|
|
| version 1.35, 2001/03/10 12:12:51 | version 1.185, 2003/10/21 05:11:00 |
|---|---|
| Line 1 | Line 1 |
| /* | /** @file |
| $Id$ | Parser: string class. @see untalength_t.C. |
| Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) | |
| Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) | |
| */ | */ |
| #include <string.h> | static const char* IDENT_STRING_C="$Date$"; |
| #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_table.h" | |
| #include "pa_dictionary.h" | |
| #include "pa_charset.h" | |
| const String String::Empty; | |
| // cord lib extension | |
| #ifndef DOXYGEN | |
| typedef struct { | |
| ssize_t countdown; | |
| char 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; | |
| } | |
| // String | // helpers |
| String::String(Pool& apool) : | /// String::match uses this as replace & global search table columns |
| Pooled(apool) { | |
| last_chunk=&head; | const int MAX_MATCH_GROUPS=100; |
| head.count=CR_PREALLOCATED_COUNT; | |
| append_here=head.rows; | class String_match_table_template_columns: public ArrayString { |
| head.preallocated_link=0; | public: |
| link_row=&head.rows[head.count]; | String_match_table_template_columns() { |
| fused_rows=fsize=0; | *this+=new String("prematch"); |
| } | *this+=new String("match"); |
| *this+=new String("postmatch"); | |
| void String::expand() { | for(int i=0; i<MAX_MATCH_GROUPS; i++) { |
| int new_chunk_count=last_chunk->count+last_chunk->count*CR_GROW_PERCENT/100; | *this+=new String(String::Body::Format(1+i), String::L_CLEAN); |
| last_chunk=static_cast<Chunk *>( | |
| malloc(sizeof(int)+sizeof(Chunk::Row)*new_chunk_count+sizeof(Chunk *))); | |
| last_chunk->count=new_chunk_count; | |
| link_row->link=last_chunk; | |
| append_here=last_chunk->rows; | |
| link_row=&last_chunk->rows[last_chunk->count]; | |
| link_row->link=0; | |
| } | |
| String::String(const String& src) : | |
| Pooled(src.pool()) { | |
| head.count=CR_PREALLOCATED_COUNT; | |
| int src_used_rows=src.used_rows(); | |
| if(src_used_rows<=head.count) { | |
| // all new rows fit into preallocated area | |
| int curr_chunk_rows=head.count; | |
| memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*src_used_rows); | |
| append_here=&head.rows[src_used_rows]; | |
| link_row=&head.rows[curr_chunk_rows]; | |
| } else { | |
| // warning: | |
| // heavily relies on the fact | |
| // "preallocated area is the same for all strings" | |
| // | |
| // info: | |
| // allocating only enough mem to fit src string rows | |
| // next append would allocate a new chunk | |
| // | |
| // new rows don't fit into preallocated area: splitting into two chunks | |
| // preallocated chunk src to constructing head | |
| memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*head.count); | |
| // remaining rows into new_chunk | |
| int curr_chunk_rows=src_used_rows-head.count; | |
| Chunk *new_chunk=static_cast<Chunk *>( | |
| malloc(sizeof(int)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *))); | |
| new_chunk->count=curr_chunk_rows; | |
| head.preallocated_link=new_chunk; | |
| append_here=link_row=&new_chunk->rows[new_chunk->count]; | |
| Chunk *old_chunk=src.head.preallocated_link; | |
| Chunk::Row *new_rows=new_chunk->rows; | |
| int rows_left_to_copy=new_chunk->count; | |
| while(true) { | |
| int old_count=old_chunk->count; | |
| Chunk *next_chunk=old_chunk->rows[old_count].link; | |
| if(next_chunk) { | |
| // not last source chunk | |
| // taking it all | |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*old_count); | |
| new_rows+=old_count; | |
| rows_left_to_copy-=old_count; | |
| old_chunk=next_chunk; | |
| } else { | |
| // the last source chunk | |
| // taking only those rows of chunk that _left_to_copy | |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*rows_left_to_copy); | |
| break; | |
| } | |
| } | } |
| } | } |
| link_row->link=0; | }; |
| fused_rows=src_used_rows; | |
| fsize=src.fsize; | 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), length); | |
| } | } |
| String& String::append(const String& src, Untaint_lang lang) { | static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) { |
| int src_used_rows=src.used_rows(); | uint& result=*static_cast<uint*>(client_data); |
| int dst_free_rows=link_row-append_here; | generic_hash_code(result, c); |
| return 0; | |
| if(src_used_rows<=dst_free_rows) { | } |
| // all new rows fit into last chunk | static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) { |
| memcpy(append_here, src.head.rows, sizeof(Chunk::Row)*src_used_rows); | uint& result=*static_cast<uint*>(client_data); |
| set_lang(append_here, lang, src_used_rows); | generic_hash_code(result, s); |
| append_here+=src_used_rows; | return 0; |
| } else { | }; |
| // not all new rows fit into last chunk: shrinking it to used part, | uint String::Body::hash_code() const { |
| int used_rows=last_chunk->count-dst_free_rows; | uint result=0; |
| //int *countp=append_here | CORD_iter5(body, 0, |
| link_row=&last_chunk->rows[last_chunk->count=used_rows]; | CORD_batched_iter_fn_generic_hash_code, |
| // allocating only enough mem to fit src string rows | CORD_batched_iter_fn_generic_hash_code, &result); |
| // next append would allocate a new chunk | return result; |
| last_chunk=static_cast<Chunk *>( | } |
| malloc(sizeof(int)+sizeof(Chunk::Row)*src_used_rows+sizeof(Chunk *))); | |
| last_chunk->count=src_used_rows; | |
| link_row->link=last_chunk; | |
| append_here=link_row=&last_chunk->rows[src_used_rows]; | |
| const Chunk *old_chunk=&src.head; | |
| Chunk::Row *new_rows=last_chunk->rows; | |
| int rows_left_to_copy=src_used_rows; | |
| while(true) { | |
| int old_count=old_chunk->count; | |
| Chunk *next_chunk=old_chunk->rows[old_count].link; | |
| if(next_chunk) { | |
| // not last source chunk | |
| // taking it all | |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*old_count); | |
| set_lang(new_rows, lang, old_count); | |
| new_rows+=old_count; | |
| rows_left_to_copy-=old_count; | |
| old_chunk=next_chunk; | |
| } else { | |
| // the last source chunk | |
| // taking only those rows of chunk that _left_to_copy | |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*rows_left_to_copy); | |
| set_lang(new_rows, lang, rows_left_to_copy); | |
| break; | |
| } | |
| } | |
| link_row->link=0; | |
| } | |
| fused_rows+=src_used_rows; | |
| fsize+=src.fsize; | |
| return *this; | // String methods |
| String::String(const char* cstr, size_t helper_length, bool tainted): body(CORD_EMPTY) { | |
| append_help_length(cstr, helper_length, tainted?L_TAINTED:L_CLEAN); | |
| } | } |
| void String::set_lang(Chunk::Row *row, Untaint_lang lang, size_t size) { | String::String(const String::C cstr, bool tainted): body(CORD_EMPTY) { |
| if(lang==PASS_APPENDED) | append_know_length(cstr.str, cstr.length, tainted?L_TAINTED:L_CLEAN); |
| return; | |
| while(size--) { | |
| Untaint_lang& item_lang=(row++)->item.lang; | |
| if(item_lang==YES) // tainted? need untaint language assignment | |
| item_lang=lang; // assign untaint language | |
| } | |
| } | } |
| String& String::real_append(STRING_APPEND_PARAMS) { | String& String::append_know_length(const char* str, size_t known_length, Language lang) { |
| if(!src) | if(!known_length) |
| return *this; | return *this; |
| if(!size) | |
| size=strlen(src); | // first: langs |
| if(!size) | 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 *this; |
| if(chunk_is_full()) | return append_know_length(str, known_length, lang); |
| expand(); | } |
| 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; | |
| append_here->item.ptr=src; | // first: langs |
| fsize+=append_here->item.size=size; | langs.append(body, lang, known_length); |
| append_here->item.lang=tainted?/*Untaint_lang::*/YES:Untaint_lang::NO; | // next: letters themselves |
| #ifndef NO_STRING_ORIGIN | body.append_strdup_know_length(str, known_length); |
| append_here->item.origin.file=file; | |
| append_here->item.origin.line=line; | |
| #endif | |
| append_here++; fused_rows++; | |
| ASSERT_STRING_INVARIANT(*this); | |
| return *this; | return *this; |
| } | } |
| char *String::cstr() const { | /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab" |
| char *result=static_cast<char *>(malloc(size()+1)); | String& String::mid(size_t substr_begin, size_t substr_end) const { |
| String& result=*new String; | |
| size_t self_length=length(); | |
| substr_begin=min(substr_begin, self_length); | |
| substr_end=min(max(substr_end, substr_begin), self_length); | |
| size_t substr_length=substr_end-substr_begin; | |
| if(!substr_length) | |
| return result; | |
| // first: their langs | |
| result.langs.append(result.body, langs, substr_begin, substr_length); | |
| // next: letters themselves | |
| result.body=body.mid(substr_begin, substr_length); | |
| // SAPI::log("piece of '%s' from %d to %d is '%s'", | |
| //cstr(), substr_begin, substr_end, result.cstr()); | |
| ASSERT_STRING_INVARIANT(result); | |
| return result; | |
| } | |
| char *copy_here=result; | size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const { |
| const Chunk *chunk=&head; | size_t substr_length=substr.length(); |
| do { | while(true) { |
| const Chunk::Row *row=chunk->rows; | size_t substr_begin=body.pos(substr, this_offset); |
| for(int i=0; i<chunk->count; i++) { | |
| if(row==append_here) | if(substr_begin==CORD_NOT_FOUND) |
| goto break2; | return STRING_NOT_FOUND; |
| switch(row->item.lang) { | if(langs.check_lang(lang, substr_begin, substr_length)) |
| case NO: | return substr_begin; |
| case YES: // for VString.get_double of tainted values | |
| case AS_IS: | this_offset=substr_begin+substr_length; |
| memcpy(copy_here, row->item.ptr, row->item.size); | } |
| break; | } |
| case HTML_TYPO: | |
| memset(copy_here, '?', row->item.size); | size_t String::pos(const String& substr, |
| break; | size_t this_offset, Language lang) const { |
| default: | return pos(substr.body, this_offset, lang); |
| THROW(0,0, | } |
| this, | |
| "unknown untaint language #%d of %d piece", | void String::split(ArrayString& result, |
| static_cast<int>(row->item.lang), | size_t& pos_after, |
| i); | const char* delim, |
| } | Language lang, int limit) const { |
| copy_here+=row->item.size; | size_t self_length=length(); |
| row++; | if(size_t delim_length=strlen(delim)) { |
| int pos_before; | |
| // while we have 'delim'... | |
| for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) { | |
| result+=&mid(pos_after, pos_before); | |
| pos_after=pos_before+delim_length; | |
| } | } |
| chunk=row->link; | // last piece |
| } while(chunk); | if(pos_after<self_length && limit) { |
| break2: | result+=&mid(pos_after, self_length); |
| *copy_here=0; | pos_after=self_length; |
| return result; | } |
| } else { // empty delim | |
| result+=this; | |
| pos_after+=self_length; | |
| } | |
| } | } |
| uint String::hash_code() const { | void String::split(ArrayString& result, |
| uint result=0; | size_t& pos_after, |
| const String& delim, Language lang, | |
| int limit) const { | |
| if(!delim.is_empty()) { | |
| int pos_before; | |
| // while we have 'delim'... | |
| for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) { | |
| result+=&mid(pos_after, pos_before); | |
| pos_after=pos_before+delim.length(); | |
| } | |
| // last piece | |
| if(pos_after<length() && limit) { | |
| result+=&mid(pos_after, length()); | |
| pos_after=length(); | |
| } | |
| } else { // empty delim | |
| result+=this; | |
| pos_after+=length(); | |
| } | |
| } | |
| const Chunk *chunk=&head; | static void regex_options(const String* options, int *result, bool& need_pre_post_match){ |
| do { | struct Regex_option { |
| const Chunk::Row *row=chunk->rows; | const char* keyL; |
| for(int i=0; i<chunk->count; i++) { | const char* keyU; |
| if(row==append_here) | int clear, set; |
| goto break2; | int *result; |
| bool *flag; | |
| result=Hash::generic_code(result, row->item.ptr, row->item.size); | } regex_option[]={ |
| row++; | {"i", "I", 0, PCRE_CASELESS, result}, // a=A |
| } | {"s", "S", 0, PCRE_DOTALL, result}, // \n\n$ [default] |
| chunk=row->link; | {"x", "U", 0, PCRE_EXTENDED, result}, // whitespace in regex ignored |
| } while(chunk); | {"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$ |
| break2: | {"g", "G", 0, true, result+1}, // many rows |
| return result; | {"'", 0, 0, 0, 0, &need_pre_post_match}, |
| {0} | |
| }; | |
| result[0]=PCRE_EXTRA | PCRE_DOTALL | PCRE_DOLLAR_ENDONLY; | |
| result[1]=0; | |
| if(options && !options->is_empty()) | |
| for(Regex_option *o=regex_option; o->keyL; o++) | |
| if(options->pos(o->keyL)!=STRING_NOT_FOUND | |
| || (o->keyU && options->pos(o->keyU)!=STRING_NOT_FOUND)) { | |
| if(o->flag) | |
| *o->flag=true; | |
| else { // result | |
| *o->result &= ~o->clear; | |
| *o->result |= o->set; | |
| } | |
| } | |
| } | } |
| int String::cmp(const String& src) const { | Table* String::match(Charset& source_charset, |
| const Chunk *a_chunk=&head; | const String& regexp, |
| const Chunk *b_chunk=&src.head; | const String* options, |
| const Chunk::Row *a_row=a_chunk->rows; | Row_action row_action, void *info, |
| const Chunk::Row *b_row=b_chunk->rows; | bool& just_matched) const { |
| int a_offset=0; | if(regexp.is_empty()) |
| int b_offset=0; | throw Exception(0, |
| Chunk::Row *a_end=append_here; | 0, |
| Chunk::Row *b_end=src.append_here; | "regexp is empty"); |
| int a_countdown=a_chunk->count; | |
| int b_countdown=b_chunk->count; | const char* pattern=regexp.cstr(); |
| bool a_break=false; | const char* errptr; |
| bool b_break=false; | int erroffset; |
| int result; | bool need_pre_post_match=false; |
| int option_bits[2]={0}; regex_options(options, option_bits, need_pre_post_match); | |
| bool global=option_bits[1]!=0; | |
| pcre *code=pcre_compile(pattern, option_bits[0], | |
| &errptr, &erroffset, | |
| source_charset.pcre_tables); | |
| if(!code) | |
| throw Exception(0, | |
| ®exp.mid(erroffset, regexp.length()), | |
| "regular expression syntax error - %s", errptr); | |
| int subpatterns=pcre_info(code, 0, 0); | |
| if(subpatterns<0) { | |
| pcre_free(code); | |
| throw Exception(0, | |
| ®exp, | |
| "pcre_info error (%d)", | |
| subpatterns); | |
| } | |
| const char* subject=cstr(); | |
| size_t subject_length=strlen(subject); | |
| const int oveclength=(1/*match*/+MAX_MATCH_GROUPS)*3; | |
| int ovector[oveclength]; | |
| // create table | |
| Table::Action_options table_options; | |
| Table& table=*new Table(string_match_table_template, table_options); | |
| int exec_option_bits=0; | |
| int prestart=0; | |
| int poststart=0; | |
| int postfinish=length(); | |
| while(true) { | while(true) { |
| a_break=a_row==a_end; | int exec_substrings=pcre_exec(code, 0, |
| b_break=b_row==b_end; | subject, subject_length, prestart, |
| if(a_break || b_break) | exec_option_bits, ovector, oveclength); |
| if(exec_substrings==PCRE_ERROR_NOMATCH) { | |
| pcre_free(code); | |
| row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info); | |
| if(global || subpatterns) | |
| return &table; // global or with subpatterns=true+result | |
| else { | |
| just_matched=false; return 0; // not global=no result | |
| } | |
| } | |
| if(exec_substrings<0) { | |
| pcre_free(code); | |
| throw Exception(0, | |
| ®exp, | |
| "regular expression execute error (%d)", | |
| exec_substrings); | |
| } | |
| int prefinish=ovector[0]; | |
| poststart=ovector[1]; | |
| ArrayString* row=new ArrayString; | |
| 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++) { | |
| // -1:-1 case handled peacefully by mid() itself | |
| *row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value | |
| } | |
| row_action(table, row, prestart, prefinish, poststart, postfinish, info); | |
| if(!global || prestart==poststart) { // not global | going to hang | |
| pcre_free(code); | |
| row_action(table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info); | |
| return &table; | |
| } | |
| prestart=poststart; | |
| /* | |
| if(option_bits[0] & PCRE_MULTILINE) | |
| exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL | |
| */ | |
| } | |
| } | |
| 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(); | |
| char *dest=new_cstr; | |
| if(source_charset.isUTF8()) { | |
| switch(kind) { | |
| case CC_UPPER: | |
| change_case_UTF8((const XMLByte*)new_cstr, (XMLByte*)new_cstr, UTF8CaseToUpper); | |
| break; | break; |
| case CC_LOWER: | |
| change_case_UTF8((const XMLByte*)new_cstr, (XMLByte*)new_cstr, UTF8CaseToLower); | |
| break; | |
| default: | |
| assert(!"unknown change case kind"); | |
| break; // never | |
| } | |
| } else { | |
| const unsigned char *tables=source_charset.pcre_tables; | |
| int size_diff= | const unsigned char *a; |
| (a_row->item.size-a_offset)- | const unsigned char *b; |
| (b_row->item.size-b_offset); | switch(kind) { |
| case CC_UPPER: | |
| if(size_diff==0) { // a has same size as b | a=tables+lcc_offset; |
| result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, a_row->item.size-a_offset); | b=tables+fcc_offset; |
| if(result) | break; |
| return result; | case CC_LOWER: |
| a_row++; a_countdown--; a_offset=0; | a=tables+lcc_offset; |
| b_row++; b_countdown--; b_offset=0; | b=0; |
| } else if (size_diff>0) { // a longer | break; |
| result=memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, b_row->item.size-b_offset); | default: |
| if(result) | assert(!"unknown change case kind"); |
| return result; | a=b=0; // calm, compiler |
| a_offset+=b_row->item.size-b_offset; | break; // never |
| 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); | unsigned char index; |
| if(result) | for(const char* current=new_cstr; index=(unsigned char)*current; current++) { |
| return result; | unsigned char c=a[index]; |
| b_offset+=a_row->item.size-a_offset; | if(b) |
| a_row++; a_countdown--; a_offset=0; | c=b[c]; |
| } | |
| *dest++=(char)c; | |
| if(!a_countdown) { | |
| a_chunk=a_row->link; | |
| a_row=a_chunk->rows; | |
| a_countdown=a_chunk->count; | |
| } | |
| if(!b_countdown) { | |
| b_chunk=b_row->link; | |
| b_row=b_chunk->rows; | |
| b_countdown=b_chunk->count; | |
| } | } |
| } | } |
| if(a_break==b_break) // ended simultaneously | result.langs=langs; |
| result=0; | result.body=new_cstr; |
| else if(a_break) // first bytes equal, but a ended before b | |
| result=-1; | return result; |
| } | |
| const String& String::replace(const Dictionary& dict) const { | |
| String& result=*new String(); | |
| const char* old_cstr=cstr(); | |
| const char* prematch_begin=old_cstr; | |
| const char* current=old_cstr; | |
| while(*current) { | |
| if(Dictionary::Subst subst=dict.first_that_begins(current)) { | |
| // prematch | |
| if(size_t prematch_length=current-prematch_begin) { | |
| result.langs.append(result.body, langs, prematch_begin-old_cstr, prematch_length); | |
| result.body.append_strdup_know_length(prematch_begin, prematch_length); | |
| } | |
| // match | |
| // skip 'a' in 'current'; move prematch_begin | |
| current+=subst.from_length; prematch_begin=current; | |
| if(const String* b=subst.to) // are there any b? | |
| result<<*b; | |
| } else // simply advance | |
| current++; | |
| } | |
| // postmatch | |
| if(size_t postmatch_length=current-prematch_begin) { | |
| result.langs.append(result.body, langs, prematch_begin-old_cstr, postmatch_length); | |
| result.body.append_strdup_know_length(prematch_begin, postmatch_length); | |
| } | |
| ASSERT_STRING_INVARIANT(result); | |
| return result; | |
| } | |
| double String::as_double() const { | |
| double result; | |
| const char *str=cstr(); | |
| while(*str && isspace(*str)) | |
| str++; | |
| if(!*str) | |
| return 0; | |
| char *error_pos; | |
| // 0xABC | |
| if(str[0]=='0') | |
| if(str[1]=='x' || str[1]=='X') | |
| result=(double)(unsigned long)strtol(str, &error_pos, 0); | |
| else | |
| result=(double)strtod(str+1/*skip leading 0*/, &error_pos); | |
| else | |
| result=(double)strtod(str, &error_pos); | |
| while(char c=*error_pos++) | |
| if(!isspace(c)) | |
| throw Exception("number.format", | |
| this, | |
| "invalid number (double)"); | |
| return result; | |
| } | |
| int String::as_int() const { | |
| int result; | |
| const char *str=cstr(); | |
| while(*str && isspace(*str)) | |
| str++; | |
| if(!*str) | |
| return 0; | |
| char *error_pos; | |
| // 0xABC | |
| if(str[0]=='0') | |
| if(str[1]=='x' || str[1]=='X') | |
| result=(int)(unsigned long)strtol(str, &error_pos, 0); | |
| else | |
| result=(int)strtol(str+1/*skip leading 0*/, &error_pos, 0); | |
| else | else |
| result=+1; | result=(int)strtol(str, &error_pos, 0); |
| while(char c=*error_pos++) | |
| if(!isspace(c)) | |
| throw Exception("number.format", | |
| this, | |
| "invalid number (int)"); | |
| return result; | return result; |
| } | } |
| bool String::operator == (char* b_ptr) const { | static int serialize_body_char(char c, char** cur) { |
| size_t b_size=b_ptr?strlen(b_ptr):0; | *((*cur)++)=c; |
| if(size() != b_size) | 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 | |
| memcpy(*cur, &alang, sizeof(alang)); *cur+=sizeof(alang); | |
| // length | |
| 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 buf_length= | |
| prolog_length //1 | |
| +sizeof(size_t) //2 | |
| +fragments_count*(sizeof(char)+sizeof(size_t)) //3 | |
| +body.length() //4 | |
| +1; // for zero terminator used in deserialize | |
| String::Cm result(new(PointerFreeGC) char[buf_length], buf_length); | |
| // 1: prolog | |
| char *cur=result.str+prolog_length; | |
| // 2: langs.count | |
| memcpy(cur, &fragments_count, sizeof(fragments_count)); cur+=sizeof(fragments_count); | |
| // 3: lang info | |
| langs.for_each(body, serialize_lang_piece, &cur); | |
| // 4: letters | |
| body.for_each(serialize_body_char, serialize_body_piece, &cur); | |
| // 5: zero terminator | |
| *cur=0; | |
| return result; | |
| } | |
| bool String::deserialize(size_t prolog_length, void *buf, size_t buf_length) { | |
| if(buf_length<=prolog_length) | |
| return false; | return false; |
| buf_length-=prolog_length; | |
| buf_length-=1; // 5: zero terminator | |
| const Chunk *a_chunk=&head; | // 1: prolog |
| const Chunk::Row *a_row=a_chunk->rows; | const char* cur=(const char* )buf+prolog_length; |
| int a_offset=0; | |
| int b_offset=0; | |
| Chunk::Row *a_end=append_here; | |
| int a_countdown=a_chunk->count; | |
| bool a_break=false; | |
| bool b_break=false; | |
| while(true) { | |
| int size_diff= | |
| (a_row->item.size-a_offset)- | |
| (b_size-b_offset); | |
| if(size_diff==0) { // a has same size as b | // 2: langs.count |
| if(memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, a_row->item.size-a_offset)!=0) | if(buf_length<sizeof(size_t)) // langs.count don't fit? |
| return false; | return false; |
| a_row++; a_countdown--; a_offset=0; | size_t fragments_count=*reinterpret_cast<const size_t*>(cur); cur+=sizeof(size_t); |
| b_break=true; | buf_length-=sizeof(size_t); |
| } else if (size_diff>0) { // a longer | |
| if(memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, b_size-b_offset)!=0) | if(fragments_count) { |
| return false; | // 3: lang info |
| a_offset+=b_size-b_offset; | size_t total_length=0; |
| b_break=true; | for(size_t f=0; f<fragments_count; f++) { |
| } else { // b longer | size_t piece_length=sizeof(char)+sizeof(size_t); |
| if(memcmp(a_row->item.ptr+a_offset, b_ptr+b_offset, a_row->item.size-a_offset)!=0) | if(buf_length<piece_length) // lang+length |
| return false; | return false; |
| b_offset+=a_row->item.size-a_offset; | |
| a_row++; a_countdown--; a_offset=0; | |
| } | |
| a_break=a_row==a_end; | Language lang=*reinterpret_cast<const Language *>(cur); cur+=sizeof(char); |
| if(a_break || b_break) | size_t fragment_length=*reinterpret_cast<const size_t*>(cur); cur+=sizeof(size_t); |
| break; | langs.append(total_length, lang, fragment_length); |
| total_length+=fragment_length; | |
| if(!a_countdown) { | buf_length-=piece_length; |
| a_chunk=a_row->link; | |
| a_row=a_chunk->rows; | |
| a_countdown=a_chunk->count; | |
| } | } |
| // 4: letters | |
| if(buf_length!=total_length) | |
| return false; | |
| // serialize wrote extra zero byte there, we can rely on that | |
| body=String::Body(cur, buf_length); | |
| } | } |
| return a_break==b_break; | |
| ASSERT_STRING_INVARIANT(*this); | |
| return true; | |
| } | |
| const char* String::Body::v() const { | |
| return CORD_to_const_char_star(body); | |
| } | |
| const char* String::Languages::v() const { | |
| if(opt.is_not_just_lang) | |
| return CORD_to_const_char_star(langs); | |
| else | |
| return (const char*)&langs; | |
| } | |
| const char* String::v() const { | |
| #define LIMIT_VIEW 20 | |
| char* buf=(char*)malloc(MAX_STRING); | |
| const char*body_view=body.v(); | |
| const char*langs_view=langs.v(); | |
| snprintf(buf, MAX_STRING, | |
| "%d:%.*s%s} " | |
| "{%d:%s", | |
| langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"", | |
| strlen(body_view), body_view | |
| ); | |
| return buf; | |
| #undef LIMIT_VIEW | |
| } | } |