|
|
| version 1.172.2.16, 2003/02/17 09:56:15 | version 1.193, 2004/01/30 09:56:49 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser: string class. @see untasize_t.C. | Parser: string class. @see untalength_t.C. |
| Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) | Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) |
| Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) | Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) |
| */ | */ |
| static const char* IDENT_STRING_C="$Date$"; | static const char * const IDENT_STRING_C="$Date$"; |
| #include "pcre.h" | #include "pcre.h" |
| Line 15 static const char* IDENT_STRING_C="$Date | Line 15 static const char* IDENT_STRING_C="$Date |
| #include "pa_dictionary.h" | #include "pa_dictionary.h" |
| #include "pa_charset.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(*str)) | |
| str++; | |
| if(!*str) | |
| return 0; | |
| int result; | |
| 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 | |
| result=(int)strtol(str, &error_pos, 0); | |
| while(char c=*error_pos++) | |
| if(!isspace(c)) | |
| throw Exception("number.format", | |
| problem_source, | |
| problem_source?"invalid number (int)": "'%s' is invalid number (int)", str); | |
| return result; | |
| } | |
| double pa_atod(const char* str, const String* problem_source) { | |
| if(!str) | |
| return 0; | |
| while(*str && isspace(*str)) | |
| str++; | |
| if(!*str) | |
| return 0; | |
| double result; | |
| 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", | |
| problem_source, | |
| problem_source?"invalid number (double)": "'%s' is invalid number (double)", str); | |
| return result; | |
| } | |
| // 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; | |
| } | |
| // helpers | // helpers |
| /// String::match uses this as replace & global search table columns | /// String::match uses this as replace & global search table columns |
| Line 22 static const char* IDENT_STRING_C="$Date | Line 121 static const char* IDENT_STRING_C="$Date |
| const int MAX_MATCH_GROUPS=100; | const int MAX_MATCH_GROUPS=100; |
| class String_match_table_template_columns: public ArrayString { | class String_match_table_template_columns: public ArrayString { |
| Pool pool; | |
| public: | public: |
| String_match_table_template_columns() { | String_match_table_template_columns() { |
| *this+=StringPtr(new String("prematch")); | *this+=new String("prematch"); |
| *this+=StringPtr(new String("match")); | *this+=new String("match"); |
| *this+=StringPtr(new String("postmatch")); | *this+=new String("postmatch"); |
| for(int i=0; i<MAX_MATCH_GROUPS; i++) { | for(int i=0; i<MAX_MATCH_GROUPS; i++) { |
| char *cname=new(pool) char[3/*strlen("100")*/+1/*terminating 0*/]; | *this+=new String(String::Body::Format(1+i), String::L_CLEAN); |
| StringPtr sname(new String(cname, sprintf(cname, "%d", 1+i))); | |
| *this+=sname; | |
| } | } |
| } | } |
| }; | }; |
| Table string_match_table_template( | Table string_match_table_template(new String_match_table_template_columns); |
| Exception::undefined_source, | |
| Table::columns_type(new String_match_table_template_columns)); | |
| // methods | |
| String::String(const char* src, size_t src_size, bool tainted): Array<String_fragment>(1), fsize(0) { | |
| if(src) | |
| if(tainted) | |
| APPEND_TAINTED(src, src_size, 0, 0); | |
| else | |
| APPEND_CLEAN(src, src_size, 0, 0); | |
| } | |
| String::String(const String& src): Array<String_fragment>(src.count()) { | |
| append(src, UL_PASS_APPENDED); | |
| } | |
| String& String::real_append(STRING_APPEND_PARAMS) { | |
| if(!src) | |
| return *this; | |
| if(!size) | |
| size=strlen(src); | |
| if(!size) | |
| return *this; | |
| if(is_full()) | // String::Body methods |
| expand(fdelta); | |
| String_fragment *fragment=&felements[fused++]; | |
| fragment->ptr=src; | |
| fragment->size=size; | |
| fragment->lang=lang; | |
| #ifndef NO_STRING_ORIGIN | |
| fragment->origin.file=file; | |
| fragment->origin.line=line; | |
| #endif | |
| fsize+=size; | String::Body String::Body::Format(int value) { |
| char local[MAX_NUMBER]; | |
| return *this; | size_t length=snprintf(local, MAX_NUMBER, "%d", value); |
| return String::Body(pa_strdup(local, length), length); | |
| } | } |
| uint String::hash_code() const { | 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::hash_code() const { | |
| uint result=0; | uint result=0; |
| STRING_FOREACH_FRAGMENT( | CORD_iter5(body, 0, |
| result=generic_hash_code(result, fragment->ptr, fragment->size); | CORD_batched_iter_fn_generic_hash_code, |
| ); | CORD_batched_iter_fn_generic_hash_code, &result); |
| return result; | return result; |
| } | } |
| /// @todo move 'lang' skipping to pos | // String methods |
| int String::cmp(int& partial, const String& src, | |
| size_t this_offset, Untaint_lang lang) const { | |
| partial=-1; | |
| size_t a_size=size(); | |
| this_offset=min(this_offset, a_size-1); | |
| const String_fragment *a_current=felements; | |
| const String_fragment *b_current=src.felements; | |
| size_t a_offset=this_offset; | |
| size_t b_offset=0; | |
| String_fragment *a_end=felements+fused; | |
| String_fragment *b_end=src.felements+src.fused; | |
| int result; | |
| size_t pos=0; | |
| bool a_break; | String::String(const char* cstr, size_t helper_length, bool tainted): body(CORD_EMPTY) { |
| bool b_break; | append_help_length(cstr, helper_length, tainted?L_TAINTED:L_CLEAN); |
| while(!( | } |
| (a_break=(a_current==a_end)) | String::String(const String::C cstr, bool tainted): body(CORD_EMPTY) { |
| || (b_break=(b_current==b_end)))) { | append_know_length(cstr.str, cstr.length, tainted?L_TAINTED:L_CLEAN); |
| if(pos+a_current->size > this_offset) { | |
| if(lang!=UL_UNSPECIFIED && a_current->lang>(String_UL)lang) | |
| return -1; // wrong lang -- bail out | |
| int size_diff= | |
| (a_current->size-a_offset)- | |
| (b_current->size-b_offset); | |
| if(size_diff==0) { // a has same size as b | |
| result=memcmp(a_current->ptr+a_offset, b_current->ptr+b_offset, | |
| a_current->size-a_offset); | |
| if(result) | |
| return result; | |
| pos+=a_current->size; | |
| a_current++; a_offset=0; | |
| b_current++; b_offset=0; | |
| } else if(size_diff>0) { // a longer | |
| result=memcmp(a_current->ptr+a_offset, b_current->ptr+b_offset, | |
| b_current->size-b_offset); | |
| if(result) | |
| return result; | |
| a_offset+=b_current->size-b_offset; | |
| b_current++; b_offset=0; | |
| } else { // b longer | |
| result=memcmp(a_current->ptr+a_offset, b_current->ptr+b_offset, | |
| a_current->size-a_offset); | |
| if(result) | |
| return result; | |
| b_offset+=a_current->size-a_offset; | |
| pos+=a_current->size; | |
| a_current++; a_offset=0; | |
| } | |
| } else { | |
| a_offset-=a_current->size; | |
| pos+=a_current->size; | |
| a_current++; | |
| } | |
| } | |
| 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 { | |
| partial=2; return +1; | |
| } | |
| } | } |
| /// @todo move 'lang' skipping to pos | String& String::append_know_length(const char* str, size_t known_length, Language lang) { |
| int String::cmp(int& partial, const char* b_ptr, size_t src_size, | if(!known_length) |
| size_t this_offset, Untaint_lang lang) const { | return *this; |
| partial=-1; | |
| size_t a_size=size(); | // first: langs |
| size_t b_size=src_size?src_size:b_ptr?strlen(b_ptr):0; | langs.append(body, lang, known_length); |
| this_offset=min(this_offset, a_size-1); | // next: letters themselves |
| body.append_know_length(str, known_length); | |
| const String_fragment *a_current=felements; | |
| size_t a_offset=this_offset; | ASSERT_STRING_INVARIANT(*this); |
| size_t b_offset=0; | return *this; |
| String_fragment *a_end=felements+fused; | |
| size_t pos=0; | |
| bool a_break; | |
| bool b_break=b_size==0; | |
| while(!( | |
| (a_break=(a_current==a_end)) | |
| || b_break)) { | |
| if(pos+a_current->size > this_offset) { | |
| if(lang!=UL_UNSPECIFIED && a_current->lang>(String_UL)lang) | |
| return -1; // wrong lang -- bail out | |
| int size_diff= | |
| (a_current->size-a_offset)- | |
| (b_size-b_offset); | |
| if(size_diff==0) { // a has same size as b | |
| if(int result=memcmp(a_current->ptr+a_offset, b_ptr+b_offset, | |
| a_current->size-a_offset)!=0) | |
| return result; | |
| pos+=a_current->size; | |
| a_current++; a_offset=0; | |
| b_break=true; | |
| } else if (size_diff>0) { // a longer | |
| if(int result=memcmp(a_current->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_current->ptr+a_offset, b_ptr+b_offset, | |
| a_current->size-a_offset)!=0) | |
| return result; | |
| b_offset+=a_current->size-a_offset; | |
| pos+=a_current->size; | |
| a_current++; a_offset=0; | |
| } | |
| } else { | |
| a_offset-=a_current->size; | |
| pos+=a_current->size; | |
| a_current++; | |
| } | |
| } | |
| 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 { | |
| partial=2; return +1; | |
| } | |
| } | } |
| 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; | |
| #ifndef NO_STRING_ORIGIN | return append_know_length(str, known_length, lang); |
| const String_fragment::Origin& String::origin() const { | |
| if(is_empty()) { | |
| static const String_fragment::Origin empty_origin={"empty string"}; | |
| return empty_origin; | |
| } | |
| // 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 | |
| String_fragment::Origin& first_origin=felements[0].origin; | |
| return first_origin.file ? first_origin : felements[fused-1].origin; | |
| } | } |
| #endif | 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); | |
| StringPtr String::mid(size_t start, size_t finish) const { | ASSERT_STRING_INVARIANT(*this); |
| StringPtr result(new String()); | return *this; |
| } | |
| start=min(start, size()); | /// @todo check in doc: whether it documents NOW bad situation "abc".mid(-1, 3) =were?="ab" |
| finish=max(start, 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_FRAGMENT( | result.langs.append(result.body, langs, substr_begin, substr_length); |
| size_t item_finish=pos+fragment->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? | // SAPI::log("piece of '%s' from %d to %d is '%s'", |
| size_t offset=started?start-pos:0; | //cstr(), substr_begin, substr_end, result.cstr()); |
| size_t size=finished?finish-pos:fragment->size; | ASSERT_STRING_INVARIANT(result); |
| result->APPEND( | |
| fragment->ptr+offset, size-offset, | |
| fragment->lang, | |
| fragment->origin.file, fragment->origin.line); | |
| if(finished) | |
| goto break2; | |
| } | |
| pos+=fragment->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, | 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, 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; | |
| } | } |
| return -1; | |
| } | } |
| int String::pos(const char* substr, size_t substr_size, | size_t String::pos(const String& substr, |
| int result, Untaint_lang lang) const { | size_t this_offset, Language lang) const { |
| size_t self_size=size(); | return pos(substr.body, this_offset, lang); |
| for(; size_t(result)<self_size; result++) { | |
| int partial; cmp(partial, substr, substr_size, result, lang); | |
| if( | |
| partial==0 || // full match | |
| partial==2) // 'substr' starts 'this'+'result' | |
| return result; | |
| } | |
| return -1; | |
| } | } |
| void String::split(ArrayString& 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) { | Language lang, int limit) const { |
| size_t self_size=size(); | size_t self_length=length(); |
| if(delim_size) { | if(size_t delim_length=strlen(delim)) { |
| size_t pos_after=pos_after_ref?*pos_after_ref:0; | size_t pos_before; |
| int pos_before; | |
| // while we have 'delim'... | // while we have 'delim'... |
| for(; (pos_before=pos(delim, delim_size, pos_after, lang))>=0 && limit; limit--) { | 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+=StringPtr(this); | result+=this; |
| if(pos_after_ref) | pos_after+=self_length; |
| *pos_after_ref+=self_size; | |
| } | } |
| } | } |
| void String::split(ArrayString& 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) { | int limit) const { |
| 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+=StringPtr(this); | result+=this; |
| if(pos_after_ref) | pos_after+=length(); |
| *pos_after_ref+=size(); | |
| } | } |
| } | } |
| static void regex_options(StringPtr options, int *result, bool& need_pre_post_match){ | static void regex_options(const String* options, int *result, bool& need_pre_post_match){ |
| struct Regex_option { | struct Regex_option { |
| const char* keyL; | const char* keyL; |
| const char* keyU; | const char* keyU; |
| Line 356 static void regex_options(StringPtr opti | Line 299 static void regex_options(StringPtr opti |
| int *result; | int *result; |
| bool *flag; | bool *flag; |
| } regex_option[]={ | } regex_option[]={ |
| {"i", "I", 0, PCRE_CASELESS, result}, // a=A | {"i", "I", 0, PCRE_CASELESS, result, 0}, // a=A |
| {"s", "S", 0, PCRE_DOTALL, result}, // \n\n$ [default] | {"s", "S", 0, PCRE_DOTALL, result, 0}, // \n\n$ [default] |
| {"x", "U", 0, PCRE_EXTENDED, result}, // whitespace in regex ignored | {"x", "U", 0, PCRE_EXTENDED, result, 0}, // whitespace in regex ignored |
| {"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$ | {"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result, 0}, // ^aaa\n$^bbb\n$ |
| {"g", "G", 0, true, result+1}, // many rows | {"g", "G", 0, 1, result+1, 0}, // many rows |
| {"'", 0, 0, 0, 0, &need_pre_post_match}, | {"'", 0, 0, 0, 0, &need_pre_post_match}, |
| {0} | {0, 0, 0, 0, 0, 0} |
| }; | }; |
| result[0]=PCRE_EXTRA | PCRE_DOTALL | PCRE_DOLLAR_ENDONLY; | result[0]=PCRE_EXTRA | PCRE_DOTALL | PCRE_DOLLAR_ENDONLY; |
| result[1]=0; | result[1]=0; |
| if(options) | if(options && !options->is_empty()) |
| for(Regex_option *o=regex_option; o->keyL; o++) | for(Regex_option *o=regex_option; o->keyL; o++) |
| if(options->pos(o->keyL)>=0 | if(options->pos(o->keyL)!=STRING_NOT_FOUND |
| || (o->keyU && options->pos(o->keyU)>=0)) { | || (o->keyU && options->pos(o->keyU)!=STRING_NOT_FOUND)) { |
| if(o->flag) | if(o->flag) |
| *o->flag=true; | *o->flag=true; |
| else { // result | else { // result |
| Line 380 static void regex_options(StringPtr opti | Line 323 static void regex_options(StringPtr opti |
| } | } |
| } | } |
| TablePtr String::match(Charset& source_charset, | Table* String::match(Charset& source_charset, |
| StringPtr aorigin, | const String& regexp, |
| const String& regexp, | const String* options, |
| StringPtr options, | Row_action row_action, void *info, |
| Row_action row_action, void *info, | bool& just_matched) const { |
| bool *was_global) const { | |
| if(regexp.is_empty()) | if(regexp.is_empty()) |
| throw Exception(0, | throw Exception(0, |
| aorigin, | 0, |
| "regexp is empty"); | "regexp is empty"); |
| CharPtr pattern=regexp.cstr(); | const char* pattern=regexp.cstr(); |
| const char* errptr; | const char* errptr; |
| int erroffset; | int erroffset; |
| bool need_pre_post_match=false; | bool need_pre_post_match=false; |
| int option_bits[2]; regex_options(options, option_bits, need_pre_post_match); | int option_bits[2]={0}; regex_options(options, option_bits, need_pre_post_match); |
| if(was_global) | bool global=option_bits[1]!=0; |
| *was_global=option_bits[1]!=0; | pcre *code=pcre_compile(pattern, option_bits[0], |
| pcre *code=pcre_compile(pattern.get(), option_bits[0], | |
| &errptr, &erroffset, | &errptr, &erroffset, |
| source_charset.pcre_tables); | source_charset.pcre_tables); |
| if(!code) | if(!code) |
| throw Exception(0, | throw Exception(0, |
| regexp.mid(erroffset, regexp.size()), | ®exp.mid(erroffset, regexp.length()), |
| "regular expression syntax error - %s", errptr); | "regular expression syntax error - %s", errptr); |
| int info_substrings=pcre_info(code, 0, 0); | int subpatterns=pcre_info(code, 0, 0); |
| if(info_substrings<0) { | if(subpatterns<0) { |
| pcre_free(code); | pcre_free(code); |
| throw Exception(0, | throw Exception(0, |
| aorigin, | ®exp, |
| "pcre_info error (%d)", | "pcre_info error (%d)", |
| info_substrings); | subpatterns); |
| } | } |
| CharPtr subject=cstr(); | const char* subject=cstr(); |
| int length=size(); | size_t subject_length=strlen(subject); |
| const int ovecsize=(1/*match*/+MAX_MATCH_GROUPS)*3; | const int oveclength=(1/*match*/+MAX_MATCH_GROUPS)*3; |
| int ovector[ovecsize]; | int ovector[oveclength]; |
| // create table | // create table |
| TablePtr result(new Table(string_match_table_template)); | Table::Action_options table_options; |
| Table& table=*new Table(string_match_table_template, table_options); | |
| int exec_option_bits=0; | 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_substrings=pcre_exec(code, 0, |
| subject.get(), length, prestart, | subject, subject_length, prestart, |
| exec_option_bits, ovector, ovecsize); | exec_option_bits, ovector, oveclength); |
| if(exec_substrings==PCRE_ERROR_NOMATCH) { | if(exec_substrings==PCRE_ERROR_NOMATCH) { |
| pcre_free(code); | pcre_free(code); |
| row_action(result, ArrayStringPtr(0)/*last time, no raw*/, 0, 0, poststart, postfinish, info); | row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info); |
| if(option_bits[1]) | if(global || subpatterns) |
| return result; // global=true+result | return &table; // global or with subpatterns=true+result |
| else | else { |
| return TablePtr(0);// not global=no result | just_matched=false; return 0; // not global=no result |
| } | |
| } | } |
| if(exec_substrings<0) { | if(exec_substrings<0) { |
| pcre_free(code); | pcre_free(code); |
| throw Exception(0, | throw Exception(0, |
| aorigin, | ®exp, |
| "regular expression execute error (%d)", | "regular expression execute error (%d)", |
| exec_substrings); | exec_substrings); |
| } | } |
| int prefinish=ovector[0]; | int prefinish=ovector[0]; |
| poststart=ovector[1]; | poststart=ovector[1]; |
| object_ptr<ArrayString> row(new ArrayString); | ArrayString* row=new ArrayString; |
| if(need_pre_post_match) { | if(need_pre_post_match) { |
| *row+=mid(0, prefinish); // .prematch column value | *row+=&mid(0, prefinish); // .prematch column value |
| *row+=mid(prefinish, poststart); // .match | *row+=&mid(prefinish, poststart); // .match |
| *row+=mid(poststart, postfinish); // .postmatch | *row+=&mid(poststart, postfinish); // .postmatch |
| } else { | } else { |
| *row+=StringPtr(0); // .prematch column value | *row+=&Empty; // .prematch column value |
| *row+=StringPtr(0); // .match | *row+=&Empty; // .match |
| *row+=StringPtr(0); // .postmatch | *row+=&Empty; // .postmatch |
| } | } |
| for(int i=1; i<exec_substrings; i++) { | for(int i=1; i<exec_substrings; 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+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value |
| } | } |
| row_action(result, row, prestart, prefinish, poststart, postfinish, info); | row_action(table, row, prestart, prefinish, poststart, postfinish, info); |
| if(!option_bits[1] || prestart==poststart) { // not global | going to hang | if(!global || prestart==poststart) { // not global | going to hang |
| pcre_free(code); | pcre_free(code); |
| row_action(result, ArrayStringPtr(0)/*last time, no row*/, 0, 0, poststart, postfinish, info); | row_action(table, 0/*last time, no row*/, 0, 0, poststart, postfinish, info); |
| return result; | return &table; |
| } | } |
| prestart=poststart; | prestart=poststart; |
| Line 485 TablePtr String::match(Charset& source_c | Line 427 TablePtr String::match(Charset& source_c |
| } | } |
| } | } |
| StringPtr String::change_case(Pool& pool, Charset& source_charset, Change_case_kind kind) { | String& String::change_case(Charset& source_charset, Change_case_kind kind) const { |
| StringPtr result(new String()); | String& result=*new String(); |
| if(is_empty()) | |
| return result; | |
| const unsigned char *tables=source_charset.pcre_tables; | 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; | |
| const unsigned char *a; | const unsigned char *a; |
| const unsigned char *b; | const unsigned char *b; |
| switch(kind) { | switch(kind) { |
| case CC_UPPER: | case CC_UPPER: |
| a=tables+lcc_offset; | a=tables+lcc_offset; |
| b=tables+fcc_offset; | b=tables+fcc_offset; |
| break; | break; |
| case CC_LOWER: | case CC_LOWER: |
| a=tables+lcc_offset; | a=tables+lcc_offset; |
| b=0; | b=0; |
| break; | break; |
| default: | default: |
| throw Exception(0, | assert(!"unknown change case kind"); |
| StringPtr(this), | a=b=0; // calm, compiler |
| "unknown change case kind #%d", | break; // never |
| static_cast<int>(kind)); // never | } |
| a=b=0; // calm, compiler | |
| break; // never | |
| } | |
| STRING_FOREACH_FRAGMENT( | |
| char *new_cstr=new(pool) char[fragment->size]; | |
| char *dest=new_cstr; | char *dest=new_cstr; |
| const char* end=fragment->ptr+fragment->size; | unsigned char index; |
| for(const char* current=fragment->ptr; current<end; current++) { | for(const char* current=new_cstr; (index=(unsigned char)*current); current++) { |
| unsigned char c=a[(unsigned char)*current]; | 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, fragment->size, | result.langs=langs; |
| fragment->lang, | result.body=new_cstr; |
| fragment->origin.file, fragment->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::replace(const Dictionary& dict) const { |
| StringPtr String::replace(Pool& pool, Dictionary& dict) const { | String& result=*new String(); |
| StringPtr result(new String()); | const char* old_cstr=cstr(); |
| CharPtr lcstr(cstr()); | const char* prematch_begin=old_cstr; |
| const char* current=lcstr.get(); | |
| const char* current=old_cstr; | |
| STRING_FOREACH_FRAGMENT( | while(*current) { |
| IFNDEF_NO_STRING_ORIGIN( | if(Dictionary::Subst subst=dict.first_that_begins(current)) { |
| const char* joined_origin_file=fragment->origin.file; | // prematch |
| const size_t joined_origin_line=fragment->origin.line; | if(size_t prematch_length=current-prematch_begin) { |
| ); | result.langs.append(result.body, langs, prematch_begin-old_cstr, prematch_length); |
| uchar joined_lang=fragment->lang; | result.body.append_strdup_know_length(prematch_begin, prematch_length); |
| const char* joined_ptr=current; | |
| // calc size | |
| size_t joined_size=0; | |
| STRING_PREPARED_FOREACH_FRAGMENT( | |
| if(fragment->lang==joined_lang) | |
| joined_size+=fragment->size; | |
| else | |
| break; // before non-ours | |
| ); | |
| current+=joined_size; | |
| // pointers are after joined piece | |
| // & one step back, see STRING_PREPARED_FOREACH_FRAGMENT | |
| --fragment; | |
| char *new_cstr=new(pool) char[(size_t)ceil(joined_size*dict.max_ratio())]; | |
| char *dest=new_cstr; | |
| while(joined_size) { | |
| // there is a fragment where first column starts 'joined_ptr' | |
| if(Table::element_type row=dict.first_that_starts(joined_ptr, joined_size)) { | |
| // get a=>b values | |
| StringPtr a=row->get(0); | |
| StringPtr b=row->get(1); | |
| // skip 'a' in 'joined_ptr' && reduce work size | |
| joined_ptr+=a->size(); joined_size-=a->size(); | |
| // write 'b' to 'dest' && skip 'b' in 'dest' | |
| b->store_to(dest, String::UL_AS_IS); dest+=b->size(); | |
| } else { | |
| // write a char to b && reduce work size | |
| *dest++=*joined_ptr++; joined_size--; | |
| } | } |
| } | |
| result->APPEND(new_cstr, dest-new_cstr, joined_lang, | // match |
| joined_origin_file, joined_origin_line); | // skip 'a' in 'current'; move prematch_begin |
| ); | current+=subst.from_length; prematch_begin=current; |
| return result; | if(const String* b=subst.to) // are there any b? |
| } | result<<*b; |
| } else // simply advance | |
| current++; | |
| } | |
| StringPtr String::join_chains(Pool& pool) const { | // postmatch |
| StringPtr result(new String()); | if(size_t postmatch_length=current-prematch_begin) { |
| result.langs.append(result.body, langs, prematch_begin-old_cstr, postmatch_length); | |
| char *pooled_buf=new(pool) char[cstr_bufsize()]; | result.body.append_strdup_know_length(prematch_begin, postmatch_length); |
| store_to(pooled_buf, String::UL_AS_IS); | } |
| const char* current=pooled_buf; | |
| STRING_FOREACH_FRAGMENT( | |
| IFNDEF_NO_STRING_ORIGIN( | |
| const char* joined_origin_file=fragment->origin.file; | |
| const size_t joined_origin_line=fragment->origin.line; | |
| ); | |
| uchar joined_lang=fragment->lang; | |
| const char* joined_ptr=current; | |
| // calc size | |
| size_t joined_size=0; | |
| STRING_PREPARED_FOREACH_FRAGMENT( | |
| if(fragment->lang==joined_lang) | |
| joined_size+=fragment->size; | |
| else | |
| break; // before non-ours | |
| ); | |
| current+=joined_size; | |
| // pointers are after joined piece | |
| // & one step back, see STRING_PREPARED_FOREACH_FRAGMENT | |
| --fragment; | |
| result->APPEND(joined_ptr, joined_size, joined_lang, | |
| joined_origin_file, joined_origin_line); | |
| ); | |
| ASSERT_STRING_INVARIANT(result); | |
| return result; | return result; |
| } | } |
| double String::as_double() { | static int serialize_body_char(char c, char** cur) { |
| double result; | *((*cur)++)=c; |
| char buf[MAX_STRING]; | return 0; // 0=continue |
| if(size()>MAX_STRING-1) | }; |
| throw Exception("number.format", | static int serialize_body_piece(const char* s, char** cur) { |
| StringPtr(this), | size_t length=strlen(s); |
| "invalid number too long a string (%u>%u)", size(), MAX_STRING-1); | memcpy(*cur, s, length); *cur+=length; |
| char *eol=store_to(buf, String::UL_AS_IS); *eol=0; | return 0; // 0=continue |
| const char* cstr=buf; | }; |
| static int serialize_lang_piece(char alang, size_t asize, char** cur) { | |
| while(*cstr && isspace(*cstr)) | // lang |
| cstr++; | **cur=alang; (*cur)++; |
| if(!*cstr) | // length [WARNING: not cast, addresses must be %4=0 on sparc] |
| return 0; | memcpy(*cur, &asize, sizeof(asize)); *cur+=sizeof(asize); |
| char *error_pos; | return 0; // 0=continue |
| // 0xABC | } |
| if(cstr[0]=='0') | String::Cm String::serialize(size_t prolog_length) const { |
| if(cstr[1]=='x' || cstr[1]=='X') | size_t fragments_count=langs.count(); |
| result=(double)(unsigned long)strtol(cstr, &error_pos, 0); | size_t buf_length= |
| else | prolog_length //1 |
| result=(double)strtod(cstr+1/*skip leading 0*/, &error_pos); | +sizeof(size_t) //2 |
| else | +fragments_count*(sizeof(char)+sizeof(size_t)) //3 |
| result=(double)strtod(cstr, &error_pos); | +body.length() //4 |
| +1; // for zero terminator used in deserialize | |
| while(char c=*error_pos++) | String::Cm result(new(PointerFreeGC) char[buf_length], buf_length); |
| if(!isspace(c)) | |
| throw Exception("number.format", | // 1: prolog |
| StringPtr(this), | char *cur=result.str+prolog_length; |
| "invalid number (double)"); | // 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; | return result; |
| } | } |
| int String::as_int() { | bool String::deserialize(size_t prolog_length, void *buf, size_t buf_length) { |
| int result; | if(buf_length<=prolog_length) |
| char buf[MAX_STRING]; | return false; |
| if(size()>MAX_STRING-1) | buf_length-=prolog_length; |
| throw Exception("number.format", | buf_length-=1; // 5: zero terminator |
| StringPtr(this), | |
| "invalid number too long a string (%u>%u)", size(), MAX_STRING-1); | |
| char *eol=store_to(buf, String::UL_AS_IS); *eol=0; | |
| const char* cstr=buf; | |
| while(*cstr && isspace(*cstr)) | |
| cstr++; | |
| if(!*cstr) | |
| return 0; | |
| char *error_pos; | |
| // 0xABC | |
| if(cstr[0]=='0') | |
| if(cstr[1]=='x' || cstr[1]=='X') | |
| result=(int)(unsigned long)strtol(cstr, &error_pos, 0); | |
| 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++) | |
| if(!isspace(c)) | |
| throw Exception("number.format", | |
| StringPtr(this), | |
| "invalid number (int)"); | |
| return result; | // 1: prolog |
| } | const char* cur=(const char* )buf+prolog_length; |
| inline void uint2uchars(uint word, uchar *bytes) { | // 2: langs.count |
| bytes[0]=word&0xFF; | size_t fragments_count; |
| bytes[1]=(word>>8)&0xFF; | if(buf_length<sizeof(fragments_count)) // langs.count don't fit? |
| bytes[2]=(word>>16)&0xFF; | |
| bytes[3]=(word>>24)&0xFF; | |
| } | |
| inline uint uchars2uint(uchar *bytes) { | |
| return bytes[3]<<24 | |
| | bytes[2]<<16 | |
| | bytes[1]<<8 | |
| | bytes[0]; | |
| } | |
| void String::serialize(Pool& pool, size_t prolog_size, char *& buf, size_t& buf_size) const { | |
| buf_size= | |
| prolog_size | |
| +fused*(sizeof(uchar)+sizeof(size_t)) | |
| +size(); | |
| buf=new(pool) char[buf_size]; | |
| char *cur=buf+prolog_size; | |
| STRING_FOREACH_FRAGMENT( | |
| // lang | |
| memcpy(cur, &fragment->lang, sizeof(fragment->lang)); | |
| cur+=sizeof(fragment->lang); | |
| // size | |
| // bug on some sparc platform [you can't work with integers on odd pointers] | |
| // forces us to use byte array instead | |
| uchar bytes[4]; | |
| uint2uchars(fragment->size, bytes); | |
| memcpy(cur, &bytes, sizeof(bytes)); cur+=sizeof(bytes); | |
| // bytes | |
| memcpy(cur, fragment->ptr, fragment->size); | |
| cur+=fragment->size; | |
| ); | |
| } | |
| bool String::deserialize(size_t prolog_size, void *buf, size_t buf_size, const char* file) { | |
| if(buf_size<=prolog_size) | |
| 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; | |
| // 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); | |
| // uchar needed to prevent propagating 0x80 bit to upper bytes | |
| langs.append(total_length, (String::Language)(uchar)lang, fragment_length); | |
| total_length+=fragment_length; | |
| char *cur=(char *)buf+prolog_size; | buf_length-=piece_length; |
| buf_size-=prolog_size; | } |
| while(buf_size) { | // 4: letters |
| if(sizeof(uchar)+sizeof(size_t)>buf_size) // lang+size | if(buf_length!=total_length) |
| return false; | return false; |
| String_UL lang=*(String_UL *)(cur); | // serialize wrote extra zero byte there, we can rely on that |
| size_t size=uchars2uint((uchar *)cur); | body=String::Body(cur, buf_length); |
| } | |
| size_t piece_size=sizeof(uchar)+sizeof(size_t)+size; | ASSERT_STRING_INVARIANT(*this); |
| if(piece_size>buf_size) // buffer overrun, can be on incomplete cache files | return true; |
| return false; | } |
| const char* ptr=(const char*)(cur+sizeof(uchar)+sizeof(size_t)); | const char* String::Body::v() const { |
| APPEND(ptr, size, lang, file, 0); | 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 | |
| ); | |
| cur+=piece_size; | return buf; |
| buf_size-=piece_size; | #undef LIMIT_VIEW |
| } | |
| return true; | |
| } | } |