--- parser3/src/main/pa_globals.C 2001/09/17 16:49:15 1.74 +++ parser3/src/main/pa_globals.C 2002/08/19 09:26:53 1.138 @@ -1,29 +1,36 @@ /** @file Parser: globals. - Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - - Author: Alexander Petrosyan (http://design.ru/paf) + Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char *RCSId="$Id: pa_globals.C,v 1.74 2001/09/17 16:49:15 parser Exp $"; -#if _MSC_VER -# pragma warning(disable:4291) // disable warning -// "no matching operator delete found; memory will not be freed if initialization throws an exception -#endif +static const char* IDENT_GLOBALS_C="$Date: 2002/08/19 09:26:53 $"; -#include -#include +#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 "pa_string.h" #include "pa_hash.h" #include "pa_sql_driver_manager.h" -#include "pa_xslt_stylesheet_manager.h" #include "pa_dictionary.h" +#include "pa_stylesheet_manager.h" +#include "pa_sapi.h" +#include "pa_cache_managers.h" +#include "pa_charsets.h" +#include "pa_charset.h" +#include "pa_threads.h" -String *user_html_name; String *content_type_name; +String *charset_name; String *body_name; String *value_name; String *expires_name; @@ -32,34 +39,24 @@ 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 *conf_method_name; 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 *exception_var_name; +String *exception_type_part_name; +String *exception_source_part_name; +String *exception_comment_part_name; +String *exception_handled_part_name; + +String *charsets_name; String *mime_types_name; String *vfile_mime_type_name; String *origins_mode_name; @@ -67,20 +64,31 @@ String *origins_mode_name; String *class_path_name; String *switch_data_name; -String *case_default_value; + +String *cache_data_name; String *sql_limit_name; String *sql_offset_name; String *sql_default_name; +String *charset_UTF8_name; + String *hash_default_element_name; +Table *string_match_table_template; + +Table *date_calendar_table_template; + Hash *untaint_lang_name2enum; -Dictionary *default_typo_dict; +Charset *utf8_charset; 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; @@ -107,7 +115,191 @@ static void setup_hex_value() { 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 +xmlFileRead (void * context, char * buffer, int len) { + return(fread(&buffer[0], 1, len, (FILE *) context)); +} + +/** + * xmlFileWrite: + * @context: the I/O context + * @buffer: where to drop data + * @len: number of bytes to write + * + * Write @len bytes from @buffer to the I/O channel. + * + * Returns the number of bytes written + */ +static int +xmlFileWrite (void * context, const char * buffer, int len) { + return(fwrite(&buffer[0], 1, len, (FILE *) context)); +} + +/** + * xmlFileClose: + * @context: the I/O context + * + * Close an I/O channel + */ +static int +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 + if(cache_managers) + cache_managers->~Cache_managers(); + + charsets->~Charsets(); + + } 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(Pool& pool) { + pool.register_cleanup(pa_globals_destroy, 0); + #undef NEW #define NEW new(pool) @@ -115,8 +307,8 @@ void pa_globals_init(Pool& pool) { setup_hex_value(); // names - user_html_name=NEW String(pool, USER_HTML_NAME); content_type_name=NEW String(pool, CONTENT_TYPE_NAME); + charset_name=NEW String(pool, CHARSET_NAME); body_name=NEW String(pool, BODY_NAME); value_name=NEW String(pool, VALUE_NAME); expires_name=NEW String(pool, EXPIRES_NAME); @@ -125,97 +317,175 @@ void pa_globals_init(Pool& pool) { 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); - + conf_method_name=NEW String(pool, CONF_METHOD_NAME); 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); + exception_var_name=NEW String(pool, EXCEPTION_VAR_NAME); + exception_type_part_name=NEW String(pool, EXCEPTION_TYPE_PART_NAME); + exception_source_part_name=NEW String(pool, EXCEPTION_SOURCE_PART_NAME); + exception_comment_part_name=NEW String(pool, EXCEPTION_COMMENT_PART_NAME); + exception_handled_part_name=NEW String(pool, EXCEPTION_HANDLED_PART_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); + charsets_name=NEW String(pool, CHARSETS_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); + + //^cache + cache_data_name=NEW String(pool, CACHE_DATA_NAME); // sql sql_limit_name=NEW String(pool, SQL_LIMIT_NAME); sql_offset_name=NEW String(pool, SQL_OFFSET_NAME); sql_default_name=NEW String(pool, SQL_DEFAULT_NAME); + // charsets + charset_UTF8_name=NEW String(pool, CHARSET_UTF8_NAME); + // hash hash_default_element_name=NEW String(pool, HASH_DEFAULT_ELEMENT_NAME); - // 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-spec", FILE_SPEC); - 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(xml, "xml", XML); - 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; \ + #define ULN(cstr, LANG) \ + untaint_lang_name2enum->put(*NEW String(pool, cstr), (int)String::UL_##LANG); + ULN("as-is", AS_IS); + ULN("file-spec", FILE_SPEC); + ULN("http-header", HTTP_HEADER); + ULN("mail-header", MAIL_HEADER); + ULN("uri", URI); + ULN("table", TABLE); + ULN("sql", SQL); + ULN("js", JS); + ULN("xml", XML); + ULN("html", HTML); + ULN("optimized-html", HTML|String::UL_OPTIMIZE_BIT); + + // string_match_table_template + { + Array *columns=NEW Array(pool); + *columns+=NEW String(pool, STRING_PRE_MATCH_NAME); + *columns+=NEW String(pool, STRING_MATCH_NAME); + *columns+=NEW String(pool, STRING_POST_MATCH_NAME); + for(int i=1; i<=MAX_STRING_MATCH_TABLE_COLUMNS; i++) { + char *column=(char *)pool.malloc(MAX_NUMBER); + snprintf(column, MAX_NUMBER, "%d", i); + *columns+=NEW String(pool, column); // .i column name } - DT_ROW("<", "<"); - DT_ROW(">", ">"); - DT_ROW("\"", """); - DT_ROW("&", "&"); - DT_ROW("\\n\\n", "

"); - DT_ROW("\\n", "
"); - default_typo_dict=NEW Dictionary(*default_typo_table); + string_match_table_template=NEW Table(pool, 0, columns); + } - // SQL driver manager - SQL_driver_manager=NEW SQL_Driver_manager(pool); + // date_calendar_table_template + { + Array *columns=NEW Array(pool); + for(int i=0; i<=6; i++) { + char *column=(char *)pool.malloc(MAX_NUMBER); + snprintf(column, MAX_NUMBER, "%d", i); + *columns+=NEW String(pool, column); // .i column name + } + *columns+=NEW String(pool, DATE_CALENDAR_WEEKNO_NAME); + *columns+=NEW String(pool, DATE_CALENDAR_WEEKYEAR_NAME); + date_calendar_table_template=NEW Table(pool, 0, columns); + } + + // charsets + charsets=NEW Charsets(pool); + charsets->put(*charset_UTF8_name, + utf8_charset=NEW Charset(pool, *charset_UTF8_name, 0/*no file=system*/)); + + + // Status registration, must be initialized before all registrants + cache_managers=NEW Cache_managers(pool); - // Xerces & Xalan + // SQL driver manager + cache_managers->put(*NEW String(pool, "sql"), + SQL_driver_manager=NEW SQL_Driver_manager(pool)); - // Use the static initializers to initialize the Xalan-C++ and Xerces-C++ platforms. - // You must initialize Xerces-C++ once per process - XMLPlatformUtils::Initialize(); - XalanTransformer::initialize(); +#ifdef XML + // initializing xml libs - // XSLT stylesheet driver manager - XSLT_stylesheet_manager=NEW XSLT_Stylesheet_manager(pool); + /* 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, + xmlFileRead, 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