--- parser3/src/classes/string.C 2001/03/12 09:08:46 1.7 +++ parser3/src/classes/string.C 2003/02/14 12:59:09 1.125.2.7 @@ -1,48 +1,574 @@ -/* - Parser - Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - Author: Alexander Petrosyan (http://design.ru/paf) +/** @file + Parser: @b string parser class. - $Id: string.C,v 1.7 2001/03/12 09:08:46 paf Exp $ + Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) */ +static const char* IDENT_STRING_C="$Date: 2003/02/14 12:59:09 $"; + +#include "classes.h" +#include "pa_vmethod_frame.h" + #include "pa_request.h" -#include "_string.h" #include "pa_vdouble.h" #include "pa_vint.h" +#include "pa_vtable.h" +#include "pa_vbool.h" +#include "pa_string.h" +#include "pa_sql_connection.h" +#include "pa_dictionary.h" +#include "pa_vmethod_frame.h" + +// class + +class MString : public Methoded { +public: + MString(); +public: // Methoded + bool used_directly() { return true; } +}; -// global var -VClass *string_class; +// global variable + +MethodedPtr string_class(new MString); // methods -static void _length(Request& r, const String&, Array *) { +static void _length(Request& r, StringPtr method_name, MethodParams& ) { + Pool& pool=r.pool(); + double result=GET_SELF(r, VString).string()->size(); + r.write_no_lang(ValuePtr(new VDouble(result))); +} + +static void _int(Request& r, StringPtr method_name, MethodParams& params) { + Pool& pool=r.pool(); + StringPtr self_string=r.get_self()->get_string(&pool); + int converted; + ValuePtr default_code=params.count()>0?params.as_junction(0, "default must be int") + :ValuePtr(0); // (default) + try { + if(!self_string || self_string->is_empty()) + throw Exception("parser.runtime", + method_name, + "parameter is empty string, error converting"); + converted=self_string->as_int(); + } catch(...) { // convert problem + if(!default_code) // we have a problem when no default + rethrow; + else + converted=r.process_to_value(default_code)->as_int(); + } + r.write_no_lang(ValuePtr(new VInt(converted))); +} + +static void _double(Request& r, StringPtr method_name, MethodParams& params) { + Pool& pool=r.pool(); + StringPtr self_string=GET_SELF(r, VString).string(); + double converted; + ValuePtr default_code=params.count()>0?params.as_junction(0, "default must be double") + :ValuePtr(0); // (default) + try { + if(!self_string || self_string->is_empty()) + throw Exception("parser.runtime", + method_name, + "parameter is empty string, error converting"); + converted=self_string->as_double(); + } catch(...) { // convert problem + if(!default_code) // we have a problem when no default + rethrow; + else + converted=r.process_to_value(default_code)->as_double(); + } + + r.write_no_lang(ValuePtr(new VDouble(converted))); +} + +/*not static*/void _string_format(Request& r, StringPtr method_name, MethodParams& params) { + Pool& pool=r.pool(); + + ValuePtr fmt_maybe_code=params[0]; + // for some time due to stupid {} in original design + StringPtr fmt=r.process_to_string(fmt_maybe_code); + + const char *buf=format(pool, r.get_self()->as_double(), fmt->cstr()); + + String result; + result.APPEND_CLEAN(buf, 0, + method_name->origin().file, + method_name->origin().line); + r.write_no_lang(result); +} + +static void _left(Request& r, StringPtr /*method_name*/, MethodParams& params) { + size_t n=(size_t)params.as_int(0, "n must be int", r); + + StringPtr string=GET_SELF(r, VString).string(); + r.write_assign_lang(*string->mid(0, n)); +} + +static void _right(Request& r, StringPtr /*method_name*/, MethodParams& params) { + size_t n=(size_t)params.as_int(0, "n must be int", r); + + StringPtr string=GET_SELF(r, VString).string(); + r.write_assign_lang(*string->mid(string->size()-n, string->size())); +} + +static void _mid(Request& r, StringPtr /*method_name*/, MethodParams& params) { + StringPtr string=GET_SELF(r, VString).string(); + + size_t p=(size_t)max(0, params.as_int(0, "p must be int", r)); + size_t n=params.count()>1? + (size_t)max(0, params.as_int(1, "n must be int", r)):string->size(); + + r.write_assign_lang(*string->mid(p, p+n)); +} + +static void _pos(Request& r, StringPtr method_name, MethodParams& params) { + Pool& pool=r.pool(); + ValuePtr substr=params.as_no_junction(0, "substr must not be code"); + + StringPtr string=GET_SELF(r, VString).string(); + r.write_assign_lang(ValuePtr(new VInt(string->pos(*substr->as_string(&pool))))); +} + +static void split_list(Request& r, StringPtr method_name, + MethodParams& params, int paramIndex, + String& string, + ArrayString& result) { + Pool& pool=r.pool(); + ValuePtr delim_value=params.as_no_junction(paramIndex, "delimiter must not be code"); + + string.split(result, 0, *delim_value->as_string(&pool)); +} + +#define SPLIT_LEFT 0x0001 +#define SPLIT_RIGHT 0x0010 +#define SPLIT_HORIZONTAL 0x0100 +#define SPLIT_VERTICAL 0x1000 + +static int split_options(StringPtr options) { + struct Split_option { + const char* keyL; + const char* keyU; + int setBit; + int checkBit; + } split_option[]={ + {"l", "L", SPLIT_LEFT, SPLIT_RIGHT}, // 0xVHRL + {"r", "R", SPLIT_RIGHT, SPLIT_LEFT}, + {"h", "H", SPLIT_HORIZONTAL, SPLIT_VERTICAL}, + {"v", "V", SPLIT_VERTICAL, SPLIT_HORIZONTAL}, + {0} + }; + + int result=0; + if(options) { + for(Split_option *o=split_option; o->keyL; o++) + if(options->pos(o->keyL)>=0 + || (o->keyU && options->pos(o->keyU)>=0)) { + if(result & o->checkBit) + throw Exception("parser.runtime", + options, + "conflicting split options"); + result |= o->setBit; + } + } + + return result; +} + +static TablePtr split_vertical(Request& r, StringPtr string, ArrayString& pieces, bool right) { + Pool& pool=r.pool(); + + Table::columns_type columns(new ArrayString); + *columns+=StringPtr(new String("piece")); + + TablePtr table(new Table(string, columns, pieces.count())); + if(right) { // right + for(int i=pieces.count(); --i>=0; ) { + Table::element_type row(new ArrayString); + *row+=pieces[i]; + *table+=row; + } + } else { // left + Array_iterator i(pieces); + while(i.has_next()) { + Table::element_type row(new ArrayString); + *row+=i.next(); + *table+=row; + } + } + + return table; +} + +static TablePtr split_horizontal(Request& r, StringPtr string, ArrayString& pieces, bool right) { + Pool& pool=r.pool(); + + TablePtr table(new Table(string, Table::columns_type(0) /* nameless */)); + Table::element_type row(new ArrayString(pieces.count())); + if(right) { // right + for(int i=pieces.count(); --i>=0; ) { + *row+=pieces[i]; + } + } else { // left + Array_iterator i(pieces); + while(i.has_next()) { + *row+=i.next(); + } + } + *table+=row; + + return table; +} + +static void split_with_options(Request& r, StringPtr method_name, MethodParams& params, + int bits) { + Pool& pool=r.pool(); + StringPtr string=GET_SELF(r, VString).string(); + + ArrayString pieces; + split_list(r, method_name, params, 0, *string, pieces); + + if(!bits) { + StringPtr options(0); + if(params.count()>1) { + options=params.as_string(1, "options must not be code"); + } + bits=split_options(options); + } + + bool right=(bits & SPLIT_RIGHT) != 0; + bool horizontal=(bits & SPLIT_HORIZONTAL) !=0; + TablePtr table=horizontal?split_horizontal(r, string, pieces, right) + :split_vertical(r, string, pieces, right); + + r.write_no_lang(ValuePtr(new VTable(table))); +} +static void _split(Request& r, StringPtr method_name, MethodParams& params) { + split_with_options(r, method_name, params, 0 /* maybe-determine from param #2 */); +} +static void _lsplit(Request& r, StringPtr method_name, MethodParams& params) { + split_with_options(r, method_name, params, SPLIT_LEFT); +} +static void _rsplit(Request& r, StringPtr method_name, MethodParams& params) { + split_with_options(r, method_name, params, SPLIT_RIGHT); +} + +static void search_action(TablePtr table, Table::element_type row, int, int, int, int, void *) { + if(row) + *table+=row; +} + +#ifndef DOXYGEN +struct Replace_action_info { + Request *request; + StringPtr origin; + StringPtr src; String* dest; + VTablePtr vtable; + ValuePtr replacement_code; +}; +#endif +/// @todo they can do $global[$result] there, getting pointer to later-invalid local var, kill this +static void replace_action(TablePtr table, ArrayStringPtr row, + int prestart, int prefinish, + int poststart, int postfinish, + void *info) { + Replace_action_info& ai=*static_cast(info); + if(row) { // begin&middle + // piece from last match['prestart'] to beginning of this match['prefinish'] + if(prestart!=prefinish) + *ai.dest << ai.src->mid(prestart, prefinish);//ai.dest->APPEND_CONST("-"); + // store found parts in one-record VTable + if(table->count()) // middle + table->put(0, row); + else // begin + *table+=row; + { // execute 'replacement_code' in 'table' context + ai.vtable->set_table(table); + + *ai.dest << ai.request->process_to_string(ai.replacement_code); + } + } else // end + *ai.dest << ai.src->mid(poststart, postfinish); +} + +/// @todo use pcre:study somehow +static void _match(Request& r, StringPtr method_name, MethodParams& params) { Pool& pool=r.pool(); - Value& value=*new(pool) VDouble(pool, r.self->as_string().size()); - r.wcontext->write(value, String::Untaint_lang::NO /*always object, not string*/); + ValuePtr regexp=params.as_no_junction(0, "regexp must not be code"); + + StringPtr options= + params.count()>1? + params.as_no_junction(1, "options must not be code")->as_string(&pool):StringPtr(0); + + Temp_lang temp_lang(r, String::UL_PASS_APPENDED); + StringPtr src=GET_SELF(r, VString).string(); + if(params.count()<3) { // search + bool was_global; + TablePtr table=src->match(r.charsets.source(), + method_name, + *regexp->as_string(&pool), options, + search_action, 0, + &was_global); + ValuePtr result; + // matched + // not (just matched[3=pre/match/post], no substrings) or Global search + if(table->columns()->count()>3 || was_global) + result=ValuePtr(new VTable(table/*TODO: clone this when table would be stacked!*/)); // table of pre/match/post+substrings + else + result=ValuePtr(new VBool(table)); + r.write_assign_lang(result); + } else { // replace + ValuePtr replacement_code=params.as_junction(2, "replacement param must be code"); + + String result; + VTablePtr vtable(new VTable); + Replace_action_info info; + info.request=&r; + info.origin=method_name; + info.src=src; + info.dest=&result; + info.vtable=vtable; + info.replacement_code=replacement_code; + Temp_value_element temp_match_var( + *replacement_code->get_junction()->method_frame, + match_var_name, vtable); + src->match(r.charsets.source(), + method_name, + *r.process_to_string(regexp), options, + replace_action, &info); + r.write_assign_lang(result); + } } -static void _int(Request& r, const String&, Array *) { +static void change_case(Request& r, StringPtr method_name, MethodParams& params, + String::Change_case_kind kind) { Pool& pool=r.pool(); - Value& value=*new(pool) VInt(pool, static_cast(r.self->get_double())); - r.wcontext->write(value, String::Untaint_lang::NO /*always object, not string*/); + StringPtr src=GET_SELF(r, VString).string(); + + r.write_assign_lang(*src->change_case(pool, r.charsets.source(), kind)); +} +static void _upper(Request& r, StringPtr method_name, MethodParams& params) { + change_case(r, method_name, params, String::CC_UPPER); +} +static void _lower(Request& r, StringPtr method_name, MethodParams& params) { + change_case(r, method_name, params, String::CC_LOWER); } -static void _double(Request& r, const String&, Array *) { +#ifndef DOXYGEN +class String_sql_event_handlers: public SQL_Driver_query_event_handlers { + //Pool& pool; + StringPtr statement_string; const char* statement_cstr; + bool got_column; +public: + bool got_cell; + StringPtr result; +public: + String_sql_event_handlers(//Pool& apool, + StringPtr astatement_string, const char* astatement_cstr): + //pool(apool), + statement_string(astatement_string), + statement_cstr(astatement_cstr), + result(new String) {} + + bool add_column(SQL_Error& error, void *ptr, size_t size) { + if(got_column) { + error=SQL_Error("parser.runtime", + //statement_string, + "result must contain exactly one column"); + return true; + } + got_column=true; + return false; + } + bool before_rows(SQL_Error& /*error*/ ) { /* ignore */ return false; } + bool add_row(SQL_Error& /*error*/) { /* ignore */ return false; } + bool add_row_cell(SQL_Error& error, void *ptr, size_t size) { + if(got_cell) { + error=SQL_Error("parser.runtime", + //statement_string, + "result must not contain more then one row"); + return true; + } + + try { + got_cell=true; + result->APPEND_TAINTED((const char* )ptr, size, statement_cstr, 0); + return false; + } catch(...) { + error=SQL_Error("exception occured in String_sql_event_handlers::add_row_cell"); + return true; + } + } +}; +#endif +StringPtr sql_result_string(Request& r, StringPtr method_name, MethodParams& params, + HashStringValue*& options, ValuePtr& default_code) { Pool& pool=r.pool(); - Value& value=*new(pool) VDouble(pool, r.self->get_double()); - r.wcontext->write(value, String::Untaint_lang::NO /*always object, not string*/); + + ValuePtr statement=params.as_junction(0, "statement must be code"); + + ulong limit=0; + ulong offset=0; + default_code=ValuePtr(0); + if(params.count()>1) { + ValuePtr voptions=params.as_no_junction(1, "options must be hash, not code"); + if(!voptions->is_string()) + if(options=voptions->get_hash(method_name)) { + if(ValuePtr vlimit=options->get(sql_limit_name)) + limit=(ulong)r.process_to_value(vlimit)->as_double(); + if(ValuePtr voffset=options->get(sql_offset_name)) + offset=(ulong)r.process_to_value(voffset)->as_double(); + if(default_code=options->get(sql_default_name)) { + if(JunctionPtr default_junction=default_code->get_junction()) + ;//default_junction->change_context(statement.get_junction()); + else + throw Exception("parser.runtime", + method_name, + "default option must be code"); + } + } else + throw Exception("parser.runtime", + method_name, + "options must be hash"); + } else + options=0; + + Temp_lang temp_lang(r, String::UL_SQL); + StringPtr statement_string=r.process_to_string(statement); + const char* statement_cstr= + statement_string->cstr(pool, String::UL_UNSPECIFIED, r.connection(method_name)); + String_sql_event_handlers handlers(statement_string, statement_cstr); + r.connection(method_name)->query( + statement_cstr, offset, limit, + handlers, + statement_string); + + if(!handlers.got_cell) + return 0; // no lines, caller should return second param[default value] + + return handlers.result; } -void initialize_string_class(Pool& pool, VClass& vclass) { +static void _sql(Request& r, StringPtr method_name, MethodParams& params) { + Pool& pool=r.pool(); + + HashStringValue* options; + ValuePtr default_code; + StringPtr string=sql_result_string(r, method_name, params, options, default_code); + if(!string) { + if(default_code) { + string=r.process_to_string(default_code); + if(!string) + string=StringPtr(new String); + } else + throw Exception("parser.runtime", + method_name, + "produced no result, but no default option specified"); + } + + r.write_assign_lang(*string); +} + +static void _replace(Request& r, StringPtr method_name, MethodParams& params) { + Pool& pool=r.pool(); + StringPtr src=GET_SELF(r, VString).string(); + + Table* table=params.as_no_junction(0, "parameter must not be code")->get_table(); + if(!table) + throw Exception("parser.runtime", + method_name, + "parameter must be table"); + + Dictionary dict2(TablePtr(table), 1/*am not sure why using the =default parameter causes compile errors in replace*/ ); + r.write_assign_lang(*src->replace(pool, dict2)); +} + +static void _save(Request& r, StringPtr method_name, MethodParams& params) { + StringPtr file_name=params.as_string(params.count()-1, + "file name must be string"); + + StringPtr src=GET_SELF(r, VString).string(); + + bool do_append=false; + if(params.count()>1) { + StringPtr mode=params.as_string(0, "mode must be string"); + if(*mode=="append") + do_append=true; + else + throw Exception("parser.runtime", + mode, + "unknown mode, must be 'append'"); + } + + // write + CharPtr buf=src->cstr(String::UL_UNSPECIFIED, r.connection(StringPtr(0)/*no error if none*/)); + file_write(r.absolute(file_name), + buf, strlen(buf), true, do_append); +} + +static void _normalize(Request& r, StringPtr method_name, MethodParams& /*params*/) { + StringPtr src=GET_SELF(r, VString).string(); + + r.write_assign_lang(*src->join_chains(r.pool())); +} + +// constructor + +MString::MString(): Methoded("string") { // ^string.length[] - vclass.add_native_method("length", _length, 0, 0); + add_native_method("length", Method::CT_DYNAMIC, _length, 0, 0); // ^string.int[] - vclass.add_native_method("int", _int, 0, 0); - + // ^string.int(default) + add_native_method("int", Method::CT_DYNAMIC, _int, 0, 1); // ^string.double[] - vclass.add_native_method("double", _double, 0, 0); -} + // ^string.double(default) + add_native_method("double", Method::CT_DYNAMIC, _double, 0, 1); + + // ^string.format{format} + add_native_method("format", Method::CT_DYNAMIC, _string_format, 1, 1); + + // ^string.left(n) + add_native_method("left", Method::CT_DYNAMIC, _left, 1, 1); + // ^string.right(n) + add_native_method("right", Method::CT_DYNAMIC, _right, 1, 1); + // ^string.mid(p;n) + add_native_method("mid", Method::CT_DYNAMIC, _mid, 1, 2); + + // ^string.pos[substr] + add_native_method("pos", Method::CT_DYNAMIC, _pos, 1, 1); + // ^string.split[delim] + // ^string.split[delim][options] + add_native_method("split", Method::CT_DYNAMIC, _split, 1, 2); + // old names for backward compatibility + // ^string.lsplit[delim] + add_native_method("lsplit", Method::CT_DYNAMIC, _lsplit, 1, 1); + // ^string.rsplit[delim] + add_native_method("rsplit", Method::CT_DYNAMIC, _rsplit, 1, 1); + + // ^string.match[regexp][options] + // ^string.match[regexp][options]{replacement-code} + add_native_method("match", Method::CT_DYNAMIC, _match, 1, 3); + + // ^string.toupper[] + add_native_method("upper", Method::CT_DYNAMIC, _upper, 0, 0); + // ^string.tolower[] + add_native_method("lower", Method::CT_DYNAMIC, _lower, 0, 0); + + // ^sql[query] + // ^sql[query][$.limit(1) $.offset(2) $.default[n/a]] + add_native_method("sql", Method::CT_STATIC, _sql, 1, 2); + + // ^string.replace[table] + add_native_method("replace", Method::CT_DYNAMIC, _replace, 1, 1); + + // ^string.save[file] + add_native_method("save", Method::CT_DYNAMIC, _save, 1, 2); + + // ^string.normalize[] + add_native_method("normalize", Method::CT_DYNAMIC, _normalize, 0, 0); +}