--- parser3/src/main/pa_globals.C 2001/10/27 13:00:10 1.86 +++ parser3/src/main/pa_globals.C 2002/01/23 13:07:53 1.108 @@ -2,25 +2,40 @@ Parser: globals. Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - Author: Alexander Petrosyan (http://design.ru/paf) + Author: Alexander Petrosyan (http://paf.design.ru) - $Id: pa_globals.C,v 1.86 2001/10/27 13:00:10 paf Exp $ + $Id: pa_globals.C,v 1.108 2002/01/23 13:07:53 paf Exp $ */ +#include "pa_config_includes.h" + +#ifdef XML +//#include "libxml/parser.h" +//#include "libxslt/xslt.h" +//#include "libxslt/libxslt.h" +#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_charset_manager.h" #include "pa_sapi.h" +#include "pa_cache_managers.h" +#include "pa_charsets.h" +#include "pa_charset.h" +#include "pa_threads.h" #ifdef DB2 #include "pa_db_manager.h" #endif -String *user_html_name; String *content_type_name; String *charset_name; String *body_name; @@ -50,7 +65,6 @@ String *string_pre_match_name; String *string_match_name; String *string_post_match_name; -String *defaults_name; String *charsets_name; String *mime_types_name; String *vfile_mime_type_name; @@ -65,11 +79,15 @@ String *sql_limit_name; String *sql_offset_name; String *sql_default_name; +String *charset_UTF8_name; + String *hash_default_element_name; Hash *untaint_lang_name2enum; -Dictionary *default_typo_dict; +GdomeDOMImplementation *domimpl; + +Charset *utf8_charset; short hex_value[0x100]; @@ -99,34 +117,101 @@ static void setup_hex_value() { hex_value['f'] = 15; } -void pa_globals_destroy(void *) { - try { - // SQL driver manager - if(SQL_driver_manager) - SQL_driver_manager->~SQL_Driver_manager(); +#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 +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); -#ifdef DB2 - // DB driver manager - if(DB_manager) - DB_manager->~DB_Manager(); + 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; +} #endif - + +void pa_globals_destroy(void *) { + try { #ifdef XML - // XSLT stylesheet driver manager - if(stylesheet_manager) - stylesheet_manager->~Stylesheet_manager(); + GdomeException exc; + gdome_di_unref (domimpl, &exc); #endif - - // Charset manager - if(charset_manager) - charset_manager->~Charset_manager(); + if(cache_managers) + cache_managers->~Cache_managers(); + charsets->~Charsets(); + } catch(const Exception& e) { SAPI::die("pa_globals_destroy failed: %s", e.comment()); } } - +/// @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); @@ -137,7 +222,6 @@ void pa_globals_init(Pool& pool) { setup_hex_value(); // names - user_html_name=NEW String(pool, USER_HTML_NAME); content_type_name=NEW String(pool, CONTENT_TYPE_NAME); charset_name=NEW String(pool, CHARSET_NAME); body_name=NEW String(pool, BODY_NAME); @@ -168,7 +252,6 @@ void pa_globals_init(Pool& pool) { string_post_match_name=NEW String(pool, STRING_POST_MATCH_NAME); - defaults_name=NEW String(pool, DEFAULTS_NAME); charsets_name=NEW String(pool, CHARSETS_NAME); mime_types_name=NEW String(pool, MIME_TYPES_NAME); vfile_mime_type_name=NEW String(pool, VFILE_MIME_TYPE_NAME); @@ -185,85 +268,110 @@ void pa_globals_init(Pool& pool) { sql_offset_name=NEW String(pool, SQL_OFFSET_NAME); sql_default_name=NEW String(pool, SQL_DEFAULT_NAME); + // charsets + charset_UTF8_name=NEW String(pool, CHARSET_UTF8_NAME); + // hash hash_default_element_name=NEW String(pool, HASH_DEFAULT_ELEMENT_NAME); // hashes untaint_lang_name2enum=NEW Hash(pool); - #define ULN(var, cstr, LANG) \ - String var(pool, cstr); \ - untaint_lang_name2enum->put(var, (int)String::UL_##LANG); - ULN(as_is, "as-is", AS_IS); - ULN(file_name, "file-spec", FILE_SPEC); - ULN(http_header, "http-header", HTTP_HEADER); - ULN(mail_header, "mail-header", MAIL_HEADER); - ULN(uri, "uri", URI); - ULN(table, "table", TABLE); - ULN(sql, "sql", SQL); - ULN(js, "js", JS); - ULN(xml, "xml", XML); - ULN(html, "html", HTML); - ULN(user_html, "user-html", USER_HTML); - - // tables - Table *default_typo_table=NEW Table(pool, 0, 0); - #define DT_ROW(from, to) { \ - Array *row=NEW Array(pool); \ - *row+=NEW String(pool, from); \ - *row+=NEW String(pool, to); \ - *default_typo_table+=row; \ - } - DT_ROW("<", "<"); - DT_ROW(">", ">"); - DT_ROW("\"", """); - DT_ROW("&", "&"); - DT_ROW("\\n\\n", "

"); - DT_ROW("\\n", "
"); - default_typo_dict=NEW Dictionary(*default_typo_table); + #define ULN(cstr, LANG) \ + untaint_lang_name2enum->put(*NEW String(pool, cstr), (int)String::UL_##LANG); + ULN("as-is", AS_IS); + ULN("file-spec", FILE_SPEC); + ULN("http-header", HTTP_HEADER); + ULN("mail-header", MAIL_HEADER); + ULN("uri", URI); + ULN("table", TABLE); + ULN("sql", SQL); + ULN("js", JS); + ULN("xml", XML); + ULN("html", HTML); + ULN("optimized-html", HTML|String::UL_OPTIMIZE_BIT); + + // charsets + charsets=NEW Charsets(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); // SQL driver manager - SQL_driver_manager=NEW SQL_Driver_manager(pool); + cache_managers->put(*NEW String(pool, "sql"), + SQL_driver_manager=NEW SQL_Driver_manager(pool)); #ifdef DB2 // DB driver manager - DB_manager=NEW DB_Manager(pool); + cache_managers->put(*NEW String(pool, "db"), + DB_manager=NEW DB_Manager(pool)); #endif #ifdef XML - // XSLT stylesheet driver manager - stylesheet_manager=NEW Stylesheet_manager(pool); -#endif + // initializing xml libs - // Charset manager - charset_manager=NEW Charset_manager(pool); + /* First I get a DOMImplementation reference */ + domimpl = gdome_di_mkref (); + /* + * Register the EXSLT extensions and the test module + */ + exsltRegisterAll(); + xsltRegisterTestModule(); + xmlDefaultSAXHandlerInit(); + /* + * disable CDATA from being built in the document tree + */ + // never added yet xmlDefaultSAXHandler.cdataBlock = NULL; + + /* + * Initialization function for the XML parser. + * This is not reentrant. Call once before processing in case of + * use in multithreaded programs. + */ + xmlInitParser(); + + // 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); + + // XSLT stylesheet manager + cache_managers->put(*NEW String(pool, "stylesheet"), + stylesheet_manager=NEW Stylesheet_manager(pool)); +#endif } #if defined(XML) && defined(_MSC_VER) -# define XERCES_BUILD_WIN32_BUILD "/parser3project/win32xml/xml-xerces/c/Build/Win32/VC6" -# define XALAN_BUILD_WIN32_BUILD "/parser3project/win32xml/xml-xalan/c/Build/Win32/VC6" +# 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_so/libxslt.lib") # ifdef _DEBUG -# pragma comment(lib, XERCES_BUILD_WIN32_BUILD "/Debug/xerces-c_1D.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Debug/XSLTD.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Debug/XPathD.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Debug/XalanTransformerD.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Debug/XalanDOMD.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Debug/DOMSupportD.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Debug/PlatformSupportD.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Debug/XMLSupportD.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Debug/XercesParserLiaisonD.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Debug/XalanSourceTreeD.lib") +# pragma comment(lib, XML_LIBS "/gdome2-0.7.0/win32/dsp/Debug/libgdome.lib") # else -# pragma comment(lib, XERCES_BUILD_WIN32_BUILD "/Release/xerces-c_1.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Release/XalanSourceTree.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Release/PlatformSupport.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Release/XalanDOM.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Release/DOMSupport.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Release/XalanTransformer.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Release/XercesParserLiaison.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Release/XMLSupport.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Release/XPath.lib") -# pragma comment(lib, XALAN_BUILD_WIN32_BUILD "/Release/XSLT.lib") +# pragma comment(lib, XML_LIBS "/gdome2-0.7.0/win32/dsp/Release/libgdome.lib") # endif #endif @@ -276,4 +384,3 @@ void pa_globals_init(Pool& pool) { # pragma comment(lib, LIBDB2_WIN32_BUILD "/release/libdb.lib") # endif #endif -