|
|
| version 1.69, 2001/08/06 16:18:26 | version 1.216, 2024/11/11 05:28:00 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser: globals. | Parser: globals. |
| Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) | Copyright (c) 2001-2024 Art. Lebedev Studio (http://www.artlebedev.com) |
| Authors: Konstantin Morshnev <moko@design.ru>, Alexandr Petrosian <paf@design.ru> | |
| Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf) | |
| */ | */ |
| static const char *RCSId="$Id$"; | |
| #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_globals.h" |
| #include "pa_string.h" | #include "pa_socks.h" |
| #include "pa_hash.h" | #include "pa_sapi.h" |
| #include "pa_sql_driver_manager.h" | #include "pa_xml_io.h" |
| #include "pa_dictionary.h" | #include "pa_common.h" |
| #include "pa_symbols.h" | |
| String *user_html_name; | |
| String *content_type_name; | #include "pa_cache_managers.h" |
| String *body_name; | |
| String *value_name; | #include "ltdl.h" |
| String *expires_name; | #include "pa_vregex.h" |
| String *path_name; | |
| String *name_name; | volatile const char * IDENT_PA_GLOBALS_C="$Id$" IDENT_PA_GLOBALS_H IDENT_PA_SAPI_H; |
| String *size_name; | |
| String *text_name; | // defines |
| String *exception_method_name; | //#define PA_DEBUG_XML_GC_MEMORY |
| String *post_process_method_name; | |
| // globals | |
| String *content_disposition_name; | |
| String *content_disposition_filename_name; | short hex_value[0x100] = { |
| 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
| String *unnamed_name; | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
| String *empty_string; | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
| 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 0, 0, 0, 0, 0, 0, | |
| String *auto_method_name; | 0, 10, 11, 12, 13, 14, 15, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
| String *main_method_name; | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
| 0, 10, 11, 12, 13, 14, 15, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
| String *main_class_name; | }; |
| String *result_var_name; | THREAD_LOCAL Request* thread_request=NULL; // every TLS should be referenced elsewhere, or GC will collect it |
| String *match_var_name; | |
| String *string_pre_match_name; | void pa_register_thread_request(Request& r) { |
| String *string_match_name; | thread_request=&r; |
| String *string_post_match_name; | } |
| String *defaults_name; | /// retrives request set by pa_set_request function, useful in contextless places [slow] |
| String *ctype_name; | Request& pa_thread_request() { |
| String *ctype_white_space_name; | if(!thread_request) |
| String *ctype_digit_name; | throw Exception(0, 0, "request is not initialized"); |
| String *ctype_hex_digit_name; | return *thread_request; |
| String *ctype_letter_name; | } |
| String *ctype_word_name; | |
| String *ctype_lowercase_name; | |
| String *mime_types_name; | #ifdef XML |
| String *vfile_mime_type_name; | |
| String *origins_mode_name; | class XML_Generic_error_info { |
| public:/*internal, actually*/ | |
| String *class_path_name; | char buf[MAX_STRING*5]; |
| size_t used; | |
| String *switch_data_name; | public: |
| String *case_default_value; | XML_Generic_error_info() { |
| buf[used=0]=0; | |
| } | |
| Hash *untaint_lang_name2enum; | const char* get() { |
| return used? buf: 0; | |
| Dictionary *default_typo_dict; | } |
| }; | |
| short hex_value[0x100]; | |
| THREAD_LOCAL XML_Generic_error_info* xml_generic_error_info = NULL; | |
| static void setup_hex_value() { | |
| memset(hex_value, 0, sizeof(hex_value)); | static void xmlParserGenericErrorFunc(void * /*ctx*/, const char* msg, ...) { |
| hex_value['0'] = 0; | XML_Generic_error_info* p; |
| hex_value['1'] = 1; | |
| hex_value['2'] = 2; | if(!(p=xml_generic_error_info)) // occupy empty one |
| hex_value['3'] = 3; | p=xml_generic_error_info=new(PointerFreeGC) XML_Generic_error_info; |
| hex_value['4'] = 4; | |
| hex_value['5'] = 5; | va_list args; |
| hex_value['6'] = 6; | va_start(args, msg); |
| hex_value['7'] = 7; | p->used+=vsnprintf(p->buf+p->used, sizeof(p->buf)-p->used, msg, args); |
| hex_value['8'] = 8; | va_end(args); |
| hex_value['9'] = 9; | } |
| hex_value['A'] = 10; | |
| hex_value['B'] = 11; | bool xmlHaveGenericErrors() { |
| hex_value['C'] = 12; | return xml_generic_error_info!=0; |
| hex_value['D'] = 13; | } |
| hex_value['E'] = 14; | |
| hex_value['F'] = 15; | const char* xmlGenericErrors() { |
| hex_value['a'] = 10; | if(XML_Generic_error_info *p=xml_generic_error_info) { |
| hex_value['b'] = 11; | xml_generic_error_info=0; |
| hex_value['c'] = 12; | return p->get(); |
| hex_value['d'] = 13; | } |
| hex_value['e'] = 14; | |
| hex_value['f'] = 15; | return 0; // no errors for our thread_id registered |
| } | } |
| void pa_globals_init(Pool& pool) { | #endif // XML |
| #undef NEW | |
| #define NEW new(pool) | #ifdef XML |
| // hex value | static char *pa_GC_strdup(const char *s) { |
| setup_hex_value(); | if(!s) |
| return 0; | |
| // names | |
| user_html_name=NEW String(pool, USER_HTML_NAME); | size_t size=strlen(s)+1; |
| content_type_name=NEW String(pool, CONTENT_TYPE_NAME); | char *result=(char *)GC_MALLOC_ATOMIC(size); |
| body_name=NEW String(pool, BODY_NAME); | |
| value_name=NEW String(pool, VALUE_NAME); | if(!result) |
| expires_name=NEW String(pool, EXPIRES_NAME); | pa_fail_alloc("duplicate XML string",size); |
| path_name=NEW String(pool, PATH_NAME); | |
| name_name=NEW String(pool, NAME_NAME); | memcpy(result, s, size); |
| size_name=NEW String(pool, SIZE_NAME); | #ifdef PA_DEBUG_XML_GC_MEMORY |
| text_name=NEW String(pool, TEXT_NAME); | fprintf(stderr, "pa_GC_strdup(%p=%s, length=%d)=0x%p\n", s, s, size, result); |
| #endif | |
| exception_method_name=NEW String(pool, EXCEPTION_METHOD_NAME); | return result; |
| post_process_method_name=NEW String(pool, POST_PROCESS_METHOD_NAME); | } |
| content_disposition_name=NEW String(pool, CONTENT_DISPOSITION_NAME); | #ifdef PA_DEBUG_XML_GC_MEMORY |
| content_disposition_filename_name=NEW String(pool, CONTENT_DISPOSITION_FILENAME_NAME); | |
| static void* pa_gc_malloc_log(size_t size){ | |
| unnamed_name=NEW String(pool, UNNAMED_NAME); | void *p=GC_MALLOC(size); |
| empty_string=NEW String(pool); | fprintf(stderr, "pa_gc_malloc_log(%d)=0x%p\n", size, p); |
| return p; | |
| 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); | static void* pa_gc_malloc_atomic_log(size_t size){ |
| void *p=GC_MALLOC_ATOMIC(size); | |
| result_var_name=NEW String(pool, RESULT_VAR_NAME); | fprintf(stderr, "pa_gc_malloc_atomic_log(%d)=0x%p\n", size, p); |
| match_var_name=NEW String(pool, MATCH_VAR_NAME); | return p; |
| 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); | static void* pa_gc_realloc_log(void *ptr, size_t size){ |
| void *p=GC_REALLOC(ptr, size); | |
| fprintf(stderr, "pa_gc_realloc_log(0x%p, %d)=0x%p\n", ptr, size, p); | |
| defaults_name=NEW String(pool, DEFAULTS_NAME); | return p; |
| 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); | static void pa_gc_free_log(void *p){ |
| ctype_hex_digit_name=NEW String(pool, CTYPE_HEX_DIGIT_NAME); | fprintf(stderr, "pa_gc_free_log(0x%p)\n", p); |
| ctype_letter_name=NEW String(pool, CTYPE_LETTER_NAME); | GC_FREE(p); |
| 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); | #else |
| vfile_mime_type_name=NEW String(pool, VFILE_MIME_TYPE_NAME); | |
| origins_mode_name=NEW String(pool, ORIGINS_MODE_NAME); | inline void *check(void *result, const char *where, size_t size) { |
| if(!result) | |
| class_path_name=NEW String(pool, CLASS_PATH_NAME); | pa_fail_alloc(where, size); |
| return result; | |
| //^switch ^case | } |
| switch_data_name=NEW String(pool, SWITCH_DATA_NAME); | |
| case_default_value=NEW String(pool, CASE_DEFAULT_VALUE); | static void* pa_gc_malloc_nonull(size_t size) { |
| return check(GC_MALLOC(size), "allocate XML compsite memory", size); | |
| // hashes | } |
| untaint_lang_name2enum=NEW Hash(pool); | |
| #define ULN(var, cstr, LANG) \ | static void* pa_gc_malloc_atomic_nonull(size_t size) { |
| String var(pool, cstr); \ | return check(GC_MALLOC_ATOMIC(size), "allocate XML atomic memory", size); |
| untaint_lang_name2enum->put(var, (int)String::UL_##LANG); | } |
| ULN(as_is, "as-is", AS_IS); | |
| ULN(file_name, "file-name", FILE_NAME); | static void* pa_gc_realloc_nonull(void* ptr, size_t size) { |
| ULN(http_header, "http-header", HTTP_HEADER); | return check(GC_REALLOC(ptr, size), "reallocate XML memory", size); |
| ULN(mail_header, "mail-header", MAIL_HEADER); | } |
| ULN(uri, "uri", URI); | |
| ULN(table, "table", TABLE); | static void pa_gc_free_maybeignore(void* ptr) { |
| ULN(sql, "sql", SQL); | GC_FREE(ptr); |
| ULN(js, "js", JS); | } |
| ULN(html, "html", HTML); | |
| ULN(user_html, "user-html", USER_HTML); | #endif |
| // tables | #endif // XML |
| Table *default_typo_table=NEW Table(pool, 0, 0); | |
| #define DT_ROW(from, to) { \ | #ifdef HAVE_PCRE2 |
| Array *row=NEW Array(pool); \ | pcre2_general_context* VRegex::fgen_ctxt; |
| *row+=NEW String(pool, from); \ | |
| *row+=NEW String(pool, to); \ | static void* pa_pcre_malloc(size_t size, void *ptr){ |
| *default_typo_table+=row; \ | return pa_malloc(size); |
| } | } |
| DT_ROW("<", "<"); | |
| DT_ROW(">", ">"); | static void pa_pcre_free(void *ptr, void *tag){ |
| DT_ROW("\"", """); | pa_free(ptr); |
| DT_ROW("&", "&"); | } |
| DT_ROW("\\n\\n", "<p>"); | #endif |
| DT_ROW("\\n", "<br>"); | |
| default_typo_dict=NEW Dictionary(*default_typo_table); | |
| 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 | |
| #ifdef HAVE_PCRE2 | |
| VRegex::fgen_ctxt=pcre2_general_context_create(pa_pcre_malloc, pa_pcre_free, NULL); | |
| #else | |
| pcre_malloc=pa_malloc; | |
| pcre_free=pa_free; | |
| #endif | |
| // cord | |
| CORD_oom_fn=pa_CORD_oom_fn; | |
| } | |
| /** | |
| @test hint on one should call this for each thread xmlSubstituteEntitiesDefault(1); | |
| */ | |
| void pa_globals_init() { | |
| #ifndef PA_DEBUG_DISABLE_GC | |
| // Dont collect unless explicitly requested. This is quicker (~30% ), | |
| // but less memory-efficient(~8%), so deciding for speed. | |
| GC_disable(); | |
| // as we log allocation errors, we don't need default gc warnings (without timestamp and URI) | |
| GC_set_warn_proc(GC_ignore_warn_proc); | |
| #endif | |
| // init socks | |
| pa_socks_init(); | |
| // global variables | |
| cache_managers=new Cache_managers; | |
| // in various libraries | |
| gc_substitute_memory_management_functions(); | |
| #ifdef SYMBOLS_CACHING | |
| // symbols cache | |
| Symbols::init(); | |
| #endif | |
| #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); | |
| // FILE *f=fopen("xslt.log", "wt"); | |
| // xsltSetGenericDebugFunc(f/*stderr*/, 0); | |
| pa_xml_io_init(); | |
| #endif | |
| // SQL_Driver_manager | // init *_class variables |
| SQL_driver_manager=NEW SQL_Driver_manager(pool); | methoded_array(); |
| } | } |
| static bool is_dlinited=false; | |
| void pa_globals_done() { | |
| delete cache_managers; | |
| cache_managers=0; | |
| if(is_dlinited) | |
| lt_dlexit(); | |
| pa_socks_done(); | |
| } | |
| 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 | |
| #define PREFIX "../../../../win32/" | |
| #ifdef _DEBUG | |
| #define CONFIGURATION "Debug" | |
| #else | |
| #define CONFIGURATION "Release" | |
| #endif | |
| #ifdef _WIN64 | |
| #define PLATFORM_64 "x64/" | |
| #define PLATFORM_32 "" | |
| #else | |
| #define PLATFORM_64 "" | |
| #define PLATFORM_32 "win32/" | |
| #endif | |
| #pragma comment(lib, PREFIX "pcre/" PLATFORM_64 CONFIGURATION "/pcre.lib") | |
| #ifndef PA_DEBUG_DISABLE_GC | |
| #pragma comment(lib, PREFIX "gc/" PLATFORM_64 CONFIGURATION "/gc.lib") | |
| #endif // PA_DEBUG_DISABLE_GC | |
| #ifdef XML | |
| #define GNOME_LIBS PREFIX "gnome" | |
| #define LIB_XML PREFIX "gnome/libxml2-x.x.x/" PLATFORM_64 PLATFORM_32 CONFIGURATION "/lib/" | |
| #define LIB_XSLT PREFIX "gnome/libxslt-x.x.x/" PLATFORM_64 PLATFORM_32 CONFIGURATION "/lib/" | |
| #ifdef XML_STATIC | |
| #pragma comment(lib, LIB_XML "libxml2_a.lib") | |
| #pragma comment(lib, LIB_XSLT "libxslt_a.lib") | |
| #pragma comment(lib, LIB_XSLT "libexslt_a.lib") | |
| #else | |
| #pragma comment(lib, LIB_XML "libxml2.lib") | |
| #pragma comment(lib, LIB_XSLT "libxslt.lib") | |
| #pragma comment(lib, LIB_XSLT "libexslt.lib") | |
| #endif | |
| #endif // XML | |
| // defines for VS2015 to link with gc/xml libs compiled in the previous VS versions | |
| #if _MSC_VER >= 1900 | |
| #pragma comment(lib,"legacy_stdio_definitions.lib") | |
| #endif | |
| #endif // _MSC_VER |