--- parser3/src/main/pa_globals.C 2001/03/14 09:12:05 1.2 +++ parser3/src/main/pa_globals.C 2003/01/31 12:34:37 1.152.2.10 @@ -1,77 +1,402 @@ -/* - 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) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) */ +static const char* IDENT_GLOBALS_C="$Date: 2003/01/31 12:34:37 $"; + +#include "pa_config_includes.h" + +#ifdef XML +#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_string.h" +#include "pa_sapi.h" +#include "pa_threads.h" + +#ifdef XML +#include "pa_stylesheet_manager.h" +#endif + +#define AUTO_FILE_NAME "auto.p" + +#define BODY_NAME "body" +#define DOWNLOAD_NAME "download" + +#define CONF_METHOD_NAME "conf" +#define AUTO_METHOD_NAME "auto" + +#define MAIN_CLASS_NAME "MAIN" + +#define DEFAULTS_NAME "DEFAULTS" +#define CHARSETS_NAME "CHARSETS" +#define MIME_TYPES_NAME "MIME-TYPES" +#define ORIGINS_MODE_NAME "ORIGINS" + +#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 DATE_CALENDAR_WEEKNO_NAME "week" +#define DATE_CALENDAR_WEEKYEAR_NAME "year" + +#define CONTENT_DISPOSITION_NAME "content-disposition" +#define CONTENT_DISPOSITION_VALUE "attachment" +#define CONTENT_DISPOSITION_FILENAME_NAME "filename" + +#define SWITCH_DATA_NAME "SWITCH-DATA" + +#define CACHE_DATA_NAME "CACHE-DATA" + +#define CLASS_PATH_NAME "CLASS_PATH" + +#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" + +#define HTTP_METHOD_NAME "method" +#define HTTP_TIMEOUT_NAME "timeout" +#define HTTP_HEADERS_NAME "headers" +#define HTTP_ANY_STATUS_NAME "any-status" +#define FILE_STATUS_NAME "status" + +StringPtr body_name(new String(BODY_NAME)); +StringPtr download_name(new String(DOWNLOAD_NAME)); + +StringPtr content_disposition_name(new String(CONTENT_DISPOSITION_NAME)); +StringPtr content_disposition_value(new String(CONTENT_DISPOSITION_VALUE)); +StringPtr content_disposition_filename_name(new String(CONTENT_DISPOSITION_FILENAME_NAME)); + +StringPtr conf_method_name(new String(CONF_METHOD_NAME)); +StringPtr auto_method_name(new String(AUTO_METHOD_NAME)); + +StringPtr main_class_name(new String(MAIN_CLASS_NAME)); + +StringPtr match_var_name(new String(MATCH_VAR_NAME)); + +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)); + +StringPtr charsets_name(new String(CHARSETS_NAME)); +StringPtr mime_types_name(new String(MIME_TYPES_NAME)); +StringPtr origins_mode_name(new String(ORIGINS_MODE_NAME)); + +StringPtr class_path_name(new String(CLASS_PATH_NAME)); + +//^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)); + +// http +StringPtr http_method_name(new String(HTTP_METHOD_NAME)); +StringPtr http_timeout_name(new String(HTTP_TIMEOUT_NAME)); +StringPtr http_headers_name(new String(HTTP_HEADERS_NAME)); +StringPtr http_any_status_name(new String(HTTP_ANY_STATUS_NAME)); +StringPtr file_status_name(new String(FILE_STATUS_NAME)); + +short hex_value[0x100]; + +#ifdef XML +GdomeDOMImplementation *domimpl; +#endif + +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['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; +} + +#ifdef XML + +const int MAX_CONCURRENT_XML_GENERIC_ERROR_THREADS=10; + +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; +} + +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 + + // 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; + } + + 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); + + break; + } +} + +bool xmlHaveGenericErrors() { + pa_thread_t thread_id=pa_get_thread_id(); + + SYNCHRONIZED; // find blocked + + return xml_generic_error_info(thread_id)!=0; +} + +const char* xmlGenericErrors() { + 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; + + const char* result=p->message; + + // free slot up + memset(p, 0, sizeof(*p)); + + // it is up to caller to free it + return result; +} + +/** + * 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); +} + + +/** + * 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]); + +#ifdef WIN32 + fd = fopen(path, "rb"); +#else + fd = fopen(path, "r"); +#endif /* WIN32 */ + return((void *) fd); +} + +/** + * 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)); +} + +/** + * xmlFileClose: + * @context: the I/O context + * + * Close an I/O channel + */ +static int +pa_xmlFileClose (void * context) { + return ( ( fclose((FILE *) context) == EOF ) ? -1 : 0 ); } + +#endif + +void pa_globals_destroy(void *) { + try { +#ifdef XML + GdomeException exc; + gdome_di_unref (domimpl, &exc); +#endif + } catch(const Exception& e) { + SAPI::die("pa_globals_destroy failed: %s", e.comment()); + } +} + + +/// @test hint on one should call this for each thread xmlSubstituteEntitiesDefault(1); +void pa_globals_init() { + + // 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 + */ + // 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; + + 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)); +#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") +# 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") +# 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") +# endif +#endif