--- parser3/src/main/pa_globals.C 2003/02/04 15:15:03 1.152.2.14 +++ parser3/src/main/pa_globals.C 2005/08/05 13:03:01 1.176.8.1 @@ -1,15 +1,16 @@ /** @file Parser: globals. - Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char* IDENT_GLOBALS_C="$Date: 2003/02/04 15:15:03 $"; +static const char * const IDENT_GLOBALS_C="$Date: 2005/08/05 13:03:01 $"; #include "pa_config_includes.h" #ifdef XML +#include "libxml/xmlversion.h" #include "libxslt/extensions.h" #include "libxslt/xsltutils.h" extern "C" { @@ -17,66 +18,28 @@ extern "C" { }; #endif +#include "pcre.h" + #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" -#ifdef XML -#include "pa_stylesheet_manager.h" -#endif - - - - -#define DEFAULTS_NAME "DEFAULTS" - -#define MATCH_VAR_NAME "match" - -#define EXCEPTION_VAR_NAME "exception" -#define EXCEPTION_TYPE_PART_NAME "type" -#define EXCEPTION_SOURCE_PART_NAME "source" -#define EXCEPTION_COMMENT_PART_NAME "comment" -#define EXCEPTION_HANDLED_PART_NAME "handled" - - - - -#define SWITCH_DATA_NAME "SWITCH-DATA" - -#define CACHE_DATA_NAME "CACHE-DATA" - -#define SQL_LIMIT_NAME "limit" -#define SQL_OFFSET_NAME "offset" -#define SQL_DEFAULT_NAME "default" -#define SQL_DISTINCT_NAME "distinct" - -#define HASH_DEFAULT_ELEMENT_NAME "_default" +#include "pa_cache_managers.h" +// defines -StringPtr match_var_name(new String(MATCH_VAR_NAME)); +//#define PA_DEBUG_XML_GC_MEMORY -StringPtr exception_var_name(new String(EXCEPTION_VAR_NAME)); -StringPtr exception_type_part_name(new String(EXCEPTION_TYPE_PART_NAME)); -StringPtr exception_source_part_name(new String(EXCEPTION_SOURCE_PART_NAME)); -StringPtr exception_comment_part_name(new String(EXCEPTION_COMMENT_PART_NAME)); -StringPtr exception_handled_part_name(new String(EXCEPTION_HANDLED_PART_NAME)); +// 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 - -//^switch ^case -StringPtr switch_data_name(new String(SWITCH_DATA_NAME)); - -//^cache -StringPtr cache_data_name(new String(CACHE_DATA_NAME)); - -// sql -StringPtr sql_limit_name(new String(SQL_LIMIT_NAME)); -StringPtr sql_offset_name(new String(SQL_OFFSET_NAME)); -StringPtr sql_default_name(new String(SQL_DEFAULT_NAME)); -StringPtr sql_distinct_name(new String(SQL_DISTINCT_NAME)); - -// hash -StringPtr hash_default_element_name(new String(HASH_DEFAULT_ELEMENT_NAME)); +// globals short hex_value[0x100]; @@ -110,170 +73,261 @@ static void setup_hex_value() { hex_value['f'] = 15; } -#ifdef XML -const int MAX_CONCURRENT_XML_GENERIC_ERROR_THREADS=10; +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()); +} -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; +#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 -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 +#ifdef XML - // 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; - } +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; + } +}; - 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); +static Hash xml_generic_error_infos; - break; +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(); + pa_thread_t thread_id=pa_get_thread_id(); SYNCHRONIZED; // find blocked - return xml_generic_error_info(thread_id)!=0; + return xml_generic_error_infos.get(thread_id)!=0; } const char* xmlGenericErrors() { - pa_thread_t thread_id=pa_get_thread_id(); + 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; + if(XML_Generic_error_info *p=xml_generic_error_infos.get(thread_id)) { + xml_generic_error_infos.remove(thread_id); + return p->get(); + } - const char* result=p->message; + return 0; // no errors for our thread_id registered +} - // free slot up - memset(p, 0, sizeof(*p)); +#endif - // it is up to caller to free it - return result; +void pa_globals_destroy(void *) { + try { +#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::abort("pa_globals_destroy failed: %s", e.comment()); + } } -/** - * 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); -} +#ifdef XML -/** - * 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]); +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; +} -#ifdef WIN32 - fd = fopen(path, "rb"); +#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 - fd = fopen(path, "r"); -#endif /* WIN32 */ - return((void *) fd); + 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 -/** - * 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 -pa_xmlFileRead (void * context, char * buffer, int len) { - return(fread(&buffer[0], 1, len, (FILE *) context)); +inline void *check(void *result, const char *where, size_t size) { + if(!result) + SAPI::abort("out of memory (while %s [size=%d])", where, size); + + 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) { +#ifdef PA_WORKAROUND_BUGGY_MALLOCATOMIC_IN_LIBXML_GC_MEMORY + return check(pa_gc_malloc(size), "allocating XML composite memory (asked atomic)", size); +#else + return check(pa_gc_malloc_atomic(size), "allocating XML atomic memory", size); +#endif +} +static void* pa_gc_realloc_nonull(void* ptr, size_t size) { + return check(pa_gc_realloc(ptr, size), "reallocating XML memory", size); } -/** - * xmlFileClose: - * @context: the I/O context - * - * Close an I/O channel - */ -static int -pa_xmlFileClose (void * context) { - return ( ( fclose((FILE *) context) == EOF ) ? -1 : 0 ); +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_globals_destroy(void *) { - try { +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 - GdomeException exc; - gdome_di_unref (domimpl, &exc); + // 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 - } catch(const Exception& e) { - SAPI::die("pa_globals_destroy failed: %s", e.comment()); - } -} + // pcre + pcre_malloc=pa_gc_malloc; + pcre_free=pa_gc_free; -/// @test hint on one should call this for each thread xmlSubstituteEntitiesDefault(1); + + // 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(); @@ -281,19 +335,19 @@ void pa_globals_init() { #ifdef XML // initializing xml libs - /* First I get a DOMImplementation reference */ + /* 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; - + /* + * 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 @@ -314,7 +368,7 @@ void pa_globals_init() { 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; + xmlLoadExtDtdDefaultValue |= XML_COMPLETE_ATTRS; // validate each document after load/create (?) //xmlDoValidityCheckingDefaultValue = 1; @@ -322,35 +376,85 @@ void pa_globals_init() { //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, - pa_xmlFileRead, pa_xmlFileClose); - - // XSLT stylesheet manager - cache_managers->put(*NEW String(pool, "stylesheet"), - stylesheet_manager=NEW Stylesheet_manager(pool)); + pa_xml_io_init(); #endif } -#if defined(XML) && defined(_MSC_VER) -# define GNOME_LIBS "/parser3project/win32xml/win32/gnome" -# pragma comment(lib, GNOME_LIBS "/glib/lib/libglib-1.3-11.lib") +void pa_globals_done() { + delete cache_managers; cache_managers=0; +} + +#ifdef _MSC_VER + +#ifndef PA_DEBUG_DISABLE_GC +# define GC_LIB "../../../../win32/gc" # ifdef _DEBUG -# pragma comment(lib, GNOME_LIBS "/libxml2-x.x.x/win32/dsp/libxml2_so_debug/libxml2.lib") -# pragma comment(lib, GNOME_LIBS "/libxslt-x.x.x/win32/dsp/libexslt_so_debug/libexslt.lib") -# pragma comment(lib, GNOME_LIBS "/libxslt-x.x.x/win32/dsp/libxslt_so_debug/libxslt.lib") -# pragma comment(lib, GNOME_LIBS "/gdome2-x.x.x/win32/dsp/Debug/libgdome.lib") +# pragma comment(lib, GC_LIB "/Debug/gc.lib") # else -# pragma comment(lib, GNOME_LIBS "/libxml2-x.x.x/win32/dsp/libxml2_so_release/libxml2.lib") -# pragma comment(lib, GNOME_LIBS "/libxslt-x.x.x/win32/dsp/libexslt_so_release/libexslt.lib") -# pragma comment(lib, GNOME_LIBS "/libxslt-x.x.x/win32/dsp/libxslt_so_release/libxslt.lib") -# pragma comment(lib, GNOME_LIBS "/gdome2-x.x.x/win32/dsp/Release/libgdome.lib") +# pragma comment(lib, GC_LIB "/Release/gc.lib") +# endif + +#endif + +#ifdef XML +# define GNOME_LIBS "../../../../win32/gnome" +# pragma comment(lib, GNOME_LIBS "/glib/lib/libglib-1.3-11.lib") +# ifdef _DEBUG + +# 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