1da2e3ebdSchin /***********************************************************************
2da2e3ebdSchin * *
3da2e3ebdSchin * This software is part of the ast package *
4*b30d1939SAndy Fiddaman * Copyright (c) 1992-2012 AT&T Intellectual Property *
5da2e3ebdSchin * and is licensed under the *
6*b30d1939SAndy Fiddaman * Eclipse Public License, Version 1.0 *
77c2fbfb3SApril Chin * by AT&T Intellectual Property *
8da2e3ebdSchin * *
9da2e3ebdSchin * A copy of the License is available at *
10*b30d1939SAndy Fiddaman * http://www.eclipse.org/org/documents/epl-v10.html *
11*b30d1939SAndy Fiddaman * (with md5 checksum b35adb5213ca9657e911e9befb180842) *
12da2e3ebdSchin * *
13da2e3ebdSchin * Information and Software Systems Research *
14da2e3ebdSchin * AT&T Research *
15da2e3ebdSchin * Florham Park NJ *
16da2e3ebdSchin * *
17da2e3ebdSchin * Glenn Fowler <gsf@research.att.com> *
18da2e3ebdSchin * David Korn <dgk@research.att.com> *
19da2e3ebdSchin * *
20da2e3ebdSchin ***********************************************************************/
21da2e3ebdSchin #pragma prototyped
22da2e3ebdSchin /*
23da2e3ebdSchin * David Korn
24da2e3ebdSchin * AT&T Bell Laboratories
25da2e3ebdSchin *
26da2e3ebdSchin * paste [-s] [-d delim] [file] ...
27da2e3ebdSchin *
28da2e3ebdSchin * paste lines from files together
29da2e3ebdSchin */
30da2e3ebdSchin
31da2e3ebdSchin static const char usage[] =
32*b30d1939SAndy Fiddaman "[-?\n@(#)$Id: paste (AT&T Research) 2010-06-12 $\n]"
33da2e3ebdSchin USAGE_LICENSE
34da2e3ebdSchin "[+NAME?paste - merge lines of files]"
35da2e3ebdSchin "[+DESCRIPTION?\bpaste\b concatenates the corresponding lines of a "
36da2e3ebdSchin "given input file and writes the resulting lines to standard "
37da2e3ebdSchin "output. By default \bpaste\b replaces the newline character of "
38da2e3ebdSchin "every line other than the last input file with the TAB character.]"
39da2e3ebdSchin "[+?Unless the \b-s\b option is specified, if an end-of-file is encountered "
40da2e3ebdSchin "on one or more input files, but not all input files, \bpaste\b "
41da2e3ebdSchin "behaves as if empty lines were read from the file(s) on which "
42da2e3ebdSchin "end-of-file was detected.]"
43da2e3ebdSchin "[+?Unless the \b-s\b option is specified, \bpaste\b is limited by "
44da2e3ebdSchin "the underlying operating system on how many \afile\a operands "
45da2e3ebdSchin "can be specified.]"
46da2e3ebdSchin "[+?If no \afile\a operands are given or if the \afile\a is \b-\b, \bpaste\b "
47da2e3ebdSchin "reads from standard input. The start of the file is defined as the "
48da2e3ebdSchin "current offset.]"
49da2e3ebdSchin
50da2e3ebdSchin "[s:serial?Paste the lines of one file at a time rather than one line "
51da2e3ebdSchin "from each file. In this case if the \b-d\b option is "
52da2e3ebdSchin "specified the delimiter will be reset to the first in the "
53da2e3ebdSchin "list at the beginning of each file.]"
54da2e3ebdSchin "[d:delimiters]:[list?\alist\a specifies a list of delimiters. These "
55da2e3ebdSchin "delimiters are used circularly instead of TAB to replace "
56da2e3ebdSchin "the newline character of the input lines. Unless the \b-s\b "
57da2e3ebdSchin "option is specified, the delimiter will be reset to the first "
58da2e3ebdSchin "element of \alist\a each time a line is processed from each file. "
59da2e3ebdSchin "The delimiter characters corresponding to \alist\a will be found "
60da2e3ebdSchin "by treating \alist\a as an ANSI-C string, except that the \b\\0\b "
61da2e3ebdSchin "sequence will insert the empty string instead of the null character.]"
62da2e3ebdSchin "\n"
63da2e3ebdSchin "\n[file ...]\n"
64da2e3ebdSchin "\n"
65da2e3ebdSchin "[+EXIT STATUS?]{"
66da2e3ebdSchin "[+0?All files processed successfully.]"
67da2e3ebdSchin "[+>0?An error occurred.]"
68da2e3ebdSchin "}"
69da2e3ebdSchin "[+SEE ALSO?\bcut\b(1), \bcat\b(1), \bjoin\b(1)]"
70da2e3ebdSchin ;
71da2e3ebdSchin
72da2e3ebdSchin #include <cmd.h>
73da2e3ebdSchin
743e14f97fSRoger A. Faulkner typedef struct Delim_s
753e14f97fSRoger A. Faulkner {
763e14f97fSRoger A. Faulkner const char* chr;
773e14f97fSRoger A. Faulkner size_t len;
783e14f97fSRoger A. Faulkner } Delim_t;
793e14f97fSRoger A. Faulkner
80da2e3ebdSchin /*
81da2e3ebdSchin * paste the lines of the <nstreams> defined in <streams> and put results
82da2e3ebdSchin * to <out>
83da2e3ebdSchin */
84da2e3ebdSchin
paste(int nstream,Sfio_t * streams[],Sfio_t * out,register const char * delim,int dsiz,int dlen,Delim_t * mp)853e14f97fSRoger A. Faulkner static int paste(int nstream,Sfio_t* streams[],Sfio_t *out, register const char *delim, int dsiz, int dlen, Delim_t* mp)
86da2e3ebdSchin {
87da2e3ebdSchin register const char *cp;
883e14f97fSRoger A. Faulkner register int d, n, i, z, more=1;
89da2e3ebdSchin register Sfio_t *fp;
90da2e3ebdSchin do
91da2e3ebdSchin {
92da2e3ebdSchin d = (dlen>0?0:-1);
93da2e3ebdSchin for(n=more-1,more=0; n < nstream;)
94da2e3ebdSchin {
95da2e3ebdSchin if(fp=streams[n])
96da2e3ebdSchin {
97da2e3ebdSchin if(cp = sfgetr(fp,'\n',0))
98da2e3ebdSchin {
99da2e3ebdSchin if(n==0)
100da2e3ebdSchin more = 1;
101da2e3ebdSchin else if(!more) /* first stream with output */
102da2e3ebdSchin {
1033e14f97fSRoger A. Faulkner if(dsiz == 1)
104da2e3ebdSchin sfnputc(out, *delim, n);
105da2e3ebdSchin else if(dlen>0)
106da2e3ebdSchin {
107da2e3ebdSchin for(d=n; d>dlen; d-=dlen)
1083e14f97fSRoger A. Faulkner sfwrite(out,delim,dsiz);
109da2e3ebdSchin if(d)
1103e14f97fSRoger A. Faulkner {
1113e14f97fSRoger A. Faulkner if(mp)
1123e14f97fSRoger A. Faulkner for (i = z = 0; i < d; i++)
1133e14f97fSRoger A. Faulkner z += mp[i].len;
1143e14f97fSRoger A. Faulkner else
1153e14f97fSRoger A. Faulkner z = d;
1163e14f97fSRoger A. Faulkner sfwrite(out,delim,z);
1173e14f97fSRoger A. Faulkner }
118da2e3ebdSchin }
119da2e3ebdSchin more = n+1;
120da2e3ebdSchin }
121da2e3ebdSchin if(sfwrite(out,cp,sfvalue(fp)-((n+1)<nstream)) < 0)
122da2e3ebdSchin return(-1);
123da2e3ebdSchin }
124da2e3ebdSchin else
125da2e3ebdSchin streams[n] = 0;
126da2e3ebdSchin }
127da2e3ebdSchin if(++n<nstream && more && d>=0)
128da2e3ebdSchin {
129da2e3ebdSchin register int c;
130da2e3ebdSchin if(d >= dlen)
131da2e3ebdSchin d = 0;
1323e14f97fSRoger A. Faulkner if(mp)
1333e14f97fSRoger A. Faulkner sfwrite(out,mp[d].chr,mp[d].len);
1343e14f97fSRoger A. Faulkner else if(c=delim[d])
135da2e3ebdSchin sfputc(out,c);
1363e14f97fSRoger A. Faulkner d++;
137da2e3ebdSchin }
138da2e3ebdSchin else if(n==nstream && !streams[n-1] && more)
139da2e3ebdSchin sfputc(out,'\n');
140da2e3ebdSchin }
1413e14f97fSRoger A. Faulkner } while(more);
142da2e3ebdSchin return(0);
143da2e3ebdSchin }
144da2e3ebdSchin
145da2e3ebdSchin /*
146da2e3ebdSchin * Handles paste -s, for file <in> to file <out> using delimiters <delim>
147da2e3ebdSchin */
spaste(Sfio_t * in,register Sfio_t * out,register const char * delim,int dsiz,int dlen,Delim_t * mp)1483e14f97fSRoger A. Faulkner static int spaste(Sfio_t *in,register Sfio_t* out,register const char *delim,int dsiz,int dlen,Delim_t* mp)
149da2e3ebdSchin {
150da2e3ebdSchin register const char *cp;
151da2e3ebdSchin register int d=0;
1523e14f97fSRoger A. Faulkner if((cp = sfgetr(in,'\n',0)) && sfwrite(out,cp,sfvalue(in)-1) < 0)
1533e14f97fSRoger A. Faulkner return(-1);
154da2e3ebdSchin while(cp=sfgetr(in, '\n',0))
155da2e3ebdSchin {
156da2e3ebdSchin if(dlen)
157da2e3ebdSchin {
158da2e3ebdSchin register int c;
159da2e3ebdSchin if(d >= dlen)
160da2e3ebdSchin d = 0;
1613e14f97fSRoger A. Faulkner if(mp)
1623e14f97fSRoger A. Faulkner sfwrite(out,mp[d].chr,mp[d].len);
1633e14f97fSRoger A. Faulkner else if(c=delim[d])
164da2e3ebdSchin sfputc(out,c);
1653e14f97fSRoger A. Faulkner d++;
166da2e3ebdSchin }
167da2e3ebdSchin if(sfwrite(out,cp,sfvalue(in)-1) < 0)
168da2e3ebdSchin return(-1);
169da2e3ebdSchin }
170da2e3ebdSchin sfputc(out,'\n');
171da2e3ebdSchin return(0);
172da2e3ebdSchin }
173da2e3ebdSchin
174da2e3ebdSchin int
b_paste(int argc,char ** argv,Shbltin_t * context)175*b30d1939SAndy Fiddaman b_paste(int argc, char** argv, Shbltin_t* context)
176da2e3ebdSchin {
177da2e3ebdSchin register int n, sflag=0;
178da2e3ebdSchin register Sfio_t *fp, **streams;
179da2e3ebdSchin register char *cp, *delim;
1803e14f97fSRoger A. Faulkner char *ep;
1813e14f97fSRoger A. Faulkner Delim_t *mp;
1823e14f97fSRoger A. Faulkner int dlen, dsiz;
183da2e3ebdSchin char defdelim[2];
184da2e3ebdSchin
185da2e3ebdSchin cmdinit(argc, argv, context, ERROR_CATALOG, 0);
186da2e3ebdSchin delim = 0;
187*b30d1939SAndy Fiddaman for (;;)
188da2e3ebdSchin {
189*b30d1939SAndy Fiddaman switch (optget(argv, usage))
190*b30d1939SAndy Fiddaman {
191*b30d1939SAndy Fiddaman case 'd':
192*b30d1939SAndy Fiddaman delim = opt_info.arg;
193*b30d1939SAndy Fiddaman continue;
194*b30d1939SAndy Fiddaman case 's':
195*b30d1939SAndy Fiddaman sflag++;
196*b30d1939SAndy Fiddaman continue;
197*b30d1939SAndy Fiddaman case ':':
198*b30d1939SAndy Fiddaman error(2, "%s", opt_info.arg);
199*b30d1939SAndy Fiddaman break;
200*b30d1939SAndy Fiddaman case '?':
201*b30d1939SAndy Fiddaman error(ERROR_usage(2), "%s", opt_info.arg);
202*b30d1939SAndy Fiddaman break;
203*b30d1939SAndy Fiddaman }
204da2e3ebdSchin break;
205da2e3ebdSchin }
206da2e3ebdSchin argv += opt_info.index;
207da2e3ebdSchin if(error_info.errors)
208da2e3ebdSchin error(ERROR_usage(2),"%s", optusage(NiL));
2093e14f97fSRoger A. Faulkner if(!delim || !*delim)
210da2e3ebdSchin {
2113e14f97fSRoger A. Faulkner delim = defdelim;
2123e14f97fSRoger A. Faulkner delim[0] = '\t';
2133e14f97fSRoger A. Faulkner delim[1] = 0;
2143e14f97fSRoger A. Faulkner }
215*b30d1939SAndy Fiddaman if (!(delim = strdup(delim)))
216*b30d1939SAndy Fiddaman error(ERROR_system(1), "out of space");
2173e14f97fSRoger A. Faulkner dlen = dsiz = stresc(delim);
2183e14f97fSRoger A. Faulkner mp = 0;
2193e14f97fSRoger A. Faulkner if (mbwide())
2203e14f97fSRoger A. Faulkner {
2213e14f97fSRoger A. Faulkner cp = delim;
2223e14f97fSRoger A. Faulkner ep = delim + dlen;
2233e14f97fSRoger A. Faulkner dlen = 0;
2243e14f97fSRoger A. Faulkner while (cp < ep)
2253e14f97fSRoger A. Faulkner {
2263e14f97fSRoger A. Faulkner mbchar(cp);
2273e14f97fSRoger A. Faulkner dlen++;
2283e14f97fSRoger A. Faulkner }
2293e14f97fSRoger A. Faulkner if(dlen < dsiz)
2303e14f97fSRoger A. Faulkner {
2313e14f97fSRoger A. Faulkner if (!(mp = newof(0, Delim_t, dlen, 0)))
232*b30d1939SAndy Fiddaman {
233*b30d1939SAndy Fiddaman free(delim);
2343e14f97fSRoger A. Faulkner error(ERROR_system(1), "out of space");
235*b30d1939SAndy Fiddaman }
2363e14f97fSRoger A. Faulkner cp = delim;
2373e14f97fSRoger A. Faulkner dlen = 0;
2383e14f97fSRoger A. Faulkner while (cp < ep)
2393e14f97fSRoger A. Faulkner {
2403e14f97fSRoger A. Faulkner mp[dlen].chr = cp;
2413e14f97fSRoger A. Faulkner mbchar(cp);
2423e14f97fSRoger A. Faulkner mp[dlen].len = cp - mp[dlen].chr;
2433e14f97fSRoger A. Faulkner dlen++;
2443e14f97fSRoger A. Faulkner }
2453e14f97fSRoger A. Faulkner }
246da2e3ebdSchin }
247da2e3ebdSchin if(cp = *argv)
248da2e3ebdSchin {
249da2e3ebdSchin n = argc - opt_info.index;
250da2e3ebdSchin argv++;
251da2e3ebdSchin }
252da2e3ebdSchin else
253da2e3ebdSchin n = 1;
254da2e3ebdSchin if(!sflag)
255da2e3ebdSchin {
256da2e3ebdSchin if (!(streams = (Sfio_t**)stakalloc(n*sizeof(Sfio_t*))))
257da2e3ebdSchin error(ERROR_exit(1), "out of space");
258da2e3ebdSchin n = 0;
259da2e3ebdSchin }
260da2e3ebdSchin do
261da2e3ebdSchin {
262da2e3ebdSchin if(!cp || streq(cp,"-"))
263da2e3ebdSchin fp = sfstdin;
264da2e3ebdSchin else if(!(fp = sfopen(NiL,cp,"r")))
265da2e3ebdSchin error(ERROR_system(0),"%s: cannot open",cp);
266da2e3ebdSchin if(fp && sflag)
267da2e3ebdSchin {
2683e14f97fSRoger A. Faulkner if(spaste(fp,sfstdout,delim,dsiz,dlen,mp) < 0)
269da2e3ebdSchin error(ERROR_system(0),"write failed");
270da2e3ebdSchin if(fp!=sfstdin)
271da2e3ebdSchin sfclose(fp);
272da2e3ebdSchin }
2737c2fbfb3SApril Chin else if(!sflag)
274da2e3ebdSchin streams[n++] = fp;
2757c2fbfb3SApril Chin } while(cp= *argv++);
276da2e3ebdSchin if(!sflag)
277da2e3ebdSchin {
2783e14f97fSRoger A. Faulkner if(error_info.errors==0 && paste(n,streams,sfstdout,delim,dsiz,dlen,mp) < 0)
279da2e3ebdSchin error(ERROR_system(0),"write failed");
280da2e3ebdSchin while(--n>=0)
281da2e3ebdSchin if((fp=streams[n]) && fp!=sfstdin)
282da2e3ebdSchin sfclose(fp);
283da2e3ebdSchin }
2843e14f97fSRoger A. Faulkner if (mp)
2853e14f97fSRoger A. Faulkner free(mp);
286*b30d1939SAndy Fiddaman free(delim);
287da2e3ebdSchin return(error_info.errors);
288da2e3ebdSchin }
289