--- parser3/src/main/pa_globals.C	2001/04/10 10:32:10	1.53
+++ parser3/src/main/pa_globals.C	2002/05/07 07:23:10	1.124.2.2
@@ -1,33 +1,36 @@
 /** @file
 	Parser: globals.
 
-	Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com)
+	Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com)
+	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
 
-	Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)
-
-	$Id: pa_globals.C,v 1.53 2001/04/10 10:32:10 paf Exp $
+	$Id: pa_globals.C,v 1.124.2.2 2002/05/07 07:23:10 paf Exp $
 */
 
-#include "pcre.h"
+#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_table.h"
-#include "_string.h"
-#include "_double.h"
-#include "_int.h"
-#include "_table.h"
-#include "_form.h"
-#include "_response.h"
-#include "_file.h"
-#include "_random.h"
 #include "pa_sql_driver_manager.h"
-#include "_mail.h"
-#include "_image.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 *html_typo_name;
 String *content_type_name;
+String *charset_name;
 String *body_name;
 String *value_name;
 String *expires_name;
@@ -36,62 +39,55 @@ 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 *op_class_name;
 String *main_class_name;
-String *env_class_name;
-String *table_class_name;
-String *file_class_name;
-String *form_class_name;
-String *request_class_name;
-String *response_class_name;
-String *cookie_class_name;
-String *random_class_name;
-String *mail_class_name;
-String *image_class_name;
 
 String *result_var_name;
-String *string_pre_match_name;
-String *string_match_name;
-String *string_post_match_name;
-
-String *limits_name;
-String *post_max_size_name;
-
-String *defaults_name;
-String *locale_name;
-String *locale_ctype_name;
+String *match_var_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 *mail_name;
+String *origins_mode_name;
 
-String *main_sql_name;
-String *main_sql_drivers_name;
+String *class_path_name;
 
-Hash *untaint_lang_name2enum;
+String *switch_data_name;
+
+String *cache_data_name;
+
+String *sql_limit_name;
+String *sql_offset_name;
+String *sql_default_name;
+
+String *charset_UTF8_name;
 
-Table *default_typo_table;
+String *hash_default_element_name;
 
-const unsigned char *pcre_tables;
+Table *string_match_table_template;
+
+Hash *untaint_lang_name2enum;
+
+Charset *utf8_charset;
 
 short hex_value[0x100];
 
+#ifdef XML
+GdomeDOMImplementation *domimpl;
+#endif
+
 static void setup_hex_value() {
-	int i;
-	for (i=0; (i < 0x100); i++) {
-		hex_value[i] = 0;
-	}
+	memset(hex_value, 0, sizeof(hex_value));
 	hex_value['0'] = 0;	
 	hex_value['1'] = 1;	
 	hex_value['2'] = 2;	
@@ -116,7 +112,191 @@ 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; i<MAX_CONCURRENT_XML_GENERIC_ERROR_THREADS; i++) {
+		XML_Generic_error_info *p=xml_generic_error_infos+i;
+		if(p->thread_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)
 
@@ -124,8 +304,8 @@ void pa_globals_init(Pool& pool) {
 	setup_hex_value();
 
 	// names
-	html_typo_name=NEW String(pool, HTML_TYPO_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);
@@ -134,139 +314,159 @@ 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); 
-
 	auto_method_name=NEW String(pool, AUTO_METHOD_NAME);
-	main_method_name=NEW String(pool, MAIN_METHOD_NAME);
 
-	op_class_name=NEW String(pool, OP_CLASS_NAME);
 	main_class_name=NEW String(pool, MAIN_CLASS_NAME);
-	table_class_name=NEW String(pool, TABLE_CLASS_NAME);
-	file_class_name=NEW String(pool, FILE_CLASS_NAME);
-	env_class_name=NEW String(pool, ENV_CLASS_NAME);	
-	form_class_name=NEW String(pool, FORM_CLASS_NAME);	
-	request_class_name=NEW String(pool, REQUEST_CLASS_NAME);	
-	response_class_name=NEW String(pool, RESPONSE_CLASS_NAME);
-	cookie_class_name=NEW String(pool, COOKIE_CLASS_NAME);
-	random_class_name=NEW String(pool, RANDOM_CLASS_NAME);
-	mail_class_name=NEW String(pool, MAIL_CLASS_NAME);
-	image_class_name=NEW String(pool, IMAGE_CLASS_NAME);
 
 	result_var_name=NEW String(pool, RESULT_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);
+	match_var_name=NEW String(pool, MATCH_VAR_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);
 
-	limits_name=NEW String(pool, LIMITS_NAME);
-	post_max_size_name=NEW String(pool, POST_MAX_SIZE_NAME);
-
-	defaults_name=NEW String(pool, DEFAULTS_NAME);
-	locale_name=NEW String(pool, LOCALE_NAME);
-	locale_ctype_name=NEW String(pool, LOCALE_CTYPE_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);
-	mail_name=NEW String(pool, MAIL_NAME);
-	
+	origins_mode_name=NEW String(pool, ORIGINS_MODE_NAME);
 
-	main_sql_name=NEW String(pool, MAIN_SQL_NAME);
-	main_sql_drivers_name=NEW String(pool, MAIN_SQL_DRIVERS_NAME);
+	class_path_name=NEW String(pool, CLASS_PATH_NAME);
 
-	// hashes
-	untaint_lang_name2enum=NEW Hash(pool);
-	String as_is(pool, "as-is");  
-	untaint_lang_name2enum->put(as_is, (int)String::UL_AS_IS);
-	String file_name(pool, "file-name");  
-	untaint_lang_name2enum->put(file_name, (int)String::UL_FILE_NAME);
-	String http_header(pool, "http-header");  
-	untaint_lang_name2enum->put(http_header, (int)String::UL_HTTP_HEADER);
-	String mail_header(pool, "mail-header");  
-	untaint_lang_name2enum->put(mail_header, (int)String::UL_MAIL_HEADER);
-	String uri(pool, "uri");  
-	untaint_lang_name2enum->put(uri, (int)String::UL_URI);
-	String table(pool, "table");
-	untaint_lang_name2enum->put(table, (int)String::UL_TABLE);
-	String sql(pool, "sql");
-	untaint_lang_name2enum->put(sql, (int)String::UL_SQL);
-	String js(pool, "js");
-	untaint_lang_name2enum->put(js, (int)String::UL_JS);
-	String html(pool, "html");
-	untaint_lang_name2enum->put(html, (int)String::UL_HTML);
-	String html_typo(pool, "html-typo");
-	untaint_lang_name2enum->put(html_typo, (int)String::UL_HTML_TYPO);
-
-	// tables
-	default_typo_table=NEW Table(pool, 0, 0);
-	{
-		// < -> &lt;
-		{
-			Array *row=NEW Array(pool);
-			*row+=NEW String(pool, "<");	
-			*row+=NEW String(pool, "&lt;");
-			*default_typo_table+=row;
-		}
-		// > -> &gt;
-		{
-			Array *row=NEW Array(pool);
-			*row+=NEW String(pool, ">");	
-			*row+=NEW String(pool, "&gt;");
-			*default_typo_table+=row;
-		}
-		// " -> &quot;
-		{
-			Array *row=NEW Array(pool);
-			*row+=NEW String(pool, "\"");	
-			*row+=NEW String(pool, "&quot;");
-			*default_typo_table+=row;
-		}
-		// & -> &amp;
-		{
-			Array *row=NEW Array(pool);
-			*row+=NEW String(pool, "&");	
-			*row+=NEW String(pool, "&amp;");
-			*default_typo_table+=row;
-		}
-		// \n\n -> <p>
-		{
-			Array *row=NEW Array(pool);
-			*row+=NEW String(pool, "\\n\\n");	
-			*row+=NEW String(pool, "<p>");
-			*default_typo_table+=row;
-		}
-		// \n -> <br>
-		{
-			Array *row=NEW Array(pool);
-			*row+=NEW String(pool, "\\n");	
-			*row+=NEW String(pool, "<br>");
-			*default_typo_table+=row;
-		}
-	}
+	//^switch ^case
+	switch_data_name=NEW String(pool, SWITCH_DATA_NAME);
 
+	//^cache
+	cache_data_name=NEW String(pool, CACHE_DATA_NAME);
 
-	// pcre tables
-	pcre_tables=pcre_maketables();
+	// 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);
 
-	// stateless classes
-	initialize_string_class(pool, *(string_class=NEW VStateless_class(pool)));
-	initialize_double_class(pool, *(double_class=NEW VStateless_class(pool)));
-	initialize_int_class(pool, *(int_class=NEW VStateless_class(pool)));
-	initialize_table_class(pool, *(table_class=NEW VStateless_class(pool)));
-	initialize_file_class(pool, *(file_class=NEW VStateless_class(pool)));
-	initialize_response_class(pool, *(response_class=NEW VStateless_class(pool)));
-	initialize_random_class(pool, *(random_class=NEW VStateless_class(pool)));
-	initialize_mail_class(pool, *(mail_class=NEW VStateless_class(pool)));
-	initialize_image_class(pool, *(image_class=NEW VStateless_class(pool)));
+	// charsets
+	charset_UTF8_name=NEW String(pool, CHARSET_UTF8_NAME);
 
-	// stateless base classes
-	initialize_form_base_class(pool, *(form_base_class=NEW VStateless_class(pool)));  form_base_class->set_name(*form_class_name);
+	// hash
+	hash_default_element_name=NEW String(pool, HASH_DEFAULT_ELEMENT_NAME);
 
-	// SQL_Driver_manager
- 	SQL_driver_manager=NEW SQL_Driver_manager(pool);
+	// hashes
+	untaint_lang_name2enum=NEW Hash(pool);
+	#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);
+
+	// table
+	{ // create table
+		Array *string_match_table_columns=NEW Array(pool);
+		*string_match_table_columns+=NEW String(pool, STRING_PRE_MATCH_NAME);
+		*string_match_table_columns+=NEW String(pool, STRING_MATCH_NAME);
+		*string_match_table_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);
+			*string_match_table_columns+=NEW String(pool, column); // .i column name
+		}
+		string_match_table_template=NEW Table(pool, 0, string_match_table_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);
+
+	// SQL driver manager
+	cache_managers->put(*NEW String(pool, "sql"), 
+		SQL_driver_manager=NEW SQL_Driver_manager(pool));
+
+#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 
+    //never added yet xmlLoadExtDtdDefaultValue |= XML_COMPLETE_ATTRS;
+
+//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 XML_LIBS "/parser3project/win32xml"
+#	pragma comment(lib, XML_LIBS "/glib/lib/libglib-1.3-11.lib")
+#	ifdef _DEBUG
+#		pragma comment(lib, XML_LIBS "/libxml2-x.x.x/win32/dsp/libxml2_so_debug/libxml2.lib")
+#		pragma comment(lib, XML_LIBS "/libxslt-x.x.x/win32/dsp/libexslt_so_debug/libexslt.lib")
+#		pragma comment(lib, XML_LIBS "/libxslt-x.x.x/win32/dsp/libxslt_so_debug/libxslt.lib")
+#		pragma comment(lib, XML_LIBS "/gdome2-x.x.x/win32/dsp/Debug/libgdome.lib")
+#	else
+#		pragma comment(lib, XML_LIBS "/libxml2-x.x.x/win32/dsp/libxml2_so_release/libxml2.lib")
+#		pragma comment(lib, XML_LIBS "/libxslt-x.x.x/win32/dsp/libexslt_so_release/libexslt.lib")
+#		pragma comment(lib, XML_LIBS "/libxslt-x.x.x/win32/dsp/libxslt_so_release/libxslt.lib")
+#		pragma comment(lib, XML_LIBS "/gdome2-x.x.x/win32/dsp/Release/libgdome.lib")
+#	endif
+#endif