--- parser3/src/main/pa_string.C 2001/02/14 14:15:37 1.24 +++ parser3/src/main/pa_string.C 2001/04/23 13:38:31 1.78 @@ -1,44 +1,65 @@ -/* - $Id: pa_string.C,v 1.24 2001/02/14 14:15:37 paf Exp $ +/** @file + Parser: string class. @see untasize_t.C. + + Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) + + Author: Alexander Petrosyan (http://design.ru/paf) + + $Id: pa_string.C,v 1.78 2001/04/23 13:38:31 paf Exp $ */ -#include +#include "pa_config_includes.h" + +#include "pcre.h" #include "pa_pool.h" #include "pa_string.h" #include "pa_hash.h" #include "pa_exception.h" +#include "pa_common.h" +#include "pa_array.h" +#include "pa_globals.h" +#include "pa_table.h" +#include "pa_threads.h" + +//#include "pa_sapi.h" // String -String::String(Pool& apool) : +String::String(Pool& apool, const char *src, size_t src_size, bool tainted) : Pooled(apool) { - head.count=curr_chunk_rows=CR_PREALLOCATED_COUNT; + last_chunk=&head; + head.count=CR_PREALLOCATED_COUNT; append_here=head.rows; head.preallocated_link=0; - link_row=&head.rows[curr_chunk_rows]; + link_row=&head.rows[head.count]; fused_rows=fsize=0; + + if(src) + if(tainted) + APPEND_TAINTED(src, src_size, 0, 0); + else + APPEND_CLEAN(src, src_size, 0, 0); } void String::expand() { - curr_chunk_rows+=curr_chunk_rows*CR_GROW_PERCENT/100; - Chunk *chunk=static_cast( - pool.malloc(sizeof(int)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *))); - chunk->count=curr_chunk_rows; - link_row->link=chunk; - append_here=chunk->rows; - link_row=&chunk->rows[curr_chunk_rows]; + size_t new_chunk_count=last_chunk->count+last_chunk->count*CR_GROW_PERCENT/100; + last_chunk=static_cast( + malloc(sizeof(size_t)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *))); + last_chunk->count=new_chunk_count; + link_row->link=last_chunk; + append_here=last_chunk->rows; + link_row=&last_chunk->rows[last_chunk->count]; link_row->link=0; } -String::String(const String& src) : - Pooled(src.pool) { +String::String(const String& src) : Pooled(src.pool()) { head.count=CR_PREALLOCATED_COUNT; - int src_used_rows=src.used_rows(); + size_t src_used_rows=src.fused_rows; if(src_used_rows<=head.count) { - // all new rows fit into preallocated area - curr_chunk_rows=head.count; + // all new rows fit size_to preallocated area + size_t curr_chunk_rows=head.count; memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*src_used_rows); append_here=&head.rows[src_used_rows]; link_row=&head.rows[curr_chunk_rows]; @@ -51,22 +72,22 @@ String::String(const String& src) : // allocating only enough mem to fit src string rows // next append would allocate a new chunk // - // new rows don't fit into preallocated area: splitting into two chunks + // new rows don't fit size_to preallocated area: splitting size_to two chunks // preallocated chunk src to constructing head memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*head.count); - // remaining rows into new_chunk - curr_chunk_rows=src_used_rows-head.count; + // remaining rows size_to new_chunk + size_t curr_chunk_rows=src_used_rows-head.count; Chunk *new_chunk=static_cast( - pool.malloc(sizeof(int)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *))); + malloc(sizeof(size_t)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *))); new_chunk->count=curr_chunk_rows; head.preallocated_link=new_chunk; - append_here=link_row=&new_chunk->rows[curr_chunk_rows]; + append_here=link_row=&new_chunk->rows[new_chunk->count]; Chunk *old_chunk=src.head.preallocated_link; Chunk::Row *new_rows=new_chunk->rows; - int rows_left_to_copy=curr_chunk_rows; + size_t rows_left_to_copy=new_chunk->count; while(true) { - int old_count=old_chunk->count; + size_t old_count=old_chunk->count; Chunk *next_chunk=old_chunk->rows[old_count].link; if(next_chunk) { // not last source chunk @@ -88,24 +109,39 @@ String::String(const String& src) : fused_rows=src_used_rows; fsize=src.fsize; } -/* -String(const String_iterator& begin, const String_iterator& end) { - ;//TODO + +String& String::append(const String& src, Untaint_lang lang, bool forced) { + const Chunk *chunk=&src.head; + do { + const Chunk::Row *row=chunk->rows; + for(size_t i=0; icount; i++, row++) { + if(row==src.append_here) + goto break2; + + APPEND(row->item.ptr, row->item.size, + (lang!=UL_PASS_APPENDED && (row->item.lang==UL_TAINTED || forced))?lang:row->item.lang, + row->item.origin.file, row->item.origin.line); + } + chunk=row->link; + } while(chunk); +break2: + return *this; } -*/ String& String::real_append(STRING_APPEND_PARAMS) { if(!src) return *this; - int len=strlen(src); - if(!len) + if(!size) + size=strlen(src); + if(!size) return *this; if(chunk_is_full()) expand(); append_here->item.ptr=src; - fsize+=append_here->item.size=len; + fsize+=append_here->item.size=size; + append_here->item.lang=lang; #ifndef NO_STRING_ORIGIN append_here->item.origin.file=file; append_here->item.origin.line=line; @@ -115,35 +151,13 @@ String& String::real_append(STRING_APPEN return *this; } -char *String::cstr() const { - char *result=static_cast(pool.malloc(size()+1)); - - char *copy_here=result; - const Chunk *chunk=&head; - do { - const Chunk::Row *row=chunk->rows; - for(int i=0; icount; i++) { - if(row==append_here) - goto break2; - - memcpy(copy_here, row->item.ptr, row->item.size); - copy_here+=row->item.size; - row++; - } - chunk=row->link; - } while(chunk); -break2: - *copy_here=0; - return result; -} - uint String::hash_code() const { uint result=0; const Chunk *chunk=&head; do { const Chunk::Row *row=chunk->rows; - for(int i=0; icount; i++) { + for(size_t i=0; icount; i++) { if(row==append_here) goto break2; @@ -156,175 +170,419 @@ break2: return result; } -bool String::operator == (const String& src) const { - if(size() != src.size()) - return false; +/// @todo move 'lang' skipping to pos +int String::cmp(int& partial, const String& src, + size_t this_offset, Untaint_lang lang) const { + partial=-1; + this_offset=min(this_offset, size()-1); const Chunk *a_chunk=&head; const Chunk *b_chunk=&src.head; const Chunk::Row *a_row=a_chunk->rows; const Chunk::Row *b_row=b_chunk->rows; - int a_offset=0; - int b_offset=0; + size_t a_offset=this_offset; + size_t b_offset=0; Chunk::Row *a_end=append_here; Chunk::Row *b_end=src.append_here; - int a_countdown=a_chunk->count; - int b_countdown=b_chunk->count; + size_t a_countdown=a_chunk->count; + size_t b_countdown=b_chunk->count; bool a_break=false; bool b_break=false; + size_t result; + size_t pos=0; while(true) { - int size_diff= - (a_row->item.size-a_offset)- - (b_row->item.size-b_offset); - - if(size_diff==0) { // a has same size as b - if(memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, a_row->item.size-a_offset)!=0) - return false; - a_row++; a_countdown--; a_offset=0; - b_row++; b_countdown--; b_offset=0; - } else if (size_diff>0) { // a longer - if(memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, b_row->item.size-b_offset)!=0) - return false; - a_offset+=b_row->item.size-b_offset; - b_row++; b_countdown--; b_offset=0; - } else { // b longer - if(memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, a_row->item.size-a_offset)!=0) - return false; - b_offset+=a_row->item.size-a_offset; - a_row++; a_countdown--; a_offset=0; - } - a_break=a_row==a_end; b_break=b_row==b_end; if(a_break || b_break) break; + if(pos+a_row->item.size > this_offset) { + if(lang!=UL_UNSPECIFIED && a_row->item.lang!=lang) + return -1; // wrong lang -- bail out + + int size_diff= + (a_row->item.size-a_offset)- + (b_row->item.size-b_offset); + + if(size_diff==0) { // a has same size as b + result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, + a_row->item.size-a_offset); + if(result) + return result; + pos+=a_row->item.size; + a_row++; a_countdown--; a_offset=0; + b_row++; b_countdown--; b_offset=0; + } else if (size_diff>0) { // a longer + result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, + b_row->item.size-b_offset); + if(result) + return result; + a_offset+=b_row->item.size-b_offset; + b_row++; b_countdown--; b_offset=0; + } else { // b longer + result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, + a_row->item.size-a_offset); + if(result) + return result; + b_offset+=a_row->item.size-a_offset; + pos+=a_row->item.size; + a_row++; a_countdown--; a_offset=0; + } + + if(!b_countdown) { + b_chunk=b_row->link; + b_row=b_chunk->rows; + b_countdown=b_chunk->count; + } + } else { + a_offset-=a_row->item.size; + pos+=a_row->item.size; + a_row++; a_countdown--; + } + if(!a_countdown) { a_chunk=a_row->link; a_row=a_chunk->rows; a_countdown=a_chunk->count; } - if(!b_countdown) { - b_chunk=b_row->link; - b_row=b_chunk->rows; - b_countdown=b_chunk->count; - } } - return a_break==b_break; + if(a_break==b_break) { // ended simultaneously + partial=0; return 0; + } else if(a_break) { // first bytes equal, but a ended before b + partial=1; return -1; + } else { + partial=2; return +1; + } } -String& String::append(const String_iterator& begin, const String_iterator& end) { - //TODO - return *this; -} +/// @todo move 'lang' skipping to pos +int String::cmp(int& partial, const char* b_ptr, size_t src_size, + size_t this_offset, Untaint_lang lang) const { + partial=-1; + size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0; + this_offset=min(this_offset, size()-1); + + const Chunk *a_chunk=&head; + const Chunk::Row *a_row=a_chunk->rows; + size_t a_offset=this_offset; + size_t b_offset=0; + Chunk::Row *a_end=append_here; + size_t a_countdown=a_chunk->count; + bool a_break=false; + bool b_break=false; + size_t pos=0; + while(true) { + a_break=a_row==a_end; + if(a_break || b_break) + break; -// Char_types + if(pos+a_row->item.size > this_offset) { + if(lang!=UL_UNSPECIFIED && a_row->item.lang!=lang) + return -1; // wrong lang -- bail out + + int size_diff= + (a_row->item.size-a_offset)- + (b_size-b_offset); + + if(size_diff==0) { // a has same size as b + if(size_t result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, + a_row->item.size-a_offset)!=0) + return result; + pos+=a_row->item.size; + a_row++; a_countdown--; a_offset=0; + b_break=true; + } else if (size_diff>0) { // a longer + if(size_t result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, + b_size-b_offset)!=0) + return result; + a_offset+=b_size-b_offset; + b_break=true; + } else { // b longer + if(size_t result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, + a_row->item.size-a_offset)!=0) + return result; + b_offset+=a_row->item.size-a_offset; + pos+=a_row->item.size; + a_row++; a_countdown--; a_offset=0; + } + } else { + a_offset-=a_row->item.size; + pos+=a_row->item.size; + a_row++; a_countdown--; + } -Char_types::Char_types() { - memset(types, 0, sizeof(types)); + if(!a_countdown) { + a_chunk=a_row->link; + a_row=a_chunk->rows; + a_countdown=a_chunk->count; + } + } + if(a_break==b_break) { // ended simultaneously + partial=0; return 0; + } else if(a_break) { // first bytes equal, but a ended before b + partial=1; return -1; + } else { + partial=2; return +1; + } } -void Char_types::set(char from, char to, int type) { - memset(&types[static_cast(from)], type, to-from+1); +#ifndef NO_STRING_ORIGIN +const Origin& String::origin() const { + if(!fused_rows) + THROW(0, 0, + 0, + "String::origin() of empty string called"); + + // determining origin by last appended piece + // because first one frequently constant. + // ex: ^load[/file] "document_root" + "/file" + return append_here[-1].item.origin; } +#endif -// String_iterator +String& String::mid(size_t start, size_t finish) const { + start=max(0, start); + finish=min(size(), finish); + if(start==finish) + return *empty_string; -String_iterator::String_iterator(String& astring) : string(astring) { - read_here=string.head.rows; - position=string.size()==0?0:read_here->item.ptr; - link_row=reinterpret_cast(string.head.preallocated_link); -} + String& result=*NEW String(pool()); -char String_iterator::operator()() const { - return position?*position:0; -} + size_t pos=0; + const Chunk *chunk=&head; + do { + const Chunk::Row *row=chunk->rows; + for(size_t i=0; icount; pos+=row->item.size, i++, row++) { + if(row==append_here) + goto break2; -void String_iterator::skip() { - if(!position) - return; + size_t item_finish=pos+row->item.size; + if(item_finish > start) { // started now or already? + bool started=result.size()==0; // started now? + bool finished=finish <= item_finish; // finished now? + size_t offset=started?start-pos:0; + size_t size=finished?finish-pos:row->item.size; + result.APPEND( + row->item.ptr+offset, size-offset, + row->item.lang, + row->item.origin.file, row->item.origin.line); + if(finished) + goto break2; + } + } + chunk=row->link; + } while(chunk); +break2: +// SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'", + //cstr(), start, finish, result.cstr()); + return result; +} - if(++position== - read_here->item.ptr+ - read_here->item.size) { +int String::pos(const String& substr, + size_t result, Untaint_lang lang) const { + for(; resultlink; - if(!chunk) - string.pool.exception().raise(0, 0, - &string, - "String_iterator::skip() missed " - "read_here==string.append_here check"); +int String::pos(const char *substr, size_t substr_size, + size_t result, Untaint_lang lang) const { + for(; resultrows; - link_row=&chunk->rows[chunk->count]; - } - position=read_here->item.ptr; +void String::split(Array& result, + size_t* pos_after_ref, + const char *delim, size_t delim_size, + Untaint_lang lang, int limit) const { + if(delim_size) { + size_t pos_after=pos_after_ref?*pos_after_ref:0; + int pos_before; + // while we have 'delim'... + for(; (pos_before=pos(delim, delim_size, pos_after, lang))>=0 && limit; limit--) { + result+=&mid(pos_after, pos_before); + pos_after=pos_before+delim_size; + } + // last piece + if(pos_after=0 && limit; limit--) { + result+=&mid(pos_after, pos_before); + pos_after=pos_before+delim.size(); + } + // last piece + if(pos_after( - memchr(position, c, read_here->item.size-(position-read_here->item.ptr)))) { - position=found; - return true; - } +static void regex_options(char *options, int *result){ + struct Regex_option { + char key; + int clear, set; + int *result; + } regex_option[]={ + {'i', 0, PCRE_CASELESS, result}, // a=A + {'s', 0, PCRE_DOTALL, result}, // \n\n$ + {'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}, + }; + 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; + } +} - // next row - if(++read_here==string.append_here) { - position=0; - return false; - } - if(read_here==link_row) { - String::Chunk *chunk=link_row->link; - if(!chunk) - string.pool.exception().raise(0, 0, - &string, - "String_iterator::skip_to(char) missed " - "read_here==string.append_here check"); +/** + returns true if fills table. + table format is defined and fixed[can be used by others]: + @verbatim + pre-match/match/post-match/1/2/3/... + @endverbatim +*/ +bool String::match(const unsigned char *pcre_tables, + const String *aorigin, + const String& regexp, + const String *options, + Table **table, + Row_action row_action, void *info) const { + + if(!regexp.size()) + THROW(0, 0, + aorigin, + "regexp is empty"); + const char *pattern=regexp.cstr(UL_AS_IS); + const char *errptr; + int erroffset; + int option_bits[2]; regex_options(options?options->cstr():0, option_bits); + pcre *code=pcre_compile(pattern, option_bits[0], + &errptr, &erroffset, + pcre_tables); + + if(!code) + THROW(0, 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(0, 0, + aorigin, + "pcre_info error (%d)", + info_substrings); + } - read_here=chunk->rows; - link_row=&chunk->rows[chunk->count]; + int startoffset=0; + const char *subject=cstr(UL_AS_IS); + int length=strlen(subject); + int ovecsize; + int *ovector=(int *)malloc(sizeof(int)* + (ovecsize=(1/*match*/+info_substrings)*3)); + + { // create table + Array& columns=*NEW Array(pool()); + columns+=string_pre_match_name; + columns+=string_match_name; + columns+=string_post_match_name; + for(int i=1; i<=info_substrings; i++) { + char *column=(char *)malloc(MAX_NUMBER); + snprintf(column, MAX_NUMBER, "%d", i); + columns+=NEW String(pool(), column); // .i column name } - position=read_here->item.ptr; + *table=NEW Table(pool(), aorigin, &columns); } -} - -int String_iterator::skip_to(Char_types& types) { - if(!position) - return false; + int exec_option_bits=0; while(true) { - int countdown=read_here->item.size-(position-read_here->item.ptr); - for(; countdown--; position++) - if(int type=types.get(*position)) - return type; - - // next row - if(++read_here==string.append_here) { - position=0; - return -1; - } - if(read_here==link_row) { - String::Chunk *chunk=link_row->link; - if(!chunk) - string.pool.exception().raise(0, 0, - &string, - "String_iterator::skip_to(Char_type) missed " - "read_here==string.append_here check"); - - read_here=chunk->rows; - link_row=&chunk->rows[chunk->count]; + int exec_substrings=pcre_exec(code, 0, + subject, length, startoffset, + 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); + return option_bits[1]!=0; // global=true+table, not global=false + } + + if(exec_substrings<0) { + (*pcre_free)(code); + THROW(0, 0, + aorigin, + "regular expression execute error (%d)", + exec_substrings); + } + + Array& row=*NEW Array(pool()); + row+=&mid(0, ovector[0]); // .pre-match column value + row+=&mid(ovector[0], ovector[1]); // .match + row+=&mid(ovector[1], size()); // .post-match + + for(int i=1; iitem.ptr; + +/* + if(option_bits[0] & PCRE_MULTILINE) + exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL +*/ } }