--- parser3/src/main/pa_globals.C 2001/03/14 09:12:05 1.2 +++ parser3/src/main/pa_globals.C 2017/02/07 22:00:43 1.198 @@ -1,77 +1,365 @@ -/* - Parser - Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - Author: Alexander Petrosyan (http://design.ru/paf) +/** @file + Parser: globals. - $Id: pa_globals.C,v 1.2 2001/03/14 09:12:05 paf Exp $ + Copyright (c) 2001-2017 Art. Lebedev Studio (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) */ +#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 "_string.h" -#include "_double.h" -#include "_int.h" -#include "_table.h" -#include "_form.h" -#include "_env.h" - -String *exception_method_name; - -String *unnamed_name; -String *empty_string; - -String *auto_method_name; -String *main_method_name; - -String *root_class_name; -String *main_class_name; -String *env_class_name; -String *table_class_name; -String *form_class_name; - -Hash *untaint_lang_name2enum; - - -void globals_init(Pool& pool) { - #define NEW_STRING(name, value) name=new(pool) String(pool); name->APPEND_CONST(value) - #define LOCAL_STRING(name, value) String name(pool); name.APPEND_CONST(value) - - // names - NEW_STRING(exception_method_name, EXCEPTION_METHOD_NAME); - - NEW_STRING(unnamed_name, UNNAMED_NAME); - empty_string=new(pool) String(pool); - - NEW_STRING(auto_method_name, AUTO_METHOD_NAME); - NEW_STRING(main_method_name, MAIN_METHOD_NAME); - - NEW_STRING(root_class_name, ROOT_CLASS_NAME); - NEW_STRING(main_class_name, MAIN_CLASS_NAME); - NEW_STRING(table_class_name, TABLE_CLASS_NAME); - NEW_STRING(env_class_name, ENV_CLASS_NAME); - NEW_STRING(form_class_name, FORM_CLASS_NAME); - - // hashes - untaint_lang_name2enum=new(pool) Hash(pool); - LOCAL_STRING(as_is, "as-is"); - untaint_lang_name2enum->put(as_is, (int)String::Untaint_lang::AS_IS); - LOCAL_STRING(table, "table"); - untaint_lang_name2enum->put(table, (int)String::Untaint_lang::TABLE); - LOCAL_STRING(sql, "sql"); - untaint_lang_name2enum->put(sql, (int)String::Untaint_lang::SQL); - LOCAL_STRING(js, "js"); - untaint_lang_name2enum->put(js, (int)String::Untaint_lang::JS); - LOCAL_STRING(html, "html"); - untaint_lang_name2enum->put(html, (int)String::Untaint_lang::HTML); - LOCAL_STRING(html_typo, "html-typo"); - untaint_lang_name2enum->put(html_typo, (int)String::Untaint_lang::HTML_TYPO); - - // read-only stateless classes - initialize_string_class(pool, *(string_class=new(pool) VStateless_class(pool))); string_class->freeze(); - initialize_double_class(pool, *(double_class=new(pool) VStateless_class(pool))); double_class->freeze(); - initialize_int_class(pool, *(int_class=new(pool) VStateless_class(pool))); int_class->freeze(); - initialize_table_class(pool, *(table_class=new(pool) VStateless_class(pool))); table_class->freeze(); - - // read-only stateless base classes - initialize_env_base_class(pool, *(env_base_class=new(pool) VStateless_class(pool))); env_base_class->set_name(*env_class_name); env_base_class->freeze(); - initialize_form_base_class(pool, *(form_base_class=new(pool) VStateless_class(pool))); form_base_class->set_name(*form_class_name); form_base_class->freeze(); +#include "pa_sapi.h" +#include "pa_threads.h" +#include "pa_xml_io.h" +#include "pa_common.h" +#include "pa_symbols.h" + +#include "pa_cache_managers.h" + +#include "ltdl.h" +#include "pcre.h" + +volatile const char * IDENT_PA_GLOBALS_C="$Id: pa_globals.C,v 1.198 2017/02/07 22:00:43 moko Exp $" IDENT_PA_GLOBALS_H IDENT_PA_SAPI_H; + +// defines + +//#define PA_DEBUG_XML_GC_MEMORY + +// globals + +short hex_value[0x100]={0}; + +static void setup_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['9'] = 9; + hex_value['A'] = 10; + hex_value['B'] = 11; + hex_value['C'] = 12; + hex_value['D'] = 13; + hex_value['E'] = 14; + hex_value['F'] = 15; + hex_value['a'] = 10; + hex_value['b'] = 11; + hex_value['c'] = 12; + hex_value['d'] = 13; + hex_value['e'] = 14; + hex_value['f'] = 15; +} + +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(); + +#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 +} + +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 + +// 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") +extern "C" { FILE _iob[3] = { *stdin, *stdout, *stderr }; } +#endif + +#endif // _MSC_VER