--- parser3/src/main/pa_dictionary.C 2002/03/27 15:30:36 1.14 +++ parser3/src/main/pa_dictionary.C 2012/03/16 09:24:13 1.26 @@ -1,77 +1,77 @@ /** @file Parser: dictionary class impl. - Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2012 Art. Lebedev Studio (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) - - $Id: pa_dictionary.C,v 1.14 2002/03/27 15:30:36 paf Exp $ */ #include "pa_dictionary.h" #include "pa_exception.h" -void Dictionary::add_first(Array::Item *value, void *info) { - Dictionary& self=*static_cast(info); - Array *row=static_cast(value); - // get a=>b values - const String& a=*row->get_string(0); - const String& b=*row->get_string(1); - // empty 'a' check - if(a.is_empty()) { - Pool& pool=self.pool(); - throw Exception("parser.runtime", - &a, - "dictionary table 'from' column elements must not be empty"); - } - unsigned char c=(unsigned char)a.first_char(); - if(!self.starting_line_of[c]) - self.starting_line_of[c]=self.constructor_line; - - double ratio=((double)b.size())/a.size(); - if(ratio>self.fmax_ratio) - self.fmax_ratio=ratio; +volatile const char * IDENT_PA_DICTIONARY_C="$Id: pa_dictionary.C,v 1.26 2012/03/16 09:24:13 moko Exp $" IDENT_PA_DICTIONARY_H; + +Dictionary::Dictionary(Table& atable): substs(atable.count()) { + // clear starting_lines + memset(starting_line_of, 0, sizeof(starting_line_of)); + // grab first letters of first column of a table + constructor_line=1; - self.constructor_line++; + for(Array_iterator i(atable); i.has_next(); ) { + ArrayString* row=i.next(); + + append_subst( + row->get(0), + (row->count()>1) ? row->get(1) : 0, + "dictionary table 'from' column elements must not be empty" + ); + } } -Dictionary::Dictionary(Table& atable, double amin_ratio) : Pooled(atable.pool()), - table(atable), - fmax_ratio(amin_ratio) { +Dictionary::Dictionary(const String& from, const String& to): substs(1) { // clear starting_lines memset(starting_line_of, 0, sizeof(starting_line_of)); - // grab first letters of first column of a table - constructor_line=1; table.for_each(add_first, this); + constructor_line=1; + + append_subst(&from, &to); +} + + +void Dictionary::append_subst(const String* from, const String* to, const char* exception_cstr) { + if(from->is_empty()) + throw Exception(PARSER_RUNTIME, + 0, + exception_cstr ? exception_cstr : "'from' must not be empty"); + + // record simplier 'from' for quick comparisons in 'starts' extremely-tight-callback + substs+=Dictionary::Subst(from->cstr(), (to && !to->is_empty()) ? to : 0); + + unsigned char c=(unsigned char)from->first_char(); + if(!starting_line_of[c]) + starting_line_of[c]=constructor_line; + + constructor_line++; } #ifndef DOXYGEN -struct First_that_starts_info { +struct First_that_begins_info { int line; - const char *src; size_t src_size; + const char* str; }; #endif -static void *starts(Array::Item *value, void *info) { - First_that_starts_info& i=*static_cast(info); +static bool starts(Dictionary::Subst subst, First_that_begins_info* info) { // skip irrelevant lines - if(i.line>1) { - --i.line; + if(info->line>1) { + --info->line; return 0; } - Array *row=static_cast(value); - int partial; - row->get_string(0)->cmp(partial, i.src, i.src_size); - return ( - partial==0 || // full match - partial==1) // typo left column starts 'src' - ?value:0; + return strncmp(subst.from, info->str, subst.from_length)==0; } - -void* Dictionary::first_that_starts(const char *src, size_t src_size) const { - First_that_starts_info info; - if(info.line=starting_line_of[(unsigned char)*src]) { - info.src=src; - info.src_size=src_size; - return table.first_that(starts, &info); +Dictionary::Subst Dictionary::first_that_begins(const char* str) const { + First_that_begins_info info; + if((info.line=starting_line_of[(unsigned char)*str])) { + info.str=str; + return substs.first_that(starts, &info); } else return 0; }