|
|
| version 1.78, 2001/04/23 13:38:31 | version 1.280, 2024/11/10 20:46:08 |
|---|---|
| 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) 2001-2024 Art. Lebedev Studio (http://www.artlebedev.com) |
| Authors: Konstantin Morshnev <moko@design.ru>, Alexandr Petrosian <paf@design.ru> | |
| Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf) | |
| $Id$ | |
| */ | */ |
| #include "pa_config_includes.h" | |
| #include "pcre.h" | |
| #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_threads.h" | #include "pa_dictionary.h" |
| #include "pa_charset.h" | |
| #include "pa_vregex.h" | |
| volatile const char * IDENT_PA_STRING_C="$Id$" IDENT_PA_STRING_H; | |
| const String String::Empty; | |
| #define COMPILE_ASSERT(x) extern int assert_checker[(x) ? 1 : -1] | |
| COMPILE_ASSERT(sizeof(String::Languages) == sizeof(CORD)); | |
| // pa_atoui is based on Manuel Novoa III _strto_l for uClibc | |
| template<typename T> inline T pa_ato_any(const char *str, int base, const String* problem_source,const T max){ | |
| T result = 0; | |
| const char *pos = str; | |
| while (isspace(*pos)) /* skip leading whitespace */ | |
| ++pos; | |
| if (base == 16 && *pos == '0') { /* handle option prefix */ | |
| ++pos; | |
| if (*pos == 'x' || *pos == 'X') { | |
| ++pos; | |
| } | |
| } | |
| if (base == 0) { /* dynamic base */ | |
| base = 10; /* default is 10 */ | |
| if (*pos == '0') { | |
| ++pos; | |
| if (*pos == 'x' || *pos == 'X'){ | |
| ++pos; | |
| base=16; | |
| } | |
| } | |
| } | |
| //#include "pa_sapi.h" | if (base < 2 || base > 16) /* illegal base */ |
| throw Exception(PARSER_RUNTIME, 0, "base to must be an integer from 2 to 16"); | |
| if (*pos == '-') | |
| throw Exception("number.format", problem_source, problem_source ? "out of range (negative)" : "'%s' is out if range (negative)", str); | |
| // String | T cutoff = max / base; |
| int cutoff_digit = (int)(max - cutoff * base); | |
| String::String(Pool& apool, const char *src, size_t src_size, bool tainted) : | while(true) { |
| Pooled(apool) { | int digit; |
| last_chunk=&head; | |
| head.count=CR_PREALLOCATED_COUNT; | if ((*pos >= '0') && (*pos <= '9')) { |
| append_here=head.rows; | digit = (*pos - '0'); |
| head.preallocated_link=0; | } else if (*pos >= 'a') { |
| link_row=&head.rows[head.count]; | digit = (*pos - 'a' + 10); |
| fused_rows=fsize=0; | } else if (*pos >= 'A') { |
| digit = (*pos - 'A' + 10); | |
| if(src) | } else break; |
| if(tainted) | |
| APPEND_TAINTED(src, src_size, 0, 0); | if (digit >= base) { |
| else | break; |
| APPEND_CLEAN(src, src_size, 0, 0); | } |
| } | |
| ++pos; | |
| void String::expand() { | |
| size_t new_chunk_count=last_chunk->count+last_chunk->count*CR_GROW_PERCENT/100; | /* adjust number, with overflow check */ |
| last_chunk=static_cast<Chunk *>( | if ((result > cutoff) || ((result == cutoff) && (digit > cutoff_digit))) { |
| malloc(sizeof(size_t)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *))); | throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str); |
| last_chunk->count=new_chunk_count; | } else { |
| link_row->link=last_chunk; | result = result * base + digit; |
| append_here=last_chunk->rows; | } |
| link_row=&last_chunk->rows[last_chunk->count]; | } |
| link_row->link=0; | |
| while(char c=*pos++) | |
| if(!isspace(c)) | |
| throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is an invalid number (int)", str); | |
| return result; | |
| } | } |
| String::String(const String& src) : Pooled(src.pool()) { | unsigned int pa_atoui(const char *str, int base, const String* problem_source){ |
| head.count=CR_PREALLOCATED_COUNT; | if(!str) |
| return 0; | |
| return pa_ato_any<unsigned int>(str, base, problem_source, UINT_MAX); | |
| } | |
| uint64_t pa_atoul(const char *str, int base, const String* problem_source){ | |
| if(!str) | |
| return 0; | |
| return pa_ato_any<uint64_t>(str, base, problem_source, ULLONG_MAX); | |
| } | |
| int pa_atoi(const char* str, int base, const String* problem_source) { | |
| if(!str) | |
| return 0; | |
| while(isspace(*str)) | |
| str++; | |
| if(!*str) | |
| return 0; | |
| const char *str_copy=str; | |
| bool negative=false; | |
| if(str[0]=='-') { | |
| negative=true; | |
| str++; | |
| if(!*str || isspace(*str)) | |
| throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is an invalid number (int)", str_copy); | |
| } else if(str[0]=='+') { | |
| str++; | |
| if(!*str || isspace(*str)) | |
| throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is an invalid number (int)", str_copy); | |
| } | |
| unsigned int result=pa_atoui(str, base, problem_source); | |
| if(negative && result <= ((unsigned int)(-(1+INT_MIN)))+1) | |
| return -(int)result; | |
| size_t src_used_rows=src.fused_rows; | if(result<=INT_MAX) |
| if(src_used_rows<=head.count) { | return (int)result; |
| // all new rows fit size_to preallocated area | |
| size_t curr_chunk_rows=head.count; | throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str_copy); |
| 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]; | double pa_atod(const char* str, const String* problem_source /* never null */) { |
| } else { | if(!str) |
| // warning: | return 0; |
| // heavily relies on the fact | |
| // "preallocated area is the same for all strings" | while(isspace(*str)) |
| // | str++; |
| // info: | |
| // allocating only enough mem to fit src string rows | if(!*str) |
| // next append would allocate a new chunk | return 0; |
| // | |
| // new rows don't fit size_to preallocated area: splitting size_to two chunks | bool negative=false; |
| // preallocated chunk src to constructing head | if(str[0]=='-') { |
| memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*head.count); | negative=true; |
| // remaining rows size_to new_chunk | str++; |
| size_t curr_chunk_rows=src_used_rows-head.count; | if(!*str || isspace(*str)) |
| Chunk *new_chunk=static_cast<Chunk *>( | throw Exception("number.format", problem_source, "invalid number (double)"); |
| malloc(sizeof(size_t)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *))); | } else if(str[0]=='+') { |
| new_chunk->count=curr_chunk_rows; | str++; |
| head.preallocated_link=new_chunk; | if(!*str || isspace(*str)) |
| append_here=link_row=&new_chunk->rows[new_chunk->count]; | throw Exception("number.format", problem_source, "invalid number (double)"); |
| } | |
| Chunk *old_chunk=src.head.preallocated_link; | |
| Chunk::Row *new_rows=new_chunk->rows; | double result; |
| size_t rows_left_to_copy=new_chunk->count; | if(str[0]=='0') { |
| while(true) { | if(str[1]=='x' || str[1]=='X') { |
| size_t old_count=old_chunk->count; | // 0xABC |
| Chunk *next_chunk=old_chunk->rows[old_count].link; | result=(double)pa_atoul(str, 0, problem_source); |
| if(next_chunk) { | return negative ? -result : result; |
| // not last source chunk | } else { |
| // taking it all | // skip leading 0000, to disable octal interpretation |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*old_count); | do str++; while(*str=='0'); |
| 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; | |
| fused_rows=src_used_rows; | char *error_pos; |
| fsize=src.fsize; | result=strtod(str, &error_pos); |
| } | |
| while(const char c=*error_pos++) | |
| String& String::append(const String& src, Untaint_lang lang, bool forced) { | if(!isspace(c)) |
| const Chunk *chunk=&src.head; | throw Exception("number.format", problem_source, "invalid number (double)"); |
| do { | |
| const Chunk::Row *row=chunk->rows; | return negative ? -result : result; |
| for(size_t i=0; i<chunk->count; 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) { | // cord lib extension |
| if(!src) | |
| return *this; | #ifndef DOXYGEN |
| if(!size) | typedef struct { |
| size=strlen(src); | ssize_t countdown; |
| if(!size) | int target; /* Character we're looking for */ |
| return *this; | } chr_data; |
| if(chunk_is_full()) | |
| expand(); | |
| append_here->item.ptr=src; | |
| 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; | |
| #endif | #endif |
| append_here++; fused_rows++; | |
| return *this; | static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data) |
| { | |
| 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'*/); | |
| } | } |
| uint String::hash_code() const { | static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data) |
| uint result=0; | { |
| int* result=(int*)client_data; | |
| (*result)++; | |
| return(0); // 0=continue | |
| } | |
| const Chunk *chunk=&head; | size_t CORD_block_count(CORD x) |
| do { | { |
| const Chunk::Row *row=chunk->rows; | size_t result=0; |
| for(size_t i=0; i<chunk->count; i++) { | CORD_block_iter(x, 0, CORD_block_count_proc, &result); |
| 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; | return result; |
| } | } |
| /// @todo move 'lang' skipping to pos | // helpers |
| int String::cmp(int& partial, const String& src, | |
| size_t this_offset, Untaint_lang lang) const { | /// String::match uses this as replace & global search table columns |
| partial=-1; | |
| this_offset=min(this_offset, size()-1); | const int MAX_MATCH_GROUPS=100; |
| const Chunk *a_chunk=&head; | class String_match_table_template_columns: public ArrayString { |
| const Chunk *b_chunk=&src.head; | public: |
| const Chunk::Row *a_row=a_chunk->rows; | String_match_table_template_columns() { |
| const Chunk::Row *b_row=b_chunk->rows; | *this+=new String("prematch"); |
| size_t a_offset=this_offset; | *this+=new String("match"); |
| size_t b_offset=0; | *this+=new String("postmatch"); |
| Chunk::Row *a_end=append_here; | for(int i=0; i<MAX_MATCH_GROUPS; i++) { |
| Chunk::Row *b_end=src.append_here; | *this+=new String(pa_uitoa(1+i), String::L_CLEAN); |
| 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; | static Table &string_match_table_template(){ |
| size_t pos=0; | static Table *singleton=NULL; |
| while(true) { | if(!singleton) |
| a_break=a_row==a_end; | singleton=new Table(new String_match_table_template_columns); |
| b_break=b_row==b_end; | return *singleton; |
| if(a_break || b_break) | } |
| break; | |
| if(pos+a_row->item.size > this_offset) { | // String::Body methods |
| if(lang!=UL_UNSPECIFIED && a_row->item.lang!=lang) | |
| return -1; // wrong lang -- bail out | String::Body String::Body::trim(String::Trim_kind kind, const char* chars, size_t* out_start, size_t* out_length, Charset* source_charset) const { |
| size_t our_length=length(); | |
| int size_diff= | if(!our_length) |
| (a_row->item.size-a_offset)- | return *this; |
| (b_row->item.size-b_offset); | |
| // check if any UTF-8 in chars | |
| if(size_diff==0) { // a has same size as b | bool fast=true; |
| result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, | if(chars && source_charset && source_charset->isUTF8()){ |
| a_row->item.size-a_offset); | const char* pos=chars; |
| if(result) | while(unsigned char c=*pos++) |
| return result; | if(c>127){ |
| pos+=a_row->item.size; | fast=false; |
| a_row++; a_countdown--; a_offset=0; | break; |
| 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); | |
| } | |
| } | |
| // from right.. | |
| if(kind!=TRIM_START) { | |
| CORD_pos pos; set_pos(pos, end-1); | |
| 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); | |
| } | |
| } | |
| } else { | |
| const XMLByte* src_begin=(const XMLByte*)cstr(); | |
| const XMLByte* src_end=src_begin+our_length; | |
| // from left... | |
| if(kind!=TRIM_END) { | |
| while(src_begin<src_end){ | |
| uint char_length=1; | |
| const XMLByte* ptr=src_begin; | |
| // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3 | |
| while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0)) | |
| char_length++; | |
| bool found=false; | |
| for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++) | |
| if(strncmp(chars_byte, (const char*)ptr, char_length)==0){ | |
| found=true; | |
| break; | |
| } | |
| if(found){ | |
| start+=char_length; | |
| if(start==our_length) | |
| return 0; // all chars are empty, just return empty string | |
| } else | |
| break; | |
| } | } |
| } 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, (const char *)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; | |
| } | } |
| /// @todo move 'lang' skipping to pos | struct CORD_pos_info { |
| int String::cmp(int& partial, const char* b_ptr, size_t src_size, | const char* chars; |
| size_t this_offset, Untaint_lang lang) const { | size_t left; |
| partial=-1; | size_t pos; |
| 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) { | // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk |
| if(lang!=UL_UNSPECIFIED && a_row->item.lang!=lang) | static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) { |
| return -1; // wrong lang -- bail out | if(info->pos < info->left){ |
| info->pos=STRING_NOT_FOUND; | |
| int size_diff= | return 1; |
| (a_row->item.size-a_offset)- | } |
| (b_size-b_offset); | if(strchr(info->chars, c)) |
| return 1; | |
| if(size_diff==0) { // a has same size as b | --(info->pos); |
| if(size_t result=memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, | return 0; |
| 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--; | |
| } | |
| if(!a_countdown) { | size_t String::Body::strrpbrk(const char* chars, size_t left, size_t right) const { |
| a_chunk=a_row->link; | if(is_empty() || !chars || !strlen(chars)) |
| a_row=a_chunk->rows; | return STRING_NOT_FOUND; |
| a_countdown=a_chunk->count; | CORD_pos_info info={chars, left, right}; |
| } | if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rpos, &info)) |
| return info.pos; | |
| else | |
| return STRING_NOT_FOUND; | |
| } | |
| // can be called only for IS_FUNCTION(CORD) which is used in String::Body::rskipchars | |
| static int CORD_iter_fn_rskip(char c, CORD_pos_info* info) { | |
| if(info->pos < info->left) { | |
| info->pos=STRING_NOT_FOUND; | |
| return 1; | |
| } | } |
| if(a_break==b_break) { // ended simultaneously | if(!strchr(info->chars, c)) |
| partial=0; return 0; | return 1; |
| } else if(a_break) { // first bytes equal, but a ended before b | --(info->pos); |
| partial=1; return -1; | return 0; |
| } | |
| size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const { | |
| if(is_empty() || !chars || !strlen(chars)) | |
| return STRING_NOT_FOUND; | |
| CORD_pos_info info={chars, left, right}; | |
| if(CORD_riter4(body, right, (CORD_iter_fn)CORD_iter_fn_rskip, &info)) | |
| return info.pos; | |
| else | |
| return STRING_NOT_FOUND; | |
| } | |
| // String methods | |
| String& String::append_know_length(const char* str, size_t known_length, Language lang) { | |
| if(!known_length) | |
| return *this; | |
| // first: langs | |
| langs.append(body, lang, known_length); | |
| // next: letters themselves | |
| body.append_know_length(str, known_length); | |
| ASSERT_STRING_INVARIANT(*this); | |
| return *this; | |
| } | |
| 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; | |
| return append_know_length(str, known_length, lang); | |
| } | |
| String::String(int value, const char *format) : langs(L_CLEAN){ | |
| char buf[MAX_NUMBER]; | |
| body.append_strdup_know_length(buf, snprintf(buf, MAX_NUMBER, format, value)); | |
| } | |
| String& String::append_strdup(const char* str, size_t helper_length, Language lang) { | |
| size_t known_length=helper_length?helper_length:strlen(str); | |
| if(!known_length) | |
| return *this; | |
| // first: langs | |
| langs.append(body, lang, known_length); | |
| // next: letters themselves | |
| body.append_strdup_know_length(str, known_length); | |
| ASSERT_STRING_INVARIANT(*this); | |
| return *this; | |
| } | |
| struct CORD_length_info { | |
| size_t len; | |
| size_t skip; | |
| }; | |
| int CORD_batched_len(const char* s, CORD_length_info* info){ | |
| info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s)); | |
| return 0; | |
| } | |
| // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid | |
| int CORD_batched_len(const char c, CORD_length_info* info){ | |
| if (info->skip==0){ | |
| info->len++; | |
| info->skip = lengthUTF8Char(c)-1; | |
| } else { | } else { |
| partial=2; return +1; | info->skip--; |
| } | } |
| return 0; | |
| } | } |
| #ifndef NO_STRING_ORIGIN | size_t String::length(Charset& charset) const { |
| const Origin& String::origin() const { | if(charset.isUTF8()){ |
| if(!fused_rows) | CORD_length_info info = {0, 0}; |
| THROW(0, 0, | body.for_each<CORD_length_info *>(CORD_batched_len, CORD_batched_len, &info); |
| 0, | return info.len; |
| "String::origin() of empty string called"); | } else |
| return body.length(); | |
| // 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& String::mid(size_t start, size_t finish) const { | /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab" |
| start=max(0, start); | String& String::mid(size_t substr_begin, size_t substr_end) const { |
| finish=min(size(), finish); | String& result=*new String; |
| if(start==finish) | |
| return *empty_string; | size_t self_length=length(); |
| substr_begin=min(substr_begin, self_length); | |
| String& result=*NEW String(pool()); | substr_end=min(max(substr_end, substr_begin), self_length); |
| size_t substr_length=substr_end-substr_begin; | |
| size_t pos=0; | if(!substr_length) |
| const Chunk *chunk=&head; | return result; |
| do { | |
| const Chunk::Row *row=chunk->rows; | // first: their langs |
| for(size_t i=0; i<chunk->count; pos+=row->item.size, i++, row++) { | result.langs.append(result.body, langs, substr_begin, substr_length); |
| if(row==append_here) | // next: letters themselves |
| goto break2; | result.body=body.mid(substr_begin, substr_length); |
| size_t item_finish=pos+row->item.size; | ASSERT_STRING_INVARIANT(result); |
| 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; | 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; | |
| } | |
| } | |
| size_t String::pos(const String& substr, size_t this_offset, Language lang) const { | |
| return pos(substr.body, this_offset, lang); | |
| } | |
| size_t String::pos(Charset& charset, const String& substr, 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; | |
| } | } |
| return -1; | |
| } | } |
| void String::split(Array& result, | void String::split(ArrayString& result, size_t pos_after, const char* delim, Language lang) const { |
| size_t* pos_after_ref, | if(is_empty()) |
| const char *delim, size_t delim_size, | return; |
| Untaint_lang lang, int limit) const { | size_t self_length=length(); |
| if(delim_size) { | if(size_t delim_length=strlen(delim)) { |
| size_t pos_after=pos_after_ref?*pos_after_ref:0; | size_t pos_before; |
| int pos_before; | |
| // while we have 'delim'... | // while we have 'delim'... |
| for(; (pos_before=pos(delim, delim_size, pos_after, lang))>=0 && limit; limit--) { | while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) { |
| result+=&mid(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) |
| result+=&mid(pos_after, size()); | result+=&mid(pos_after, self_length); |
| pos_after=size(); | |
| } | |
| 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_ref+=size(); | |
| } | } |
| } | } |
| void String::split(Array& result, | void String::split(ArrayString& result, size_t pos_after, const String& delim, Language lang) const { |
| size_t* pos_after_ref, | if(is_empty()) |
| const String& delim, Untaint_lang lang, | return; |
| int limit) const { | if(!delim.is_empty()) { |
| if(delim.size()) { | size_t pos_before; |
| size_t pos_after=pos_after_ref?*pos_after_ref:0; | |
| int pos_before; | |
| // while we have 'delim'... | // while we have 'delim'... |
| for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) { | while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) { |
| result+=&mid(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()) |
| result+=&mid(pos_after, size()); | result+=&mid(pos_after, length()); |
| pos_after=size(); | |
| } | |
| 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_ref+=size(); | |
| } | } |
| } | } |
| static void regex_options(char *options, int *result){ | Table* String::match(VRegex* vregex, Row_action row_action, void *info, int& matches_count) const { |
| 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; | |
| } | |
| } | |
| /** | // vregex->info(); // I have no idea what does it for? |
| returns true if fills table. | |
| table format is defined and fixed[can be used by others]: | bool need_pre_post_match=vregex->is_pre_post_match_needed(); |
| @verbatim | bool global=vregex->is_global_search(); |
| 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); | |
| } | |
| 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 | |
| } | |
| *table=NEW Table(pool(), aorigin, &columns); | |
| } | |
| int exec_option_bits=0; | const char* subject=cstr(); |
| size_t subject_length=length(); | |
| const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3; /* 1/3 is used as workspace by pcre_exec() */ | |
| 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(); | |
| int action_was_executed=-1; | |
| while(true) { | while(true) { |
| int exec_substrings=pcre_exec(code, 0, | int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart); |
| subject, length, startoffset, | |
| exec_option_bits, ovector, ovecsize); | if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception |
| break; | |
| if(exec_substrings==PCRE_ERROR_NOMATCH) { | |
| (*pcre_free)(code); | int prefinish=ovector[0]; |
| (*row_action)(**table, 0/*last time, no row*/, 0, 0, info); | poststart=ovector[1]; |
| return option_bits[1]!=0; // global=true+table, not global=false | |
| } | if (prestart==poststart && action_was_executed==1){ |
| prestart++; | |
| if(exec_substrings<0) { | action_was_executed=0; |
| (*pcre_free)(code); | continue; |
| THROW(0, 0, | } |
| aorigin, | |
| "regular expression execute error (%d)", | ArrayString* row=new ArrayString(3); |
| exec_substrings); | if(need_pre_post_match) { |
| } | *row+=&mid(0, prefinish); // .prematch column value |
| *row+=&mid(prefinish, poststart); // .match | |
| Array& row=*NEW Array(pool()); | *row+=&mid(poststart, postfinish); // .postmatch |
| row+=&mid(0, ovector[0]); // .pre-match column value | } else { |
| row+=&mid(ovector[0], ovector[1]); // .match | *row+=&Empty; // .prematch column value |
| row+=&mid(ovector[1], size()); // .post-match | *row+=&Empty; // .match |
| *row+=&Empty; // .postmatch | |
| } | |
| for(int i=1; i<exec_substrings; i++) { | for(int i=1; i<exec_result; i++) { |
| // -1:-1 case handled peacefully by mid() itself | // -1:-1 case handled peacefully by mid() itself |
| row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value | *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 |
| } | } |
| (*row_action)(**table, &row, startoffset, ovector[0], info); | matches_count++; |
| row_action(table, row, prestart - !action_was_executed, prefinish, poststart, postfinish, info); | |
| if(!global || (size_t)poststart>=subject_length) // last step, avoid prestart++ after last char | |
| break; | |
| prestart=poststart; | |
| action_was_executed=1; | |
| } | |
| 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; | |
| if(!option_bits[1] || !(startoffset=ovector[1])) { // not global | going to hang | const unsigned char *a; |
| (*pcre_free)(code); | const unsigned char *b; |
| (*row_action)(**table, 0/*last time, no row*/, 0, 0, info); | switch(kind) { |
| return true; | 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; |
| if(option_bits[0] & PCRE_MULTILINE) | } |
| exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL | |
| */ | const String& String::escape(Charset& source_charset) const { |
| if(is_empty()) | |
| return *this; | |
| return Charset::escape(*this, source_charset); | |
| } | |
| #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \ | |
| result.langs.append(result.body, langs, langs_offset, length); \ | |
| result.body.append_strdup_know_length(from_cstr, length); | |
| const String& String::replace(const Dictionary& dict) const { | |
| if(!dict.count() || is_empty()) | |
| return *this; | |
| String& result=*new String(); | |
| const char* old_cstr=cstr(); | |
| const char* prematch_begin=old_cstr; | |
| if(dict.count()==1) { | |
| // optimized simple case | |
| Dictionary::Subst subst=dict.get(0); | |
| while(const char* p=strstr(prematch_begin, subst.from)) { | |
| // prematch | |
| if(size_t prematch_length=p-prematch_begin) { | |
| STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, prematch_length) | |
| } | |
| // match | |
| prematch_begin=p+subst.from_length; | |
| if(const String* b=subst.to) // are there any b? | |
| result<<*b; | |
| } | |
| } else { | |
| 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) { | |
| STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, 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++; | |
| } | |
| } | |
| if(prematch_begin==old_cstr) // not modified | |
| return *this; | |
| // postmatch | |
| if(size_t postmatch_length=old_cstr+length()-prematch_begin) { | |
| STRING_APPEND(result, prematch_begin, langs, prematch_begin-old_cstr, 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(String::C(cur, body_length)); | |
| 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; | |
| } | |
| void String::Body::dump() const { | |
| CORD_dump(body); | |
| } | |
| const char* String::Languages::visualize() const { | |
| if(opt.is_not_just_lang) | |
| return CORD_to_const_char_star(langs, 0); | |
| else | |
| return 0; | |
| } | |
| void String::Languages::dump() const { | |
| if(opt.is_not_just_lang) | |
| CORD_dump(langs); | |
| else | |
| puts((const char*)&langs); | |
| } | |
| void String::dump() const { | |
| body.dump(); | |
| langs.dump(); | |
| } | |
| static char *n_chars(char c, size_t length){ | |
| char *result=(char *)pa_malloc_atomic(length+1); | |
| memset(result, c, length); | |
| result[length] = '\0'; | |
| return result; | |
| } | |
| char* String::visualize_langs() const { | |
| return is_not_just_lang() ? pa_strdup(langs.visualize()) : n_chars((char)just_lang(), length()); | |
| } | |
| 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; | |
| } | } |