|
|
| version 1.162, 2002/06/26 10:35:16 | version 1.246, 2015/04/06 22:27:26 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser: string class. @see untasize_t.C. | Parser: string class. @see untalength_t.C. |
| Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) | Copyright (c) 2001-2012 Art. Lebedev Studio (http://www.artlebedev.com) |
| Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) | Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) |
| $Id$ | |
| */ | */ |
| #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_dictionary.h" | #include "pa_dictionary.h" |
| #include "pa_charset.h" | #include "pa_charset.h" |
| #include "pa_vregex.h" | |
| #define DEBUG_STRING_APPENDS_VS_EXPANDS | volatile const char * IDENT_PA_STRING_C="$Id$" IDENT_PA_STRING_H; |
| const String String::Empty; | |
| #ifdef DEBUG_STRING_APPENDS_VS_EXPANDS | |
| ulong string_piece_appends=0; | |
| #endif | |
| String& String::OnPool(Pool& apool, const char *local_src, size_t src_size, bool tainted) { | // pa_atoui is based on Manuel Novoa III _strto_l for uClibc |
| if(local_src && *local_src) { | |
| if(src_size==0) | |
| src_size=strlen(local_src); | |
| char *pooled_src=(char *)apool.malloc(src_size); | |
| memcpy(pooled_src, local_src, src_size); | |
| return *new(apool) String(apool, pooled_src, src_size, tainted); | |
| } else | |
| return *new(apool) String(apool); | |
| } | |
| String::String(Pool& apool, const char *src, size_t src_size, bool tainted) : | |
| Pooled(apool) { | |
| last_chunk=&head.chunk; | |
| head.chunk.count=CR_PREALLOCATED_COUNT; | |
| append_here=head.chunk.rows; | |
| if(src) | unsigned int pa_atoui(const char *str, int base, const String* problem_source){ |
| if(tainted) | unsigned int result = 0; |
| APPEND_TAINTED(src, src_size, 0, 0); | const char *pos = str; |
| else | |
| APPEND_CLEAN(src, src_size, 0, 0); | |
| } | |
| String::String(const String& src) : | while (isspace(*pos)) /* skip leading whitespace */ |
| Pooled(src.pool()) { | ++pos; |
| last_chunk=&head.chunk; | |
| head.chunk.count=CR_PREALLOCATED_COUNT; | |
| append_here=head.chunk.rows; | |
| append(src, UL_UNSPECIFIED); | if (base == 16 && *pos == '0') { /* handle option prefix */ |
| } | ++pos; |
| if (*pos == 'x' || *pos == 'X') { | |
| ++pos; | |
| } | |
| } | |
| size_t String::size() const { | if (base == 0) { /* dynamic base */ |
| size_t result=0; | base = 10; /* default is 10 */ |
| STRING_FOREACH_ROW( | if (*pos == '0') { |
| result+=row->item.size; | ++pos; |
| ); | if (*pos == 'x' || *pos == 'X'){ |
| return result; | ++pos; |
| } | base=16; |
| } | |
| } | |
| } | |
| if (base < 2 || base > 16) { /* illegal base */ | |
| throw Exception(PARSER_RUNTIME, 0, "base to must be an integer from 2 to 16"); | |
| } | |
| unsigned int cutoff = UINT_MAX / base; | |
| int cutoff_digit = UINT_MAX - cutoff * base; | |
| while(true) { | |
| int digit; | |
| if ((*pos >= '0') && (*pos <= '9')) { | |
| digit = (*pos - '0'); | |
| } else if (*pos >= 'a') { | |
| digit = (*pos - 'a' + 10); | |
| } else if (*pos >= 'A') { | |
| digit = (*pos - 'A' + 10); | |
| } else break; | |
| if (digit >= base) { | |
| break; | |
| } | |
| ++pos; | |
| /* adjust number, with overflow check */ | |
| if ((result > cutoff) || ((result == cutoff) && (digit > cutoff_digit))) { | |
| throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str); | |
| } else { | |
| result = result * base + digit; | |
| } | |
| } | |
| while(char c=*pos++) | |
| if(!isspace(c)) | |
| throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is invalid number (int)", str); | |
| /// @todo not very optimal | |
| uint String::used_rows() const { | |
| uint result=0; | |
| STRING_FOREACH_ROW( | |
| result++; | |
| ); | |
| return result; | return result; |
| } | } |
| void String::expand() { | |
| uint new_chunk_count=last_chunk->count+CR_GROW_COUNT; | int pa_atoi(const char* str, const String* problem_source) { |
| if(new_chunk_count>max_integral(Chunk::count_type)) | if(!str) |
| new_chunk_count=max_integral(Chunk::count_type); | return 0; |
| Chunk *new_chunk=static_cast<Chunk *>(malloc( | while(isspace(*str)) |
| sizeof(Chunk)// count+interpadding(?)+rows[CR_PREALLOCATED_COUNT]+tailpadding(??) | str++; |
| -sizeof(Chunk::rows_type) // PREALLOCATED rows | |
| +sizeof(Chunk::Row)*new_chunk_count // neaded rows | if(!*str) |
| +sizeof(Chunk *) // link size | return 0; |
| , 10)); | |
| new_chunk->rows[new_chunk->count=new_chunk_count].link=0; | bool negative=false; |
| last_chunk->rows[last_chunk->count].link=new_chunk; | if(str[0]=='-') { |
| negative=true; | |
| str++; | |
| } else if(str[0]=='+') { | |
| str++; | |
| } | |
| unsigned int result=pa_atoui(str, 0, problem_source); | |
| if(negative && result <= ((unsigned int)(-(1+INT_MIN)))+1) | |
| return -(int)result; | |
| last_chunk=new_chunk; | if(result<=INT_MAX) |
| append_here=last_chunk->rows; | return (int)result; |
| throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str); | |
| } | } |
| String& String::real_append(STRING_APPEND_PARAMS) { | double pa_atod(const char* str, const String* problem_source) { |
| if(!last_chunk) // growth stopped [we're appended as string to somebody] | if(!str) |
| throw Exception(0, | return 0; |
| this, | |
| "string growth stopped (append cstr)"); | |
| if(!src) | |
| return *this; | |
| if(!size) | |
| size=strlen(src); | |
| if(!size) | |
| return *this; | |
| #ifdef DEBUG_STRING_APPENDS_VS_EXPANDS | while(isspace(*str)) |
| string_piece_appends++; | str++; |
| #endif | |
| // manually unrolled to avoid extra check | if(!*str) |
| while(size>max_integral(Chunk::Row::item_size_type)) { | return 0; |
| if(chunk_is_full()) | |
| expand(); | |
| append_here->item.ptr=src; | |
| append_here->item.size=max_integral(Chunk::Row::item_size_type); | |
| append_here->item.lang=lang; | |
| #ifndef NO_STRING_ORIGIN | |
| append_here->item.origin.file=file; | |
| append_here->item.origin.line=line; | |
| #endif | |
| append_here++; | |
| src+=max_integral(Chunk::Row::item_size_type); | bool negative=false; |
| size-=max_integral(Chunk::Row::item_size_type); | if(str[0]=='-') { |
| negative=true; | |
| str++; | |
| } else if(str[0]=='+') { | |
| str++; | |
| } | } |
| if(chunk_is_full()) | double result; |
| expand(); | if(str[0]=='0') |
| if(str[1]=='x' || str[1]=='X'){ | |
| // 0xABC | |
| result=(double)pa_atoui(str, 0, problem_source); | |
| return negative ? -result : result; | |
| } else { | |
| // skip leading 0000, to disable octal interpretation | |
| do str++; while(*str=='0'); | |
| } | |
| append_here->item.ptr=src; | char *error_pos; |
| append_here->item.size=size; | result=strtod(str, &error_pos); |
| append_here->item.lang=lang; | |
| #ifndef NO_STRING_ORIGIN | |
| append_here->item.origin.file=file; | |
| append_here->item.origin.line=line; | |
| #endif | |
| append_here++; | |
| return *this; | while(char c=*error_pos++) |
| if(!isspace((unsigned char)c)) | |
| throw Exception("number.format", problem_source, problem_source ? "invalid number (double)" : "'%s' is invalid number (double)", str); | |
| return negative ? -result : result; | |
| } | } |
| char String::first_char() const { | // cord lib extension |
| if(is_empty()) | |
| throw Exception(0, | |
| this, | |
| "getting first char of empty string"); | |
| return *head.chunk.rows[0].item.ptr; | #ifndef DOXYGEN |
| typedef struct { | |
| ssize_t countdown; | |
| int target; /* Character we're looking for */ | |
| } chr_data; | |
| #endif | |
| 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; | |
| } | } |
| uint String::hash_code() const { | // helpers |
| uint result=0; | |
| STRING_FOREACH_ROW( | /// String::match uses this as replace & global search table columns |
| result=Hash::generic_code(result, row->item.ptr, row->item.size); | |
| ); | const int MAX_MATCH_GROUPS=100; |
| return result; | |
| 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); | |
| } | |
| } | |
| }; | |
| Table string_match_table_template(new String_match_table_template_columns); | |
| // String::Body methods | |
| String::Body String::Body::Format(int value) { | |
| char local[MAX_NUMBER]; | |
| size_t length=snprintf(local, MAX_NUMBER, "%d", value); | |
| return String::Body(pa_strdup(local, length)); | |
| } | } |
| /// @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) |
| size_t a_size=size(); | return *this; |
| this_offset=min(this_offset, a_size-1); | |
| // check if any UTF-8 in chars | |
| const Chunk *a_chunk=&head.chunk; | bool fast=true; |
| const Chunk *b_chunk=&src.head.chunk; | if(chars && source_charset && source_charset->isUTF8()){ |
| const Chunk::Row *a_row=a_chunk->rows; | const char* pos=chars; |
| const Chunk::Row *b_row=b_chunk->rows; | while(unsigned char c=*pos++) |
| size_t a_offset=this_offset; | if(c>127){ |
| size_t b_offset=0; | fast=false; |
| Chunk::Row *a_end=append_here; | break; |
| Chunk::Row *b_end=src.append_here; | |
| uint a_countdown=a_chunk->count; | |
| uint b_countdown=b_chunk->count; | |
| int result; | |
| size_t pos=0; | |
| bool a_break=a_size==0; | |
| bool b_break=src.is_empty(); | |
| if(!(a_break || b_break)) while(true) { | |
| 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_break=b_row==b_end) { | |
| a_break=a_row==a_end; | |
| break; | |
| } | } |
| if(!b_countdown) { | } |
| b_chunk=b_row->link; | |
| b_row=b_chunk->rows; | size_t start=0; |
| b_countdown=b_chunk->count; | size_t end=our_length; |
| 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_break=a_row==a_end) { | // from right.. |
| b_break=b_row==b_end; | if(kind!=TRIM_START) { |
| break; | CORD_pos pos; set_pos(pos, end-1); |
| } | while(true) { |
| if(!a_countdown) { | char c=CORD_pos_fetch(pos); |
| a_chunk=a_row->link; | if(strchr(chars, c)) { |
| a_row=a_chunk->rows; | if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible |
| a_countdown=a_chunk->count; | 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 | // from left... |
| int String::cmp(int& partial, const char* b_ptr, size_t src_size, | if(kind!=TRIM_END) { |
| size_t this_offset, Untaint_lang lang) const { | while(src_begin<src_end){ |
| partial=-1; | uint char_length=1; |
| size_t a_size=size(); | const XMLByte* ptr=src_begin; |
| size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0; | // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3 |
| this_offset=min(this_offset, a_size-1); | while(++src_begin<=src_end && (*src_begin>127 && *src_begin<0xC0)) |
| char_length++; | |
| const Chunk *a_chunk=&head.chunk; | |
| const Chunk::Row *a_row=a_chunk->rows; | bool found=false; |
| size_t a_offset=this_offset; | for(const char* chars_byte=chars; chars_byte=strchr(chars_byte, *ptr); chars_byte++) |
| size_t b_offset=0; | if(strncmp(chars_byte, (const char*)ptr, char_length)==0){ |
| Chunk::Row *a_end=append_here; | found=true; |
| uint a_countdown=a_chunk->count; | break; |
| size_t pos=0; | } |
| bool a_break=a_size==0; | if(found){ |
| bool b_break=b_size==0; | start+=char_length; |
| if(!(a_break || b_break)) while(true) { | if(start==our_length) |
| if(pos+a_row->item.size > this_offset) { | return 0; // all chars are empty, just return empty string |
| if(lang!=UL_UNSPECIFIED && a_row->item.lang>lang) | } else |
| return -1; // wrong lang -- bail out | break; |
| int size_diff= | |
| (a_row->item.size-a_offset)- | |
| (b_size-b_offset); | |
| if(size_diff==0) { // a has same size as b | |
| if(int 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(int 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(int 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--; | |
| } | } |
| a_break=a_row==a_end; | // from right.. |
| if(a_break || b_break) | if(kind!=TRIM_START) { |
| break; | while(src_begin<src_end){ |
| if(!a_countdown) { | uint char_length=1; |
| a_chunk=a_row->link; | // searching first UTF-8 byte: http://tools.ietf.org/html/rfc3629#section-3 |
| a_row=a_chunk->rows; | while(src_begin<=--src_end && (*src_end>127 && *src_end<0xC0)) |
| a_countdown=a_chunk->count; | 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; | |
| } | |
| struct CORD_pos_info { | |
| const char* chars; | |
| size_t left; | |
| size_t pos; | |
| }; | |
| // can be called only for IS_FUNCTION(CORD) which is used in String::Body::strrpbrk | |
| static int CORD_iter_fn_rpos(char c, CORD_pos_info* info) { | |
| if(info->pos < info->left){ | |
| info->pos=STRING_NOT_FOUND; | |
| return 1; | |
| } | |
| if(strchr(info->chars, c)) | |
| return 1; | |
| --(info->pos); | |
| return 0; | |
| } | |
| size_t String::Body::strrpbrk(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_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(!strchr(info->chars, c)) | |
| return 1; | |
| --(info->pos); | |
| return 0; | |
| } | } |
| #ifndef NO_STRING_ORIGIN | size_t String::Body::rskipchars(const char* chars, size_t left, size_t right) const { |
| const Origin& String::origin() const { | if(is_empty() || !chars || !strlen(chars)) |
| if(is_empty()) { | return STRING_NOT_FOUND; |
| static const Origin empty_origin={"empty string"}; | CORD_pos_info info={chars, left, right}; |
| return empty_origin; | 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 { | |
| info->skip--; | |
| } | } |
| return 0; | |
| // determining origin by first piece or last appended piece | |
| // because any of them can be constant=without origin: | |
| // ex: ^load[/file] "document_root" + "/file" | |
| // when last peice is constant, | |
| // ex: parser_root_auto_path{dynamic} / auto.p{const} | |
| // using first piece | |
| Origin& first_origin=head.chunk.rows[0].item.origin; | |
| return first_origin.file ? first_origin : append_here[-1].item.origin; | |
| } | } |
| #endif | |
| String& String::mid(size_t start, size_t finish) const { | size_t String::length(Charset& charset) const { |
| String& result=*NEW String(pool()); | 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(); | |
| } | |
| start=max(size_t(0), start); | /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab" |
| finish=min(size(), finish); | String& String::mid(size_t substr_begin, size_t substr_end) const { |
| if(start==finish) | 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; | return result; |
| size_t pos=0; | // first: their langs |
| STRING_FOREACH_ROW( | result.langs.append(result.body, langs, substr_begin, substr_length); |
| size_t item_finish=pos+row->item.size; | // next: letters themselves |
| if(item_finish > start) { // started now or already? | result.body=body.mid(substr_begin, substr_length); |
| bool started=result.is_empty(); // started now? | |
| bool finished=finish <= item_finish; // finished now? | ASSERT_STRING_INVARIANT(result); |
| 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; | |
| } | |
| pos+=row->item.size; | |
| ); | |
| 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) |
| int result, Untaint_lang lang) const { | String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const { |
| size_t self_size=size(); | String& result=*new String; |
| for(; size_t(result)<self_size; result++) { | |
| int partial; cmp(partial, substr, result, lang); | size_t self_length=(helper_length)?helper_length:length(charset); |
| if( | |
| partial==0 || // full match | if(!self_length) |
| partial==2) // 'substr' starts 'this'+'result' | 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 { |
| int result, Untaint_lang lang) const { | size_t substr_length=substr.length(); |
| size_t self_size=size(); | while(true) { |
| for(; size_t(result)<self_size; result++) { | size_t substr_begin=body.pos(substr, this_offset); |
| int partial; cmp(partial, substr, substr_size, result, lang); | |
| if( | if(substr_begin==CORD_NOT_FOUND) |
| partial==0 || // full match | return STRING_NOT_FOUND; |
| partial==2) // 'substr' starts 'this'+'result' | |
| 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_ref, | size_t& pos_after, |
| const char *delim, size_t delim_size, | const char* delim, |
| Untaint_lang lang, int limit) const { | Language lang, int limit) const { |
| size_t self_size=size(); | if(is_empty()) |
| if(delim_size) { | return; |
| size_t pos_after=pos_after_ref?*pos_after_ref:0; | size_t self_length=length(); |
| int pos_before; | 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+=&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<self_size && limit) { | if(pos_after<self_length && limit) { |
| result+=&mid(pos_after, self_size); | result+=&mid(pos_after, self_length); |
| pos_after=self_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+=self_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(is_empty()) | |
| return; | |
| if(!delim.is_empty()) { | if(!delim.is_empty()) { |
| 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, pos_after, lang))>=0 && limit; limit--) { | for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) { |
| 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() && limit) { |
| result+=&mid(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(); | |
| } | } |
| } | } |
| static void regex_options(const String *options, int *result, bool& need_pre_post_match){ | Table* String::match(VRegex* vregex, |
| struct Regex_option { | Row_action row_action, void *info, |
| const char *keyL; | int& matches_count) const { |
| const char *keyU; | |
| int clear, set; | |
| int *result; | |
| bool *flag; | |
| } regex_option[]={ | |
| {"i", "I", 0, PCRE_CASELESS, result}, // a=A | |
| {"s", "S", 0, PCRE_DOTALL, result}, // \n\n$ [default] | |
| {"x", "U", 0, PCRE_EXTENDED, result}, // whitespace in regex ignored | |
| {"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$ | |
| {"g", "G", 0, true, result+1}, // many rows | |
| {"'", 0, 0, 0, 0, &need_pre_post_match}, | |
| {0} | |
| }; | |
| result[0]=PCRE_EXTRA | PCRE_DOTALL; | |
| result[1]=0; | |
| if(options) | |
| for(Regex_option *o=regex_option; o->keyL; o++) | |
| if(options->pos(o->keyL)>=0 | |
| || (o->keyU && options->pos(o->keyU)>=0)) { | |
| if(o->flag) | |
| *o->flag=true; | |
| else { // result | |
| *o->result &= ~o->clear; | |
| *o->result |= o->set; | |
| } | |
| } | |
| } | |
| /// @todo make replacement Table stacked | // vregex->info(); // I have no idea what does it for? |
| bool String::match( | |
| const String *aorigin, | bool need_pre_post_match=vregex->is_pre_post_match_needed(); |
| const String& regexp, | bool global=vregex->is_global_search(); |
| const String *options, | |
| Table **table, | |
| Row_action row_action, void *info, | |
| bool *was_global) const { | |
| if(regexp.is_empty()) | |
| throw Exception(0, | |
| aorigin, | |
| "regexp is empty"); | |
| const char *pattern=regexp.cstr(); | |
| const char *errptr; | |
| int erroffset; | |
| bool need_pre_post_match=false; | |
| int option_bits[2]; regex_options(options, option_bits, need_pre_post_match); | |
| if(was_global) | |
| *was_global=option_bits[1]!=0; | |
| pcre *code=pcre_compile(pattern, option_bits[0], | |
| &errptr, &erroffset, | |
| pool().get_source_charset().pcre_tables); | |
| if(!code) | |
| throw Exception(0, | |
| ®exp.mid(erroffset, regexp.size()), | |
| "regular expression syntax error - %s", errptr); | |
| int info_substrings=pcre_info(code, 0, 0); | |
| if(info_substrings<0) { | |
| pcre_free(code); | |
| throw Exception(0, | |
| aorigin, | |
| "pcre_info error (%d)", | |
| info_substrings); | |
| } | |
| const char *subject=cstr(); | |
| int length=strlen(subject); | |
| const int ovecsize=(1/*match*/+MAX_STRING_MATCH_TABLE_COLUMNS)*3; | |
| int ovector[ovecsize]; | |
| // create table | const char* subject=cstr(); |
| *table=NEW Table(pool(), *string_match_table_template); | 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 exec_option_bits=0; | |
| int prestart=0; | int prestart=0; |
| int poststart=0; | int poststart=0; |
| int postfinish=size(); | int postfinish=length(); |
| 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, prestart, | |
| exec_option_bits, ovector, ovecsize); | |
| if(exec_substrings==PCRE_ERROR_NOMATCH) { | |
| pcre_free(code); | |
| row_action(**table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info); | |
| return option_bits[1]!=0; // global=true+table, not global=false | |
| } | |
| if(exec_substrings<0) { | if(exec_result<0) // only PCRE_ERROR_NOMATCH might be here, other negative results cause an exception |
| pcre_free(code); | break; |
| throw Exception(0, | |
| aorigin, | |
| "regular expression execute error (%d)", | |
| exec_substrings); | |
| } | |
| int prefinish=ovector[0]; | int prefinish=ovector[0]; |
| poststart=ovector[1]; | poststart=ovector[1]; |
| Array& row=*NEW Array(pool()); | |
| row+=need_pre_post_match?&mid(0, prefinish):0; // .prematch column value | if (prestart==poststart && subject[poststart]=='\n'){ |
| row+=need_pre_post_match?&mid(prefinish, poststart):0; // .match | prestart++; |
| row+=need_pre_post_match?&mid(poststart, postfinish):0; // .postmatch | 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_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, prestart, prefinish, poststart, postfinish, info); | matches_count++; |
| row_action(table, row, prestart, prefinish, poststart, postfinish, info); | |
| if(!option_bits[1] || prestart==poststart) { // not global | going to hang | if(!global || prestart==poststart) // last step |
| pcre_free(code); | break; |
| row_action(**table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info); | |
| return true; | |
| } | |
| prestart=poststart; | |
| /* | prestart=poststart; |
| if(option_bits[0] & PCRE_MULTILINE) | |
| exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL | |
| */ | |
| } | } |
| row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info); | |
| return vregex->is_just_count() ? 0 : &table; | |
| } | } |
| String& String::change_case(Pool& pool, | String& String::change_case(Charset& source_charset, Change_case_kind kind) const { |
| Change_case_kind kind) const { | String& result=*new String(); |
| const unsigned char *tables=pool.get_source_charset().pcre_tables; | if(is_empty()) |
| String& result=*new(pool) String(pool); | return result; |
| const unsigned char *a; | char* new_cstr=cstrm(); |
| const unsigned char *b; | |
| switch(kind) { | if(source_charset.isUTF8()) { |
| case CC_UPPER: | size_t new_cstr_len=length(); |
| a=tables+lcc_offset; | switch(kind) { |
| b=tables+fcc_offset; | case CC_UPPER: |
| break; | change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToUpper); |
| case CC_LOWER: | break; |
| a=tables+lcc_offset; | case CC_LOWER: |
| b=0; | change_case_UTF8((const XMLByte*)new_cstr, new_cstr_len, (XMLByte*)new_cstr, new_cstr_len, UTF8CaseToLower); |
| break; | break; |
| default: | default: |
| throw Exception(0, | assert(!"unknown change case kind"); |
| this, | break; // never |
| "unknown change case kind #%d", | } |
| static_cast<int>(kind)); // never | |
| a=b=0; // calm, compiler | } else { |
| break; // never | 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 | |
| } | |
| STRING_FOREACH_ROW( | |
| char *new_cstr=(char *)pool.malloc(row->item.size, 12); | |
| char *dest=new_cstr; | char *dest=new_cstr; |
| const char *src=row->item.ptr; | unsigned char index; |
| for(int size=row->item.size; size--; src++) { | for(const char* current=new_cstr; (index=(unsigned char)*current); current++) { |
| unsigned char c=a[(unsigned char)*src]; | unsigned char c=a[index]; |
| if(b) | if(b) |
| c=b[c]; | c=b[c]; |
| *dest++=(char)c; | *dest++=(char)c; |
| } | } |
| } | |
| result.APPEND(new_cstr, row->item.size, | result.langs=langs; |
| row->item.lang, | result.body=new_cstr; |
| row->item.origin.file, row->item.origin.line); | |
| ); | |
| return result; | return result; |
| } | } |
| /// @test if in some piece were found no dict words, append it, not it's duplicate | const String& String::escape(Charset& source_charset) const { |
| String& String::replace(Pool& pool, Dictionary& dict) const { | if(is_empty()) |
| // return reconstruct(pool).replace_in_reconstructed(pool, dict); | return *this; |
| String& result=*new(pool) String(pool); | |
| return Charset::escape(*this, source_charset); | |
| STRING_FOREACH_ROW( | } |
| const char *src=row->item.ptr; | |
| size_t src_size=row->item.size; | #define STRING_APPEND(result, from_cstr, langs, langs_offset, length) \ |
| char *new_cstr=(char *)pool.malloc((size_t)ceil(src_size*dict.max_ratio()), 14); | result.langs.append(result.body, langs, langs_offset, length); \ |
| char *dest=new_cstr; | result.body.append_strdup_know_length(from_cstr, length); |
| while(src_size) { | |
| // there is a row where first column starts 'src' | const String& String::replace(const Dictionary& dict) const { |
| if(Table::Item *item=dict.first_that_starts(src, src_size)) { | if(!dict.count() || is_empty()) |
| // get a=>b values | return *this; |
| const String& a=*static_cast<Array *>(item)->get_string(0); | |
| const String& b=*static_cast<Array *>(item)->get_string(1); | String& result=*new String(); |
| // skip 'a' in 'src' && reduce work size | const char* old_cstr=cstr(); |
| src+=a.size(); src_size-=a.size(); | const char* prematch_begin=old_cstr; |
| // write 'b' to 'dest' && skip 'b' in 'dest' | |
| b.store_to(dest); dest+=b.size(); | if(dict.count()==1) { |
| } else { | // optimized simple case |
| // write a char to b && reduce work size | |
| *dest++=*src++; src_size--; | 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; | |
| } | } |
| result.APPEND(new_cstr, dest-new_cstr, row->item.lang, | } else { |
| row->item.origin.file, row->item.origin.line); | |
| ); | 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; | return result; |
| } | } |
| String& String::join_chains(Pool& pool, char** acstr) const { | static int serialize_body_char(char c, char** cur) { |
| char *lcstr=cstr(); | *((*cur)++)=c; |
| const char *current=lcstr; | return 0; // 0=continue |
| }; | |
| String& result=*new(pool) String(pool); | static int serialize_body_piece(const char* s, char** cur) { |
| STRING_FOREACH_ROW( | size_t length=strlen(s); |
| IFNDEF_NO_STRING_ORIGIN( | memcpy(*cur, s, length); *cur+=length; |
| const char *joined_origin_file=row->item.origin.file; | return 0; // 0=continue |
| const size_t joined_origin_line=row->item.origin.line; | }; |
| ); | static int serialize_lang_piece(char alang, size_t asize, char** cur) { |
| uchar joined_lang=row->item.lang; | // lang |
| const char *joined_ptr=current; | **cur=alang; (*cur)++; |
| // calc size | // length [WARNING: not cast, addresses must be %4=0 on sparc] |
| size_t joined_size=0; | memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize); |
| STRING_PREPARED_FOREACH_ROW(*this, | |
| if(row->item.lang==joined_lang) | return 0; // 0=continue |
| joined_size+=row->item.size; | } |
| else | String::Cm String::serialize(size_t prolog_length) const { |
| break; // before non-ours | size_t fragments_count=langs.count(); |
| ); | size_t body_length=body.length(); |
| current+=joined_size; | size_t buf_length= |
| prolog_length //1 | |
| // pointers are after joined piece | +sizeof(size_t) //2 |
| // & one step back, see STRING_FOREACH_ROW | +body_length //3 |
| --row; ++countdown; | +1 // 4 for zero terminator used in deserialize |
| +sizeof(size_t) //5 | |
| result.APPEND(joined_ptr, joined_size, joined_lang, | +fragments_count*(sizeof(char)+sizeof(size_t)); //6 |
| joined_origin_file, joined_origin_line); | |
| ); | 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); | |
| if(acstr) | |
| *acstr=lcstr; | |
| return result; | 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; | |
| double String::as_double() const { | // 1: prolog |
| double result; | const char* cur=(const char* )buf+prolog_size; |
| const char *cstr; | |
| char buf[MAX_NUMBER]; | |
| if(head.chunk.rows+1==append_here) { | |
| int size=min(head.chunk.rows[0].item.size, MAX_NUMBER-1); | |
| memcpy(buf, head.chunk.rows[0].item.ptr, size); | |
| buf[size]=0; | |
| cstr=buf; | |
| } else | |
| cstr=this->cstr(); | |
| while(*cstr && isspace(*cstr)) | |
| cstr++; | |
| if(!*cstr) | |
| return 0; | |
| char *error_pos; | // 2: chars.count |
| // 0xABC | size_t body_length; |
| if(cstr[0]=='0') | if(in_buf<sizeof(body_length)) // body.length don't fit? |
| if(cstr[1]=='x' || cstr[1]=='X') | return false; |
| result=(double)(unsigned long)strtol(cstr, &error_pos, 0); | // [WARNING: not cast, addresses must be %4=0 on sparc] |
| else | memcpy(&body_length, cur, sizeof(body_length)); cur+=sizeof(body_length); |
| result=(double)strtod(cstr+1/*skip leading 0*/, &error_pos); | in_buf-=sizeof(body_length); |
| else | |
| result=(double)strtod(cstr, &error_pos); | |
| while(char c=*error_pos++) | if(in_buf<body_length+1) // letters+terminator don't fit? |
| if(!isspace(c)) | return false; |
| throw Exception("number.format", | // 4: zero terminator |
| this, | if(cur[body_length] != 0) // in place? |
| "invalid number (double)"); | 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; | |
| return result; | // 5: langs.count |
| } | size_t fragments_count; |
| int String::as_int() const { | if(in_buf<sizeof(fragments_count)) // langs.count don't fit? |
| int result; | return false; |
| const char *cstr; | // [WARNING: not cast, addresses must be %4=0 on sparc] |
| char buf[MAX_NUMBER]; | memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count); |
| if(head.chunk.rows+1==append_here) { | in_buf-=sizeof(fragments_count); |
| int size=min(head.chunk.rows[0].item.size, MAX_NUMBER-1); | |
| memcpy(buf, head.chunk.rows[0].item.ptr, size); | if(fragments_count) { |
| buf[size]=0; | // 6: lang info |
| cstr=buf; | size_t total_length=0; |
| } else | for(size_t f=0; f<fragments_count; f++) { |
| cstr=this->cstr(); | char lang; |
| while(*cstr && isspace(*cstr)) | size_t fragment_length; |
| cstr++; | size_t piece_length=sizeof(lang)+sizeof(fragment_length); |
| if(!*cstr) | if(in_buf<piece_length) // lang+length |
| return 0; | 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; | |
| } | |
| char *error_pos; | if(total_length!=body_length) // length(all language fragments) vs length(letters) |
| // 0xABC | return false; |
| if(cstr[0]=='0') | } |
| if(cstr[1]=='x' || cstr[1]=='X') | if(in_buf!=0) // some strange extra bytes |
| result=(int)(unsigned long)strtol(cstr, &error_pos, 0); | return false; |
| else | |
| result=(int)strtol(cstr+1/*skip leading 0*/, &error_pos, 0); | |
| else | |
| result=(int)strtol(cstr, &error_pos, 0); | |
| while(char c=*error_pos++) | ASSERT_STRING_INVARIANT(*this); |
| if(!isspace(c)) | return true; |
| throw Exception("number.format", | } |
| this, | |
| "invalid number (int)"); | |
| return result; | const char* String::Body::v() const { |
| return CORD_to_const_char_star(body, length()); | |
| } | |
| void String::Body::dump() const { | |
| CORD_dump(body); | |
| } | } |
| inline void ushort2uchars(ushort word, uchar& byte1, uchar& byte2) { | const char* String::Languages::v() const { |
| byte1=word&0xFF; | if(opt.is_not_just_lang) |
| byte2=word>>8; | return CORD_to_const_char_star(langs, 0); |
| } | else |
| inline ushort uchars2ushort(uchar byte1, uchar byte2) { | return (const char*)&langs; |
| return (byte2<<8) | byte1; | } |
| } | void String::Languages::dump() const { |
| /* @todo maybe network order worth spending some effort? | if(opt.is_not_just_lang) |
| don't bothering myself with network byte order, | CORD_dump(langs); |
| am not planning to be able to move resulting file across platforms | else |
| for now | puts((const char*)&langs); |
| */ | } |
| void String::serialize(size_t prolog_size, void *& buf, size_t& buf_size) const { | const char* String::v() const { |
| buf_size= | const uint LIMIT_VIEW=20; |
| prolog_size | char* buf=(char*)pa_malloc(MAX_STRING); |
| +used_rows()*(sizeof(uchar)+sizeof(ushort)) | const char*body_view=body.v(); |
| +size(); | const char*langs_view=langs.v(); |
| buf=malloc(buf_size,15); | snprintf(buf, MAX_STRING, |
| char *cur=(char *)buf+prolog_size; | "%d:%.*s%s} " |
| "{%d:%s", | |
| STRING_FOREACH_ROW( | langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"", |
| // lang | strlen(body_view), body_view |
| memcpy(cur, &row->item.lang, sizeof(uchar)); | |
| cur+=sizeof(uchar); | |
| // size | |
| uchar byte1; uchar byte2; | |
| ushort2uchars(row->item.size, byte1, byte2); | |
| memcpy(cur, &byte1, sizeof(uchar)); cur+=sizeof(uchar); | |
| memcpy(cur, &byte2, sizeof(uchar)); cur+=sizeof(uchar); | |
| // bytes | |
| memcpy(cur, row->item.ptr, row->item.size); | |
| cur+=row->item.size; | |
| ); | ); |
| return buf; | |
| } | } |
| bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file) { | |
| if(buf_size<=prolog_size) | |
| return false; | |
| char *cur=(char *)buf+prolog_size; | void String::dump() const { |
| buf_size-=prolog_size; | body.dump(); |
| langs.dump(); | |
| } | |
| while(buf_size) { | const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const { |
| if(sizeof(uchar)+sizeof(ushort)>buf_size) // lang+size | if(is_empty()) |
| return false; | return *this; |
| uchar lang=*(uchar *)(cur); | size_t substr_begin, substr_length; |
| ushort size=uchars2ushort( | Body new_body=body.trim(kind, chars, &substr_begin, &substr_length, source_charset); |
| *(uchar*)(cur+sizeof(uchar)*1), | if(new_body==body) // we received unchanged pointer, do likewise |
| *(uchar*)(cur+sizeof(uchar)*2) | return *this; |
| ); | // new_body differs from body, adjust langs along |
| size_t piece_size=sizeof(uchar)+sizeof(ushort)+size; | String& result=*new String; |
| if(piece_size>buf_size) // buffer overrun, can be on incomplete cache files | if(!new_body) // body.trim produced empty result |
| return false; | return result; |
| // body.trim produced nonempty result | |
| const char *ptr=(const char*)(cur+sizeof(uchar)*3); | // first: their langs |
| APPEND(ptr, size, lang, file, 0); | result.langs.append(result.body, langs, substr_begin, substr_length); |
| // next: letters themselves | |
| result.body=new_body; | |
| cur+=piece_size; | ASSERT_STRING_INVARIANT(result); |
| buf_size-=piece_size; | return result; |
| } | |
| return true; | |
| } | } |