|
|
| version 1.22, 2001/02/13 10:50:23 | version 1.201, 2004/10/07 09:22:01 |
|---|---|
| Line 1 | Line 1 |
| /* | /** @file |
| $Id$ | Parser: string class. @see untalength_t.C. |
| Copyright (c) 2001-2004 ArtLebedev Group (http://www.artlebedev.com) | |
| Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) | |
| */ | */ |
| #include <string.h> | static const char * const 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; | |
| int pa_atoi(const char* str, const String* problem_source) { | |
| if(!str) | |
| return 0; | |
| while(*str && isspace((unsigned char)*str)) | |
| str++; | |
| if(!*str) | |
| return 0; | |
| int result; | |
| char *error_pos; | |
| bool negative=false; | |
| if(str[0]=='-') { | |
| negative=true; | |
| str++; | |
| } else if(str[0]=='+') { | |
| str++; | |
| } | |
| // 0xABC | |
| if(str[0]=='0') | |
| if(str[1]=='x' || str[1]=='X') | |
| result=(int)(unsigned long)strtol(str, &error_pos, 0); | |
| else { | |
| // skip leading 0000, to disable octal interpretation | |
| do str++; while(*str=='0'); | |
| result=(int)strtol(str, &error_pos, 0); | |
| } | |
| else | |
| result=(int)strtol(str, &error_pos, 0); | |
| if(negative) | |
| result=-result; | |
| while(char c=*error_pos++) | |
| if(!isspace((unsigned char)c)) | |
| throw Exception("number.format", | |
| problem_source, | |
| problem_source?"invalid number (int)": "'%s' is invalid number (int)", str); | |
| // String | return result; |
| } | |
| String::String(Pool& apool) : | double pa_atod(const char* str, const String* problem_source) { |
| Pooled(apool) { | if(!str) |
| head.count=curr_chunk_rows=CR_PREALLOCATED_COUNT; | return 0; |
| append_here=head.rows; | |
| head.preallocated_link=0; | while(*str && isspace((unsigned char)*str)) |
| link_row=&head.rows[curr_chunk_rows]; | str++; |
| fused_rows=fsize=0; | if(!*str) |
| } | return 0; |
| void String::expand() { | double result; |
| curr_chunk_rows+=curr_chunk_rows*CR_GROW_PERCENT/100; | char *error_pos; |
| Chunk *chunk=static_cast<Chunk *>( | bool negative=false; |
| pool.malloc(sizeof(int)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *))); | if(str[0]=='-') { |
| chunk->count=curr_chunk_rows; | negative=true; |
| link_row->link=chunk; | str++; |
| append_here=chunk->rows; | } else if(str[0]=='+') { |
| link_row=&chunk->rows[curr_chunk_rows]; | str++; |
| link_row->link=0; | } |
| } | // 0xABC |
| if(str[0]=='0') | |
| String::String(const String& src) : | if(str[1]=='x' || str[1]=='X') |
| Pooled(src.pool) { | result=(double)(unsigned long)strtol(str, &error_pos, 0); |
| head.count=CR_PREALLOCATED_COUNT; | else { |
| // skip leading 0000, to disable octal interpretation | |
| int src_used_rows=src.used_rows(); | do str++; while(*str=='0'); |
| if(src_used_rows<=head.count) { | result=(double)strtod(str, &error_pos); |
| // all new rows fit into preallocated area | } |
| curr_chunk_rows=head.count; | else |
| memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*src_used_rows); | result=(double)strtod(str, &error_pos); |
| append_here=&head.rows[src_used_rows]; | if(negative) |
| link_row=&head.rows[curr_chunk_rows]; | result=-result; |
| } else { | |
| // warning: | while(char c=*error_pos++) |
| // heavily relies on the fact | if(!isspace((unsigned char)c)) |
| // "preallocated area is the same for all strings" | throw Exception("number.format", |
| // | problem_source, |
| // info: | problem_source?"invalid number (double)": "'%s' is invalid number (double)", str); |
| // allocating only enough mem to fit src string rows | |
| // next append would allocate a new chunk | return result; |
| // | } |
| // new rows don't fit into preallocated area: splitting into two chunks | |
| // preallocated chunk src to constructing head | // cord lib extension |
| memcpy(head.rows, src.head.rows, sizeof(Chunk::Row)*head.count); | |
| // remaining rows into new_chunk | #ifndef DOXYGEN |
| curr_chunk_rows=src_used_rows-head.count; | typedef struct { |
| Chunk *new_chunk=static_cast<Chunk *>( | ssize_t countdown; |
| pool.malloc(sizeof(int)+sizeof(Chunk::Row)*curr_chunk_rows+sizeof(Chunk *))); | char target; /* Character we're looking for */ |
| new_chunk->count=curr_chunk_rows; | } chr_data; |
| head.preallocated_link=new_chunk; | #endif |
| append_here=link_row=&new_chunk->rows[curr_chunk_rows]; | static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data) |
| { | |
| Chunk *old_chunk=src.head.preallocated_link; | register chr_data * d = (chr_data *)client_data; |
| Chunk::Row *new_rows=new_chunk->rows; | |
| int rows_left_to_copy=curr_chunk_rows; | if (d -> countdown<=0) return(2); |
| while(true) { | d -> countdown -= size; |
| int old_count=old_chunk->count; | if (c > d -> target) return(1); |
| Chunk *next_chunk=old_chunk->rows[old_count].link; | return(0); |
| if(next_chunk) { | } |
| // not last source chunk | int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c) |
| // taking it all | { |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*old_count); | chr_data d; |
| new_rows+=old_count; | |
| rows_left_to_copy-=old_count; | d.countdown = n; |
| d.target = c; | |
| old_chunk=next_chunk; | return(CORD_block_iter(x, i, CORD_range_contains_chr_greater_then_proc, &d) == 1/*alternatives: 0 normally ended, 2=struck 'n'*/); |
| } else { | } |
| // the last source chunk | |
| // taking only those rows of chunk that _left_to_copy | static int CORD_block_count_proc(char /*c*/, size_t /*size*/, void* client_data) |
| memcpy(new_rows, old_chunk->rows, sizeof(Chunk::Row)*rows_left_to_copy); | { |
| break; | int* result=(int*)client_data; |
| } | (*result)++; |
| return(0); // 0=continue | |
| } | |
| size_t CORD_block_count(CORD x) | |
| { | |
| size_t result=0; | |
| CORD_block_iter(x, 0, CORD_block_count_proc, &result); | |
| return result; | |
| } | |
| // helpers | |
| /// String::match uses this as replace & global search table columns | |
| const int MAX_MATCH_GROUPS=100; | |
| class String_match_table_template_columns: public ArrayString { | |
| public: | |
| String_match_table_template_columns() { | |
| *this+=new String("prematch"); | |
| *this+=new String("match"); | |
| *this+=new String("postmatch"); | |
| for(int i=0; i<MAX_MATCH_GROUPS; i++) { | |
| *this+=new String(String::Body::Format(1+i), String::L_CLEAN); | |
| } | } |
| } | } |
| 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::real_append(STRING_APPEND_PARAMS) { | String::Body String::Body::trim(String::Trim_kind kind, const char* chars, |
| if(!src) | size_t* out_start, size_t* out_length) const { |
| return *this; | size_t our_length=length(); |
| int len=strlen(src); | if(!our_length) |
| if(!len) | |
| return *this; | return *this; |
| if(!chars) | |
| chars=" \t\n"; // white space | |
| if(chunk_is_full()) | size_t start=0; |
| expand(); | size_t end=our_length; |
| // 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; | |
| append_here->item.ptr=src; | CORD_next(pos); |
| fsize+=append_here->item.size=len; | } |
| #ifndef NO_STRING_ORIGIN | } |
| append_here->item.origin.file=file; | // from right.. |
| append_here->item.origin.line=line; | if(kind!=TRIM_START) { |
| #endif | CORD_pos pos; set_pos(pos, end-1); |
| append_here++; fused_rows++; | 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; | |
| return *this; | CORD_prev(pos); |
| } | } |
| } | |
| char *String::cstr() const { | if(start==0 && end==our_length) // nobody moved a thing |
| char *result=static_cast<char *>(pool.malloc(size()+1)); | return *this; |
| char *copy_here=result; | if(out_start) |
| const Chunk *chunk=&head; | *out_start=start; |
| do { | size_t new_length=end-start; |
| const Chunk::Row *row=chunk->rows; | if(out_length) |
| for(int i=0; i<chunk->count; i++) { | *out_length=new_length; |
| if(row==append_here) | |
| goto break2; | return mid(start, new_length); |
| } | |
| memcpy(copy_here, row->item.ptr, row->item.size); | |
| copy_here+=row->item.size; | static int CORD_batched_iter_fn_generic_hash_code(char c, void * client_data) { |
| row++; | uint& result=*static_cast<uint*>(client_data); |
| } | generic_hash_code(result, c); |
| chunk=row->link; | return 0; |
| } while(chunk); | } |
| break2: | static int CORD_batched_iter_fn_generic_hash_code(const char* s, void * client_data) { |
| *copy_here=0; | uint& result=*static_cast<uint*>(client_data); |
| generic_hash_code(result, s); | |
| return 0; | |
| }; | |
| uint String::Body::hash_code() const { | |
| uint result=0; | |
| CORD_iter5(body, 0, | |
| CORD_batched_iter_fn_generic_hash_code, | |
| CORD_batched_iter_fn_generic_hash_code, &result); | |
| return result; | return result; |
| } | } |
| uint String::hash_code() const { | // String methods |
| uint result=0; | |
| const Chunk *chunk=&head; | String::String(const char* cstr, size_t helper_length, bool tainted): body(CORD_EMPTY) { |
| do { | append_help_length(cstr, helper_length, tainted?L_TAINTED:L_CLEAN); |
| const Chunk::Row *row=chunk->rows; | } |
| for(int i=0; i<chunk->count; i++) { | String::String(const String::C cstr, bool tainted): body(CORD_EMPTY) { |
| if(row==append_here) | append_know_length(cstr.str, cstr.length, tainted?L_TAINTED:L_CLEAN); |
| goto break2; | |
| result=Hash::generic_code(result, row->item.ptr, row->item.size); | |
| row++; | |
| } | |
| chunk=row->link; | |
| } while(chunk); | |
| break2: | |
| return result; | |
| } | } |
| bool String::operator == (const String& src) const { | String& String::append_know_length(const char* str, size_t known_length, Language lang) { |
| if(size() != src.size()) | if(!known_length) |
| return false; | return *this; |
| const Chunk *a_chunk=&head; | |
| const Chunk *b_chunk=&src.head; | |
| const Chunk::Row *a_row=a_chunk->rows; | |
| const Chunk::Row *b_row=b_chunk->rows; | |
| int a_offset=0; | |
| int b_offset=0; | |
| Chunk::Row *a_end=append_here; | |
| Chunk::Row *b_end=src.append_here; | |
| int a_countdown=a_chunk->count; | |
| int b_countdown=b_chunk->count; | |
| bool a_break=false; | |
| bool b_break=false; | |
| while(true) { | |
| int size_diff= | |
| (a_row->item.size-a_offset)- | |
| (b_row->item.size-b_offset); | |
| if(size_diff==0) { // a has same size as b | // first: langs |
| if(memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, a_row->item.size-a_offset)!=0) | langs.append(body, lang, known_length); |
| return false; | // next: letters themselves |
| a_row++; a_countdown--; a_offset=0; | body.append_know_length(str, known_length); |
| b_row++; b_countdown--; b_offset=0; | |
| } else if (size_diff>0) { // a longer | |
| if(memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, b_row->item.size-b_offset)!=0) | |
| return false; | |
| a_offset+=b_row->item.size-b_offset; | |
| b_row++; b_countdown--; b_offset=0; | |
| } else { // b longer | |
| if(memcmp(a_row->item.ptr+a_offset, b_row->item.ptr+b_offset, a_row->item.size-a_offset)!=0) | |
| return false; | |
| b_offset+=a_row->item.size-a_offset; | |
| a_row++; a_countdown--; a_offset=0; | |
| } | |
| a_break=a_row==a_end; | ASSERT_STRING_INVARIANT(*this); |
| b_break=b_row==b_end; | return *this; |
| if(a_break || b_break) | } |
| break; | 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; | |
| if(!a_countdown) { | return append_know_length(str, known_length, lang); |
| a_chunk=a_row->link; | |
| a_row=a_chunk->rows; | |
| a_countdown=a_chunk->count; | |
| } | |
| if(!b_countdown) { | |
| b_chunk=b_row->link; | |
| b_row=b_chunk->rows; | |
| b_countdown=b_chunk->count; | |
| } | |
| } | |
| return a_break==b_break; | |
| } | } |
| 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; | |
| String& String::append(const String_iterator& begin, const String_iterator& end) { | // first: langs |
| //TODO | langs.append(body, lang, known_length); |
| // next: letters themselves | |
| body.append_strdup_know_length(str, known_length); | |
| ASSERT_STRING_INVARIANT(*this); | |
| return *this; | return *this; |
| } | } |
| // Char_types | /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab" |
| String& String::mid(size_t substr_begin, size_t substr_end) const { | |
| String& result=*new String; | |
| size_t self_length=length(); | |
| substr_begin=min(substr_begin, self_length); | |
| substr_end=min(max(substr_end, substr_begin), self_length); | |
| size_t substr_length=substr_end-substr_begin; | |
| if(!substr_length) | |
| return result; | |
| // first: their langs | |
| result.langs.append(result.body, langs, substr_begin, substr_length); | |
| // next: letters themselves | |
| result.body=body.mid(substr_begin, substr_length); | |
| Char_types::Char_types() { | ASSERT_STRING_INVARIANT(result); |
| memset(types, 0, sizeof(types)); | return result; |
| } | } |
| // String_iterator | size_t String::pos(const String::Body substr, size_t this_offset, Language lang) const { |
| size_t substr_length=substr.length(); | |
| while(true) { | |
| size_t substr_begin=body.pos(substr, this_offset); | |
| if(substr_begin==CORD_NOT_FOUND) | |
| return STRING_NOT_FOUND; | |
| if(langs.check_lang(lang, substr_begin, substr_length)) | |
| return substr_begin; | |
| this_offset=substr_begin+substr_length; | |
| } | |
| } | |
| String_iterator::String_iterator(String& astring) : string(astring) { | size_t String::pos(const String& substr, |
| read_here=string.head.rows; | size_t this_offset, Language lang) const { |
| position=string.size()==0?0:read_here->item.ptr; | return pos(substr.body, this_offset, lang); |
| link_row=reinterpret_cast<String::Chunk::Row*>(string.head.preallocated_link); | } |
| void String::split(ArrayString& result, | |
| size_t& pos_after, | |
| const char* delim, | |
| Language lang, int limit) const { | |
| size_t self_length=length(); | |
| if(size_t delim_length=strlen(delim)) { | |
| size_t 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<self_length && limit) { | |
| result+=&mid(pos_after, self_length); | |
| pos_after=self_length; | |
| } | |
| } else { // empty delim | |
| result+=this; | |
| pos_after+=self_length; | |
| } | |
| } | } |
| char String_iterator::operator()() const { | void String::split(ArrayString& result, |
| return position?*position:0; | size_t& pos_after, |
| const String& delim, Language lang, | |
| int limit) const { | |
| if(!delim.is_empty()) { | |
| size_t 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(); | |
| } | |
| } | } |
| void String_iterator::skip() { | static void regex_options(const String* options, int *result, bool& need_pre_post_match){ |
| if(!position) | struct Regex_option { |
| return; | const char* keyL; |
| const char* keyU; | |
| int clear, set; | |
| int *result; | |
| bool *flag; | |
| } regex_option[]={ | |
| {"i", "I", 0, PCRE_CASELESS, result, 0}, // a=A | |
| {"s", "S", 0, PCRE_DOTALL, result, 0}, // \n\n$ [default] | |
| {"x", "U", 0, PCRE_EXTENDED, result, 0}, // whitespace in regex ignored | |
| {"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result, 0}, // ^aaa\n$^bbb\n$ | |
| {"g", "G", 0, 1, result+1, 0}, // many rows | |
| {"'", 0, 0, 0, 0, &need_pre_post_match}, | |
| {0, 0, 0, 0, 0, 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; | |
| } | |
| } | |
| } | |
| if(++position== | Table* String::match(Charset& source_charset, |
| read_here->item.ptr+ | const String& regexp, |
| read_here->item.size) { | const String* options, |
| Row_action row_action, void *info, | |
| bool& just_matched) const { | |
| if(regexp.is_empty()) | |
| throw Exception(0, | |
| 0, | |
| "regexp is empty"); | |
| const char* pattern=regexp.cstr(); | |
| const char* errptr; | |
| int erroffset; | |
| 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); | |
| } | |
| // next row | const char* subject=cstr(); |
| if(++read_here==string.append_here) { | size_t subject_length=strlen(subject); |
| position=0; | const int oveclength=(1/*match*/+MAX_MATCH_GROUPS)*3; |
| return; | 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) { | |
| int exec_substrings=pcre_exec(code, 0, | |
| subject, subject_length, prestart, | |
| 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(read_here==link_row) { | |
| String::Chunk *chunk=link_row->link; | |
| if(!chunk) | |
| string.pool.exception().raise(0, 0, | |
| &string, | |
| "String_iterator::skip() missed " | |
| "read_here==string.append_here check"); | |
| read_here=chunk->rows; | if(exec_substrings<0) { |
| link_row=&chunk->rows[chunk->count]; | 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; | |
| } | } |
| position=read_here->item.ptr; | prestart=poststart; |
| /* | |
| if(option_bits[0] & PCRE_MULTILINE) | |
| exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL | |
| */ | |
| } | } |
| } | } |
| bool String_iterator::skip_to(char c) { | String& String::change_case(Charset& source_charset, Change_case_kind kind) const { |
| if(!position) | String& result=*new String(); |
| return false; | if(is_empty()) |
| return result; | |
| char* new_cstr=cstrm(); | |
| size_t new_cstr_len=length(); | |
| if(source_charset.isUTF8()) { | |
| 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; | |
| while(true) { | const unsigned char *a; |
| if(char *found=static_cast<char *>( | const unsigned char *b; |
| memchr(position, c, read_here->item.size-(position-read_here->item.ptr)))) { | switch(kind) { |
| position=found; | case CC_UPPER: |
| return true; | 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]; | |
| // next row | *dest++=(char)c; |
| if(++read_here==string.append_here) { | |
| position=0; | |
| return false; | |
| } | } |
| if(read_here==link_row) { | } |
| String::Chunk *chunk=link_row->link; | result.langs=langs; |
| if(!chunk) | result.body=new_cstr; |
| string.pool.exception().raise(0, 0, | |
| &string, | |
| "String_iterator::skip_to(char) missed " | |
| "read_here==string.append_here check"); | |
| read_here=chunk->rows; | return result; |
| link_row=&chunk->rows[chunk->count]; | } |
| } | |
| position=read_here->item.ptr; | const String& String::replace(const Dictionary& dict) const { |
| String& result=*new String(); | |
| const char* old_cstr=cstr(); | |
| const char* prematch_begin=old_cstr; | |
| const char* current=old_cstr; | |
| while(*current) { | |
| if(Dictionary::Subst subst=dict.first_that_begins(current)) { | |
| // prematch | |
| if(size_t prematch_length=current-prematch_begin) { | |
| result.langs.append(result.body, langs, prematch_begin-old_cstr, prematch_length); | |
| result.body.append_strdup_know_length(prematch_begin, prematch_length); | |
| } | |
| // match | |
| // skip 'a' in 'current'; move prematch_begin | |
| current+=subst.from_length; prematch_begin=current; | |
| if(const String* b=subst.to) // are there any b? | |
| result<<*b; | |
| } else // simply advance | |
| current++; | |
| } | } |
| // postmatch | |
| if(size_t postmatch_length=current-prematch_begin) { | |
| result.langs.append(result.body, langs, prematch_begin-old_cstr, postmatch_length); | |
| result.body.append_strdup_know_length(prematch_begin, postmatch_length); | |
| } | |
| ASSERT_STRING_INVARIANT(result); | |
| return result; | |
| } | |
| static int serialize_body_char(char c, char** cur) { | |
| *((*cur)++)=c; | |
| return 0; // 0=continue | |
| }; | |
| static int serialize_body_piece(const char* s, char** cur) { | |
| size_t length=strlen(s); | |
| memcpy(*cur, s, length); *cur+=length; | |
| return 0; // 0=continue | |
| }; | |
| static int serialize_lang_piece(char alang, size_t asize, char** cur) { | |
| // lang | |
| **cur=alang; (*cur)++; | |
| // length [WARNING: not cast, addresses must be %4=0 on sparc] | |
| memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize); | |
| return 0; // 0=continue | |
| } | |
| String::Cm String::serialize(size_t prolog_length) const { | |
| size_t fragments_count=langs.count(); | |
| size_t 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 [WARNING: not cast, addresses must be %4=0 on sparc] | |
| 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; | |
| buf_length-=prolog_length; | |
| buf_length-=1; // 5: zero terminator | |
| int String_iterator::skip_to(Char_types& types) { | // 1: prolog |
| if(!position) | const char* cur=(const char* )buf+prolog_length; |
| // 2: langs.count | |
| size_t fragments_count; | |
| if(buf_length<sizeof(fragments_count)) // langs.count don't fit? | |
| return false; | return false; |
| // [WARNING: not cast, addresses must be %4=0 on sparc] | |
| memcpy(&fragments_count, cur, sizeof(fragments_count)); cur+=sizeof(fragments_count); | |
| buf_length-=sizeof(fragments_count); | |
| if(fragments_count) { | |
| // 3: 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(buf_length<piece_length) // lang+length | |
| return false; | |
| while(true) { | // lang |
| int countdown=read_here->item.size-(position-read_here->item.ptr); | lang=*cur++; |
| for(; countdown--; position++) | // length [WARNING: not cast, addresses must be %4=0 on sparc] |
| if(int type=types.get(*position)) | memcpy(&fragment_length, cur, sizeof(fragment_length)); cur+=sizeof(fragment_length); |
| return type; | |
| // uchar needed to prevent propagating 0x80 bit to upper bytes | |
| // next row | langs.append(total_length, (String::Language)(uchar)lang, fragment_length); |
| if(++read_here==string.append_here) { | total_length+=fragment_length; |
| position=0; | |
| return -1; | |
| } | |
| if(read_here==link_row) { | |
| String::Chunk *chunk=link_row->link; | |
| if(!chunk) | |
| string.pool.exception().raise(0, 0, | |
| &string, | |
| "String_iterator::skip_to(Char_type) missed " | |
| "read_here==string.append_here check"); | |
| read_here=chunk->rows; | buf_length-=piece_length; |
| link_row=&chunk->rows[chunk->count]; | |
| } | } |
| position=read_here->item.ptr; | |
| // 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); | |
| } | } |
| ASSERT_STRING_INVARIANT(*this); | |
| return true; | |
| } | |
| const char* String::Body::v() const { | |
| return CORD_to_const_char_star(body); | |
| } | |
| void String::Body::dump() const { | |
| CORD_dump(body); | |
| } | |
| const char* String::Languages::v() const { | |
| if(opt.is_not_just_lang) | |
| return CORD_to_const_char_star(langs); | |
| else | |
| return (const char*)&langs; | |
| } | |
| void String::Languages::dump() const { | |
| if(opt.is_not_just_lang) | |
| CORD_dump(langs); | |
| else | |
| puts((const char*)&langs); | |
| } | |
| const char* String::v() const { | |
| const uint LIMIT_VIEW=20; | |
| char* buf=(char*)malloc(MAX_STRING); | |
| const char*body_view=body.v(); | |
| const char*langs_view=langs.v(); | |
| snprintf(buf, MAX_STRING, | |
| "%d:%.*s%s} " | |
| "{%d:%s", | |
| langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"", | |
| strlen(body_view), body_view | |
| ); | |
| return buf; | |
| } | |
| void String::dump() const { | |
| body.dump(); | |
| langs.dump(); | |
| } | |
| const String& String::trim(String::Trim_kind kind, const char* chars) const { | |
| if(!length()) | |
| return *this; | |
| size_t substr_begin, substr_length; | |
| Body new_body=body.trim(kind, chars, &substr_begin, &substr_length); | |
| 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; | |
| } | } |