|
|
| version 1.61, 2001/04/03 08:23:08 | version 1.237, 2011/05/05 23:45:45 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser: string class. @see untasize_t.C. | Parser: string class. @see untalength_t.C. |
| Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) | Copyright (c) 2001-2009 ArtLebedev Group (http://www.artlebedev.com) |
| Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) | |
| Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf) | |
| $Id$ | |
| */ | */ |
| #include "pa_config_includes.h" | static const char * const IDENT_STRING_C="$Date$"; |
| #include "pa_pool.h" | |
| #include "pa_string.h" | #include "pa_string.h" |
| #include "pa_hash.h" | |
| #include "pa_exception.h" | #include "pa_exception.h" |
| #include "pa_common.h" | |
| #include "pa_array.h" | |
| #include "pa_globals.h" | |
| #include "pa_table.h" | #include "pa_table.h" |
| #include "pa_dictionary.h" | |
| #include "pa_charset.h" | |
| #include "pa_vregex.h" | |
| const String String::Empty; | |
| int pa_atoi(const char* str, const String* problem_source) { | |
| if(!str) | |
| return 0; | |
| while(*str && isspace((unsigned char)*str)) | |
| str++; | |
| if(!*str) | |
| return 0; | |
| int result; | |
| char *error_pos; | |
| bool negative=false; | |
| if(str[0]=='-') { | |
| negative=true; | |
| str++; | |
| } else if(str[0]=='+') { | |
| str++; | |
| } | |
| // 0xABC | |
| if(str[0]=='0') | |
| if(str[1]=='x' || str[1]=='X') | |
| result=(int)(unsigned long)strtol(str, &error_pos, 0); | |
| else { | |
| // skip leading 0000, to disable octal interpretation | |
| do str++; while(*str=='0'); | |
| result=(int)strtol(str, &error_pos, 0); | |
| } | |
| else | |
| result=(int)strtol(str, &error_pos, 0); | |
| if(negative) | |
| result=-result; | |
| while(char c=*error_pos++) | |
| if(!isspace((unsigned char)c)) | |
| throw Exception("number.format", | |
| problem_source, | |
| problem_source?"invalid number (int)": "'%s' is invalid number (int)", str); | |
| //#include "pa_sapi.h" | return result; |
| // String | |
| String::String(Pool& apool, const char *src, bool tasize_ted) : | |
| Pooled(apool) { | |
| last_chunk=&head; | |
| head.count=CR_PREALLOCATED_COUNT; | |
| append_here=head.rows; | |
| head.preallocated_link=0; | |
| link_row=&head.rows[head.count]; | |
| fused_rows=fsize=0; | |
| if(src) | |
| if(tasize_ted) | |
| APPEND_TAINTED(src, 0, 0, 0); | |
| else | |
| APPEND_CONST(src); | |
| } | |
| void String::expand() { | |
| size_t new_chunk_count=last_chunk->count+last_chunk->count*CR_GROW_PERCENT/100; | |
| last_chunk=static_cast<Chunk *>( | |
| 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()) { | double pa_atod(const char* str, const String* problem_source) { |
| head.count=CR_PREALLOCATED_COUNT; | if(!str) |
| return 0; | |
| size_t src_used_rows=src.fused_rows; | |
| if(src_used_rows<=head.count) { | while(*str && isspace((unsigned char)*str)) |
| // all new rows fit size_to preallocated area | str++; |
| size_t curr_chunk_rows=head.count; | if(!*str) |
| memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*src_used_rows); | return 0; |
| append_here=&head.rows[src_used_rows]; | |
| link_row=&head.rows[curr_chunk_rows]; | double result; |
| } else { | char *error_pos; |
| // warning: | bool negative=false; |
| // heavily relies on the fact | if(str[0]=='-') { |
| // "preallocated area is the same for all strings" | negative=true; |
| // | str++; |
| // info: | } else if(str[0]=='+') { |
| // allocating only enough mem to fit src string rows | str++; |
| // next append would allocate a new chunk | |
| // | |
| // 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 size_to new_chunk | |
| size_t curr_chunk_rows=src_used_rows-head.count; | |
| Chunk *new_chunk=static_cast<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[new_chunk->count]; | |
| Chunk *old_chunk=src.head.preallocated_link; | |
| Chunk::Row *new_rows=new_chunk->rows; | |
| size_t rows_left_to_copy=new_chunk->count; | |
| while(true) { | |
| size_t old_count=old_chunk->count; | |
| Chunk *next_chunk=old_chunk->rows[old_count].link; | |
| if(next_chunk) { | |
| // not last source chunk | |
| // taking it all | |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*old_count); | |
| new_rows+=old_count; | |
| rows_left_to_copy-=old_count; | |
| old_chunk=next_chunk; | |
| } else { | |
| // the last source chunk | |
| // taking only those rows of chunk that _left_to_copy | |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*rows_left_to_copy); | |
| break; | |
| } | |
| } | |
| } | } |
| link_row->link=0; | // 0xABC |
| fused_rows=src_used_rows; | if(str[0]=='0') |
| fsize=src.fsize; | if(str[1]=='x' || str[1]=='X') |
| } | result=(double)(unsigned long)strtol(str, &error_pos, 0); |
| else { | |
| String& String::append(const String& src, Untaint_lang lang, bool forced) { | // skip leading 0000, to disable octal interpretation |
| const Chunk *chunk=&src.head; | do str++; while(*str=='0'); |
| do { | result=(double)strtod(str, &error_pos); |
| const Chunk::Row *row=chunk->rows; | } |
| for(size_t i=0; i<chunk->count; i++, row++) { | else |
| if(row==src.append_here) | result=(double)strtod(str, &error_pos); |
| goto break2; | if(negative) |
| result=-result; | |
| APPEND(row->item.ptr, row->item.size, | |
| (lang!=UL_PASS_APPENDED && (row->item.lang==UL_TAINTED || forced))?lang:row->item.lang, | while(char c=*error_pos++) |
| row->item.origin.file, row->item.origin.line); | if(!isspace((unsigned char)c)) |
| } | throw Exception("number.format", |
| chunk=row->link; | problem_source, |
| } while(chunk); | problem_source?"invalid number (double)": "'%s' is invalid number (double)", str); |
| break2: | |
| return *this; | |
| } | |
| String& String::real_append(STRING_APPEND_PARAMS) { | return result; |
| if(!src) | } |
| return *this; | |
| if(!size) | |
| size=strlen(src); | |
| if(!size) | |
| return *this; | |
| if(chunk_is_full()) | // cord lib extension |
| expand(); | |
| append_here->item.ptr=src; | #ifndef DOXYGEN |
| fsize+=append_here->item.size=size; | typedef struct { |
| append_here->item.lang=lang; | ssize_t countdown; |
| #ifndef NO_STRING_ORIGIN | int target; /* Character we're looking for */ |
| append_here->item.origin.file=file; | } chr_data; |
| append_here->item.origin.line=line; | |
| #endif | #endif |
| append_here++; fused_rows++; | static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data) |
| { | |
| register chr_data * d = (chr_data *)client_data; | |
| if (d -> countdown<=0) return(2); | |
| d -> countdown -= size; | |
| if (c > d -> target) return(1); | |
| return(0); | |
| } | |
| int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c) | |
| { | |
| chr_data d; | |
| d.countdown = n; | |
| d.target = c; | |
| return(CORD_block_iter(x, i, CORD_range_contains_chr_greater_then_proc, &d) == 1/*alternatives: 0 normally ended, 2=struck 'n'*/); | |
| } | |
| static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data) | |
| { | |
| int* result=(int*)client_data; | |
| (*result)++; | |
| return(0); // 0=continue | |
| } | |
| size_t CORD_block_count(CORD x) | |
| { | |
| size_t result=0; | |
| CORD_block_iter(x, 0, CORD_block_count_proc, &result); | |
| return result; | |
| } | |
| // helpers | |
| /// String::match uses this as replace & global search table columns | |
| const int MAX_MATCH_GROUPS=100; | |
| class String_match_table_template_columns: public ArrayString { | |
| public: | |
| String_match_table_template_columns() { | |
| *this+=new String("prematch"); | |
| *this+=new String("match"); | |
| *this+=new String("postmatch"); | |
| for(int i=0; i<MAX_MATCH_GROUPS; i++) { | |
| *this+=new String(String::Body::Format(1+i), String::L_CLEAN); | |
| } | |
| } | |
| }; | |
| return *this; | Table string_match_table_template(new String_match_table_template_columns); |
| } | |
| uint String::hash_code() const { | // String::Body methods |
| uint result=0; | |
| const Chunk *chunk=&head; | String::Body String::Body::Format(int value) { |
| do { | char local[MAX_NUMBER]; |
| const Chunk::Row *row=chunk->rows; | size_t length=snprintf(local, MAX_NUMBER, "%d", value); |
| for(size_t i=0; i<chunk->count; i++) { | return String::Body(pa_strdup(local, length)); |
| if(row==append_here) | |
| goto break2; | |
| result=Hash::generic_code(result, row->item.ptr, row->item.size); | |
| row++; | |
| } | |
| chunk=row->link; | |
| } while(chunk); | |
| break2: | |
| return result; | |
| } | } |
| /// @todo move 'lang' skipping to pos | String::Body String::Body::trim(String::Trim_kind kind, const char* chars, |
| int String::cmp(int& partial, const String& src, | size_t* out_start, size_t* out_length, Charset* source_charset) const { |
| size_t this_offset, Untaint_lang lang) const { | size_t our_length=length(); |
| partial=-1; | if(!our_length) |
| this_offset=min(this_offset, size()-1); | return *this; |
| 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; | |
| size_t a_offset=this_offset; | |
| size_t b_offset=0; | |
| Chunk::Row *a_end=append_here; | |
| Chunk::Row *b_end=src.append_here; | |
| 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) { | |
| 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) { | // check if any UTF-8 in chars |
| if(lang!=UL_UNKNOWN && a_row->item.lang!=lang) | bool fast=true; |
| return -1; // wrong lang -- bail out | if(chars && source_charset && source_charset->isUTF8()){ |
| const char* pos=chars; | |
| int size_diff= | while(unsigned char c=*pos++) |
| (a_row->item.size-a_offset)- | if(c>127){ |
| (b_row->item.size-b_offset); | fast=false; |
| break; | |
| 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; | size_t start=0; |
| b_row=b_chunk->rows; | size_t end=our_length; |
| b_countdown=b_chunk->count; | if(!chars) |
| chars=" \t\n"; // white space | |
| if(fast){ | |
| // from left... | |
| if(kind!=TRIM_END) { | |
| CORD_pos pos; set_pos(pos, 0); | |
| while(true) { | |
| char c=CORD_pos_fetch(pos); | |
| if(strchr(chars, c)) { | |
| if(++start==our_length) | |
| return 0; // all chars are empty, just return empty string | |
| } else | |
| break; | |
| CORD_next(pos); | |
| } | } |
| } else { | |
| a_offset-=a_row->item.size; | |
| pos+=a_row->item.size; | |
| a_row++; a_countdown--; | |
| } | } |
| if(!a_countdown) { | // from right.. |
| a_chunk=a_row->link; | if(kind!=TRIM_START) { |
| a_row=a_chunk->rows; | CORD_pos pos; set_pos(pos, end-1); |
| a_countdown=a_chunk->count; | while(true) { |
| char c=CORD_pos_fetch(pos); | |
| if(strchr(chars, c)) { | |
| if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible | |
| return 0; // all chars are empty, just return empty string | |
| } else | |
| break; | |
| CORD_prev(pos); | |
| } | |
| } | } |
| } | |
| 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 { | } else { |
| partial=2; return +1; | const XMLByte* src_begin=(const XMLByte*)cstr(); |
| } | const XMLByte* src_end=src_begin+our_length; |
| } | |
| /// @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; | |
| if(pos+a_row->item.size > this_offset) { | // from left... |
| if(lang!=UL_UNKNOWN && a_row->item.lang!=lang) | if(kind!=TRIM_END) { |
| return -1; // wrong lang -- bail out | while(src_begin<src_end){ |
| uint char_length=1; | |
| int size_diff= | const XMLByte* ptr=src_begin; |
| (a_row->item.size-a_offset)- | // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3 |
| (b_size-b_offset); | while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0)) |
| char_length++; | |
| 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, | bool found=false; |
| a_row->item.size-a_offset)!=0) | for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++) |
| return result; | if(strncmp(chars_byte, (const char*)ptr, char_length)==0){ |
| pos+=a_row->item.size; | found=true; |
| a_row++; a_countdown--; a_offset=0; | break; |
| 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, | if(found){ |
| b_size-b_offset)!=0) | start+=char_length; |
| return result; | if(start==our_length) |
| a_offset+=b_size-b_offset; | return 0; // all chars are empty, just return empty string |
| b_break=true; | } else |
| } else { // b longer | break; |
| 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--; | |
| } | } |
| if(!a_countdown) { | // from right.. |
| a_chunk=a_row->link; | if(kind!=TRIM_START) { |
| a_row=a_chunk->rows; | while(src_begin<src_end){ |
| a_countdown=a_chunk->count; | uint char_length=1; |
| // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3 | |
| while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0)) | |
| char_length++; | |
| bool found=false; | |
| for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *src_end); chars_byte++) | |
| if(strncmp(chars_byte, (const char*)src_end, char_length)==0){ | |
| found=true; | |
| break; | |
| } | |
| if(found){ | |
| end-=char_length; | |
| if(end==0) | |
| return 0; // all chars are empty, just return empty string | |
| } else | |
| break; | |
| } | |
| } | } |
| } | } |
| if(a_break==b_break) { // ended simultaneously | |
| partial=0; return 0; | if(start==0 && end==our_length) // nobody moved a thing |
| } else if(a_break) { // first bytes equal, but a ended before b | return *this; |
| partial=1; return -1; | |
| if(out_start) | |
| *out_start=start; | |
| size_t new_length=end-start; | |
| if(out_length) | |
| *out_length=new_length; | |
| return mid(start, new_length); | |
| } | |
| static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) { | |
| uint& result=*static_cast<uint*>(client_data); | |
| generic_hash_code(result, c); | |
| return 0; | |
| } | |
| static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) { | |
| uint& result=*static_cast<uint*>(client_data); | |
| generic_hash_code(result, s); | |
| return 0; | |
| }; | |
| uint String::Body::get_hash_code() const { | |
| #ifdef HASH_CODE_CACHING | |
| if(hash_code) | |
| return hash_code; | |
| #else | |
| uint hash_code=0; | |
| #endif | |
| if (body && CORD_IS_STRING(body)){ | |
| generic_hash_code(hash_code, body); | |
| } else { | } else { |
| partial=2; return +1; | CORD_iter5(body, 0, |
| CORD_batched_iter_fn_generic_hash_code, | |
| CORD_batched_iter_fn_generic_hash_code, &hash_code); | |
| } | } |
| return hash_code; | |
| } | } |
| #ifndef NO_STRING_ORIGIN | // String methods |
| const Origin& String::origin() const { | |
| if(!fused_rows) | String& String::append_know_length(const char* str, size_t known_length, Language lang) { |
| THROW(0, 0, | if(!known_length) |
| 0, | return *this; |
| "String::origin() of empty string called"); | |
| // first: langs | |
| // determining origin by last appended piece | langs.append(body, lang, known_length); |
| // because first one frequently constant. | // next: letters themselves |
| // ex: ^load[/file] "document_root" + "/file" | body.append_know_length(str, known_length); |
| return append_here[-1].item.origin; | |
| ASSERT_STRING_INVARIANT(*this); | |
| return *this; | |
| } | } |
| #endif | String& String::append_help_length(const char* str, size_t helper_length, Language lang) { |
| if(!str) | |
| return *this; | |
| size_t known_length=helper_length?helper_length:strlen(str); | |
| if(!known_length) | |
| return *this; | |
| String& String::piece(size_t start, size_t finish) const { | return append_know_length(str, known_length, lang); |
| start=max(0, start); | } |
| finish=min(size(), finish); | String::String(int value, char *format) : langs(L_CLEAN){ |
| if(start==finish) | char buf[MAX_NUMBER]; |
| return *empty_string; | body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value)); |
| } | |
| String& result=*NEW String(pool()); | String& String::append_strdup(const char* str, size_t helper_length, Language lang) { |
| size_t known_length=helper_length?helper_length:strlen(str); | |
| size_t pos=0; | if(!known_length) |
| const Chunk *chunk=&head; | return *this; |
| do { | |
| const Chunk::Row *row=chunk->rows; | // first: langs |
| for(size_t i=0; i<chunk->count; pos+=row->item.size, i++, row++) { | langs.append(body, lang, known_length); |
| if(row==append_here) | // next: letters themselves |
| goto break2; | body.append_strdup_know_length(str, known_length); |
| size_t item_finish=pos+row->item.size; | ASSERT_STRING_INVARIANT(*this); |
| if(item_finish > start) { // started now or already? | return *this; |
| bool started=result.size()==0; // started now? | } |
| bool finished=finish <= item_finish; // finished now? | |
| size_t offset=started?start-pos:0; | struct CORD_length_info { |
| size_t size=finished?finish-pos:row->item.size; | size_t len; |
| result.APPEND( | size_t skip; |
| row->item.ptr+offset, size-offset, | }; |
| row->item.lang, | |
| row->item.origin.file, row->item.origin.line); | int CORD_batched_len(const char* s, CORD_length_info* info){ |
| if(finished) | info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s)); return 0; |
| goto break2; | } |
| } | |
| } | // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid |
| chunk=row->link; | int CORD_batched_len(const char c, CORD_length_info* info){ |
| } while(chunk); | if (info->skip==0){ |
| break2: | info->len++; |
| // SAPI::log(pool(), "piece of '%s' from %d to %d is '%s'", | info->skip = lengthUTF8Char(c)-1; |
| //cstr(), start, finish, result.cstr()); | } else { |
| info->skip--; | |
| } | |
| return 0; | |
| } | |
| size_t String::length(Charset& charset) const { | |
| if(charset.isUTF8()){ | |
| CORD_length_info info = {0, 0}; | |
| body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info); | |
| return info.len; | |
| } else | |
| return body.length(); | |
| } | |
| /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab" | |
| String& String::mid(size_t substr_begin, size_t substr_end) const { | |
| String& result=*new String; | |
| size_t self_length=length(); | |
| substr_begin=min(substr_begin, self_length); | |
| substr_end=min(max(substr_end, substr_begin), self_length); | |
| size_t substr_length=substr_end-substr_begin; | |
| if(!substr_length) | |
| return result; | |
| // first: their langs | |
| result.langs.append(result.body, langs, substr_begin, substr_length); | |
| // next: letters themselves | |
| result.body=body.mid(substr_begin, substr_length); | |
| ASSERT_STRING_INVARIANT(result); | |
| return result; | return result; |
| } | } |
| int String::pos(const String& substr, | // from, to and helper_length in characters, not in bytes (it's important for utf-8) |
| size_t result, Untaint_lang lang) const { | String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const { |
| for(; result<size(); result++) { | String& result=*new String; |
| int partial; cmp(partial, substr, result, lang); | |
| if( | size_t self_length=(helper_length)?helper_length:length(charset); |
| partial==0 || // full match | |
| partial==2) // 'substr' starts 'this'+'result' | if(!self_length) |
| return result; | |
| from=min(min(to, from), self_length); | |
| to=min(max(to, from), self_length); | |
| size_t substr_length=to-from; | |
| if(!substr_length) | |
| return result; | |
| if(charset.isUTF8()){ | |
| const XMLByte* src_begin=(const XMLByte*)cstr(); | |
| const XMLByte* src_end=src_begin+body.length(); | |
| // convert 'from' and 'substr_length' from 'characters' to 'bytes' | |
| from=getUTF8BytePos(src_begin, src_end, from); | |
| substr_length=getUTF8BytePos(src_begin+from, src_end, substr_length); | |
| if(!substr_length) | |
| return result; | return result; |
| } | } |
| return -1; | // first: their langs |
| result.langs.append(result.body, langs, from, substr_length); | |
| // next: letters themselves | |
| result.body=body.mid(from, substr_length); | |
| ASSERT_STRING_INVARIANT(result); | |
| return result; | |
| } | } |
| int String::pos(const char *substr, size_t substr_size, | size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const { |
| size_t result, Untaint_lang lang) const { | size_t substr_length=substr.length(); |
| for(; result<size(); result++) { | while(true) { |
| int partial; cmp(partial, substr, substr_size, result, lang); | size_t substr_begin=body.pos(substr, this_offset); |
| if( | |
| partial==0 || // full match | if(substr_begin==CORD_NOT_FOUND) |
| partial==2) // 'substr' starts 'this'+'result' | return STRING_NOT_FOUND; |
| return result; | |
| if(langs.check_lang(lang, substr_begin, substr_length)) | |
| return substr_begin; | |
| this_offset=substr_begin+substr_length; | |
| } | } |
| return -1; | |
| } | } |
| void String::split(Array& result, | size_t String::pos(const String& substr, |
| size_t* pos_after_ref, | size_t this_offset, Language lang) const { |
| const char *delim, size_t delim_size, | return pos(substr.body, this_offset, lang); |
| Untaint_lang lang, int limit) const { | } |
| if(delim_size) { | |
| size_t pos_after=pos_after_ref?*pos_after_ref:0; | size_t String::pos(Charset& charset, const String& substr, |
| int pos_before; | size_t this_offset, Language lang) const { |
| if(charset.isUTF8()){ | |
| const XMLByte* srcPtr=(const XMLByte*)cstr(); | |
| const XMLByte* srcEnd=srcPtr+body.length(); | |
| // convert 'this_offset' from 'characters' to 'bytes' | |
| this_offset=getUTF8BytePos(srcPtr, srcEnd, this_offset); | |
| size_t result=pos(substr.body, this_offset, lang); | |
| return (result==CORD_NOT_FOUND) | |
| ? STRING_NOT_FOUND | |
| : getUTF8CharPos(srcPtr, srcEnd, result); // convert 'result' from 'bytes' to 'characters' | |
| } else { | |
| size_t result=pos(substr.body, this_offset, lang); | |
| return (result==CORD_NOT_FOUND) | |
| ? STRING_NOT_FOUND | |
| : result; | |
| } | |
| } | |
| void String::split(ArrayString& result, | |
| size_t& pos_after, | |
| const char* delim, | |
| Language lang, int limit) const { | |
| if(is_empty()) | |
| return; | |
| size_t self_length=length(); | |
| if(size_t delim_length=strlen(delim)) { | |
| size_t pos_before; | |
| // while we have 'delim'... | // while we have 'delim'... |
| for(; (pos_before=pos(delim, delim_size, pos_after, lang))>=0 && limit; limit--) { | for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) { |
| result+=&piece(pos_after, pos_before); | result+=&mid(pos_after, pos_before); |
| pos_after=pos_before+delim_size; | pos_after=pos_before+delim_length; |
| } | } |
| // last piece | // last piece |
| if(pos_after<size() && limit) { | if(pos_after<self_length && limit) { |
| result+=&piece(pos_after, size()); | result+=&mid(pos_after, self_length); |
| pos_after=size(); | pos_after=self_length; |
| } | } |
| if(pos_after_ref) | |
| *pos_after_ref=pos_after; | |
| } else { // empty delim | } else { // empty delim |
| result+=this; | result+=this; |
| if(pos_after_ref) | pos_after+=self_length; |
| *pos_after_ref+=size(); | |
| } | } |
| } | } |
| void String::split(Array& result, | void String::split(ArrayString& result, |
| size_t* pos_after_ref, | size_t& pos_after, |
| const String& delim, Untaint_lang lang, | const String& delim, Language lang, |
| int limit) const { | int limit) const { |
| if(delim.size()) { | if(is_empty()) |
| size_t pos_after=pos_after_ref?*pos_after_ref:0; | return; |
| int pos_before; | if(!delim.is_empty()) { |
| size_t pos_before; | |
| // while we have 'delim'... | // while we have 'delim'... |
| for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) { | for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) { |
| result+=&piece(pos_after, pos_before); | result+=&mid(pos_after, pos_before); |
| pos_after=pos_before+delim.size(); | pos_after=pos_before+delim.length(); |
| } | } |
| // last piece | // last piece |
| if(pos_after<size() && limit) { | if(pos_after<length() && limit) { |
| result+=&piece(pos_after, size()); | result+=&mid(pos_after, length()); |
| pos_after=size(); | pos_after=length(); |
| } | } |
| if(pos_after_ref) | |
| *pos_after_ref=pos_after; | |
| } else { // empty delim | } else { // empty delim |
| result+=this; | result+=this; |
| if(pos_after_ref) | pos_after+=length(); |
| *pos_after_ref+=size(); | |
| } | } |
| } | } |
| Table& String::match(const String *aorigin, | Table* String::match(VRegex* vregex, |
| const String& regexp, const String& options) const { | Row_action row_action, void *info, |
| Array *columns=0; | int& matches_count) const { |
| Table& result=*NEW Table(pool(), aorigin, columns); | |
| // vregex->info(); // I have no idea what does it for? | |
| bool need_pre_post_match=vregex->is_pre_post_match_needed(); | |
| bool global=vregex->is_global_search(); | |
| const char* subject=cstr(); | |
| size_t subject_length=length(); | |
| const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3; | |
| int ovector[ovector_size]; | |
| Table::Action_options table_options; | |
| Table& table=*new Table(string_match_table_template, table_options); | |
| int prestart=0; | |
| int poststart=0; | |
| int postfinish=length(); | |
| while(true) { | |
| int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart); | |
| if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception | |
| break; | |
| int prefinish=ovector[0]; | |
| poststart=ovector[1]; | |
| if (prestart==poststart && subject[poststart]=='\n'){ | |
| prestart++; | |
| continue; | |
| } | |
| ArrayString* row=new ArrayString(3); | |
| if(need_pre_post_match) { | |
| *row+=&mid(0, prefinish); // .prematch column value | |
| *row+=&mid(prefinish, poststart); // .match | |
| *row+=&mid(poststart, postfinish); // .postmatch | |
| } else { | |
| *row+=&Empty; // .prematch column value | |
| *row+=&Empty; // .match | |
| *row+=&Empty; // .postmatch | |
| } | |
| for(int i=1; i<exec_result; i++) { | |
| // -1:-1 case handled peacefully by mid() itself | |
| *row+=(ovector[i*2+0]>=0 && ovector[i*2+1]>0)?&mid(ovector[i*2+0], ovector[i*2+1]):new String; // .i column value | |
| } | |
| matches_count++; | |
| row_action(table, row, prestart, prefinish, poststart, postfinish, info); | |
| if(!global || prestart==poststart) // last step | |
| break; | |
| prestart=poststart; | |
| } | |
| row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info); | |
| return vregex->is_just_count() ? 0 : &table; | |
| } | |
| String& String::change_case(Charset& source_charset, Change_case_kind kind) const { | |
| String& result=*new String(); | |
| if(is_empty()) | |
| return result; | |
| char* new_cstr=cstrm(); | |
| if(source_charset.isUTF8()) { | |
| size_t new_cstr_len=length(); | |
| switch(kind) { | |
| case CC_UPPER: | |
| change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper); | |
| break; | |
| case CC_LOWER: | |
| change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower); | |
| break; | |
| default: | |
| assert(!"unknown change case kind"); | |
| break; // never | |
| } | |
| } else { | |
| const unsigned char *tables=source_charset.pcre_tables; | |
| const unsigned char *a; | |
| const unsigned char *b; | |
| switch(kind) { | |
| case CC_UPPER: | |
| a=tables+lcc_offset; | |
| b=tables+fcc_offset; | |
| break; | |
| case CC_LOWER: | |
| a=tables+lcc_offset; | |
| b=0; | |
| break; | |
| default: | |
| assert(!"unknown change case kind"); | |
| a=b=0; // calm, compiler | |
| break; // never | |
| } | |
| char *dest=new_cstr; | |
| unsigned char index; | |
| for(const char* current=new_cstr; (index=(unsigned char)*current); current++) { | |
| unsigned char c=a[index]; | |
| if(b) | |
| c=b[c]; | |
| *dest++=(char)c; | |
| } | |
| } | |
| result.langs=langs; | |
| result.body=new_cstr; | |
| return result; | |
| } | |
| const String& String::escape(Charset& source_charset) const { | |
| if(is_empty()) | |
| return *this; | |
| return Charset::escape(*this, source_charset); | |
| } | |
| const String& String::replace(const Dictionary& dict) const { | |
| String& result=*new String(); | |
| const char* old_cstr=cstr(); | |
| const char* prematch_begin=old_cstr; | |
| const char* current=old_cstr; | |
| while(*current) { | |
| if(Dictionary::Subst subst=dict.first_that_begins(current)) { | |
| // prematch | |
| if(size_t prematch_length=current-prematch_begin) { | |
| result.langs.append(result.body, langs, prematch_begin-old_cstr, prematch_length); | |
| result.body.append_strdup_know_length(prematch_begin, prematch_length); | |
| } | |
| // match | |
| // skip 'a' in 'current'; move prematch_begin | |
| current+=subst.from_length; prematch_begin=current; | |
| if(const String* b=subst.to) // are there any b? | |
| result<<*b; | |
| } else // simply advance | |
| current++; | |
| } | |
| // postmatch | |
| if(size_t postmatch_length=current-prematch_begin) { | |
| result.langs.append(result.body, langs, prematch_begin-old_cstr, postmatch_length); | |
| result.body.append_strdup_know_length(prematch_begin, postmatch_length); | |
| } | |
| ASSERT_STRING_INVARIANT(result); | |
| return result; | |
| } | |
| static int serialize_body_char(char c, char** cur) { | |
| *((*cur)++)=c; | |
| return 0; // 0=continue | |
| }; | |
| static int serialize_body_piece(const char* s, char** cur) { | |
| size_t length=strlen(s); | |
| memcpy(*cur, s, length); *cur+=length; | |
| return 0; // 0=continue | |
| }; | |
| static int serialize_lang_piece(char alang, size_t asize, char** cur) { | |
| // lang | |
| **cur=alang; (*cur)++; | |
| // length [WARNING: not cast, addresses must be %4=0 on sparc] | |
| memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize); | |
| return 0; // 0=continue | |
| } | |
| String::Cm String::serialize(size_t prolog_length) const { | |
| size_t fragments_count=langs.count(); | |
| size_t body_length=body.length(); | |
| size_t buf_length= | |
| prolog_length //1 | |
| +sizeof(size_t) //2 | |
| +body_length //3 | |
| +1 // 4 for zero terminator used in deserialize | |
| +sizeof(size_t) //5 | |
| +fragments_count*(sizeof(char)+sizeof(size_t)); //6 | |
| String::Cm result(new(PointerFreeGC) char[buf_length], buf_length); | |
| // 1: prolog | |
| char *cur=result.str+prolog_length; | |
| // 2: chars.count [WARNING: not cast, addresses must be %4=0 on sparc] | |
| memcpy(cur, &body_length, sizeof(body_length)); cur+=sizeof(body_length); | |
| // 3: letters | |
| body.for_each(serialize_body_char, serialize_body_piece, &cur); | |
| // 4: zero terminator | |
| *cur++=0; | |
| // 5: langs.count [WARNING: not cast, addresses must be %4=0 on sparc] | |
| memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count); | |
| // 6: lang info | |
| langs.for_each(body, serialize_lang_piece, &cur); | |
| return result; | |
| } | |
| bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size) { | |
| size_t in_buf=buf_size; | |
| if(in_buf<=prolog_size) | |
| return false; | |
| in_buf-=prolog_size; | |
| // 1: prolog | |
| const char* cur=(const char* )buf+prolog_size; | |
| // 2: chars.count | |
| size_t body_length; | |
| if(in_buf<sizeof(body_length)) // body.length don't fit? | |
| return false; | |
| // [WARNING: not cast, addresses must be %4=0 on sparc] | |
| memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length); | |
| in_buf-=sizeof(body_length); | |
| if(in_buf<body_length+1) // letters+terminator don't fit? | |
| return false; | |
| // 4: zero terminator | |
| if(cur[body_length] != 0) // in place? | |
| return false; | |
| // 3: letters | |
| body=String::Body(*cur?cur:0); | |
| #ifdef STRING_LENGTH_CACHING | |
| body.set_length(body_length); | |
| #endif | |
| cur+=body_length+1; | |
| in_buf-=body_length+1; | |
| // 5: langs.count | |
| size_t fragments_count; | |
| if(in_buf<sizeof(fragments_count)) // langs.count don't fit? | |
| return false; | |
| // [WARNING: not cast, addresses must be %4=0 on sparc] | |
| memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count); | |
| in_buf-=sizeof(fragments_count); | |
| if(fragments_count) { | |
| // 6: lang info | |
| size_t total_length=0; | |
| for(size_t f=0; f<fragments_count; f++) { | |
| char lang; | |
| size_t fragment_length; | |
| size_t piece_length=sizeof(lang)+sizeof(fragment_length); | |
| if(in_buf<piece_length) // lang+length | |
| return false; | |
| // lang | |
| lang=*cur++; | |
| // length [WARNING: not cast, addresses must be %4=0 on sparc] | |
| memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length); | |
| size_t combined_length=total_length+fragment_length; | |
| if(combined_length>body_length) | |
| return false; // file curruption | |
| // uchar needed to prevent propagating 0x80 bit to upper bytes | |
| langs.append(total_length, (String::Language)(uchar)lang, fragment_length); | |
| total_length=combined_length; | |
| in_buf-=piece_length; | |
| } | |
| if(total_length!=body_length) // length(all language fragments) vs length(letters) | |
| return false; | |
| } | |
| if(in_buf!=0) // some strange extra bytes | |
| return false; | |
| ASSERT_STRING_INVARIANT(*this); | |
| return true; | |
| } | |
| const char* String::Body::v() const { | |
| return CORD_to_const_char_star(body, length()); | |
| } | |
| void String::Body::dump() const { | |
| CORD_dump(body); | |
| } | |
| const char* String::Languages::v() const { | |
| if(opt.is_not_just_lang) | |
| return CORD_to_const_char_star(langs, 0); | |
| else | |
| return (const char*)&langs; | |
| } | |
| void String::Languages::dump() const { | |
| if(opt.is_not_just_lang) | |
| CORD_dump(langs); | |
| else | |
| puts((const char*)&langs); | |
| } | |
| const char* String::v() const { | |
| const uint LIMIT_VIEW=20; | |
| char* buf=(char*)malloc(MAX_STRING); | |
| const char*body_view=body.v(); | |
| const char*langs_view=langs.v(); | |
| snprintf(buf, MAX_STRING, | |
| "%d:%.*s%s} " | |
| "{%d:%s", | |
| langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"", | |
| strlen(body_view), body_view | |
| ); | |
| return buf; | |
| } | |
| void String::dump() const { | |
| body.dump(); | |
| langs.dump(); | |
| } | |
| const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const { | |
| if(is_empty()) | |
| return *this; | |
| size_t substr_begin, substr_length; | |
| Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset); | |
| if(new_body==body) // we received unchanged pointer, do likewise | |
| return *this; | |
| // new_body differs from body, adjust langs along | |
| String& result=*new String; | |
| if(!new_body) // body.trim produced empty result | |
| return result; | |
| // body.trim produced nonempty result | |
| // first: their langs | |
| result.langs.append(result.body, langs, substr_begin, substr_length); | |
| // next: letters themselves | |
| result.body=new_body; | |
| ASSERT_STRING_INVARIANT(result); | |
| return result; | return result; |
| } | } |