml_generic_error_infos)); + xmlSetGenericErrorFunc(0, xmlParserGenericErrorFunc); + xsltSetGenericErrorFunc(0, xmlParserGenericErrorFunc); +// FILE *f=fopen("y:\\xslt.log", "wt"); +// xsltSetGenericDebugFunc(f/*stderr*/, 0); + + // http://localhost/abc -> $ENV{DOCUMENT_ROOT}/abc | ./abc + xmlRegisterInputCallbacks( + xmlFileMatchLocalhost, xmlFileOpenLocalhost, + xmlFileRead, xmlFileClose); -/* // XSLT stylesheet manager cache_managers->put(*NEW String(pool, "stylesheet"), stylesheet_manager=NEW Stylesheet_manager(pool)); - */ #endif } #if defined(XML) && defined(_MSC_VER) # define XML_LIBS "/parser3project/win32xml" # pragma comment(lib, XML_LIBS "/glib/lib/libglib-1.3-11.lib") -# pragma comment(lib, XML_LIBS "/libxml2-2.4.12/win32/dsp/libxml2_so/libxml2.lib") -# pragma comment(lib, XML_LIBS "/libxslt-1.0.9/win32/dsp/libexslt_so/libexslt.lib") -# pragma comment(lib, XML_LIBS "/libxslt-1.0.9/win32/dsp/libxslt_a/libxslt.lib") -# ifdef _DEBUG -# pragma comment(lib, XML_LIBS "/gdome2-0.7.0/win32/dsp/Debug/libgdome.lib") -# else -# pragma comment(lib, XML_LIBS "/gdome2-0.7.0/win32/dsp/Release/libgdome.lib") -# endif -#endif - - -#if defined(DB2) && defined(_MSC_VER) -# define LIBDB2_WIN32_BUILD "/parser3project/win32db/lib" # ifdef _DEBUG -# pragma comment(lib, LIBDB2_WIN32_BUILD "/debug/libdb.lib") +# pragma comment(lib, XML_LIBS "/libxml2-x.x.x/win32/dsp/libxml2_so_debug/libxml2.lib") +# pragma comment(lib, XML_LIBS "/libxslt-x.x.x/win32/dsp/libexslt_so_debug/libexslt.lib") +# pragma comment(lib, XML_LIBS "/libxslt-x.x.x/win32/dsp/libxslt_so_debug/libxslt.lib") +# pragma comment(lib, XML_LIBS "/gdome2-x.x.x/win32/dsp/Debug/libgdome.lib") # else -# pragma comment(lib, LIBDB2_WIN32_BUILD "/release/libdb.lib") +# pragma comment(lib, XML_LIBS "/libxml2-x.x.x/win32/dsp/libxml2_so_release/libxml2.lib") +# pragma comment(lib, XML_LIBS "/libxslt-x.x.x/win32/dsp/libexslt_so_release/libexslt.lib") +# pragma comment(lib, XML_LIBS "/libxslt-x.x.x/win32/dsp/libxslt_so_release/libxslt.lib") +# pragma comment(lib, XML_LIBS "/gdome2-x.x.x/win32/dsp/Release/libgdome.lib") # endif #endif