suckless
/
sbase
mirror of git://git.suckless.org/sbase
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
sbase/tr.c

304 lines
6.2 KiB
C
Raw Normal View History

Add license remark to tr.c
2015-01-20 15:26:08 +00:00
/* See LICENSE file for copyright and license details. */
Add Unicode character class support Thinking about it long enough, the solution seems almost trivial.
2015-01-11 19:26:20 +00:00
#include <wctype.h>
Add the tr program including man page
2013-11-15 16:25:10 +00:00
#include <stdio.h>
#include <stdlib.h>
Fix coding style It was about damn time. Consistency is very important in such a big codebase.
2014-11-13 17:29:30 +00:00
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
#include "utf.h"
Add the tr program including man page
2013-11-15 16:25:10 +00:00
#include "util.h"
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
static int cflag = 0;
static int dflag = 0;
static int sflag = 0;
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
struct range {
Rune start;
Rune end;
size_t quant;
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
};
Staticise some symbols in tr(1)
2015-01-10 14:21:09 +00:00
static struct {
Add Unicode character class support Thinking about it long enough, the solution seems almost trivial.
2015-01-11 19:26:20 +00:00
char *name;
int (*check)(wint_t);
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
} classes[] = {
Add Unicode character class support Thinking about it long enough, the solution seems almost trivial.
2015-01-11 19:26:20 +00:00
{ "alnum", iswalnum },
{ "alpha", iswalpha },
{ "blank", iswblank },
{ "cntrl", iswcntrl },
{ "digit", iswdigit },
{ "graph", iswgraph },
{ "lower", iswlower },
Add proper casts and fix a small error
2015-01-11 19:29:27 +00:00
{ "print", iswprint },
Add Unicode character class support Thinking about it long enough, the solution seems almost trivial.
2015-01-11 19:26:20 +00:00
{ "punct", iswpunct },
{ "space", iswspace },
{ "upper", iswupper },
{ "xdigit", iswxdigit },
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
};
Add Unicode character class support Thinking about it long enough, the solution seems almost trivial.
2015-01-11 19:26:20 +00:00
static struct range *set1 = NULL;
static size_t set1ranges = 0;
static int (*set1check)(wint_t) = NULL;
static struct range *set2 = NULL;
static size_t set2ranges = 0;
static int (*set2check)(wint_t) = NULL;
Add the tr program including man page
2013-11-15 16:25:10 +00:00
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
static size_t
rangelen(struct range r)
Add the tr program including man page
2013-11-15 16:25:10 +00:00
{
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
return (r.end - r.start + 1) * r.quant;
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
}
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
static size_t
setlen(struct range *set, size_t setranges)
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
{
tr: fix signed/unsigned warnings
2015-01-10 16:38:28 +00:00
size_t len = 0, i;
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
for (i = 0; i < setranges; i++)
len += rangelen(set[i]);
Wrap mbtowc to check for errors
2014-04-12 18:50:51 +00:00
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
return len;
Wrap mbtowc to check for errors
2014-04-12 18:50:51 +00:00
}
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
static int
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
rstrmatch(Rune *r, char *s, size_t n)
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
{
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
size_t i;
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
for (i = 0; i < n; i++)
if (r[i] != s[i])
return 0;
return 1;
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
}
Resolve escape characters in tr(1) This is one aspect which I think has blown up the complexity of many tr-implementations around today. Instead of complicating the set-theory-based parser itself (he should still be relying on one rune per char, not multirunes), I added a preprocessor, which basically scans the code for upcoming '\'s, reads what he finds, substitutes the real character onto '\'s index and shifts the entire following array so there are no "holes". What is left to reflect on is what to do with octal sequences. I have a local implementation here, which works fine, but imho, given tr is already so focused on UTF-8, we might as well ignore POSIX at this point and rather implement the unicode UTF-8 code points, which are way more contemporary and future-proof. Reading in \uC3A4 as a an array of 0xC3 and 0xA4 is not the issue, but I'm still struggling to find a way to turn it into a well-formed byte sequence. Hit me with a mail if you have a simple solution for that.
2015-01-15 10:51:58 +00:00
static size_t
resolveescapes(Rune *r, size_t len)
{
Add octal sequence support to tr(1)
2015-01-24 21:43:46 +00:00
size_t i, off, m, factor, q;
Resolve escape characters in tr(1) This is one aspect which I think has blown up the complexity of many tr-implementations around today. Instead of complicating the set-theory-based parser itself (he should still be relying on one rune per char, not multirunes), I added a preprocessor, which basically scans the code for upcoming '\'s, reads what he finds, substitutes the real character onto '\'s index and shifts the entire following array so there are no "holes". What is left to reflect on is what to do with octal sequences. I have a local implementation here, which works fine, but imho, given tr is already so focused on UTF-8, we might as well ignore POSIX at this point and rather implement the unicode UTF-8 code points, which are way more contemporary and future-proof. Reading in \uC3A4 as a an array of 0xC3 and 0xA4 is not the issue, but I'm still struggling to find a way to turn it into a well-formed byte sequence. Hit me with a mail if you have a simple solution for that.
2015-01-15 10:51:58 +00:00
Fix segmentation fault in tr(1) and make the parser stricter.
2015-01-24 22:00:34 +00:00
for (i = 0; i < len; i++) {
Resolve escape characters in tr(1) This is one aspect which I think has blown up the complexity of many tr-implementations around today. Instead of complicating the set-theory-based parser itself (he should still be relying on one rune per char, not multirunes), I added a preprocessor, which basically scans the code for upcoming '\'s, reads what he finds, substitutes the real character onto '\'s index and shifts the entire following array so there are no "holes". What is left to reflect on is what to do with octal sequences. I have a local implementation here, which works fine, but imho, given tr is already so focused on UTF-8, we might as well ignore POSIX at this point and rather implement the unicode UTF-8 code points, which are way more contemporary and future-proof. Reading in \uC3A4 as a an array of 0xC3 and 0xA4 is not the issue, but I'm still struggling to find a way to turn it into a well-formed byte sequence. Hit me with a mail if you have a simple solution for that.
2015-01-15 10:51:58 +00:00
if (r[i] != '\\')
continue;
off = 0;
switch (r[i + 1]) {
case '\\': r[i] = '\\'; off++; break;
case 'a': r[i] = '\a'; off++; break;
case 'b': r[i] = '\b'; off++; break;
case 'f': r[i] = '\f'; off++; break;
case 'n': r[i] = '\n'; off++; break;
case 'r': r[i] = '\r'; off++; break;
case 't': r[i] = '\t'; off++; break;
case 'v': r[i] = '\v'; off++; break;
Add octal sequence support to tr(1)
2015-01-24 21:43:46 +00:00
case '\0':
eprintf("tr: null escape sequence\n");
default:
/* "\O[OO]" octal escape */
for (m = i + 1; m < i + 1 + 3 && m < len; m++)
if (r[m] < '0' || r[m] > '7')
break;
if (m == i + 1)
Fix segmentation fault in tr(1) and make the parser stricter.
2015-01-24 22:00:34 +00:00
eprintf("tr: invalid escape sequence '\\%c'\n", r[i + 1]);
Add octal sequence support to tr(1)
2015-01-24 21:43:46 +00:00
off += m - i - 1;
for (--m, q = 0, factor = 1; m > i; m--) {
q += (r[m] - '0') * factor;
factor *= 8;
}
r[i] = q;
Resolve escape characters in tr(1) This is one aspect which I think has blown up the complexity of many tr-implementations around today. Instead of complicating the set-theory-based parser itself (he should still be relying on one rune per char, not multirunes), I added a preprocessor, which basically scans the code for upcoming '\'s, reads what he finds, substitutes the real character onto '\'s index and shifts the entire following array so there are no "holes". What is left to reflect on is what to do with octal sequences. I have a local implementation here, which works fine, but imho, given tr is already so focused on UTF-8, we might as well ignore POSIX at this point and rather implement the unicode UTF-8 code points, which are way more contemporary and future-proof. Reading in \uC3A4 as a an array of 0xC3 and 0xA4 is not the issue, but I'm still struggling to find a way to turn it into a well-formed byte sequence. Hit me with a mail if you have a simple solution for that.
2015-01-15 10:51:58 +00:00
}
for (m = i + 1; m <= len - off; m++)
r[m] = r[m + off];
len -= off;
}
return len;
}
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
static size_t
Add Unicode character class support Thinking about it long enough, the solution seems almost trivial.
2015-01-11 19:26:20 +00:00
makeset(char *str, struct range **set, int (**check)(wint_t))
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
{
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
Rune *rstr;
size_t len, i, j, m, n;
Add Unicode character class support Thinking about it long enough, the solution seems almost trivial.
2015-01-11 19:26:20 +00:00
size_t q, setranges = 0;
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
int factor, base;
/* rstr defines at most len ranges */
len = chartorunearr(str, &rstr);
Resolve escape characters in tr(1) This is one aspect which I think has blown up the complexity of many tr-implementations around today. Instead of complicating the set-theory-based parser itself (he should still be relying on one rune per char, not multirunes), I added a preprocessor, which basically scans the code for upcoming '\'s, reads what he finds, substitutes the real character onto '\'s index and shifts the entire following array so there are no "holes". What is left to reflect on is what to do with octal sequences. I have a local implementation here, which works fine, but imho, given tr is already so focused on UTF-8, we might as well ignore POSIX at this point and rather implement the unicode UTF-8 code points, which are way more contemporary and future-proof. Reading in \uC3A4 as a an array of 0xC3 and 0xA4 is not the issue, but I'm still struggling to find a way to turn it into a well-formed byte sequence. Hit me with a mail if you have a simple solution for that.
2015-01-15 10:51:58 +00:00
len = resolveescapes(rstr, len);
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
*set = emalloc(len * sizeof(**set));
for (i = 0; i < len; i++) {
if (rstr[i] == '[') {
j = i;
nextbrack:
if (j == len)
goto literal;
for (m = j; m < len; m++)
if (rstr[m] == ']') {
j = m;
break;
}
if (j == i)
goto literal;
/* CLASSES [=EQUIV=] (skip) */
if (j - i > 3 && rstr[i + 1] == '=' && rstr[m - 1] == '=') {
i = j;
continue;
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
}
Add the tr program including man page
2013-11-15 16:25:10 +00:00
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
/* CLASSES [:CLASS:] */
if (j - i > 3 && rstr[i + 1] == ':' && rstr[m - 1] == ':') {
for (n = 0; n < LEN(classes); n++) {
if (rstrmatch(rstr + i + 2, classes[n].name, j - i - 3)) {
Add Unicode character class support Thinking about it long enough, the solution seems almost trivial.
2015-01-11 19:26:20 +00:00
*check = classes[n].check;
return 0;
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
}
}
Fix mapping a class to a simple set and improve error-reporting It's standard behaviour to map a whole class of matched objects to the last element of a given simple set2 instead of just passing it through. Also, error out more strictly when the user gives us bogus sets.
2015-01-11 23:03:48 +00:00
eprintf("Invalid character class.\n");
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
}
Add the tr program including man page
2013-11-15 16:25:10 +00:00
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
/* REPEAT [_*n] (only allowed in set2) */
if (j - i > 2 && rstr[i + 2] == '*' && set1ranges > 0) {
/* check if right side of '*' is a number */
q = 0;
factor = 1;
base = (rstr[i + 3] == '0') ? 8 : 10;
for (n = j - 1; n > i + 2; n--) {
if (rstr[n] < '0' && rstr[n] > '9') {
n = 0;
break;
}
q += (rstr[n] - '0') * factor;
factor *= base;
}
if (n == 0) {
j = m + 1;
goto nextbrack;
}
(*set)[setranges].start = rstr[i + 1];
(*set)[setranges].end = rstr[i + 1];
(*set)[setranges].quant = q ? q : setlen(set1, set1ranges);
setranges++;
i = j;
continue;
}
Add the tr program including man page
2013-11-15 16:25:10 +00:00
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
j = m + 1;
goto nextbrack;
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
}
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
literal:
/* RANGES [_-__-_], _-__-_ */
/* LITERALS _______ */
(*set)[setranges].start = rstr[i];
if (i < len - 2 && rstr[i + 1] == '-' && rstr[i + 2] >= rstr[i])
i += 2;
(*set)[setranges].end = rstr[i];
(*set)[setranges].quant = 1;
setranges++;
Add the tr program including man page
2013-11-15 16:25:10 +00:00
}
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
free(rstr);
return setranges;
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
}
static void
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
usage(void)
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
{
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
eprintf("usage: %s [-cCds] set1 [set2]\n", argv0);
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
}
Add the tr program including man page
2013-11-15 16:25:10 +00:00
int
main(int argc, char *argv[])
{
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
Rune r = 0, lastrune = 0;
Prevail order
2015-01-10 19:55:37 +00:00
size_t off1, off2, i, m;
Add the tr program including man page
2013-11-15 16:25:10 +00:00
ARGBEGIN {
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
case 'c':
case 'C':
cflag = 1;
break;
tr: add dflag, error with usage() on invalid flag combination Signed-off-by: Hiltjo Posthuma <hiltjo@codemadness.org>
2014-04-09 12:12:34 +00:00
case 'd':
dflag = 1;
break;
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
case 's':
sflag = 1;
Improved tr - Added support for character ranges ( a-z ) - Added support for complementary charset ( -c ), only in delete mode - Added support for octal escape sequences - Unicode now only works when there are no octal escape sequences, otherwise behavior is not predictable at first sight. - tr now supports null characters in the input - Does not yet have support for character classes ( [:upper:] )
2014-07-14 22:49:42 +00:00
break;
Add the tr program including man page
2013-11-15 16:25:10 +00:00
default:
usage();
} ARGEND;
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
if (argc < 1 || argc > 2 || (argc == 1 && dflag == sflag))
Add the tr program including man page
2013-11-15 16:25:10 +00:00
usage();
Add Unicode character class support Thinking about it long enough, the solution seems almost trivial.
2015-01-11 19:26:20 +00:00
set1ranges = makeset(argv[0], &set1, &set1check);
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
if (argc == 2)
Add Unicode character class support Thinking about it long enough, the solution seems almost trivial.
2015-01-11 19:26:20 +00:00
set2ranges = makeset(argv[1], &set2, &set2check);
if (dflag == sflag && !set2ranges && !set2check)
Fix mapping a class to a simple set and improve error-reporting It's standard behaviour to map a whole class of matched objects to the last element of a given simple set2 instead of just passing it through. Also, error out more strictly when the user gives us bogus sets.
2015-01-11 23:03:48 +00:00
eprintf("set2 must be non-empty.\n");
if (set2check && set2check != iswlower && set2check != iswupper)
eprintf("set2 can only be the 'lower' or 'upper' class.\n");
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
read:
if (!readrune("<stdin>", stdin, &r))
return 0;
off1 = off2 = 0;
for (i = 0; i < set1ranges; i++) {
if (set1[i].start <= r && r <= set1[i].end) {
if (dflag && !cflag)
goto read;
if (sflag) {
if (r == lastrune)
goto read;
else
goto write;
}
for (m = 0; m < i; m++)
off1 += rangelen(set1[m]);
off1 += r - set1[m].start;
if (off1 > setlen(set2, set2ranges) - 1) {
r = set2[set2ranges - 1].end;
goto write;
}
for (m = 0; m < set2ranges; m++) {
if (off2 + rangelen(set2[m]) > off1) {
m++;
break;
}
off2 += rangelen(set2[m]);
}
m--;
r = set2[m].start + (off1 - off2) / set2[m].quant;
Add the tr program including man page
2013-11-15 16:25:10 +00:00
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
goto write;
}
Add the tr program including man page
2013-11-15 16:25:10 +00:00
}
Add proper casts and fix a small error
2015-01-11 19:29:27 +00:00
if (set1check && set1check((wint_t)r)) {
Add Unicode character class support Thinking about it long enough, the solution seems almost trivial.
2015-01-11 19:26:20 +00:00
if (dflag && !cflag)
goto read;
if (sflag) {
if (r == lastrune)
goto read;
else
goto write;
}
if (set1check == iswupper && set2check == iswlower)
Add proper casts and fix a small error
2015-01-11 19:29:27 +00:00
r = towlower((wint_t)r);
Fix mapping a class to a simple set and improve error-reporting It's standard behaviour to map a whole class of matched objects to the last element of a given simple set2 instead of just passing it through. Also, error out more strictly when the user gives us bogus sets.
2015-01-11 23:03:48 +00:00
else if (set1check == iswlower && set2check == iswupper)
Add proper casts and fix a small error
2015-01-11 19:29:27 +00:00
r = towupper((wint_t)r);
Fix mapping a class to a simple set and improve error-reporting It's standard behaviour to map a whole class of matched objects to the last element of a given simple set2 instead of just passing it through. Also, error out more strictly when the user gives us bogus sets.
2015-01-11 23:03:48 +00:00
else if (set2ranges > 0)
r = set2[set2ranges - 1].end;
else
eprintf("Misaligned character classes.\n");
Add Unicode character class support Thinking about it long enough, the solution seems almost trivial.
2015-01-11 19:26:20 +00:00
}
Rewrite tr(1) in a sane way tr(1) always used to be a saddening part of sbase, which was inherently broken and crufted. But to be fair, the POSIX-standard doesn't make it very simple. Given the current version was unfixable and broken by design, I sat down and rewrote tr(1) very close to the concept of set theory and the POSIX-standard with a few exceptions: - UTF-8: not allowed in POSIX, but in my opinion a must. This finally allows you to work with UTF-8 streams without problems or unexpected behaviour. - Equivalence classes: Left out, even GNU coreutils ignore them and depending on LC_COLLATE, which sucks. - Character classes: No experiments or environment-variable-trickery. Just plain definitions derived from the POSIX- standard, working as expected. I tested this thoroughly, but expect problems to show up in some way given the wide range of input this program has to handle. The only thing left on the TODO is to add support for literal expressions ('\n', '\t', '\001', ...) and probably rethinking the way [_*n] is unnecessarily restricted to string2.
2015-01-09 19:36:27 +00:00
if (dflag && cflag)
goto read;
if (dflag && sflag && r == lastrune)
goto read;
write:
lastrune = r;
writerune("<stdout>", stdout, &r);
goto read;
Add the tr program including man page
2013-11-15 16:25:10 +00:00
}