lib/strutil.h

   1 #ifndef MC_STRUTIL_H
   2 #define MC_STRUTIL_H
   3
   4 #include "lib/global.h"         /* include glib.h */
   5
   6 #include <sys/types.h>
   7 #include <string.h>
   8 #include <assert.h>             /* assert() */
   9
  10 /* Header file for strutil.c, strutilascii.c, strutil8bit.c, strutilutf8.c.
  11  * There are two sort of functions:
  12  * 1. functions for working with growing strings and conversion strings between
  13  *    different encodings.
  14  *    (implemented directly in strutil.c)
  15  * 2. functions, that hide differences between encodings derived from ASCII.
  16  *    (implemented separately in strutilascii.c, strutil8bit.c, strutilutf8.c)
  17  * documentation is made for UTF-8 version of functions.
  18  */
  19
  20 /* invalid strings
  21  * function, that works with invalid strings are marked with "I"
  22  * in documentation
  23  * invalid bytes of string are handled as one byte characters with width 1, they
  24  * are displayed as questionmarks, I-maked comparing functions try to keep
  25  * the original value of these bytes.
  26  */
  27
  28 /* combining characters
  29  * displaynig: all handled as zero with characters, expect combing character
  30  * at the begin of string, this character has with one (space add before),
  31  * so str_term_width is not good for computing width of singles characters
  32  * (never return zero, expect emtpy string)
  33  * for compatibility are strings composed before displaynig
  34  * comparing: comparing decompose all string before comparing, n-compare
  35  * functions do not work as is usual, because same strings do not have to be
  36  * same length in UTF-8. So they return 0 if one string is prefix of the other
  37  * one.
  38  * str_prefix is used to determine, how many characters from one string are
  39  * prefix in second string. However, str_prefix return number of characters in
  40  * decompose form. (used in do_search (screen.c))
  41  */
  42
  43 /*** typedefs(not structures) and defined constants **********************************************/
  44
  45 #define IS_FIT(x) ((x) & 0x0010)
  46 #define MAKE_FIT(x) ((x) | 0x0010)
  47 #define HIDE_FIT(x) ((x) & 0x000f)
  48
  49 #define INVALID_CONV ((GIConv) (-1))
  50
  51 /*** enums ***************************************************************************************/
  52
  53 /* results of conversion function
  54  */
  55 typedef enum
  56 {
  57     /* success means, that convertion has been finished successully
  58      */
  59     ESTR_SUCCESS = 0,
  60     /* problem means, that not every characters was successfully converted (They are
  61      * replaced with questionmark). So is impossible convert string back.
  62      */
  63     ESTR_PROBLEM = 1,
  64     /* failure means, that conversion is not possible (example: wrong encoding
  65      * of input string)
  66      */
  67     ESTR_FAILURE = 2
  68 } estr_t;
  69
  70 /* alignment strings on terminal
  71  */
  72 typedef enum
  73 {
  74     J_LEFT = 0x01,
  75     J_RIGHT = 0x02,
  76     J_CENTER = 0x03,
  77     /* if there is enough space for string on terminal,
  78      * string is centered otherwise is aligned to left */
  79     J_CENTER_LEFT = 0x04,
  80     /* fit alignment, if string is to long, is truncated with '~' */
  81     J_LEFT_FIT = 0x11,
  82     J_RIGHT_FIT = 0x12,
  83     J_CENTER_FIT = 0x13,
  84     J_CENTER_LEFT_FIT = 0x14
  85 } align_crt_t;
  86
  87 /*** structures declarations (and typedefs of structures)*****************************************/
  88
  89 /* all functions in str_class must be defined for every encoding */
  90 struct str_class
  91 {
  92     gchar *(*conv_gerror_message) (GError * error, const char *def_msg);
  93       /*I*/ estr_t (*vfs_convert_to) (GIConv coder, const char *string, int size, GString * buffer);
  94       /*I*/ void (*insert_replace_char) (GString * buffer);
  95     int (*is_valid_string) (const char *);
  96       /*I*/ int (*is_valid_char) (const char *, size_t);
  97       /*I*/ void (*cnext_char) (const char **);
  98     void (*cprev_char) (const char **);
  99     void (*cnext_char_safe) (const char **);
 100       /*I*/ void (*cprev_char_safe) (const char **);
 101       /*I*/ int (*cnext_noncomb_char) (const char **text);
 102       /*I*/ int (*cprev_noncomb_char) (const char **text, const char *begin);
 103       /*I*/ int (*isspace) (const char *);
 104       /*I*/ int (*ispunct) (const char *);
 105       /*I*/ int (*isalnum) (const char *);
 106       /*I*/ int (*isdigit) (const char *);
 107       /*I*/ int (*isprint) (const char *);
 108       /*I*/ int (*iscombiningmark) (const char *);
 109       /*I*/ int (*length) (const char *);
 110       /*I*/ int (*length2) (const char *, int);
 111       /*I*/ int (*length_noncomb) (const char *);
 112       /*I*/ int (*toupper) (const char *, char **, size_t *);
 113     int (*tolower) (const char *, char **, size_t *);
 114     void (*fix_string) (char *);
 115       /*I*/ const char *(*term_form) (const char *);
 116       /*I*/ const char *(*fit_to_term) (const char *, int, align_crt_t);
 117       /*I*/ const char *(*term_trim) (const char *text, int width);
 118       /*I*/ const char *(*term_substring) (const char *, int, int);
 119       /*I*/ int (*term_width1) (const char *);
 120       /*I*/ int (*term_width2) (const char *, size_t);
 121       /*I*/ int (*term_char_width) (const char *);
 122       /*I*/ const char *(*trunc) (const char *, int);
 123       /*I*/ int (*offset_to_pos) (const char *, size_t);
 124       /*I*/ int (*column_to_pos) (const char *, size_t);
 125       /*I*/ char *(*create_search_needle) (const char *, int);
 126     void (*release_search_needle) (char *, int);
 127     const char *(*search_first) (const char *, const char *, int);
 128     const char *(*search_last) (const char *, const char *, int);
 129     int (*compare) (const char *, const char *);
 130       /*I*/ int (*ncompare) (const char *, const char *);
 131       /*I*/ int (*casecmp) (const char *, const char *);
 132       /*I*/ int (*ncasecmp) (const char *, const char *);
 133       /*I*/ int (*prefix) (const char *, const char *);
 134       /*I*/ int (*caseprefix) (const char *, const char *);
 135       /*I*/ char *(*create_key) (const char *text, int case_sen);
 136       /*I*/ char *(*create_key_for_filename) (const char *text, int case_sen);
 137       /*I*/ int (*key_collate) (const char *t1, const char *t2, int case_sen);
 138       /*I*/ void (*release_key) (char *key, int case_sen);
 139   /*I*/};
 140
 141 /*** global variables defined in .c file *********************************************************/
 142
 143 /* standard convertors */
 144 extern GIConv str_cnv_to_term;
 145 extern GIConv str_cnv_from_term;
 146 /* from terminal encoding to terminal encoding */
 147 extern GIConv str_cnv_not_convert;
 148
 149 /*** declarations of public functions ************************************************************/
 150
 151 struct str_class str_utf8_init (void);
 152 struct str_class str_8bit_init (void);
 153 struct str_class str_ascii_init (void);
 154
 155 /* create convertor from "from_enc" to terminal encoding
 156  * if "from_enc" is not supported return INVALID_CONV
 157  */
 158 GIConv str_crt_conv_from (const char *);
 159
 160 /* create convertor from terminal encoding to "to_enc"
 161  * if "to_enc" is not supported return INVALID_CONV
 162  */
 163 GIConv str_crt_conv_to (const char *);
 164
 165 /* close convertor, do not close str_cnv_to_term, str_cnv_from_term,
 166  * str_cnv_not_convert
 167  */
 168 void str_close_conv (GIConv);
 169
 170 /* return on of not used buffers (.used == 0) or create new
 171  * returned buffer has set .used to 1
 172  */
 173
 174 /* convert string using coder, result of conversion is appended at end of buffer
 175  * return ESTR_SUCCESS if there was no problem.
 176  * otherwise return  ESTR_PROBLEM or ESTR_FAILURE
 177  */
 178 estr_t str_convert (GIConv, const char *, GString *);
 179 estr_t str_nconvert (GIConv, const char *, int, GString *);
 180
 181 /* convert GError message (which in UTF-8) to terminal charset
 182  * def_char is used if result of error->str conversion if ESTR_FAILURE
 183  * return new allocated null-terminated string, which is need to be freed
 184  * I
 185  */
 186 gchar *str_conv_gerror_message (GError * error, const char *def_msg);
 187
 188 /* return only ESTR_SUCCESS or ESTR_FAILURE, because vfs must be able to convert
 189  * result to original string. (so no replace with questionmark)
 190  * if coder is str_cnv_from_term or str_cnv_not_convert, string is only copied,
 191  * so is possible to show file, that is not valid in terminal encoding
 192  */
 193 estr_t str_vfs_convert_from (GIConv, const char *, GString *);
 194
 195 /* if coder is str_cnv_to_term or str_cnv_not_convert, string is only copied,
 196  * does replace with questionmark
 197  * I
 198  */
 199 estr_t str_vfs_convert_to (GIConv, const char *, int, GString *);
 200
 201 /* printf functin for str_buffer, append result of printf at the end of buffer
 202  */
 203 void str_printf (GString *, const char *, ...);
 204
 205 /* add standard replacement character in terminal encoding
 206  */
 207 void str_insert_replace_char (GString *);
 208
 209 /* init strings and set terminal encoding,
 210  * if is termenc NULL, detect terminal encoding
 211  * create all str_cnv_* and set functions for terminal encoding
 212  */
 213 void str_init_strings (const char *termenc);
 214
 215 /* free all str_buffer and all str_cnv_*
 216  */
 217 void str_uninit_strings (void);
 218
 219 /* try convert characters in ch to output using conv
 220  * ch_size is size of ch, can by (size_t)(-1) (-1 only for ASCII
 221  *     compatible encoding, for other must be set)
 222  * return ESTR_SUCCESS if conversion was successfully,
 223  * ESTR_PROBLEM if ch contains only part of characters,
 224  * ESTR_FAILURE if conversion is not possible
 225  */
 226 estr_t str_translate_char (GIConv conv, const char *ch, size_t ch_size,
 227                            char *output, size_t out_size);
 228
 229 /* test, if text is valid in terminal encoding
 230  * I
 231  */
 232 int str_is_valid_string (const char *text);
 233
 234 /* test, if first char of ch is valid
 235  * size, how many bytes characters occupied, could be (size_t)(-1)
 236  * return 1 if it is valid, -1 if it is invalid or -2 if it is only part of
 237  * multibyte character
 238  * I
 239  */
 240 int str_is_valid_char (const char *ch, size_t size);
 241
 242 /* return next characters after text, do not call on the end of string
 243  */
 244 char *str_get_next_char (char *text);
 245 const char *str_cget_next_char (const char *text);
 246
 247 /* return previous characters before text, do not call on the start of strings
 248  */
 249 char *str_get_prev_char (char *text);
 250 const char *str_cget_prev_char (const char *text);
 251
 252 /* set text to next characters, do not call on the end of string
 253  */
 254 void str_next_char (char **text);
 255 void str_cnext_char (const char **text);
 256
 257 /* set text to previous characters, do not call on the start of strings
 258  */
 259 void str_prev_char (char **text);
 260 void str_cprev_char (const char **text);
 261
 262 /* return next characters after text, do not call on the end of string
 263  * works with invalid string
 264  * I
 265  */
 266 char *str_get_next_char_safe (char *text);
 267 const char *str_cget_next_char_safe (const char *text);
 268
 269 /* return previous characters before text, do not call on the start of strings
 270  * works with invalid string
 271  * I
 272  */
 273 char *str_get_prev_char_safe (char *text);
 274 const char *str_cget_prev_char_safe (const char *text);
 275
 276 /* set text to next characters, do not call on the end of string
 277  * works with invalid string
 278  * I
 279  */
 280 void str_next_char_safe (char **text);
 281 void str_cnext_char_safe (const char **text);
 282
 283 /* set text to previous characters, do not call on the start of strings
 284  * works with invalid string
 285  * I
 286  */
 287 void str_prev_char_safe (char **text);
 288 void str_cprev_char_safe (const char **text);
 289
 290 /* set text to next noncombining characters, check the end of text
 291  * return how many characters was skipped
 292  * works with invalid string
 293  * I
 294  */
 295 int str_next_noncomb_char (char **text);
 296 int str_cnext_noncomb_char (const char **text);
 297
 298 /* set text to previous noncombining characters, search stop at begin
 299  * return how many characters was skipped
 300  * works with invalid string
 301  * I
 302  */
 303 int str_prev_noncomb_char (char **text, const char *begin);
 304 int str_cprev_noncomb_char (const char **text, const char *begin);
 305
 306 /* if first characters in ch is space, tabulator  or new lines
 307  * I
 308  */
 309 int str_isspace (const char *ch);
 310
 311 /* if first characters in ch is punctuation or symbol
 312  * I
 313  */
 314 int str_ispunct (const char *ch);
 315
 316 /* if first characters in ch is alphanum
 317  * I
 318  */
 319 int str_isalnum (const char *ch);
 320
 321 /* if first characters in ch is digit
 322  * I
 323  */
 324 int str_isdigit (const char *ch);
 325
 326 /* if first characters in ch is printable
 327  * I
 328  */
 329 int str_isprint (const char *ch);
 330
 331 /* if first characters in ch is a combining mark (only in utf-8)
 332  * combining makrs are assumed to be zero width
 333  * I
 334  */
 335 int str_iscombiningmark (const char *ch);
 336
 337 /* write lower from of fisrt characters in ch into out
 338  * decrase remain by size of returned characters
 339  * if out is not big enough, do nothing
 340  */
 341 int str_toupper (const char *ch, char **out, size_t * remain);
 342
 343 /* write upper from of fisrt characters in ch into out
 344  * decrase remain by size of returned characters
 345  * if out is not big enough, do nothing
 346  */
 347 int str_tolower (const char *ch, char **out, size_t * remain);
 348
 349 /* return length of text in characters
 350  * I
 351  */
 352 int str_length (const char *text);
 353
 354 /* return length of text in characters, limit to size
 355  * I
 356  */
 357 int str_length2 (const char *text, int size);
 358
 359 /* return length of one char
 360  * I
 361  */
 362 int str_length_char (const char *);
 363
 364 /* return length of text in characters, count only noncombining characters
 365  * I
 366  */
 367 int str_length_noncomb (const char *text);
 368
 369 /* replace all invalid characters in text with questionmark
 370  * after return, text is valid string in terminal encoding
 371  * I
 372  */
 373 void str_fix_string (char *text);
 374
 375 /* replace all invalid characters in text with questionmark
 376  * replace all unprintable characters with '.'
 377  * return static allocated string, "text" is not changed
 378  * returned string do not need to be freed
 379  * I
 380  */
 381 const char *str_term_form (const char *text);
 382
 383 /* like str_term_form, but text can be alignment to width
 384  * alignment is specified in just_mode (J_LEFT, J_LEFT_FIT, ...)
 385  * result is completed with spaces to width
 386  * I
 387  */
 388 const char *str_fit_to_term (const char *text, int width, align_crt_t just_mode);
 389
 390 /* like str_term_form, but when text is wider than width, three dots are
 391  * inserted at begin and result is completed with suffix of text
 392  * no additional spaces are inserted
 393  * I
 394  */
 395 const char *str_term_trim (const char *text, int width);
 396
 397
 398 /* like str_term_form, but return only specified substring
 399  * start - column (position) on terminal, where substring begin
 400  * result is completed with spaces to width
 401  * I
 402  */
 403 const char *str_term_substring (const char *text, int start, int width);
 404
 405 /* return width, that will be text occupied on terminal
 406  * I
 407  */
 408 int str_term_width1 (const char *text);
 409
 410 /* return width, that will be text occupied on terminal
 411  * text is limited by length in characters
 412  * I
 413  */
 414 int str_term_width2 (const char *text, size_t length);
 415
 416 /* return width, that will be character occupied on terminal
 417  * combining characters are always zero width
 418  * I
 419  */
 420 int str_term_char_width (const char *text);
 421
 422 /* convert position in characters to position in bytes
 423  * I
 424  */
 425 int str_offset_to_pos (const char *text, size_t length);
 426
 427 /* convert position on terminal to position in characters
 428  * I
 429  */
 430 int str_column_to_pos (const char *text, size_t pos);
 431
 432 /* like str_fit_to_term width just_mode = J_LEFT_FIT,
 433  * but do not insert additional spaces
 434  * I
 435  */
 436 const char *str_trunc (const char *text, int width);
 437
 438 /* create needle, that will be searched in str_search_fist/last,
 439  * so needle can be reused
 440  * in UTF-8 return normalized form of needle
 441  */
 442 char *str_create_search_needle (const char *needle, int case_sen);
 443
 444 /* free needle returned by str_create_search_needle
 445  */
 446 void str_release_search_needle (char *needle, int case_sen);
 447
 448 /* search for first occurrence of search in text
 449  */
 450 const char *str_search_first (const char *text, const char *needle, int case_sen);
 451
 452 /* search for last occurrence of search in text
 453  */
 454 const char *str_search_last (const char *text, const char *needle, int case_sen);
 455
 456 /* case sensitive compare two strings
 457  * I
 458  */
 459 int str_compare (const char *t1, const char *t2);
 460
 461 /* case sensitive compare two strings
 462  * if one string is prefix of the other string, return 0
 463  * I
 464  */
 465 int str_ncompare (const char *t1, const char *t2);
 466
 467 /* case insensitive compare two strings
 468  * I
 469  */
 470 int str_casecmp (const char *t1, const char *t2);
 471
 472 /* case insensitive compare two strings
 473  * if one string is prefix of the other string, return 0
 474  * I
 475  */
 476 int str_ncasecmp (const char *t1, const char *t2);
 477
 478 /* return, how many bytes are are same from start in text and prefix
 479  * both strings are decomposed befor comapring and return value is counted
 480  * in decomposed form, too. caling with prefix, prefix, you get size in bytes
 481  * of prefix in decomposed form,
 482  * I
 483  */
 484 int str_prefix (const char *text, const char *prefix);
 485
 486 /* case insensitive version of str_prefix
 487  * I
 488  */
 489 int str_caseprefix (const char *text, const char *prefix);
 490
 491 /* create a key that is used by str_key_collate
 492  * I
 493  */
 494 char *str_create_key (const char *text, int case_sen);
 495
 496 /* create a key that is used by str_key_collate
 497  * should aware dot '.' in text
 498  * I
 499  */
 500 char *str_create_key_for_filename (const char *text, int case_sen);
 501
 502 /* compare two string using LC_COLLATE, if is possible
 503  * if case_sen is set, comparing is case sensitive,
 504  * case_sen must be same for str_create_key, str_key_collate and str_release_key
 505  * I
 506  */
 507 int str_key_collate (const char *t1, const char *t2, int case_sen);
 508
 509 /* release_key created by str_create_key, only rigth way to release key
 510  * I
 511  */
 512 void str_release_key (char *key, int case_sen);
 513
 514 /* return 1 if codeset_name is utf8 or utf-8
 515  * I
 516  */
 517 int str_isutf8 (const char *codeset_name);
 518
 519 const char *str_detect_termencoding (void);
 520
 521 int str_verscmp (const char *s1, const char *s2);
 522
 523 /* return how many lines and columns will text occupy on terminal
 524  */
 525 void str_msg_term_size (const char *text, int *lines, int *columns);
 526
 527 /*** inline functions ****************************************************************************/
 528
 529 static inline void
 530 str_replace (char *s, char from, char to)
 531 {
 532     for (; *s != '\0'; s++)
 533     {
 534         if (*s == from)
 535             *s = to;
 536     }
 537 }
 538
 539 /*
 540  * strcpy is unsafe on overlapping memory areas, so define memmove-alike
 541  * string function.
 542  * Have sense only when:
 543  *  * dest <= src
 544  *   AND
 545  *  * dest and str are pointers to one object (as Roland Illig pointed).
 546  *
 547  * We can't use str*cpy funs here:
 548  * http://kerneltrap.org/mailarchive/openbsd-misc/2008/5/27/1951294
 549  */
 550 static inline char *
 551 str_move (char *dest, const char *src)
 552 {
 553     size_t n;
 554
 555     assert (dest <= src);
 556
 557     n = strlen (src) + 1;       /* + '\0' */
 558
 559     return (char *) memmove (dest, src, n);
 560 }
 561
 562 #endif /* MC_STRUTIL_H */