--- parser3/src/main/pa_globals.C 2001/12/28 14:06:52 1.97 +++ parser3/src/main/pa_globals.C 2013/07/05 21:09:58 1.188 @@ -1,87 +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 1.97 2001/12/28 14:06:52 paf Exp $ + Copyright (c) 2001-2012 Art. Lebedev Studio (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_sapi.h" -#include "pa_cache_managers.h" -#include "pa_charsets.h" -#include "pa_charset.h" - -#ifdef DB2 -#include "pa_db_manager.h" -#endif +#include "pa_config_includes.h" #ifdef XML -//#include "libxml/parser.h" -//#include "libxslt/xslt.h" -//#include "libxslt/libxslt.h" +#include "libxml/xmlversion.h" #include "libxslt/extensions.h" +#include "libxslt/xsltutils.h" +extern "C" { #include "libexslt/exslt.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; - -String *result_var_name; -String *match_var_name; -String *string_pre_match_name; -String *string_match_name; -String *string_post_match_name; - -String *charsets_name; -String *mime_types_name; -String *vfile_mime_type_name; -String *origins_mode_name; - -String *class_path_name; +#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 *switch_data_name; -String *case_default_value; +#include "pa_cache_managers.h" -String *sql_limit_name; -String *sql_offset_name; -String *sql_default_name; +#include "ltdl.h" +#include "pcre.h" -String *charset_UTF8_name; +volatile const char * IDENT_PA_GLOBALS_C="$Id: pa_globals.C,v 1.188 2013/07/05 21:09:58 moko Exp $" IDENT_PA_GLOBALS_H IDENT_PA_SAPI_H; -String *hash_default_element_name; +// defines -Hash *untaint_lang_name2enum; +//#define PA_DEBUG_XML_GC_MEMORY -GdomeDOMImplementation *domimpl; +//20051130 trying to remove this, author claims that fixed a lot there // 20040920 for now both workarounds needed. wait for new libxml/xsl versions +// // there is a problem with testcase, it's unstable. +// // see paf@six/bug20040920/cgi-bin/t for it-showed-bug-on-20040920-day +// #define PA_WORKAROUND_BUGGY_FREE_IN_LIBXML_GC_MEMORY +// #define PA_WORKAROUND_BUGGY_MALLOCATOMIC_IN_LIBXML_GC_MEMORY -Charset *utf8_charset; +// globals short hex_value[0x100]; @@ -111,163 +70,361 @@ static void setup_hex_value() { hex_value['f'] = 15; } -void pa_globals_destroy(void *) { - try { -#ifdef XML - GdomeException exc; - gdome_di_unref (domimpl, &exc); +THREAD_LOCAL Request* thread_request=NULL; + +void pa_register_thread_request(Request& r) { + thread_request=&r; +} +/// retrives request set by pa_set_request function, useful in contextless places [slow] +Request& pa_thread_request() { + return *thread_request; +} + +#ifdef PA_RELEASE_ASSERTS +void pa_release_assert(const char* str, const char* file, int line) { + SAPI::die("%s at %s:%d", str, file, line); +} #endif - if(cache_managers) - cache_managers->~Cache_managers(); - charsets->~Charsets(); + +#ifdef XML + +class XML_Generic_error_info { +public:/*internal, actually*/ + char buf[MAX_STRING*5]; + size_t used; +public: + XML_Generic_error_info() { + buf[used=0]=0; + } + const char* get() { + return used? buf: 0; + } +}; + +THREAD_LOCAL XML_Generic_error_info* xml_generic_error_info = NULL; + +static void xmlParserGenericErrorFunc(void * /*ctx*/, const char* msg, ...) { + XML_Generic_error_info* p; + + if(!(p=xml_generic_error_info)) // occupy empty one + p=xml_generic_error_info=new(PointerFreeGC) XML_Generic_error_info; - } catch(const Exception& e) { - SAPI::die("pa_globals_destroy failed: %s", e.comment()); + 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() { + return xml_generic_error_info!=0; +} + +const char* xmlGenericErrors() { + if(XML_Generic_error_info *p=xml_generic_error_info) { + xml_generic_error_info=0; + return p->get(); } + + return 0; // no errors for our thread_id registered } +#endif -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) + pa_fail_alloc("duplicate XML string",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*)0x84ba980,(void*)0x8969460,(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) { + __asm__("int $3"); + return true; + } + if((((int)p)&~0xFF)==0x89a7700) { + __asm__("int $3"); + 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){ +#ifdef PA_WORKAROUND_BUGGY_MALLOCATOMIC_IN_LIBXML_GC_MEMORY + void *p=pa_gc_malloc(size); + fprintf(stderr, "pa_gc_malloc_atomicFAKE_log(%d)=0x%p\n", size, p); +#else + void *p=pa_gc_malloc_atomic(size); + fprintf(stderr, "pa_gc_malloc_atomic_log(%d)=0x%p\n", size, p); +#endif + 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_log(void *p){ +#ifdef PA_WORKAROUND_BUGGY_FREE_IN_LIBXML_GC_MEMORY + fprintf(stderr, "pa_gc_freeIGNORE_log(0x%p)\n", p); +#else + fprintf(stderr, "pa_gc_free_log(0x%p)\n", p); +#endif + if(pa_looked(p)) + fprintf(stderr,"catched debug free(0x%p)\n", p); +#ifndef PA_WORKAROUND_BUGGY_FREE_IN_LIBXML_GC_MEMORY + pa_gc_free(p); +#endif +} +#else + +inline void *check(void *result, const char *where, size_t size) { + if(!result) + pa_fail_alloc(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); - - // 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(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 - 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), "allocate XML compsite memory", size); +} +static void* pa_gc_malloc_atomic_nonull(size_t size) { +#ifdef PA_WORKAROUND_BUGGY_MALLOCATOMIC_IN_LIBXML_GC_MEMORY + return check(pa_gc_malloc(size), "allocate XML composite memory (asked atomic)", size); +#else + return check(pa_gc_malloc_atomic(size), "allocate XML atomic memory", size); +#endif +} +static void* pa_gc_realloc_nonull(void* ptr, size_t size) { + return check(pa_gc_realloc(ptr, size), "reallocate XML memory", size); +} + +static void pa_gc_free_maybeignore( + void* +#ifndef PA_WORKAROUND_BUGGY_FREE_IN_LIBXML_GC_MEMORY + ptr +#endif + ) { +#ifndef PA_WORKAROUND_BUGGY_FREE_IN_LIBXML_GC_MEMORY + pa_gc_free(ptr); +#endif +} + +#endif #endif +void pa_CORD_oom_fn(void) { + pa_fail_alloc("expand string", 0); +} + +/** + @todo gc: libltdl: substitute lt_dlmalloc & co +*/ +static void gc_substitute_memory_management_functions() { + // in libxml & libxslt +#ifdef XML + // 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_maybeignore, + /*xmlMallocFunc */pa_gc_malloc_nonull, + /*xmlMallocFunc */pa_gc_malloc_atomic_nonull, + /*xmlReallocFunc */pa_gc_realloc_nonull, + /*xmlStrdupFunc */pa_GC_strdup); +#endif + +#else + xmlMemSetup( + /*xmlFreeFunc */pa_gc_free_maybeignore, + /*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 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 - */ - xmlDefaultSAXHandler.cdataBlock = NULL; - -/* - // XSLT stylesheet manager - cache_managers->put(*NEW String(pool, "stylesheet"), - stylesheet_manager=NEW Stylesheet_manager(pool)); - */ + /* + * 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 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") +static bool is_dlinited=false; + +void pa_globals_done() { + delete cache_managers; cache_managers=0; + if(is_dlinited) + lt_dlexit(); +} + +void pa_dlinit() { + if(!is_dlinited){ + if(lt_dlinit()) + throw Exception(0,0,"prepare to dynamic libary loading failed, %s", lt_dlerror()); + is_dlinited=true; + } +} + +#ifdef _MSC_VER + +#ifndef PA_DEBUG_DISABLE_GC +# define GC_LIB "../../../../win32/gc" # ifdef _DEBUG -# pragma comment(lib, XML_LIBS "/gdome2-0.7.0/win32/dsp/Debug/gdome.lib") +# pragma comment(lib, GC_LIB "/Debug/gc.lib") # else -# pragma comment(lib, XML_LIBS "/gdome2-0.7.0/win32/dsp/Release/gdome.lib") +# pragma comment(lib, GC_LIB "/Release/gc.lib") # endif + #endif +#define GNOME_LIBS "../../../../win32/gnome" + +#ifdef WITH_MAILRECEIVE +# pragma comment(lib, GNOME_LIBS "/glib/lib/libglib-1.3-11.lib") +#endif -#if defined(DB2) && defined(_MSC_VER) -# define LIBDB2_WIN32_BUILD "/parser3project/win32db/lib" +#ifdef XML # 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 + +#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 + # endif #endif + +#endif