--- parser3/src/main/pa_string.C 2002/02/28 09:35:41 1.147 +++ parser3/src/main/pa_string.C 2002/04/19 08:28:35 1.154 @@ -4,7 +4,7 @@ Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) - $Id: pa_string.C,v 1.147 2002/02/28 09:35:41 paf Exp $ + $Id: pa_string.C,v 1.154 2002/04/19 08:28:35 paf Exp $ */ #include "pcre.h" @@ -29,9 +29,9 @@ ulong string_piece_appends=0; String::String(Pool& apool, const char *src, size_t src_size, bool tainted) : Pooled(apool) { - last_chunk=&head; - head.count=CR_PREALLOCATED_COUNT; - append_here=head.rows; + last_chunk=&head.chunk; + head.chunk.count=CR_PREALLOCATED_COUNT; + append_here=head.chunk.rows; if(src) if(tainted) @@ -42,9 +42,9 @@ String::String(Pool& apool, const char * String::String(const String& src) : Pooled(src.pool()) { - last_chunk=&head; - head.count=CR_PREALLOCATED_COUNT; - append_here=head.rows; + last_chunk=&head.chunk; + head.chunk.count=CR_PREALLOCATED_COUNT; + append_here=head.chunk.rows; append(src, UL_UNSPECIFIED); } @@ -70,8 +70,12 @@ void String::expand() { if(new_chunk_count>max_integral(Chunk::count_type)) new_chunk_count=max_integral(Chunk::count_type); - Chunk *new_chunk=static_cast( - malloc(sizeof(Chunk::count_type)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *), 10)); + Chunk *new_chunk=static_cast(malloc( + sizeof(Chunk)// count+interpadding(?)+rows[CR_PREALLOCATED_COUNT]+tailpadding(??) + -sizeof(Chunk::rows_type) // PREALLOCATED rows + +sizeof(Chunk::Row)*new_chunk_count // neaded rows + +sizeof(Chunk *) // link size + , 10)); new_chunk->rows[new_chunk->count=new_chunk_count].link=0; last_chunk->rows[last_chunk->count].link=new_chunk; @@ -81,7 +85,7 @@ void String::expand() { String& String::real_append(STRING_APPEND_PARAMS) { if(!last_chunk) // growth stopped [we're appended as string to somebody] - throw Exception(0, 0, + throw Exception(0, this, "string growth stopped (append cstr)"); @@ -131,11 +135,11 @@ String& String::real_append(STRING_APPEN char String::first_char() const { if(is_empty()) - throw Exception(0, 0, + throw Exception(0, this, "getting first char of empty string"); - return *head.rows[0].item.ptr; + return *head.chunk.rows[0].item.ptr; } uint String::hash_code() const { @@ -153,8 +157,8 @@ int String::cmp(int& partial, const Stri size_t a_size=size(); this_offset=min(this_offset, a_size-1); - const Chunk *a_chunk=&head; - const Chunk *b_chunk=&src.head; + const Chunk *a_chunk=&head.chunk; + const Chunk *b_chunk=&src.head.chunk; const Chunk::Row *a_row=a_chunk->rows; const Chunk::Row *b_row=b_chunk->rows; size_t a_offset=this_offset; @@ -243,7 +247,7 @@ int String::cmp(int& partial, const char size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0; this_offset=min(this_offset, a_size-1); - const Chunk *a_chunk=&head; + const Chunk *a_chunk=&head.chunk; const Chunk::Row *a_row=a_chunk->rows; size_t a_offset=this_offset; size_t b_offset=0; @@ -320,7 +324,7 @@ const Origin& String::origin() const { // when last peice is constant, // ex: parser_root_auto_path{dynamic} / auto.p{const} // using first piece - Origin& first_origin=head.rows[0].item.origin; + Origin& first_origin=head.chunk.rows[0].item.origin; return first_origin.file ? first_origin : append_here[-1].item.origin; } #endif @@ -437,29 +441,35 @@ void String::split(Array& result, } } -static void regex_options(char *options, int *result){ +static void regex_options(const String *options, int *result, bool& need_pre_post_match){ struct Regex_option { - char key; + const char *keyL; + const char *keyU; int clear, set; int *result; + bool *flag; } regex_option[]={ - {'i', 0, PCRE_CASELESS, result}, // a=A - {'s', 0, PCRE_DOTALL, result}, // \n\n$ [default] - {'x', 0, PCRE_EXTENDED, result}, // whitespace in regex ignored - {'m', PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$ - {'g', 0, true, result+1}, // many rows - {0}, + {"i", "I", 0, PCRE_CASELESS, result}, // a=A + {"s", "S", 0, PCRE_DOTALL, result}, // \n\n$ [default] + {"x", "U", 0, PCRE_EXTENDED, result}, // whitespace in regex ignored + {"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$ + {"g", "G", 0, true, result+1}, // many rows + {"'", 0, 0, 0, 0, &need_pre_post_match}, + {0} }; result[0]=PCRE_EXTRA | PCRE_DOTALL; result[1]=0; if(options) - for(Regex_option *o=regex_option; o->key; o++) - if( - strchr(options, o->key) || - strchr(options, toupper(o->key))) { - *(o->result)&=~o->clear; - *(o->result)|=o->set; + for(Regex_option *o=regex_option; o->keyL; o++) + if(options->pos(o->keyL)>=0 + || (o->keyU && options->pos(o->keyU)>=0)) { + if(o->flag) + *o->flag=true; + else { // result + *o->result &= ~o->clear; + *o->result |= o->set; + } } } @@ -473,13 +483,15 @@ bool String::match( bool *was_global) const { if(regexp.is_empty()) - throw Exception(0, 0, + throw Exception(0, aorigin, "regexp is empty"); + const char *pattern=regexp.cstr(); const char *errptr; int erroffset; - int option_bits[2]; regex_options(options?options->cstr():0, option_bits); + bool need_pre_post_match=false; + int option_bits[2]; regex_options(options, option_bits, need_pre_post_match); if(was_global) *was_global=option_bits[1]!=0; pcre *code=pcre_compile(pattern, option_bits[0], @@ -487,20 +499,19 @@ bool String::match( pool().get_source_charset().pcre_tables); if(!code) - throw Exception(0, 0, + throw Exception(0, ®exp.mid(erroffset, regexp.size()), "regular expression syntax error - %s", errptr); int info_substrings=pcre_info(code, 0, 0); if(info_substrings<0) { pcre_free(code); - throw Exception(0, 0, + throw Exception(0, aorigin, "pcre_info error (%d)", info_substrings); } - int startoffset=0; const char *subject=cstr(); int length=strlen(subject); int ovecsize; @@ -521,43 +532,48 @@ bool String::match( } int exec_option_bits=0; + int prestart=0; + int poststart=0; + int postfinish=size(); while(true) { int exec_substrings=pcre_exec(code, 0, - subject, length, startoffset, + subject, length, prestart, exec_option_bits, ovector, ovecsize); if(exec_substrings==PCRE_ERROR_NOMATCH) { pcre_free(code); - row_action(**table, 0/*last time, no row*/, 0, 0, info); + row_action(**table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info); return option_bits[1]!=0; // global=true+table, not global=false } if(exec_substrings<0) { pcre_free(code); - throw Exception(0, 0, + throw Exception(0, aorigin, "regular expression execute error (%d)", exec_substrings); } + int prefinish=ovector[0]; + poststart=ovector[1]; Array& row=*NEW Array(pool()); - row+=&mid(0, ovector[0]); // .prematch column value - row+=&mid(ovector[0], ovector[1]); // .match - row+=&mid(ovector[1], size()); // .postmatch + row+=need_pre_post_match?&mid(0, prefinish):0; // .prematch column value + row+=need_pre_post_match?&mid(prefinish, poststart):0; // .match + row+=need_pre_post_match?&mid(poststart, postfinish):0; // .postmatch for(int i=1; i(kind)); // never @@ -655,9 +671,11 @@ void String::join_chain(Pool& pool, } } -String& String::reconstruct(Pool& pool) const { - //_asm int 3; +/// @test if in some piece were found no dict words, append it, not it's duplicate +String& String::replace(Pool& pool, Dictionary& dict) const { +// return reconstruct(pool).replace_in_reconstructed(pool, dict); String& result=*new(pool) String(pool); + STRING_FOREACH_ROW( uchar joined_lang; const char *joined_ptr; @@ -668,20 +686,9 @@ IFNDEF_NO_STRING_ORIGIN( ); join_chain(pool, chunk, row, countdown, joined_lang, joined_ptr, joined_size); - - result.APPEND(joined_ptr, joined_size, joined_lang, - joined_origin_file, joined_origin_line); - ); - - return result; -}; - -String& String::replace_in_reconstructed(Pool& pool, Dictionary& dict) const { - //_asm int 3; - String& result=*new(pool) String(pool); - STRING_FOREACH_ROW( - const char *src=row->item.ptr; - size_t src_size=row->item.size; + + const char *src=joined_ptr; + size_t src_size=joined_size; char *new_cstr=(char *)pool.malloc((size_t)ceil(src_size*dict.max_ratio()), 14); char *dest=new_cstr; while(src_size) { @@ -700,24 +707,19 @@ String& String::replace_in_reconstructed } } - result.APPEND(new_cstr, dest-new_cstr, - row->item.lang, - row->item.origin.file, row->item.origin.line); + result.APPEND(new_cstr, dest-new_cstr, joined_lang, + joined_origin_file, joined_origin_line); ); return result; } -String& String::replace(Pool& pool, Dictionary& dict) const { - return reconstruct(pool).replace_in_reconstructed(pool, dict); -} - double String::as_double() const { double result; const char *cstr; char buf[MAX_NUMBER]; - if(head.rows+1==append_here) { - int size=min(head.rows[0].item.size, MAX_NUMBER-1); - memcpy(buf, head.rows[0].item.ptr, size); + if(head.chunk.rows+1==append_here) { + int size=min(head.chunk.rows[0].item.size, MAX_NUMBER-1); + memcpy(buf, head.chunk.rows[0].item.ptr, size); buf[size]=0; cstr=buf; } else @@ -733,7 +735,7 @@ double String::as_double() const { result=(double)strtod(cstr, &error_pos); if(*error_pos/*not EOS*/) - throw Exception(0, 0, + throw Exception("number.format", this, "invalid number (double)"); @@ -743,9 +745,9 @@ int String::as_int() const { int result; const char *cstr; char buf[MAX_NUMBER]; - if(head.rows+1==append_here) { - int size=min(head.rows[0].item.size, MAX_NUMBER-1); - memcpy(buf, head.rows[0].item.ptr, size); + if(head.chunk.rows+1==append_here) { + int size=min(head.chunk.rows[0].item.size, MAX_NUMBER-1); + memcpy(buf, head.chunk.rows[0].item.ptr, size); buf[size]=0; cstr=buf; } else @@ -761,7 +763,7 @@ int String::as_int() const { result=(int)strtol(cstr, &error_pos, 0); if(*error_pos/*not EOS*/) - throw Exception(0, 0, + throw Exception("number.format", this, "invalid number (int)"); @@ -802,26 +804,32 @@ void String::serialize(size_t prolog_siz cur+=row->item.size; ); } -void String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file) { +bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file) { if(buf_size<=prolog_size) - return; + return false; char *cur=(char *)buf+prolog_size; buf_size-=prolog_size; while(buf_size) { - uchar lang=*(uchar *)(cur); - + if(sizeof(uchar)+sizeof(ushort)>buf_size) // lang+size + return false; + + uchar lang=*(uchar *)(cur); ushort size=uchars2ushort( *(uchar*)(cur+sizeof(uchar)*1), *(uchar*)(cur+sizeof(uchar)*2) ); + size_t piece_size=sizeof(uchar)+sizeof(ushort)+size; + if(piece_size>buf_size) // buffer overrun, can be on incomplete cache files + return false; + const char *ptr=(const char*)(cur+sizeof(uchar)*3); APPEND(ptr, size, lang, file, 0); - size_t piece_size=sizeof(uchar)+sizeof(ushort)+size; cur+=piece_size; buf_size-=piece_size; } + return true; }