/usr/include/kannel/gwlib/octstr.h is in kannel-dev 1.4.3-2+b2.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 | /* ====================================================================
* The Kannel Software License, Version 1.0
*
* Copyright (c) 2001-2009 Kannel Group
* Copyright (c) 1998-2001 WapIT Ltd.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
*
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the
* distribution.
*
* 3. The end-user documentation included with the redistribution,
* if any, must include the following acknowledgment:
* "This product includes software developed by the
* Kannel Group (http://www.kannel.org/)."
* Alternately, this acknowledgment may appear in the software itself,
* if and wherever such third-party acknowledgments normally appear.
*
* 4. The names "Kannel" and "Kannel Group" must not be used to
* endorse or promote products derived from this software without
* prior written permission. For written permission, please
* contact org@kannel.org.
*
* 5. Products derived from this software may not be called "Kannel",
* nor may "Kannel" appear in their name, without prior written
* permission of the Kannel Group.
*
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESSED OR IMPLIED
* WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
* DISCLAIMED. IN NO EVENT SHALL THE KANNEL GROUP OR ITS CONTRIBUTORS
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY,
* OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
* OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
* ====================================================================
*
* This software consists of voluntary contributions made by many
* individuals on behalf of the Kannel Group. For more information on
* the Kannel Group, please see <http://www.kannel.org/>.
*
* Portions of this software are based upon software originally written at
* WapIT Ltd., Helsinki, Finland for the Kannel project.
*/
/*
* octstr.h - Octet strings
*
* This header file declares an abstract data type, Octstr, for storing
* and manipulating octet strings: strings of arbitrary binary data in
* 8-bit bytes. Unlike C strings, they can contain the NUL byte ('\0').
* Conceptually, they consist of a sequence of octets (bytes) and the
* length of the sequence. There are various basic operations on octet
* strings: concatenating, comparing, printing, etc.
*
* Octet strings come in two flavors: mutable and immutable. Mutable
* octet strings are the normal kind and they can be modified and
* otherwise manipulated at will. Immutable octet strings are meant to
* be wrappers around a C string literal. They may not be modified, though
* they may be destroyed.
*
* Immutable octet strings are meant to simplify usage of octet strings
* together with C strings by reducing the number of octstr_* functions.
* For example, we need a function for searching one string within another.
* There needs to be different flavors of this: exact search, case-insensitive
* search, and a search limited to the first N octets of the strings.
* If in each of these one of the arguments may be either an octet string
* or a C string, the number of functions doubles. Thus, we use immutable
* strings instead:
*
* octstr_search(os, octstr_imm("foo"), 0)
*
* The above looks like a memory leak, but it is not. Each immutable
* octet string (i.e., with the same C string literal pointer) is really
* created only the first time, and octstr_destroy won't destroy it,
* either. The immutable octet strings are destroyed automatically when
* the process ends.
*
* See comments below for explanations on individual functions. Note that
* all functions use gw_malloc and friends, so they won't return if the
* memory allocations fail. Octet string functions are thread safe, as
* long as they only one thread at a time operates on each octet string.
*/
#ifndef OCTSTR_H
#define OCTSTR_H
#include <stdio.h>
#include <stdarg.h>
#include "list.h"
typedef struct Octstr Octstr;
/*
* Initialize the Octstr subsystem.
*/
void octstr_init(void);
/*
* Shut down the Octstr subsystem.
*/
void octstr_shutdown(void);
/*
* Create an octet string from a NUL-terminated C string. Return pointer to
* the new object.
*/
Octstr *octstr_create_real(const char *cstr, const char *file, long line,
const char *func);
#define octstr_create(cstr) \
(Octstr*)gw_claim_area(octstr_create_real((cstr), __FILE__, __LINE__, __func__))
/*
* Create an octet string from arbitrary binary data. The length of the
* data is given, so it can contain NUL characters.
*/
Octstr *octstr_create_from_data_real(const char *data, long len, const char *file,
long line, const char *func);
#define octstr_create_from_data(data, len) \
(Octstr*)gw_claim_area(octstr_create_from_data_real((data), (len), __FILE__, __LINE__, __func__))
#define octstr_create_from_data_trace(data, len, file, line, func) \
(Octstr*)gw_claim_area(octstr_create_from_data_real(data, len, file, line, func))
/*
* Create an immutable octet string from a C string literal. The
* C string literal MUST NOT be modified and it MUST exist until the
* octet string is destroyed. The immutable octet string need not be
* destroyed - it is destroyed automatically when octstr_shutdown is
* called. In fact, octstr_destroy is a no-op for immutables.
*/
Octstr *octstr_imm(const char *cstr);
/*
* Destroy an octet string, freeing all memory it uses. A NULL argument
* is ignored.
*/
void octstr_destroy(Octstr *ostr);
/*
* Destroy an octet string. Wrapper around octstr_destroy that is callable
* via gwlist_destroy.
*/
void octstr_destroy_item(void *os);
/*
* Return the length of (number of octets in) an object string.
*/
long octstr_len(const Octstr *ostr);
/*
* Create a new octet string by copying part of an existing one. Return
* pointer to the new object. If `from' is after end of `ostr', an empty
* octet string is created. If `from+len' is after the end of `ostr',
* `len' is reduced appropriately.
*/
Octstr *octstr_copy_real(const Octstr *ostr, long from, long len, const char *file,
long line, const char *func);
#define octstr_copy(ostr, from, len) \
gw_claim_area(octstr_copy_real((ostr), (from), (len), __FILE__, __LINE__, __func__))
/*
* Copy all of an octet string.
*/
Octstr *octstr_duplicate_real(const Octstr *ostr, const char *file, long line,
const char *func);
#define octstr_duplicate(ostr) \
gw_claim_area(octstr_duplicate_real((ostr), __FILE__, __LINE__, __func__))
/*
* Create a new octet string by catenating two existing ones. Return
* pointer to the new object.
*/
Octstr *octstr_cat(Octstr *ostr1, Octstr *ostr2);
/*
* Return value of octet at a given position in an octet string. The returned
* value has a range of 0..255 for valid positions, and -1 if `pos' is
* after the end of the octet string.
*/
int octstr_get_char(const Octstr *ostr, long pos);
/*
* Replace a single, existing character in an octet string. Operation cannot
* fail: if pos is not inside the string, the operation will silently be
* ignored.
*/
void octstr_set_char(Octstr *ostr, long pos, int ch);
/*
* Copy bytes from octet string into array.
*/
void octstr_get_many_chars(char *buf, Octstr *ostr, long pos, long len);
/*
* Return pointer to contents of octet string as a NUL-terminated C string.
* This is guaranteed to have a NUL character at the end, but it is not
* guaranteed (how could it?) to not contain NUL characters elsewhere.
* The pointer points directly into the internal buffer of the octet
* string, and must not be modified, and must not be used after any
* octstr_* function that modifies the octet string is called after this
* one. It is meant for printing debug messages easily.
*
* If the octet string is empty, an empty C string is returned, not NULL.
*/
char *octstr_get_cstr_real(const Octstr *ostr, const char *file, long line,
const char *func);
#define octstr_get_cstr(ostr) \
(octstr_get_cstr_real(ostr, __FILE__, __LINE__, __func__))
/*
* Append characters from printable hexadecimal format at the tail of
* an octet string. "78797a" or "78797A" would be converted to "xyz"
* and then appended.
*/
void octstr_append_from_hex(Octstr *ostr, char *hex);
/* Convert the octet string in-place to printable hexadecimal format.
* "xyz" would be converted to "78797a". If the uppercase
* flag is set, 'A' through 'F' are used instead of 'a' through 'f'.
*/
void octstr_binary_to_hex(Octstr *ostr, int uppercase);
/* Convert the octet string in-place from printable hexadecimal
* format to binary. "78797a" or "78797A" would be converted to "xyz".
* If the string is not in the expected format, return -1 and leave
* the string unchanged. If all was fine, return 0. */
int octstr_hex_to_binary(Octstr *ostr);
/* Base64-encode the octet string in-place, using the MIME base64
* encoding defined in RFC 2045. Note that the result may be
* multi-line and is always terminated with a CR LF sequence. */
void octstr_binary_to_base64(Octstr *ostr);
/* Base64-decode the octet string in-place, using the MIME base64
* encoding defined in RFC 2045. */
void octstr_base64_to_binary(Octstr *ostr);
/* Parse a number at position 'pos' in 'ostr', using the same rules as
* strtol uses regarding 'base'. Skip leading whitespace.
*
* Return the position of the first character after the number,
* or -1 if there was an error. Return the length of the octet string
* if the number ran to the end of the string.
*
* Assign the number itself to the location pointed to by 'number', if
* there was no error.
*
* Possible errno values in case of an error:
* ERANGE The number did not fit in a long.
* EINVAL No digits of the appropriate base were found.
*/
long octstr_parse_long(long *number, Octstr *ostr, long pos, int base);
/* As above but parses and assigns double number. */
long octstr_parse_double(double *number, Octstr *ostr, long pos);
/* Run the 'filter' function over each character in the specified range.
* Return 1 if the filter returned true for all characters, otherwise 0.
* The octet string is not changed.
* For example: ok = octstr_check_range(o, 1, 10, gw_isdigit);
*/
typedef int (*octstr_func_t)(int);
int octstr_check_range(Octstr *ostr, long pos, long len,
octstr_func_t filter);
/* Run the 'map' function over each character in the specified range,
* replacing each character with the return value of that function.
* For example: octstr_convert_range(o, 1, 10, tolower);
*/
void octstr_convert_range(Octstr *ostr, long pos, long len,
octstr_func_t map);
/*
* Use the octstr_convert_range() with make_printable() to ensure
* every char in the octstr can be printed in the current locale. Each
* character that is NOT printable is converted to a '.' (dot).
*/
void octstr_convert_printable(Octstr *ostr);
/*
* Compare two octet strings, returning 0 if they are equal, negative if
* `ostr1' is less than `ostr2' (when compared octet-value by octet-value),
* and positive if greater.
*/
int octstr_compare(const Octstr *ostr1, const Octstr *ostr2);
/*
* Like octstr_compare, except compares bytes without case sensitivity.
* Note that this probably doesn't work for Unicode, but should work
* for such 8-bit character sets as are supported by libc.
*/
int octstr_case_compare(const Octstr *ostr1, const Octstr *ostr2);
/*
* as above, but comparing is done only up to n bytes
*/
int octstr_ncompare(const Octstr *ostr1, const Octstr *ostr2, long n);
/*
* Same as octstr_compare, but compares the content of the octet string to
* a C string.
*/
int octstr_str_compare(const Octstr *ostr1, const char *str);
/*
* Like octstr_str_compare, except compares bytes without case sensitifity.
*/
int octstr_str_case_compare(const Octstr *ostr1, const char *str);
/*
* Same as octstr_str_compare, but comparing is done only up to n bytes.
*/
int octstr_str_ncompare(const Octstr *ostr, const char *str, long n);
/*
* Write contents of octet string to a file. Return -1 for error, 0 for OK.
*/
int octstr_print(FILE *f, Octstr *ostr);
/*
* Search the character from octet string starting from position pos. Returns
* the position (index) of the char in string, -1 if not found.
*/
int octstr_search_char(const Octstr *ostr, int ch, long pos);
/*
* Search several character from octet string starting from position pos. Returns
* the position (index) of the first char found in string, -1 if none was found.
*/
int octstr_search_chars(const Octstr *ostr, const Octstr *chars, long pos);
/*
* Search for the octet string 'needle' in the octet string 'haystack'.
* Return the start position (index) of 'needle' in 'haystack'.
* Return -1 if not found.
*/
int octstr_search(const Octstr *haystack, const Octstr *needle, long pos);
/*
* Like octstr_search, but ignores 8-bit byte case.
*/
int octstr_case_search(const Octstr *haystack, const Octstr *needle, long pos);
/*
* Like octstr_case_search, but searchs only first n octets.
*/
int octstr_case_nsearch(const Octstr *haystack, const Octstr *needle, long pos, long n);
/*
* Write contents of octet string to a file, in human readable form.
* Return -1 for error, 0 for OK. Octets that are not printable characters
* are printed using C-style escape notation.
*/
int octstr_pretty_print(FILE *f, Octstr *ostr);
/*
* Write contents of octet string to a socket. Return -1 for error, 0 for OK.
*/
int octstr_write_to_socket(int socket, Octstr *ostr);
/*
* Write contents of octet string starting at 'from' to a
* non-blocking file descriptor.
* Return the number of octets written. Return -1 for error.
* It is possible for this function to write only part of the octstr.
*/
long octstr_write_data(Octstr *ostr, int fd, long from);
/*
* Read available data from socket and return it as an octstr.
* Block if no data is available. If a lot of data is available,
* read only up to an internal limit.
* Return -1 for error.
*/
int octstr_append_from_socket(Octstr *ostr, int socket);
/*
* Insert one octet string into another. `pos' gives the position
* in `ostr1' where `ostr2' should be inserted.
*/
void octstr_insert(Octstr *ostr1, const Octstr *ostr2, long pos);
/*
* Insert characters from C array into an octet string. `pos'
* gives the position in `ostr' where `data' should be inserted. `len'
* gives the number of characters in `data'.
* If the given `pos' is greater than the length of the input octet string,
* it is set to that length, resulting in an append.
*/
void octstr_insert_data(Octstr *ostr, long pos, const char *data, long len);
/*
* Similar as previous, expect that now a single character is inserted.
*/
void octstr_insert_char(Octstr *ostr, long pos, const char c);
/*
* Append characters from C array at the tail of an octet string.
*/
void octstr_append_data(Octstr *ostr, const char *data, long len);
/*
* Append a second octstr to the first.
*/
void octstr_append(Octstr *ostr1, const Octstr *ostr2);
/*
* Append a normal C string at the tail of an octet string.
*/
void octstr_append_cstr(Octstr *ostr, const char *cstr);
/*
* Append a single character at the tail of an octet string.
*/
void octstr_append_char(Octstr *ostr, int ch);
/*
* Truncate octet string at `new_len'. If new_len is same or more
* than current, do nothing.
*/
void octstr_truncate(Octstr *ostr, int new_len);
/*
* Strip white space from start and end of a octet string.
*/
void octstr_strip_blanks(Octstr *ostr);
/*
* Strip CR and LF from start and end of a octet string.
*/
void octstr_strip_crlfs(Octstr *ostr);
/*
* Strip non-alphanums from start and end of a octet string.
*/
void octstr_strip_nonalphanums(Octstr *ostr);
/*
* Shrink consecutive white space characters into one space.
*/
void octstr_shrink_blanks(Octstr *ostr);
/*
* Delete part of an octet string.
*/
void octstr_delete(Octstr *ostr1, long pos, long len);
/*
* Read the contents of a named file to an octet string. Return pointer to
* octet string.
*/
Octstr *octstr_read_file(const char *filename);
/*
* Read the contents of a file descriptor pipe to an octet string.
* Return pointer to octet string.
*/
Octstr *octstr_read_pipe(FILE *f);
/*
* Split an octet string into words at whitespace, and return a list
* containing the new octet strings.
*/
List *octstr_split_words(const Octstr *ostr);
/*
* Split an octet string into substrings at every occurence of `sep'.
* Return List with the substrings.
*/
List *octstr_split(const Octstr *os, const Octstr *sep);
/*
* Compare two octet strings in a manner suitable for gwlist_search.
*/
int octstr_item_match(void *item, void *pattern);
/*
* Same as above, except compares bytes without case sensitivity
*/
int octstr_item_case_match(void *item, void *pattern);
/*
* Print debugging information about octet string. This is abstracted to the
* various log levels we have: GW_DEBUG, GW_INFO, GW_WARNING, GW_ERROR
*
* If a third parameter in the argument list is given, we will dump the
* octstr in that log level instead of the default GW_DEBUG level.
*/
void octstr_dump_real(const Octstr *ostr, int level, ...);
#define octstr_dump(ostr, level, ...) \
octstr_dump_real(ostr, level, GW_DEBUG, ##__VA_ARGS__)
/*
* Write the contents of an octet string to the debug log.
* Keep it on one line if the octet string is short and printable,
* otherwise use a hex dump.
*/
void octstr_dump_short(Octstr *ostr, int level, const char *name);
/*
* decode url-encoded octstr in-place.
* Return 0 if all went fine, or -1 if there was some garbage
*/
int octstr_url_decode(Octstr *ostr);
/*
* URL encode the argument string in place.
*/
void octstr_url_encode(Octstr *ostr);
/*
* Treat the octstr as an unsigned array of bits, most significant bit
* first, and return the indicated bit range as an integer. numbits
* must not be larger than 32. Bits beyond the end of the string will
* be read as 0.
*/
long octstr_get_bits(Octstr *ostr, long bitpos, int numbits);
/*
* Treat the octstr as an unsigned array of bits, most significant bit
* first, and set the indicated bit range to the given value. numbits
* must not be larger than 32. The value must fit in that number of bits.
* The string will be extended with 0-valued octets as necessary to hold
* the indicated bit range.
*/
void octstr_set_bits(Octstr *ostr, long bitpos, int numbits,
unsigned long value);
/*
* Encode value in WSP's uintvar format, and append it to the octstr
*/
void octstr_append_uintvar(Octstr *ostr, unsigned long value);
/*
* Decode a value in WSP's uintvar format at position pos of the octstr,
* and put the result in *value. Return the position after the uintvar.
* Return -1 if there is not a valid uintvar at pos.
*/
long octstr_extract_uintvar(Octstr *ostr, unsigned long *value, long pos);
/*
* Append the decimal representation of the given value to ostr
*/
void octstr_append_decimal(Octstr *ostr, long value);
/*
* Create a new octet string based on a printf-like (but not identical)
* format string, and a list of other arguments. The format string is
* a C string for convenience, but this may change later.
*
* The syntax for the format string is as follows:
*
* % [-] [0] [width] [. prec] [type] conversion
*
* where [] denotes optional parts and the various parts have the
* following meanings:
*
* - add padding to the right, instead of the left of the field
*
* 0 pad with zeroes, not spaces
*
* width minimum output width; non-negative integer or '*', indicating
* that the next argument is an int and gives the width
*
* . a dot to indicate that precision follows
*
* prec precision: maximum length of strings, maximum number of
* decimals for floating point numbers; non-negative integer
* or '*' indicating that the next argument is an int and
* gives the precision
*
* type type of integer argument: either h (for short int) or
* l (for long int); may only be used with conversion 'd'
*
* conversion
* how the field is to be converted, also implicitly defines
* the type of the next argument; one of
*
* d int (unless type says otherwise)
* output as a decimal integer
*
* e, f, g double
* output in various formats of floating
* point, see printf(3) for details
*
* s char *
* output as character string
*
* S Octstr *
* output as character string, except '\0'
* inside the string is included in the
* output
*
* E Octstr *
* output as character string, except that
* contents are URL-encoded when need to. Note
* that trunctae is done afterwards and can
* cut escape '%EE' in half
*
* H Octstr *
* output as character string, except that
* contents are HEX-encoded in uppercase
*/
Octstr *octstr_format(const char *fmt, ...);
/*
* Like octstr_format, but takes the argument list as a va_list.
*/
Octstr *octstr_format_valist_real(const char *fmt, va_list args);
#define octstr_format_valist(fmt, args) gw_claim_area(octstr_format_valist_real(fmt, args))
/*
* Like octstr_format, but appends output to an existing octet
* string, instead of creating a new one.
*/
void octstr_format_append(Octstr *os, const char *fmt, ...);
/*
* Compute a hash key value for an octet string by adding all the
* octets together.
*/
unsigned long octstr_hash_key(Octstr *ostr);
/*
* return an Octstr encoded in charset named tocode created from the data
* in the Octstr orig that is encoded in the charset fromcode.
*/
int octstr_recode(Octstr *tocode, Octstr *fromcode, Octstr *orig);
/*
* Strip all occurence of char ch from start of Octstr
*/
void octstr_strip_char(Octstr *text, char ch);
/*
* Check if ostr is numeric
*/
int octstr_isnum(Octstr *ostr1);
/*
* Replace all occurences of needle with repl within haystack
*/
void octstr_replace(Octstr *haystack, Octstr *needle, Octstr *repl);
/*
* Symbolize hex string '78797a' becomes '%78%79%7a'
*/
int octstr_symbolize(Octstr *ostr);
/*
* Remove all occurrences of 'needle' within 'haystack'.
*/
void octstr_delete_matching(Octstr *haystack, Octstr *needle);
/*
* Return 1, if octstr 'os' contains only hex chars, 0 otherwise.
*/
int octstr_is_all_hex(Octstr *os);
/*
* make data HTML safe by converting appropriate characters to HTML entities.
* conversion is done in place
*/
void octstr_convert_to_html_entities(Octstr* input);
/*
* convert HTML safe data back to binary data by replacing HTML entities with their
* respective character values.
* conversion is done in place
*/
void octstr_convert_from_html_entities(Octstr* input);
#endif
|