Add UTF8-support to paste(1) and refactor parallel()

Hopefully it now is clearer what the heck is going on.
This commit is contained in:
FRIGN 2015-01-29 20:15:50 +01:00
parent 7b6d918b78
commit cdbc0d5035
2 changed files with 115 additions and 163 deletions

2
README
View File

@ -49,7 +49,7 @@ The following tools are implemented ('*' == finished, '#' == UTF-8 support,
=* nice yes none =* nice yes none
= nl no -d, -f, -h, -l, -n, -p, -v, -w = nl no -d, -f, -h, -l, -n, -p, -v, -w
=* nohup yes none =* nohup yes none
paste yes none # paste yes none
= printenv non-posix none = printenv non-posix none
printf stolen stolen printf stolen stolen
=* pwd yes none =* pwd yes none

272
paste.c
View File

@ -4,20 +4,116 @@
#include <stdio.h> #include <stdio.h>
#include <string.h> #include <string.h>
#include <unistd.h> #include <unistd.h>
#include <wchar.h>
#include "utf.h"
#include "util.h" #include "util.h"
typedef struct { struct fdescr {
FILE *fp; FILE *fp;
const char *name; const char *name;
} Fdescr; };
static size_t unescape(wchar_t *); static size_t
static wint_t in(Fdescr *); resolveescapes(char *s)
static void out(wchar_t); {
static void sequential(Fdescr *, int, const wchar_t *, size_t); size_t len, i, off, m;
static void parallel(Fdescr *, int, const wchar_t *, size_t);
len = strlen(s);
for (i = 0; i < len; i++) {
if (s[i] != '\\')
continue;
off = 0;
switch (s[i + 1]) {
case '\\': s[i] = '\\'; off++; break;
case 'a': s[i] = '\a'; off++; break;
case 'b': s[i] = '\b'; off++; break;
case 'f': s[i] = '\f'; off++; break;
case 'n': s[i] = '\n'; off++; break;
case 'r': s[i] = '\r'; off++; break;
case 't': s[i] = '\t'; off++; break;
case 'v': s[i] = '\v'; off++; break;
case '\0':
eprintf("paste: null escape sequence in delimiter\n");
default:
eprintf("paste: invalid escape sequence '\\%c' in "
"delimiter\n", s[i + 1]);
}
for (m = i + 1; m <= len - off; m++)
s[m] = s[m + off];
len -= off;
}
return len;
}
static void
sequential(struct fdescr *dsc, int fdescrlen, Rune *delim, size_t delimlen)
{
Rune c, last;
size_t i, d;
for (i = 0; i < fdescrlen; i++) {
d = 0;
last = 0;
while (readrune(dsc[i].name, dsc[i].fp, &c)) {
if (last == '\n') {
if (delim[d] != '\0')
writerune("<stdout>", stdout, &delim[d]);
d = (d + 1) % delimlen;
}
if (c != '\n')
writerune("<stdout>", stdout, &c);
last = c;
}
if (last == '\n')
writerune("<stdout>", stdout, &last);
}
}
static void
parallel(struct fdescr *dsc, int fdescrlen, Rune *delim, size_t delimlen)
{
Rune c, d;
size_t i, m;
ssize_t last;
nextline:
last = -1;
for (i = 0; i < fdescrlen; i++) {
d = delim[i % delimlen];
c = 0;
for (; readrune(dsc[i].name, dsc[i].fp, &c) ;) {
for (m = last + 1; m < i; m++)
writerune("<stdout>", stdout, &(delim[m % delimlen]));
last = i;
if (c == '\n') {
if (i != fdescrlen - 1)
c = d;
writerune("<stdout>", stdout, &c);
break;
}
writerune("<stdout>", stdout, &c);
}
if (c == 0 && last != -1) {
if (i == fdescrlen - 1)
putchar('\n');
else
writerune("<stdout>", stdout, &d);
last++;
}
}
if (last != -1)
goto nextline;
}
static void static void
usage(void) usage(void)
@ -28,14 +124,11 @@ usage(void)
int int
main(int argc, char *argv[]) main(int argc, char *argv[])
{ {
const char *adelim = NULL; struct fdescr *dsc;
Rune *delim;
size_t i, len;
int seq = 0; int seq = 0;
wchar_t *delim = NULL; char *adelim = "\t";
size_t len;
Fdescr *dsc = NULL;
int i;
setlocale(LC_CTYPE, "");
ARGBEGIN { ARGBEGIN {
case 's': case 's':
@ -51,20 +144,9 @@ main(int argc, char *argv[])
if (argc == 0) if (argc == 0)
usage(); usage();
/* populate delimeters */ /* populate delimiters */
if (!adelim) resolveescapes(adelim);
adelim = "\t"; len = chartorunearr(adelim, &delim);
len = mbstowcs(NULL, adelim, 0);
if (len == (size_t) - 1)
eprintf("invalid delimiter\n");
delim = emalloc((len + 1) * sizeof(*delim));
mbstowcs(delim, adelim, len);
len = unescape(delim);
if (len == 0)
eprintf("no delimiters specified\n");
/* populate file list */ /* populate file list */
dsc = emalloc(argc * sizeof(*dsc)); dsc = emalloc(argc * sizeof(*dsc));
@ -76,7 +158,7 @@ main(int argc, char *argv[])
dsc[i].fp = fopen(argv[i], "r"); dsc[i].fp = fopen(argv[i], "r");
if (!dsc[i].fp) if (!dsc[i].fp)
eprintf("can't open '%s':", argv[i]); eprintf("fopen %s:", argv[i]);
dsc[i].name = argv[i]; dsc[i].name = argv[i];
} }
@ -91,135 +173,5 @@ main(int argc, char *argv[])
(void)fclose(dsc[i].fp); (void)fclose(dsc[i].fp);
} }
free(delim);
free(dsc);
return 0; return 0;
} }
static size_t
unescape(wchar_t *delim)
{
wchar_t c;
size_t i;
size_t len;
for (i = 0, len = 0; (c = delim[i++]) != '\0'; len++) {
if (c == '\\') {
switch (delim[i++]) {
case 'n':
delim[len] = '\n';
break;
case 't':
delim[len] = '\t';
break;
case '0':
delim[len] = '\0';
break;
case '\\':
delim[len] = '\\';
break;
case '\0':
default:
/* POSIX: unspecified results */
return len;
}
} else
delim[len] = c;
}
return len;
}
static wint_t
in(Fdescr *f)
{
wint_t c = fgetwc(f->fp);
if (c == WEOF && ferror(f->fp))
eprintf("'%s' read error:", f->name);
return c;
}
static void
out(wchar_t c)
{
putwchar(c);
if (ferror(stdout))
eprintf("write error:");
}
static void
sequential(Fdescr *dsc, int len, const wchar_t *delim, size_t cnt)
{
int i;
size_t d;
wint_t c, last;
for (i = 0; i < len; i++) {
d = 0;
last = WEOF;
while ((c = in(&dsc[i])) != WEOF) {
if (last == '\n') {
if (delim[d] != '\0')
out(delim[d]);
d++;
d %= cnt;
}
if (c != '\n')
out((wchar_t)c);
last = c;
}
if (last == '\n')
out((wchar_t)last);
}
}
static void
parallel(Fdescr *dsc, int len, const wchar_t *delim, size_t cnt)
{
int last, i;
wint_t c, o;
wchar_t d;
do {
last = 0;
for (i = 0; i < len; i++) {
d = delim[i % cnt];
do {
o = in(&dsc[i]);
c = o;
switch (c) {
case WEOF:
if (last == 0)
break;
o = '\n';
/* fallthrough */
case '\n':
if (i != len - 1)
o = d;
break;
default:
break;
}
if (o != WEOF) {
/* pad with delimiters up to this point */
while (++last < i) {
if (d != '\0')
out(d);
}
out((wchar_t)o);
}
} while (c != '\n' && c != WEOF);
}
} while (last > 0);
}