--- parser3/src/main/pa_globals.C 2001/08/06 16:18:26 1.69 +++ parser3/src/main/pa_globals.C 2016/05/24 11:01:19 1.195 @@ -1,83 +1,52 @@ /** @file Parser: globals. - Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - - Author: Alexander Petrosyan (http://design.ru/paf) + Copyright (c) 2001-2015 Art. Lebedev Studio (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char *RCSId="$Id: pa_globals.C,v 1.69 2001/08/06 16:18:26 parser Exp $"; + +#include "pa_config_includes.h" + +#ifdef XML +#include "libxml/xmlversion.h" +#include "libxslt/extensions.h" +#include "libxslt/xsltutils.h" +extern "C" { +#include "libexslt/exslt.h" +} +#endif #include "pa_globals.h" -#include "pa_string.h" -#include "pa_hash.h" -#include "pa_sql_driver_manager.h" -#include "pa_dictionary.h" - -String *user_html_name; -String *content_type_name; -String *body_name; -String *value_name; -String *expires_name; -String *path_name; -String *name_name; -String *size_name; -String *text_name; - -String *exception_method_name; -String *post_process_method_name; - -String *content_disposition_name; -String *content_disposition_filename_name; - -String *unnamed_name; -String *empty_string; - -String *auto_method_name; -String *main_method_name; - -String *main_class_name; - -String *result_var_name; -String *match_var_name; -String *string_pre_match_name; -String *string_match_name; -String *string_post_match_name; - -String *defaults_name; -String *ctype_name; -String *ctype_white_space_name; -String *ctype_digit_name; -String *ctype_hex_digit_name; -String *ctype_letter_name; -String *ctype_word_name; -String *ctype_lowercase_name; -String *mime_types_name; -String *vfile_mime_type_name; -String *origins_mode_name; - -String *class_path_name; - -String *switch_data_name; -String *case_default_value; +#include "pa_sapi.h" +#include "pa_threads.h" +#include "pa_xml_io.h" +#include "pa_common.h" + +#include "pa_cache_managers.h" +#include "ltdl.h" +#include "pcre.h" -Hash *untaint_lang_name2enum; +volatile const char * IDENT_PA_GLOBALS_C="$Id: pa_globals.C,v 1.195 2016/05/24 11:01:19 moko Exp $" IDENT_PA_GLOBALS_H IDENT_PA_SAPI_H; -Dictionary *default_typo_dict; +// defines -short hex_value[0x100]; +//#define PA_DEBUG_XML_GC_MEMORY + +// globals + +short hex_value[0x100]={0}; static void setup_hex_value() { - memset(hex_value, 0, sizeof(hex_value)); - hex_value['0'] = 0; - hex_value['1'] = 1; - hex_value['2'] = 2; - hex_value['3'] = 3; - hex_value['4'] = 4; - hex_value['5'] = 5; - hex_value['6'] = 6; - hex_value['7'] = 7; - hex_value['8'] = 8; + hex_value['0'] = 0; + hex_value['1'] = 1; + hex_value['2'] = 2; + hex_value['3'] = 3; + hex_value['4'] = 4; + hex_value['5'] = 5; + hex_value['6'] = 6; + hex_value['7'] = 7; + hex_value['8'] = 8; hex_value['9'] = 9; hex_value['A'] = 10; hex_value['B'] = 11; @@ -93,95 +62,293 @@ static void setup_hex_value() { hex_value['f'] = 15; } -void pa_globals_init(Pool& pool) { - #undef NEW - #define NEW new(pool) +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 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; + + 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 // XML + +#ifdef XML + +static char *pa_GC_strdup(const char *s) { + if(!s) + return 0; + + size_t size=strlen(s)+1; + char *result=(char *)pa_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; +} + +#ifdef PA_DEBUG_XML_GC_MEMORY + +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); + 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); + 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); + return p; +} + +static void pa_gc_free_log(void *p){ + fprintf(stderr, "pa_gc_free_log(0x%p)\n", p); + pa_gc_free(p); +} + +#else + +inline void *check(void *result, const char *where, size_t size) { + if(!result) + pa_fail_alloc(where, size); + 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) { + return check(pa_gc_malloc_atomic(size), "allocate XML atomic memory", size); +} + +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* ptr) { + pa_gc_free(ptr); +} + +#endif + +#endif // XML + +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 +#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 + +#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(); - // names - user_html_name=NEW String(pool, USER_HTML_NAME); - content_type_name=NEW String(pool, CONTENT_TYPE_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); - empty_string=NEW String(pool); - - 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); - - - defaults_name=NEW String(pool, DEFAULTS_NAME); - ctype_name=NEW String(pool, CTYPE_NAME); - ctype_white_space_name=NEW String(pool, CTYPE_WHITE_SPACE_NAME); - ctype_digit_name=NEW String(pool, CTYPE_DIGIT_NAME); - ctype_hex_digit_name=NEW String(pool, CTYPE_HEX_DIGIT_NAME); - ctype_letter_name=NEW String(pool, CTYPE_LETTER_NAME); - ctype_word_name=NEW String(pool, CTYPE_WORD_NAME); - ctype_lowercase_name=NEW String(pool, CTYPE_LOWERCASE_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); - - // 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-name", FILE_NAME); - 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(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); +#ifdef XML + // initializing xml libs + + // 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); - // SQL_Driver_manager - SQL_driver_manager=NEW SQL_Driver_manager(pool); +// FILE *f=fopen("xslt.log", "wt"); +// xsltSetGenericDebugFunc(f/*stderr*/, 0); + + pa_xml_io_init(); +#endif +} + +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,"preparation for dynamic library 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, GC_LIB "/Debug/gc.lib") +#else +#pragma comment(lib, GC_LIB "/Release/gc.lib") +#endif + +#endif // PA_DEBUG_DISABLE_GC + + +#ifdef XML + +#define GNOME_LIBS "../../../../win32/gnome" + +#ifdef _DEBUG +#define LIB_XML GNOME_LIBS "/libxml2-x.x.x/win32/debug/lib/" +#define LIB_XSLT GNOME_LIBS "/libxslt-x.x.x/win32/debug/lib/" +#else +#define LIB_XML GNOME_LIBS "/libxml2-x.x.x/win32/release/lib/" +#define LIB_XSLT GNOME_LIBS "/libxslt-x.x.x/win32/release/lib/" +#endif + +#ifdef LIBXML_STATIC +#pragma comment(lib, LIB_XML "libxml2_a.lib") +#else +#pragma comment(lib, LIB_XML "libxml2.lib") +#endif + +#ifdef LIBXSLT_STATIC +#pragma comment(lib, LIB_XSLT "libxslt_a.lib") +#else +#pragma comment(lib, LIB_XSLT "libxslt.lib") +#endif + +#ifdef LIBEXSLT_STATIC +#pragma comment(lib, LIB_XSLT "libexslt_a.lib") +#else +#pragma comment(lib, LIB_XSLT "libexslt.lib") +#endif + +#endif // XML + +#endif // _MSC_VER