--- parser3/src/main/pa_globals.C 2001/12/14 13:02:32 1.94 +++ parser3/src/main/pa_globals.C 2004/02/13 14:01:08 1.172 @@ -1,77 +1,46 @@ /** @file Parser: globals. - Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - Author: Alexander Petrosyan (http://paf.design.ru) - - $Id: pa_globals.C,v 1.94 2001/12/14 13:02:32 paf Exp $ + Copyright (c) 2001-2004 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) */ -#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" - -#ifdef DB2 -#include "pa_db_manager.h" -#endif - -String *content_type_name; -String *charset_name; -String *body_name; -String *value_name; -String *expires_name; -String *path_name; -String *name_name; -String *size_name; -String *text_name; - -String *exception_method_name; -String *post_process_method_name; - -String *content_disposition_name; -String *content_disposition_filename_name; - -String *unnamed_name; - -String *auto_method_name; -String *main_method_name; - -String *main_class_name; +static const char * const IDENT_GLOBALS_C="$Date: 2004/02/13 14:01:08 $"; -String *result_var_name; -String *match_var_name; -String *string_pre_match_name; -String *string_match_name; -String *string_post_match_name; +#include "pa_config_includes.h" -String *charsets_name; -String *mime_types_name; -String *vfile_mime_type_name; -String *origins_mode_name; +#ifdef XML +#include "libxml/xmlversion.h" +#include "libxslt/extensions.h" +#include "libxslt/xsltutils.h" +extern "C" { +#include "libexslt/exslt.h" +}; +#endif -String *class_path_name; +#include "pcre.h" -String *switch_data_name; -String *case_default_value; +#include "pa_globals.h" +#include "pa_string.h" +#include "pa_sapi.h" +#include "pa_threads.h" +#include "pa_xml_io.h" +#include "pa_common.h" -String *sql_limit_name; -String *sql_offset_name; -String *sql_default_name; +#include "pa_cache_managers.h" -String *hash_default_element_name; +// defines -Hash *untaint_lang_name2enum; +//#define PA_DEBUG_XML_GC_MEMORY -Dictionary *default_typo_dict; +// globals 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; @@ -98,168 +67,352 @@ static void setup_hex_value() { hex_value['f'] = 15; } + +Hash thread_request; +void pa_register_thread_request(Request& r) { + thread_request.put(pa_get_thread_id(), &r); +} +/// retrives request set by pa_set_request function, useful in contextless places [slow] +Request& pa_thread_request() { + return *thread_request.get(pa_get_thread_id()); +} + +#ifdef XML + +class XML_Generic_error_info { +public: + char buf[MAX_STRING*5]; + size_t used; +public: + XML_Generic_error_info() { + reset(); + } + void reset() { + buf[used=0]=0; + } + const char* get_and_reset() { + if(used) { + char* result=pa_strdup(buf, used); + reset(); + return result; + } else + return 0; + } +}; + +static Hash xml_generic_error_infos; + +static void xmlParserGenericErrorFunc(void * /*ctx*/, const char* msg, ...) { +//_asm int 3; + pa_thread_t thread_id=pa_get_thread_id(); + + XML_Generic_error_info* p; + { + SYNCHRONIZED; // find+fill blocked + + // first try to get existing for this thread_id + p=xml_generic_error_infos.get(thread_id); + if(!p) // occupy empty one + xml_generic_error_infos.put(thread_id, (p=new(PointerFreeGC) XML_Generic_error_info)); + } + + va_list args; + va_start(args, msg); + p->used+=vsnprintf(p->buf+p->used, sizeof(p->buf)-p->used, msg, args); + va_end(args); +} + +bool xmlHaveGenericErrors() { + pa_thread_t thread_id=pa_get_thread_id(); + + SYNCHRONIZED; // find blocked + + return xml_generic_error_infos.get(thread_id)!=0; +} + +const char* xmlGenericErrors() { + pa_thread_t thread_id=pa_get_thread_id(); + + SYNCHRONIZED; // find+free blocked + + if(XML_Generic_error_info *p=xml_generic_error_infos.get(thread_id)) + return p->get_and_reset(); + + return 0; // no errors for our thread_id registered +} + +#endif + void pa_globals_destroy(void *) { try { - if(cache_managers) - cache_managers->~Cache_managers(); - +#ifdef XML + GdomeException exc; + gdome_di_unref (domimpl, &exc); + // uncomment SAPI::abort below if adding potential-throw code here +#endif } catch(const Exception& e) { - SAPI::die("pa_globals_destroy failed: %s", e.comment()); +// SAPI::abort("pa_globals_destroy failed: %s", e.comment()); } } -void pa_globals_init(Pool& pool) { - pool.register_cleanup(pa_globals_destroy, 0); +#ifdef XML - #undef NEW - #define NEW new(pool) +static char *pa_GC_strdup(const char *s) { + if(!s) + return 0; + + size_t size=strlen(s)+1; + char *result=(char *)GC_MALLOC_ATOMIC(size); + if(!result) + SAPI::abort("out of memory (while duplicating XML string [size=%d])", size); + + memcpy(result, s, size); +#ifdef PA_DEBUG_XML_GC_MEMORY + fprintf(stderr, "pa_GC_strdup(%p=%s, length=%d)=0x%p\n", s, s, size, result); +#endif + return result; +} - // hex value - setup_hex_value(); +#ifdef PA_DEBUG_XML_GC_MEMORY +void *pa_look_for[]={(void*)0x8abe000,(void*)0x0,(void*)0x0,(void*)0x0, + (void*)0x0,(void*)0x0,(void*)0x0,(void*)0x0}; +bool pa_looked(void*p) { + for(int i=0; i<8; i++) + if(pa_look_for[i]==p) + return true; + return false; +} +static void* pa_gc_malloc_log(size_t size){ + void *p=pa_gc_malloc(size); + fprintf(stderr, "pa_gc_malloc_log(%d)=0x%p\n", size, p); +// if(pa_looked(p)) +// fprintf(stderr,"catched debug malloc(%d)=0x%p\n", size, p); + return p; + +} +static void* pa_gc_malloc_atomic_log(size_t size){ + void *p=pa_gc_malloc_atomic(size); + fprintf(stderr, "pa_gc_malloc_atomic_log(%d)=0x%p\n", size, p); +// if(pa_looked(p)) +// fprintf(stderr,"catched debug malloc atomic(%d)=0x%p\n", size, p); + return p; +} +static void* pa_gc_realloc_log(void *ptr, size_t size){ + void *p=pa_gc_realloc(ptr, size); + fprintf(stderr, "pa_gc_realloc_log(0x%p, %d)=0x%p\n", ptr, size, p); +// if(pa_looked(p)) +// fprintf(stderr,"catched debug realloc(%d)=0x%p\n", size, p); + return p; +} +//static void pa_gc_free_ignore(void *){} +static void pa_gc_free_log(void *p){ + fprintf(stderr, "pa_gc_free_log(0x%p)\n", p); +// if(pa_looked(p)) +// fprintf(stderr,"catched debug free(0x%p)\n", p); + pa_gc_free(p); +} +#else + +inline void *check(void *result, const char *where, size_t size) { + if(!result) + SAPI::abort("out of memory (while %s [size=%d])", where, size); - // names - content_type_name=NEW String(pool, CONTENT_TYPE_NAME); - charset_name=NEW String(pool, CHARSET_NAME); - body_name=NEW String(pool, BODY_NAME); - value_name=NEW String(pool, VALUE_NAME); - expires_name=NEW String(pool, EXPIRES_NAME); - path_name=NEW String(pool, PATH_NAME); - name_name=NEW String(pool, NAME_NAME); - size_name=NEW String(pool, SIZE_NAME); - text_name=NEW String(pool, TEXT_NAME); - - exception_method_name=NEW String(pool, EXCEPTION_METHOD_NAME); - post_process_method_name=NEW String(pool, POST_PROCESS_METHOD_NAME); - - content_disposition_name=NEW String(pool, CONTENT_DISPOSITION_NAME); - content_disposition_filename_name=NEW String(pool, CONTENT_DISPOSITION_FILENAME_NAME); - - unnamed_name=NEW String(pool, UNNAMED_NAME); - - auto_method_name=NEW String(pool, AUTO_METHOD_NAME); - main_method_name=NEW String(pool, MAIN_METHOD_NAME); - - main_class_name=NEW String(pool, MAIN_CLASS_NAME); - - result_var_name=NEW String(pool, RESULT_VAR_NAME); - match_var_name=NEW String(pool, MATCH_VAR_NAME); - string_pre_match_name=NEW String(pool, STRING_PRE_MATCH_NAME); - string_match_name=NEW String(pool, STRING_MATCH_NAME); - string_post_match_name=NEW String(pool, STRING_POST_MATCH_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); - origins_mode_name=NEW String(pool, ORIGINS_MODE_NAME); - - class_path_name=NEW String(pool, CLASS_PATH_NAME); - - //^switch ^case - switch_data_name=NEW String(pool, SWITCH_DATA_NAME); - case_default_value=NEW String(pool, CASE_DEFAULT_VALUE); - - // sql - sql_limit_name=NEW String(pool, SQL_LIMIT_NAME); - sql_offset_name=NEW String(pool, SQL_OFFSET_NAME); - sql_default_name=NEW String(pool, SQL_DEFAULT_NAME); - - // hash - hash_default_element_name=NEW String(pool, HASH_DEFAULT_ELEMENT_NAME); - - - // hashes - untaint_lang_name2enum=NEW Hash(pool); - #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); - - // 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); - - // Status registration, must be initialized before all registrants - cache_managers=NEW Cache_managers(pool); - - // SQL driver manager - 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)); + return result; +} +static void* pa_gc_malloc_nonull(size_t size) { + return check(pa_gc_malloc(size), "allocating XML compsite memory", size); +} +static void* pa_gc_malloc_atomic_nonull(size_t size) { + return check(pa_gc_malloc_atomic(size), "allocating XML atomic memory", size); +} +static void* pa_gc_realloc_nonull(void* ptr, size_t size) { + return check(pa_gc_realloc(ptr, size), "reallocating XML memory", size); +} + +#endif #endif +void pa_CORD_oom_fn(void) { + SAPI::abort("out of memory (while expanding string)"); +} + +/** + @todo gc: libltdl: substitute lt_dlmalloc & co +*/ +static void gc_substitute_memory_management_functions() { + // in libxml & libxslt #ifdef XML - // XSLT stylesheet driver manager - cache_managers->put(*NEW String(pool, "stylesheet"), - stylesheet_manager=NEW Stylesheet_manager(pool)); + // asking to use GC memory +#if LIBXML_VERSION >= 20507 +#ifdef PA_DEBUG_XML_GC_MEMORY + xmlGcMemSetup( + /*xmlFreeFunc */pa_gc_free_log, + /*xmlMallocFunc */pa_gc_malloc_log, + /*xmlMallocFunc */pa_gc_malloc_atomic_log, + /*xmlReallocFunc */pa_gc_realloc_log, + /*xmlStrdupFunc */pa_GC_strdup); +#else + xmlGcMemSetup( + /*xmlFreeFunc */pa_gc_free, + /*xmlMallocFunc */pa_gc_malloc_nonull, + /*xmlMallocFunc */pa_gc_malloc_atomic_nonull, + /*xmlReallocFunc */pa_gc_realloc_nonull, + /*xmlStrdupFunc */pa_GC_strdup); #endif - // Charset manager - cache_managers->put(*NEW String(pool, "charset"), - charset_manager=NEW Charset_manager(pool)); +#else + xmlMemSetup( + /*xmlFreeFunc */pa_gc_free, + /*xmlMallocFunc */pa_gc_malloc, + /*xmlReallocFunc */pa_gc_realloc, + /*xmlStrdupFunc */pa_GC_strdup); +#endif + +#endif + + // pcre + pcre_malloc=pa_gc_malloc; + pcre_free=pa_gc_free; + + + // cord + CORD_oom_fn=pa_CORD_oom_fn; +} + +/** + @test hint on one should call this for each thread xmlSubstituteEntitiesDefault(1); +*/ +void pa_globals_init() { + // global variables + cache_managers=new Cache_managers; + + + // in various libraries + gc_substitute_memory_management_functions(); + + // hex value + setup_hex_value(); + +#ifdef XML + // initializing xml libs + + /* First 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 + xmlLoadExtDtdDefaultValue |= XML_COMPLETE_ATTRS; + + // validate each document after load/create (?) + //xmlDoValidityCheckingDefaultValue = 1; + +//regretfully this not only replaces entities on parse, but also on generate xmlSubstituteEntitiesDefault(1); + // never switched this on xmlIndentTreeOutput=1; + + xmlSetGenericErrorFunc(0, xmlParserGenericErrorFunc); + xsltSetGenericErrorFunc(0, xmlParserGenericErrorFunc); +// FILE *f=fopen("y:\\xslt.log", "wt"); +// xsltSetGenericDebugFunc(f/*stderr*/, 0); + + pa_xml_io_init(); +#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" +#ifdef _MSC_VER + +#ifndef PA_DEBUG_DISABLE_GC +# define GC_LIB "../../../../win32/gc" # 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, GC_LIB "/Debug/gc.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, GC_LIB "/Release/gc.lib") # endif -#endif +#endif -#if defined(DB2) && defined(_MSC_VER) -# define LIBDB2_WIN32_BUILD "/parser3project/win32db/lib" +#ifdef XML +# define GNOME_LIBS "../../../../win32/gnome" +# pragma comment(lib, GNOME_LIBS "/glib/lib/libglib-1.3-11.lib") # ifdef _DEBUG -# pragma comment(lib, LIBDB2_WIN32_BUILD "/debug/libdb.lib") -# else -# pragma comment(lib, LIBDB2_WIN32_BUILD "/release/libdb.lib") + +# ifdef LIBXML_STATIC +# pragma comment(lib, GNOME_LIBS "/libxml2-x.x.x/win32/debug/lib/libxml2_a.lib") +# else +# pragma comment(lib, GNOME_LIBS "/libxml2-x.x.x/win32/debug/lib/libxml2.lib") +# endif + +# ifdef LIBXSLT_STATIC +# pragma comment(lib, GNOME_LIBS "/libxslt-x.x.x/win32/debug/lib/libxslt_a.lib") +# else +# pragma comment(lib, GNOME_LIBS "/libxslt-x.x.x/win32/debug/lib/libxslt.lib") +# endif +# ifdef LIBEXSLT_STATIC +# pragma comment(lib, GNOME_LIBS "/libxslt-x.x.x/win32/debug/lib/libexslt_a.lib") +# else +# pragma comment(lib, GNOME_LIBS "/libxslt-x.x.x/win32/debug/lib/libexslt.lib") +# endif + +# ifdef LIBGDOME_STATIC +# pragma comment(lib, GNOME_LIBS "/gdome2-x.x.x/win32/DebugStatic/libgdome.lib") +# else +# pragma comment(lib, GNOME_LIBS "/gdome2-x.x.x/win32/DebugDynamic/libgdome.lib") +# endif + +#else + +# ifdef LIBXML_STATIC +# pragma comment(lib, GNOME_LIBS "/libxml2-x.x.x/win32/release/lib/libxml2_a.lib") +# else +# pragma comment(lib, GNOME_LIBS "/libxml2-x.x.x/win32/release/lib/libxml2.lib") +# endif + +# ifdef LIBXSLT_STATIC +# pragma comment(lib, GNOME_LIBS "/libxslt-x.x.x/win32/release/lib/libxslt_a.lib") +# else +# pragma comment(lib, GNOME_LIBS "/libxslt-x.x.x/win32/release/lib/libxslt.lib") +# endif +# ifdef LIBEXSLT_STATIC +# pragma comment(lib, GNOME_LIBS "/libxslt-x.x.x/win32/release/lib/libexslt_a.lib") +# else +# pragma comment(lib, GNOME_LIBS "/libxslt-x.x.x/win32/release/lib/libexslt.lib") +# endif + +# ifdef LIBGDOME_STATIC +# pragma comment(lib, GNOME_LIBS "/gdome2-x.x.x/win32/ReleaseStatic/libgdome.lib") +# else +# pragma comment(lib, GNOME_LIBS "/gdome2-x.x.x/win32/ReleaseDynamic/libgdome.lib") +# endif + # endif #endif +#endif