--- parser3/src/classes/table.C	2003/04/11 11:47:39	1.172.2.14.2.29
+++ parser3/src/classes/table.C	2004/03/25 11:50:10	1.197
@@ -1,11 +1,11 @@
 /** @file
 	Parser: @b table parser class.
 
-	Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com)
+	Copyright (c) 2001-2004 ArtLebedev Group (http://www.artlebedev.com)
 	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
 */
 
-static const char* IDENT_TABLE_C="$Date: 2003/04/11 11:47:39 $";
+static const char * const IDENT_TABLE_C="$Date: 2004/03/25 11:50:10 $";
 
 #include "classes.h"
 #include "pa_vmethod_frame.h"
@@ -21,7 +21,7 @@ static const char* IDENT_TABLE_C="$Date:
 
 class MTable: public Methoded {
 public: // VStateless_class
-	Value* create_new_value() { return new VTable(); }
+	Value* create_new_value(Pool&) { return new VTable(); }
 
 public:
 	MTable();
@@ -40,6 +40,7 @@ DECLARE_CLASS_VAR(table, new MTable, 0);
 #define SQL_OFFSET_NAME "offset"
 #define SQL_DEFAULT_NAME "default"
 #define SQL_DISTINCT_NAME "distinct"
+#define TABLE_REVERSE_NAME "reverse"
 
 // globals
 
@@ -47,60 +48,90 @@ String sql_limit_name(SQL_LIMIT_NAME);
 String sql_offset_name(SQL_OFFSET_NAME);
 String sql_default_name(SQL_DEFAULT_NAME);
 String sql_distinct_name(SQL_DISTINCT_NAME);
+String table_reverse_name(TABLE_REVERSE_NAME);
 
-// methods
+// local defines
 
-static void get_copy_options(Request& r, MethodParams& params, size_t param_index,
-							 const Table& source,
-							 int& offset,
-							 int& limit) {
-
-	offset=0;
-	limit=0;
-	if(params.count()<=param_index)
-		return;
+#define COLUMN_SEPARATOR_NAME "separator"
+#define COLUMN_ENCLOSER_NAME "encloser"
 
-	Value& voptions=params.as_no_junction(param_index, "options must be hash, not code");
-	if(!voptions.is_string()) {
-		if(HashStringValue* options=voptions.get_hash()) {
-			int valid_options=0;
-			if(Value* voffset=options->get(sql_offset_name)) {
-				valid_options++;
-				if(voffset->is_string()) {
-					const String&  soffset=*voffset->get_string();
-					if(soffset == "cur")
-						offset=source.current();
-					else
-						throw Exception("parser.runtime",
-							&soffset,
-							"must be 'cur' string or expression");
-				} else 
-					offset=r.process_to_value(*voffset).as_int();
-			}
-			if(Value* vlimit=options->get(sql_limit_name)) {
-				valid_options++;
-				limit=r.process_to_value(*vlimit).as_int();
-				if(!limit) // zero limit = should be 'nothing to copy', for methods zero means 'all'
-					limit=-1; // thus fixing
-			}
-			if(valid_options!=options->count())
+
+// methods
+
+static Table::Action_options get_action_options(Request& r, MethodParams& params, 
+						const Table& source) {
+	Table::Action_options result;
+	if(!params.count())
+		return result;
+
+	Value& maybe_options=params.last();
+/* can not do it: 
+	want to enable ^table::create[$source;
+#		$.option[]
+	]
+	but there is ^table.locate[name;value]
+
+	if(maybe_options.is_string()) { // allow empty options
+		result.defined=true;
+		return result;
+	}
+*/
+	HashStringValue* options=maybe_options.get_hash();
+	if(!options)
+		return result;
+
+	result.defined=true;
+	bool defined_offset=false;
+
+	int valid_options=0;
+	if(Value* voffset=options->get(sql_offset_name)) {
+		valid_options++;
+		defined_offset=true;
+		if(voffset->is_string()) {
+			const String&  soffset=*voffset->get_string();
+			if(soffset == "cur")
+				result.offset=source.current();
+			else
 				throw Exception("parser.runtime",
-					0,
-					"called with invalid option");
-		} else
-			throw Exception("parser.runtime",
-				0,
-				"options must be hash");
+					&soffset,
+					"must be 'cur' string or expression");
+		} else 
+			result.offset=r.process_to_value(*voffset).as_int();
 	}
+	if(Value* vlimit=options->get(sql_limit_name)) {
+		valid_options++;
+		result.limit=r.process_to_value(*vlimit).as_int();
+	}
+	if(Value *vreverse=(Value *)options->get(table_reverse_name)) { 
+		valid_options++; 
+		result.reverse=r.process_to_value(*vreverse).as_bool(); 
+		if(result.reverse && !defined_offset) 
+			result.offset=source.count()-1; 
+	} 
+
+	if(valid_options!=options->count())
+		throw Exception("parser.runtime",
+			0,
+			"called with invalid option");
+
+	return result;
+}
+static void check_option_param(bool options_defined, 
+			  MethodParams& params, size_t next_param_index,
+			  const char *msg) {
+	if(next_param_index+(options_defined?1:0) != params.count())
+		throw Exception("parser.runtime",
+			0,
+			"%s", msg);
 }
 
 static void _create(Request& r, MethodParams& params) {
 	// clone/copy part?
 	if(Table *source=params[0].get_table()) {
-		int offset, limit;
-		get_copy_options(r, params, 1, *source, 
-			offset, limit);
-		GET_SELF(r, VTable).set_table(*new Table(*source, offset, limit));
+		Table::Action_options o=get_action_options(r, params, *source);
+		check_option_param(o.defined, params, 1, 
+			"too many parameters");
+		GET_SELF(r, VTable).set_table(*new Table(*source, o));
 		return;
 	}
 
@@ -147,6 +178,118 @@ static void _create(Request& r, MethodPa
 	GET_SELF(r, VTable).set_table(table);
 }
 
+struct lsplit_result {
+	char* piece;
+	char delim;
+
+	operator bool() { return piece!=0; }
+};
+
+inline lsplit_result lsplit(char* string, char delim1, char delim2) {
+	lsplit_result result;
+    if(string) {
+		char delims[]={delim1, delim2, 0};
+		if(char* v=strpbrk(string, delims)) {
+			result.delim=*v;
+			*v=0;
+			result.piece=v+1;
+			return result;
+		}
+    }
+	result.piece=0;
+	result.delim=0;
+    return result;
+}
+
+inline lsplit_result lsplit(char* *string_ref, char delim1, char delim2) {
+    lsplit_result result;
+	result.piece=*string_ref;
+	lsplit_result next=lsplit(*string_ref, delim1, delim2);
+	result.delim=next.delim;
+	*string_ref=next.piece;
+    return result;
+}
+
+static lsplit_result lsplit(char** string_ref, char delim1, char delim2, char encloser) {
+	lsplit_result result;
+
+    if(char* string=*string_ref) {
+		if(encloser && *string==encloser) {
+            string++;
+			
+			char *read;
+			char *write;
+			write=read=string;
+			char c;
+			while((c=*read++)) {
+				if(c==encloser) {
+					char n=*read;
+					if(n==encloser) // double-encloser stands for encloser
+						read++;
+					else if(n==delim1 || n==delim2) {
+						result.delim=n;
+						read++;
+						break;
+					}
+				}
+
+				*write++=c;
+			}
+			*write=0; // terminate
+			*string_ref=c? read: 0;
+			result.piece=string;
+			return result;
+		} else
+			return lsplit(string_ref, delim1, delim2);
+    }
+	result.piece=0;
+    return result;
+}
+
+static void skip_empty_and_comment_lines( char** data_ref ) {
+	if(char *data=*data_ref) {
+		while( char c=*data ) {
+			if( c== '\n' || c == '#' ) {
+				/*nowhere=*/getrow(&data); // remove empty&comment lines
+				*data_ref=data;
+				continue;
+			}
+			break;
+		}
+	}
+}
+
+struct TableSeparators {
+	char column;
+	char encloser;
+	const String* sencloser;
+
+	TableSeparators() {
+		column='\t';
+		encloser=0;
+	}
+	void load( HashStringValue& options ) {
+		if(Value* vseparator=options.get(COLUMN_SEPARATOR_NAME)) {
+			options.remove(COLUMN_SEPARATOR_NAME);
+			const String& sseparator=vseparator->as_string();
+			if(sseparator.length()!=1)
+				throw Exception("parser.runtime",
+					&sseparator,
+					"separator must be one character long");
+			column=sseparator.first_char();
+		}
+		if(Value* vencloser=options.get(COLUMN_ENCLOSER_NAME)) {
+			options.remove(COLUMN_ENCLOSER_NAME);
+			sencloser=&vencloser->as_string();
+			if(sencloser->length()!=1)
+				throw Exception("parser.runtime",
+					sencloser,
+					"encloser must be one character long");
+			encloser=sencloser->first_char();
+		}
+	}
+};
+
 static void _load(Request& r, MethodParams& params) {
 	const String&  first_param=params.as_string(0, "file name must be string");
 	int filename_param_index=0;
@@ -154,12 +297,21 @@ static void _load(Request& r, MethodPara
 	if(nameless)
 		filename_param_index++;
 	size_t options_param_index=filename_param_index+1;
-	
+
+	HashStringValue *options=0;
+	TableSeparators separators;
+	if(options_param_index<params.count()
+		&& (options=params.as_no_junction(options_param_index, "additional params must be hash").get_hash())) {
+		// cloning, so that we could change [to simplify checks on params inside file_read_text
+		options=new HashStringValue(*options);
+		separators.load(*options);
+	}
+
 	// loading text
 	char *data=file_read_text(r.charsets,
 		r.absolute(params.as_string(filename_param_index, "file name must be string")),
 		true,
-		options_param_index<params.count()?params.as_no_junction(options_param_index, "additional params must be hash").get_hash():0
+		options
 	);
 
 	// parse columns
@@ -169,81 +321,123 @@ static void _load(Request& r, MethodPara
 	} else {
 		columns=Table::columns_type(new ArrayString);
 
-		while(char *row_chars=getrow(&data)) {
-			// remove empty&comment lines
-			if(!*row_chars || *row_chars == '#')
-				continue;
-			do {
-				*columns+=new String(lsplit(&row_chars, '\t'), 0, true);
-			} while(row_chars);
-
-			break;
+		skip_empty_and_comment_lines(&data);
+		while( lsplit_result sr=lsplit(&data, separators.column, '\n', separators.encloser) ) {
+			*columns+=new String(sr.piece, 0, true);
+			if(sr.delim=='\n') 
+				break;
 		}
 	}
+	
+	Table& table=*new Table(columns);
 
 	// parse cells
-	Table& table=*new Table(columns);
-	char *row_chars;
-	while(row_chars=getrow(&data)) {
-		// remove empty&comment lines
-		if(!*row_chars || *row_chars == '#')
-			continue;
-		Table::element_type row(new ArrayString);
-		while(char *cell_chars=lsplit(&row_chars, '\t')) {
-			*row+=new String(cell_chars, 0, true);
+	Table::element_type row(new ArrayString);
+	skip_empty_and_comment_lines(&data);
+	while( lsplit_result sr=lsplit(&data, separators.column, '\n', separators.encloser) ) {
+		if(!sr.delim && !row->count()) // append last empty column [if without \n]
+			break;
+		*row+=new String(sr.piece, 0, true);
+		if(sr.delim=='\n') {
+			table+=row;
+			row=new ArrayString;
+			skip_empty_and_comment_lines(&data);
 		}
+	}
+	// last line [if without \n]
+	if(row->count())
 		table+=row;
-	};
-
+	
 	// replace any previous table value
 	GET_SELF(r, VTable).set_table(table);
 }
 
-/// @todo "x\nx" "xxx""xx"
+static void maybe_enclose( String& to, const String& from, char encloser, const String* sencloser ) {
+	if(encloser) {
+		to<<*sencloser;
+		// while we have 'encloser'...
+		size_t pos_after=0;
+		for( size_t pos_before; (pos_before=from.pos( encloser, pos_after ))!=STRING_NOT_FOUND; pos_after=pos_before+1) {
+            to<<from.mid(pos_before+1/*+found encloser*/, pos_after-pos_before);
+			to<<*sencloser; // doubling encloser
+		}
+		// last piece
+		size_t from_length=from.length();
+		if(pos_after<from_length)
+			to<<from.mid(pos_after, from_length);
+
+		to<<*sencloser;
+	} else
+		to.append(from, String::L_TABLE);
+}
+
+
 static void _save(Request& r, MethodParams& params) {
-	Value& vfile_name=params.as_no_junction(params.count()-1, "file name must not be code");
+	const String& first_arg=params.as_string(0, "first argument must not be code");
+	size_t param_index=1;
+
+	bool do_append=false;
+	bool output_column_names=true;
+
+	// mode?
+	if(first_arg=="append")
+		do_append=true;
+	else if(first_arg=="nameless")
+		output_column_names=false;
+	else
+		--param_index;
+
+	const String& file_name=params.as_string(param_index++, "file name must not be code");
+
+	TableSeparators separators;
+	if(param_index<params.count()) {
+		if(HashStringValue *options=params.as_no_junction(param_index++, "additional params must be hash").get_hash()) {
+			separators.load(*options);
+			if(options->count())
+				throw Exception("parser.runtime",
+					0,
+					"invalid option passed");
+		} else
+			throw Exception("parser.runtime",
+				0,
+				"additional params must be hash (did you spell mode parameter correctly?)");
+		
+	}
+	if(param_index<params.count())
+		throw Exception("parser.runtime",
+			0,
+			"bad mode (must be nameless or append)");
 
 	Table& table=GET_SELF(r, VTable).table();
 
-	bool do_append=false;
 	String sdata;
-	if(params.count()==1) { // named output
-		// write out names line
+	if(output_column_names) {
 		if(table.columns()) { // named table
 			for(Array_iterator<const String*> i(*table.columns()); i.has_next(); ) {
-				sdata.append(*i.next(), String::L_TABLE);
+				maybe_enclose( sdata, *i.next(), separators.encloser, separators.sencloser );
 				if(i.has_next())
 					sdata.append_know_length("\t", 1, String::L_CLEAN);
 			}
-		} else { // nameless table
+		} else { // nameless table [we were asked to output column names]
 			if(int lsize=table.count()?table[0]->count():0)
 				for(int column=0; column<lsize; column++) {
 					char *cindex_tab=new(PointerFreeGC) char[MAX_NUMBER];
 					sdata.append_know_length(cindex_tab, 
-						snprintf(cindex_tab, MAX_NUMBER, "%d\t", column),
-						String::L_CLEAN);
+						snprintf(cindex_tab, MAX_NUMBER, 
+							column<lsize-1?"%d\t":"%d", column),
+							String::L_CLEAN);
 				}
 			else
 				sdata.append_help_length("empty nameless table", 0, String::L_CLEAN);
 		}
 		sdata.append_know_length("\n", 1, String::L_CLEAN);
-	} else { // mode specified
-		const String&  mode=params.as_string(0, "mode must be string");
-		if(mode=="append")
-			do_append=true;
-		else if(mode=="nameless")
-			/*ok, already skipped names output*/;
-		else
-			throw Exception("parser.runtime",
-				&mode,
-				"unknown mode, must be 'append'");
-
 	}
+
 	// data lines
 	Array_iterator<ArrayString*> i(table);
 	while(i.has_next()) {
 		for(Array_iterator<const String*> c(*i.next()); c.has_next(); ) {
-			sdata.append(*c.next(), String::L_TABLE);
+			maybe_enclose( sdata, *c.next(), separators.encloser, separators.sencloser );
 			if(c.has_next())
 				sdata.append_know_length("\t", 1, String::L_CLEAN);
 		}
@@ -251,7 +445,7 @@ static void _save(Request& r, MethodPara
 	}
 
 	// write
-	file_write(r.absolute(vfile_name.as_string()), 
+	file_write(r.absolute(file_name), 
 		sdata.cstr(), sdata.length(), true, do_append);
 }
 
@@ -364,7 +558,8 @@ static void table_row_to_hash(Table::ele
 				table=vtable->get_table();
 			else {
 				// no? creating table of same structure as source
-				table=new Table(*info->table, 0, -1 /*no rows, just structure*/);
+				Table::Action_options table_options(0, 0);
+				table=new Table(*info->table, table_options/*no rows, just structure*/);
 				info->hash->put(*key, new VTable(table));
 			}
 			*table+=row;
@@ -434,7 +629,7 @@ static void _hash(Request& r, MethodPara
 				} else
 					throw Exception("parser.runtime",
 						0,
-						"value field(s) must be string or self_table"
+						"value field(s) must be string or table"
 					);
 			} else { // by all columns, including key
 				if(!(distinct!=D_ILLEGAL && distinct!=D_FIRST))
@@ -444,17 +639,19 @@ static void _hash(Request& r, MethodPara
 
 
 			{
-				Row_info info;
-				info.r=&r;
-				info.table=&self_table;
 				Value* key_param=&params[0];
-				info.key_code=key_param->get_junction()?key_param:0;
-				info.key_field=info.key_code?-1
-					:self_table.column_name2index(key_param->as_string(), true);
-				info.value_fields=&value_fields;
-				info.hash=&result.hash();
-				info.distinct=distinct;
-				info.row=0;
+				Row_info info={
+					&r,
+					&self_table,
+					/*key_code=*/key_param->get_junction()?key_param:0,
+					/*key_field=*/0/*filled below*/,
+					&value_fields,
+					&result.hash(),
+					distinct,
+					/*row=*/0
+				};
+				info.key_field=(info.key_code?-1
+						:self_table.column_name2index(key_param->as_string(), true));
 
 				int saved_current=self_table.current();
 				self_table.for_each(table_row_to_hash, &info);
@@ -532,48 +729,47 @@ static void _sort(Request& r, MethodPara
 	GET_SELF(r, VTable).set_table(new_table);
 }
 
-static bool _locate_expression(Request& r, MethodParams& params) {
-	if(params.count()>1)
-		throw Exception("parser.runtime", 
-			0,
-			"locate by expression has only one parameter - expression");
-
+#ifndef DOXYGEN
+struct Expression_is_true_info {
+	Request* r;
+	Value* expression_code;
+};
+#endif
+static bool expression_is_true(Table&, Expression_is_true_info* info) {
+	return info->r->process_to_value(*info->expression_code).as_bool();
+}
+static bool _locate_expression(Table& table, Table::Action_options o,
+			       Request& r, MethodParams& params) {
+	check_option_param(o.defined, params, 1,
+		"locate by expression only has parameters: expression and, maybe, options");
 	Value& expression_code=params.as_junction(0, "must be expression");
 
-	Table& table=GET_SELF(r, VTable).table();
-	int saved_current=table.current();
-	int size=table.count();
-	for(int row=0; row<size; row++) {
-		table.set_current(row);
-
-		if(r.process_to_value(expression_code).as_bool())
-			return true;
-	}
-	table.set_current(saved_current);
-	return false;
+	Expression_is_true_info info={&r, &expression_code};
+	return table.table_first_that(expression_is_true, &info, o);
 }
-static bool _locate_name_value(Request& r, MethodParams& params) {
-	if(params.count()>2)
-		throw Exception("parser.runtime", 
-			0,
-			"locate by name and value has only two parameters - name and value");
-
-	Table& table=GET_SELF(r, VTable).table();
-	return table.locate(
-		params.as_string(0, "column name must be string"),
-		params.as_string(1, "value must be string")
-	);
+static bool _locate_name_value(Table& table, Table::Action_options o,
+			       Request&, MethodParams& params) {
+	check_option_param(o.defined, params, 2,
+		"locate by locate by name has parameters: name, value and, maybe, options");
+	const String& name=params.as_string(0, "column name must be string");
+	const String& value=params.as_string(1, "value must be string");
+	return table.locate(name, value, o);
 }
 static void _locate(Request& r, MethodParams& params) {
+	Table& table=GET_SELF(r, VTable).table();
+
+	Table::Action_options o=get_action_options(r, params, table);
+
 	bool result=params[0].get_junction()?
-		_locate_expression(r, params) :
-		_locate_name_value(r, params);
+		_locate_expression(table, o, r, params) :
+		_locate_name_value(table, o, r, params);
 	r.write_no_lang(*new VBool(result));
 }
 
-static void _flip(Request& r, MethodParams& params) {
+
+static void _flip(Request& r, MethodParams&) {
 	Table& old_table=GET_SELF(r, VTable).table();
-	Table& new_table=*new Table(old_table.columns());
+	Table& new_table=*new Table(0);
 	if(size_t old_count=old_table.count())
 		if(size_t old_cols=old_table[0]->count()) 
 			for(size_t column=0; column<old_cols; column++) {
@@ -601,45 +797,41 @@ static void _append(Request& r, MethodPa
 	GET_SELF(r, VTable).table()+=row;
 }
 
+static void join_named_row(Table& src, Table* dest) {
+	Table::columns_type dest_columns=dest->columns();
+	size_t dest_columns_count=dest_columns->count();
+	Table::element_type dest_row(new ArrayString(dest_columns_count));
+	for(size_t dest_column=0; dest_column<dest_columns_count; dest_column++) {
+		const String* src_item=src.item(*dest_columns->get(dest_column));
+		*dest_row+=src_item?src_item:new String;
+	}
+	*dest+=dest_row;
+}
+static void join_nameless_row(Table& src, Table* dest) {
+	*dest+=src[src.current()];
+}
 static void _join(Request& r, MethodParams& params) {
 	Table* maybe_src=params.as_no_junction(0, "table ref must not be code").get_table();
 	if(!maybe_src)
 		throw Exception("parser.runtime", 
 			0, 
 			"source is not a table");
-
 	Table& src=*maybe_src;
+
+	Table::Action_options o=get_action_options(r, params, src);
+	check_option_param(o.defined, params, 1,
+		"invalid extra parameter");
+
 	Table& dest=GET_SELF(r, VTable).table();
 	if(&src == &dest)
 		throw Exception("parser.runtime", 
 			0, 
 			"source and destination are same table");
 
-	int offset, limit;
-	get_copy_options(r, params, 1, src, 
-		offset, limit);
-
-	if(Table::columns_type dest_columns=dest.columns()) { // dest is named
-		int saved_src_current=src.current();
-		int m=src.count()-offset;
-		if(!limit || limit>m)
-			limit=m;
-		int end=offset+limit;
-		for(int src_row=offset; src_row<end; src_row++) {
-			src.set_current(src_row);
-			Table::element_type dest_row(new ArrayString(src.count()));
-			for(size_t dest_column=0; dest_column<dest_columns->count(); dest_column++) {
-				const String* src_item=src.item(*dest_columns->get(dest_column));
-				*dest_row+=src_item?src_item:new String;
-			}
-			dest+=dest_row;
-		}
-		src.set_current(saved_src_current);
-	} else { // dest is nameless
-		size_t src_count=src.count();
-		for(size_t src_row=0; src_row<src_count; src_row++)
-			dest+=src[src_row];
-	}
+	if(dest.columns()) // dest is named
+		src.table_for_each(join_named_row, &dest, o);
+	else // dest is nameless
+		src.table_for_each(join_nameless_row, &dest, o);
 }
 
 #ifndef DOXYGEN
@@ -786,7 +978,6 @@ static void _select(Request& r, MethodPa
 		source_table.set_current(row);
 
 		bool condition=r.process_to_value(vcondition, 
-				/*0/*no name* /,*/
 				false/*don't intercept string*/).as_bool();
 
 		if(condition) // ...condition is true=
@@ -813,7 +1004,7 @@ MTable::MTable(): Methoded("table") {
 
 	// ^table.save[file]  
 	// ^table.save[nameless;file]
-	add_native_method("save", Method::CT_DYNAMIC, _save, 1, 2);
+	add_native_method("save", Method::CT_DYNAMIC, _save, 1, 3);
 
 	// ^table.count[]
 	add_native_method("count", Method::CT_DYNAMIC, _count, 0, 0);
@@ -839,7 +1030,7 @@ MTable::MTable(): Methoded("table") {
 	add_native_method("sort", Method::CT_DYNAMIC, _sort, 1, 2);
 
 	// ^table.locate[field;value]
-	add_native_method("locate", Method::CT_DYNAMIC, _locate, 1, 2);
+	add_native_method("locate", Method::CT_DYNAMIC, _locate, 1, 3);
 
 	// ^table.flip[]
 	add_native_method("flip", Method::CT_DYNAMIC, _flip, 0, 0);