|
|
| version 1.145.4.9, 2003/09/24 12:28:27 | version 1.153, 2003/11/20 16:34:25 |
|---|---|
| Line 8 | Line 8 |
| #ifndef PA_STRING_H | #ifndef PA_STRING_H |
| #define PA_STRING_H | #define PA_STRING_H |
| static const char* IDENT_STRING_H="$Date$"; | static const char * const IDENT_STRING_H="$Date$"; |
| // includes | // includes |
| Line 24 extern "C" { // cord's author forgot to | Line 24 extern "C" { // cord's author forgot to |
| /* Returns true if x does contain */ | /* Returns true if x does contain */ |
| /* char not_c at positions i..i+n. Value i,i+n must be < CORD_len(x). */ | /* char not_c at positions i..i+n. Value i,i+n must be < CORD_len(x). */ |
| int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c); | int CORD_range_contains_chr_greater_then(CORD x, size_t i, size_t n, int c); |
| size_t CORD_block_count(CORD x); | |
| // forwards | // forwards |
| Line 38 typedef Array<const String*> ArrayString | Line 39 typedef Array<const String*> ArrayString |
| /// this is result of pos functions which mean that substr were not found | /// this is result of pos functions which mean that substr were not found |
| #define STRING_NOT_FOUND ((size_t)-1) | #define STRING_NOT_FOUND ((size_t)-1) |
| class StringBody { | |
| CORD body; | |
| public: | |
| StringBody(): body(CORD_EMPTY) {} | |
| StringBody(CORD abody): body(abody) { | |
| assert(!body // no body | |
| || *body // ordinary string | |
| || body[1]==1 // CONCAT_HDR | |
| || body[1]==4 // FN_HDR | |
| || body[1]==6 // SUBSTR_HDR | |
| ); | |
| } | |
| /// WARNING: length is only HELPER length, str in ANY case should be zero-terminated | |
| StringBody(const char* str, size_t helper_length): body(CORD_EMPTY) { | |
| append_know_length(str, helper_length?helper_length:strlen(str)); | |
| } | |
| static StringBody Format(int value); | |
| void clear() { body=CORD_EMPTY; } | |
| bool operator! () const { return is_empty(); } | |
| uint hash_code() const; | |
| const char* cstr() const { return CORD_to_const_char_star(body); } | |
| char* cstrm() const { return CORD_to_char_star(body); } | |
| size_t length() const { return CORD_len(body); } | |
| bool is_empty() const { return body==CORD_EMPTY; } | |
| void append_know_length(const char *str, size_t known_length) { | |
| if(known_length) | |
| body=CORD_cat_char_star(body, str, known_length); | |
| } | |
| void append_strdup_know_length(const char* str, size_t known_length) { | |
| if(known_length) | |
| append_know_length(pa_strdup(str, known_length), known_length); | |
| } | |
| void append(char c) { body=CORD_cat_char(body, c); } | |
| StringBody& operator << (const StringBody src) { body=CORD_cat(body, src.body); return *this; } | |
| StringBody& operator << (const char* str) { append_know_length(str, strlen(str)); return *this; } | |
| // could not figure out why this operator is needed [should do this chain: string->simple->==] | |
| bool operator < (const StringBody src) const { return CORD_cmp(body, src.body)<0; } | |
| bool operator > (const StringBody src) const { return CORD_cmp(body, src.body)>0; } | |
| bool operator <= (const StringBody src) const { return CORD_cmp(body, src.body)<=0; } | |
| bool operator >= (const StringBody src) const { return CORD_cmp(body, src.body)>=0; } | |
| bool operator != (const StringBody src) const { return CORD_cmp(body, src.body)!=0; } | |
| bool operator == (const StringBody src) const { return CORD_cmp(body, src.body)==0; } | |
| int ncmp(size_t x_begin, const StringBody y, size_t y_begin, size_t size) const { | |
| return CORD_ncmp(body, x_begin, y.body, y_begin, size); | |
| } | |
| char fetch(size_t index) const { return CORD_fetch(body, index); } | |
| StringBody mid(size_t index, size_t length) const { return CORD_substr(body, index, length); } | |
| size_t pos(const char* substr, size_t offset=0) const { return CORD_str(body, offset, substr); } | |
| size_t pos(const StringBody substr, size_t offset=0) const { | |
| if(!substr.length()) | |
| return STRING_NOT_FOUND; // in this case CORD_str returns 0 [parser users got used to -1] | |
| return CORD_str(body, offset, substr.body); | |
| } | |
| size_t pos(char c, | |
| size_t offset=0) const { | |
| return CORD_chr(body, offset, c); | |
| } | |
| /* template<typename I> void for_each(int (*callback)(const char* s, I), I info) const { | |
| CORD_iter5(body, 0, 0, (CORD_batched_iter_fn)callback, info); | |
| }*/ | |
| void set_pos(CORD_pos& pos, size_t index) const { CORD_set_pos(pos, body, index); } | |
| /*StringBody normalize() const { | |
| return StringBody(CORD_balance(body)); | |
| }*/ | |
| void dump() const { | |
| CORD_dump(body); | |
| } | |
| }; | |
| template<typename T> | template<typename T> |
| inline size_t get_length(T current) { | inline size_t get_length(T current) { |
| return current; | return current; |
| } | } |
| /** | /** |
| String which knows the lang of all it's langs. | String which knows the lang of all it's langs. |
| Line 140 public: | Line 56 public: |
| /** piece is tainted or not. the lang to use when detaint | /** piece is tainted or not. the lang to use when detaint |
| remember to change String_Untaint_lang_name @ untaint.C along | remember to change String_Untaint_lang_name @ untaint.C along |
| WARNING WARNING WARNING WARNING WARNING WARNING | |
| pos function compares(<=) languages, that is used in searching | |
| for table column separator being L_CLEAN or L_AS_IS. | |
| they search for AS_IS, meaning AS_IS|CLEAN [doing <=L_AS_IS check]. | |
| letters assigned for debugging, but it's important for no language-letter | |
| come before L_AS_IS other then L_CLEAN | |
| WARNING WARNING WARNING WARNING WARNING WARNING | |
| */ | */ |
| enum Language { | enum Language { |
| L_UNSPECIFIED=0, ///< no real string has parts of this lange: it's just convinient to check when string's empty | L_UNSPECIFIED=0, ///< no real string has parts of this lange: it's just convinient to check when string's empty |
| // these two must go before others, there are checks for >L_AS_IS | // these two must go before others, there are checks for >L_AS_IS |
| L_CLEAN='1', ///< clean | L_CLEAN='0', ///< clean WARNING: read above warning before changing |
| L_AS_IS, ///< leave all characters intact | L_AS_IS='A', ///< leave all characters intact WARNING: read above warning before changing |
| L_PASS_APPENDED, | L_PASS_APPENDED='P', |
| /**< | /**< |
| leave lang built into string being appended. | leave lang built into string being appended. |
| just a flag, that value not stored | just a flag, that value not stored |
| */ | */ |
| L_TAINTED, ///< tainted, untaint lang as assigned later | L_TAINTED='T', ///< tainted, untaint lang as assigned later |
| // untaint langs. assigned by ^untaint[lang]{...} | // untaint langs. assigned by ^untaint[lang]{...} |
| L_FILE_SPEC, ///< file specification | L_FILE_SPEC='F', ///< file specification |
| L_HTTP_HEADER, ///< text in HTTP response header | L_HTTP_HEADER='h', ///< text in HTTP response header |
| L_MAIL_HEADER, ///< text in mail header | L_MAIL_HEADER='m', ///< text in mail header |
| L_URI, ///< text in uri | L_URI='U', ///< text in uri |
| L_TABLE, ///< ^table:set body | L_TABLE='L', ///< ^table:set body |
| L_SQL, ///< ^table:sql body | L_SQL='Q', ///< ^table:sql body |
| L_JS, ///< JavaScript code | L_JS='J', ///< JavaScript code |
| L_XML, ///< ^dom:set xml | L_XML='X', ///< ^dom:set xml |
| L_HTML, ///< HTML code (for editing) | L_HTML='H', ///< HTML code (for editing) |
| // READ WARNING ABOVE BEFORE ADDING ANYTHING | |
| L_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization | L_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization |
| }; | }; |
| class Languages { | union Languages { |
| union { | struct { |
| struct { | Language lang:8; |
| Language lang:8; | int is_not_just_lang:sizeof(CORD)*8-8; |
| int is_not_just_lang:sizeof(CORD)*8-8; | } opt; |
| }; | CORD langs; |
| CORD langs; | |
| }; | |
| template<typename C> | template<typename C> |
| CORD make_langs(C current) const { | CORD make_langs(C current) const { |
| return is_not_just_lang? | return opt.is_not_just_lang? |
| langs | langs |
| :CORD_chars((char)lang, get_length(current)); | :CORD_chars((char)opt.lang, get_length(current)); |
| } | } |
| CORD make_langs(size_t aoffset, size_t alength) const { | CORD make_langs(size_t aoffset, size_t alength) const { |
| return is_not_just_lang? | return opt.is_not_just_lang? |
| CORD_substr(langs, aoffset, alength) | CORD_substr(langs, aoffset, alength) |
| :CORD_chars((char)lang, alength); | :CORD_chars((char)opt.lang, alength); |
| } | } |
| /// appending when 'langs' already contain something [simple cases hanled elsewhere] | /// appending when 'langs' already contain something [simple cases hanled elsewhere] |
| Line 195 public: | Line 121 public: |
| const CORD to_nonempty_target_langs) { | const CORD to_nonempty_target_langs) { |
| assert(langs); | assert(langs); |
| if(is_not_just_lang) | if(opt.is_not_just_lang) |
| langs=CORD_cat(langs, to_nonempty_target_langs); | langs=CORD_cat(langs, to_nonempty_target_langs); |
| else { // we were "just lang" | else { // we were "just lang" |
| size_t current_size=get_length(current); | size_t current_size=get_length(current); |
| assert(current_size); | assert(current_size); |
| langs=CORD_cat( | langs=CORD_cat( |
| CORD_chars((char)lang, current_size), // first piece [making from just 'lang'] | CORD_chars((char)opt.lang, current_size), // first piece [making from just 'lang'] |
| to_nonempty_target_langs); // new piece | to_nonempty_target_langs); // new piece |
| } | } |
| } | } |
| public: | public: |
| const char* v() const; | |
| Languages(): langs(0) {} | Languages(): langs(0) {} |
| Languages(Language alang): lang(alang), is_not_just_lang(0) {} | Languages(Language alang) { |
| opt.lang=alang; | |
| opt.is_not_just_lang=0; | |
| } | |
| /// MUST be called exactly prior to modification of current [uses it's length] | /// MUST be called exactly prior to modification of current [uses it's length] |
| template<typename C> | template<typename C> |
| Line 217 public: | Line 148 public: |
| assert(alang); | assert(alang); |
| assert(asize); | assert(asize); |
| if(!is_not_just_lang) | if(!opt.is_not_just_lang) |
| if(lang) { | if(opt.lang) { |
| if(lang==alang) // same length? ignoring | if(opt.lang==alang) // same length? ignoring |
| return; | return; |
| } else { | } else { |
| lang=alang; // to uninitialized | opt.lang=alang; // to uninitialized |
| return; | return; |
| } | } |
| Line 237 public: | Line 168 public: |
| if(!langs) | if(!langs) |
| langs=src.langs; // to uninitialized | langs=src.langs; // to uninitialized |
| else if(!src.is_not_just_lang) | else if(!src.opt.is_not_just_lang) |
| append(current, src.lang, appending_length); // simplifying when simple source | append(current, src.opt.lang, appending_length); // simplifying when simple source |
| else | else |
| append(current, src.make_langs(appending_length)); | append(current, src.make_langs(appending_length)); |
| } | } |
| Line 250 public: | Line 181 public: |
| assert(alength); | assert(alength); |
| if(!langs) // to uninitialized? | if(!langs) // to uninitialized? |
| if(src.is_not_just_lang) | if(src.opt.is_not_just_lang) |
| langs=CORD_substr(src.langs, aoffset, alength); // to uninitialized complex | langs=CORD_substr(src.langs, aoffset, alength); // to uninitialized complex |
| else | else |
| lang=src.lang; // to uninitialized simple | opt.lang=src.opt.lang; // to uninitialized simple |
| else | else |
| if(!is_not_just_lang && !src.is_not_just_lang && lang==src.lang) // both simple & of same language? | if(!opt.is_not_just_lang && !src.opt.is_not_just_lang && opt.lang==src.opt.lang) // both simple & of same language? |
| return; // ignoring | return; // ignoring |
| else | else |
| append(current, src.make_langs(aoffset, alength)); | append(current, src.make_langs(aoffset, alength)); |
| Line 266 public: | Line 197 public: |
| if(alang==L_UNSPECIFIED) // ignore lang? | if(alang==L_UNSPECIFIED) // ignore lang? |
| return true; | return true; |
| if(is_not_just_lang) | if(opt.is_not_just_lang) |
| return CORD_range_contains_chr_greater_then(langs, aoffset, alength, (unsigned)alang)==0; | return CORD_range_contains_chr_greater_then(langs, aoffset, alength, (unsigned)alang)==0; |
| else | else |
| return lang<=alang; | return opt.lang<=alang; |
| } | } |
| /// @returns count of blocks | |
| /// @todo currently there can be adjucent blocks of same language. someday merge them | |
| size_t count() const { | |
| return opt.is_not_just_lang? | |
| CORD_block_count(langs) | |
| : opt.lang? | |
| 1 | |
| : 0; | |
| }; | |
| template<typename C, typename I> | template<typename C, typename I> |
| void for_each(C current, | void for_each(C current, |
| int callback(char, size_t, I), I info) const { | int callback(char, size_t, I), I info) const { |
| if(is_not_just_lang) | if(opt.is_not_just_lang) |
| CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info); | CORD_block_iter(langs, 0, (CORD_block_iter_fn)callback, info); |
| else | else |
| callback(lang, get_length(current), info); | callback(opt.lang, get_length(current), info); |
| } | } |
| bool invariant(size_t current_length) { | bool invariant(size_t current_length) { |
| if(!langs) | if(!langs) |
| return current_length==0; | return current_length==0; |
| if(is_not_just_lang) | if(opt.is_not_just_lang) |
| return CORD_len(langs)==current_length; | return CORD_len(langs)==current_length; |
| return true; // uncheckable, actually | return true; // uncheckable, actually |
| } | } |
| }; | }; |
| class Body { | |
| CORD body; | |
| public: | |
| const char* v() const; | |
| Body(): body(CORD_EMPTY) {} | |
| Body(CORD abody): body(abody) { | |
| assert(!body // no body | |
| || *body // ordinary string | |
| || body[1]==1 // CONCAT_HDR | |
| || body[1]==4 // FN_HDR | |
| || body[1]==6 // SUBSTR_HDR | |
| ); | |
| } | |
| /// WARNING: length is only HELPER length, str in ANY case should be zero-terminated | |
| Body(const char* str, size_t helper_length): body(CORD_EMPTY) { | |
| append_know_length(str, helper_length?helper_length:strlen(str)); | |
| } | |
| static Body Format(int value); | |
| void clear() { body=CORD_EMPTY; } | |
| bool operator! () const { return is_empty(); } | |
| uint hash_code() const; | |
| const char* cstr() const { return CORD_to_const_char_star(body); } | |
| char* cstrm() const { return CORD_to_char_star(body); } | |
| size_t length() const { return CORD_len(body); } | |
| bool is_empty() const { return body==CORD_EMPTY; } | |
| void append_know_length(const char *str, size_t known_length) { | |
| if(known_length) | |
| body=CORD_cat_char_star(body, str, known_length); | |
| } | |
| void append_strdup_know_length(const char* str, size_t known_length) { | |
| if(known_length) | |
| append_know_length(pa_strdup(str, known_length), known_length); | |
| } | |
| void append(char c) { body=CORD_cat_char(body, c); } | |
| Body& operator << (const Body src) { body=CORD_cat(body, src.body); return *this; } | |
| Body& operator << (const char* str) { append_know_length(str, strlen(str)); return *this; } | |
| // could not figure out why this operator is needed [should do this chain: string->simple->==] | |
| bool operator < (const Body src) const { return CORD_cmp(body, src.body)<0; } | |
| bool operator > (const Body src) const { return CORD_cmp(body, src.body)>0; } | |
| bool operator <= (const Body src) const { return CORD_cmp(body, src.body)<=0; } | |
| bool operator >= (const Body src) const { return CORD_cmp(body, src.body)>=0; } | |
| bool operator != (const Body src) const { return CORD_cmp(body, src.body)!=0; } | |
| bool operator == (const Body src) const { return CORD_cmp(body, src.body)==0; } | |
| int ncmp(size_t x_begin, const Body y, size_t y_begin, size_t size) const { | |
| return CORD_ncmp(body, x_begin, y.body, y_begin, size); | |
| } | |
| char fetch(size_t index) const { return CORD_fetch(body, index); } | |
| Body mid(size_t index, size_t length) const { return CORD_substr(body, index, length); } | |
| size_t pos(const char* substr, size_t offset=0) const { return CORD_str(body, offset, substr); } | |
| size_t pos(const Body substr, size_t offset=0) const { | |
| if(!substr.length()) | |
| return STRING_NOT_FOUND; // in this case CORD_str returns 0 [parser users got used to -1] | |
| // CORD_str checks for bad offset [CORD_chr does not] | |
| return CORD_str(body, offset, substr.body); | |
| } | |
| size_t pos(char c, | |
| size_t offset=0) const { | |
| if(offset>=length()) // CORD_chr does not check that [and ABORT's in that case] | |
| return STRING_NOT_FOUND; | |
| return CORD_chr(body, offset, c); | |
| } | |
| template<typename I> void for_each( | |
| int (*f1)(char c, I), | |
| int (*f2)(const char* s, I), | |
| I info) const { | |
| CORD_iter5(body, 0, (CORD_iter_fn)f1, (CORD_batched_iter_fn)f2, info); | |
| } | |
| void set_pos(CORD_pos& pos, size_t index) const { CORD_set_pos(pos, body, index); } | |
| /*Body normalize() const { | |
| return Body(CORD_balance(body)); | |
| }*/ | |
| }; | |
| struct C { | struct C { |
| const char *str; | const char *str; |
| size_t length; | size_t length; |
| Line 307 public: | Line 340 public: |
| private: | private: |
| Body body; ///< all characters of string | |
| Languages langs; ///< string characters lang | Languages langs; ///< string characters lang |
| StringBody body; ///< all characters of string | |
| const char* v() const; | |
| #define ASSERT_STRING_INVARIANT(string) \ | #define ASSERT_STRING_INVARIANT(string) \ |
| assert((string).langs.invariant((string).body.length())) | assert((string).langs.invariant((string).body.length())) |
| public: | public: |
| static const String Empty; | |
| explicit String(const char* cstr=0, size_t helper_length=0, bool tainted=false); | explicit String(const char* cstr=0, size_t helper_length=0, bool tainted=false); |
| explicit String(const C cstr, bool tainted=false); | explicit String(const C cstr, bool tainted=false); |
| String(StringBody abody, Language alang): body(abody), langs(alang) { | String(Body abody, Language alang): body(abody), langs(alang) { |
| assert(!body.is_empty()); | assert(!body.is_empty()); |
| ASSERT_STRING_INVARIANT(*this); | ASSERT_STRING_INVARIANT(*this); |
| } | } |
| Line 326 public: | Line 363 public: |
| } | } |
| /// for convinient hash lookup | /// for convinient hash lookup |
| operator const StringBody() const { return body; } | operator const Body() const { return body; } |
| bool is_empty() const { return body.is_empty(); } | bool is_empty() const { return body.is_empty(); } |
| size_t length() const { return body.length(); } | size_t length() const { return body.length(); } |
| /// convert to CORD. if 'lang' known, forcing 'lang' to it | /// convert to CORD. if 'lang' known, forcing 'lang' to it |
| StringBody cstr_to_string_body(Language lang=L_AS_IS, | Body cstr_to_string_body(Language lang=L_AS_IS, |
| SQL_Connection* connection=0, | SQL_Connection* connection=0, |
| const Request_charsets *charsets=0) const; | const Request_charsets *charsets=0) const; |
| Line 352 public: | Line 389 public: |
| Cm serialize(size_t prolog_size) const; | Cm serialize(size_t prolog_size) const; |
| /// appends pieces from buf to self | /// appends pieces from buf to self |
| bool deserialize(size_t prolog_size, void *buf, size_t buf_size); | bool deserialize(size_t prolog_size, void *buf, size_t buf_size); |
| /// @see StringBody::append_know_length | /// @see Body::append_know_length |
| String& append_know_length(const char* str, size_t known_length, Language lang); | String& append_know_length(const char* str, size_t known_length, Language lang); |
| /// @see StringBody::append_help_length | /// @see Body::append_help_length |
| String& append_help_length(const char* str, size_t helper_length, Language lang); | String& append_help_length(const char* str, size_t helper_length, Language lang); |
| String& append_strdup(const char* str, size_t helper_length, Language lang); | String& append_strdup(const char* str, size_t helper_length, Language lang); |
| bool operator == (const char* y) const { return body==StringBody(y); } | bool operator == (const char* y) const { return body==Body(y); } |
| bool operator != (const char* y) const { return body!=StringBody(y); } | bool operator != (const char* y) const { return body!=Body(y); } |
| /// this starts with y | /// this starts with y |
| bool starts_with(const char* y) const { | bool starts_with(const char* y) const { |
| return body.ncmp(0/*x_begin*/, StringBody(y), 0/*y_begin*/, strlen(y))==0; | return body.ncmp(0/*x_begin*/, Body(y), 0/*y_begin*/, strlen(y))==0; |
| } | } |
| /// x starts with this | /// x starts with this |
| bool this_starts(const char* x) const { | bool this_starts(const char* x) const { |
| return StringBody(x).ncmp(0/*x_begin*/, body, 0/*y_begin*/, length())==0; | return Body(x).ncmp(0/*x_begin*/, body, 0/*y_begin*/, length())==0; |
| } | } |
| String& append_to(String& dest, Language lang, bool forced) const; | String& append_to(String& dest, Language lang, bool forced) const; |
| Line 376 public: | Line 413 public: |
| } | } |
| String& operator << (const String& src) { return append(src, L_PASS_APPENDED); } | String& operator << (const String& src) { return append(src, L_PASS_APPENDED); } |
| String& operator << (const char* src) { return append_help_length(src, 0, L_AS_IS); } | String& operator << (const char* src) { return append_help_length(src, 0, L_AS_IS); } |
| String& operator << (const StringBody src) { | String& operator << (const Body src); |
| langs.append(body, L_AS_IS, src.length()); | |
| body<<src; | |
| return *this; | |
| } | |
| /// extracts first char of a string, if any | /// extracts first char of a string, if any |
| char first_char() const { | char first_char() const { |
| Line 402 public: | Line 435 public: |
| but when specified: look for substring that lies in ONE fragment in THAT lang | but when specified: look for substring that lies in ONE fragment in THAT lang |
| @return position of substr in string, -1 means "not found" [const char* version] | @return position of substr in string, -1 means "not found" [const char* version] |
| */ | */ |
| size_t pos(const StringBody substr, | size_t pos(const Body substr, |
| size_t this_offset=0, Language lang=L_UNSPECIFIED) const; | size_t this_offset=0, Language lang=L_UNSPECIFIED) const; |
| /// String version of @see pos(const char*, int, Language) | /// String version of @see pos(const char*, int, Language) |
| size_t pos(const String& substr, | size_t pos(const String& substr, |
| Line 454 private: //disabled | Line 487 private: //disabled |
| }; | }; |
| template<> | template<> |
| inline size_t get_length<StringBody>(StringBody body) { | inline size_t get_length<String::Body>(String::Body body) { |
| return body.length(); | return body.length(); |
| } | } |
| /// simple hash code of string. used by Hash | /// simple hash code of string. used by Hash |
| inline uint hash_code(const StringBody self) { | inline uint hash_code(const String::Body self) { |
| return self.hash_code(); | return self.hash_code(); |
| } | } |
| /// now that we've declared specialization we can use it | |
| inline String& String::operator << (const String::Body src) { | |
| langs.append(body, L_AS_IS, src.length()); | |
| body<<src; | |
| return *this; | |
| } | |
| #endif | #endif |