Annotation of src/usr.bin/sdiff/sdiff.c, Revision 1.27
1.27 ! ray 1: /* $OpenBSD: sdiff.c,v 1.26 2009/06/07 13:09:24 ray Exp $ */
1.1 tedu 2:
3: /*
4: * Written by Raymond Lai <ray@cyth.net>.
5: * Public domain.
6: */
7:
8: #include <sys/param.h>
9: #include <sys/queue.h>
1.17 otto 10: #include <sys/stat.h>
1.1 tedu 11: #include <sys/types.h>
12: #include <sys/wait.h>
13:
14: #include <ctype.h>
15: #include <err.h>
1.17 otto 16: #include <errno.h>
17: #include <fcntl.h>
1.1 tedu 18: #include <getopt.h>
19: #include <limits.h>
1.19 ray 20: #include <paths.h>
1.1 tedu 21: #include <stdio.h>
22: #include <stdlib.h>
23: #include <string.h>
24: #include <unistd.h>
25: #include <util.h>
26:
1.17 otto 27: #include "common.h"
1.1 tedu 28: #include "extern.h"
29:
30: #define WIDTH 130
31: /*
32: * Each column must be at least one character wide, plus three
33: * characters between the columns (space, [<|>], space).
34: */
35: #define WIDTH_MIN 5
36:
37: /* A single diff line. */
38: struct diffline {
39: SIMPLEQ_ENTRY(diffline) diffentries;
1.14 otto 40: char *left;
41: char div;
42: char *right;
1.1 tedu 43: };
44:
45: static void astrcat(char **, const char *);
1.14 otto 46: static void enqueue(char *, char, char *);
1.17 otto 47: static char *mktmpcpy(const char *);
1.14 otto 48: static void freediff(struct diffline *);
1.1 tedu 49: static void int_usage(void);
1.13 otto 50: static int parsecmd(FILE *, FILE *, FILE *);
1.1 tedu 51: static void printa(FILE *, size_t);
52: static void printc(FILE *, size_t, FILE *, size_t);
53: static void printcol(const char *, size_t *, const size_t);
1.13 otto 54: static void printd(FILE *, size_t);
1.1 tedu 55: static void println(const char *, const char, const char *);
56: static void processq(void);
57: static void prompt(const char *, const char *);
58: __dead static void usage(void);
59: static char *xfgets(FILE *);
60:
61: SIMPLEQ_HEAD(, diffline) diffhead = SIMPLEQ_HEAD_INITIALIZER(diffhead);
62: size_t line_width; /* width of a line (two columns and divider) */
63: size_t width; /* width of each column */
64: size_t file1ln, file2ln; /* line number of file1 and file2 */
1.13 otto 65: int Iflag = 0; /* ignore sets matching regexp */
1.1 tedu 66: int lflag; /* print only left column for identical lines */
67: int sflag; /* skip identical lines */
1.3 tedu 68: FILE *outfile; /* file to save changes to */
1.19 ray 69: const char *tmpdir; /* TMPDIR or /tmp */
1.1 tedu 70:
71: static struct option longopts[] = {
72: { "text", no_argument, NULL, 'a' },
73: { "ignore-blank-lines", no_argument, NULL, 'B' },
74: { "ignore-space-change", no_argument, NULL, 'b' },
75: { "minimal", no_argument, NULL, 'd' },
76: { "ignore-tab-expansion", no_argument, NULL, 'E' },
77: { "diff-program", required_argument, NULL, 'F' },
78: { "speed-large-files", no_argument, NULL, 'H' },
79: { "ignore-matching-lines", required_argument, NULL, 'I' },
1.21 ray 80: { "ignore-case", no_argument, NULL, 'i' },
1.1 tedu 81: { "left-column", no_argument, NULL, 'l' },
82: { "output", required_argument, NULL, 'o' },
83: { "strip-trailing-cr", no_argument, NULL, 'S' },
84: { "suppress-common-lines", no_argument, NULL, 's' },
85: { "expand-tabs", no_argument, NULL, 't' },
86: { "ignore-all-space", no_argument, NULL, 'W' },
87: { "width", required_argument, NULL, 'w' },
88: { NULL, 0, NULL, 0 }
89: };
90:
1.17 otto 91: /*
92: * Create temporary file if source_file is not a regular file.
93: * Returns temporary file name if one was malloced, NULL if unnecessary.
94: */
95: static char *
96: mktmpcpy(const char *source_file)
97: {
98: struct stat sb;
99: ssize_t rcount;
100: int ifd, ofd;
101: u_char buf[BUFSIZ];
102: char *target_file;
103:
104: /* Open input and output. */
105: ifd = open(source_file, O_RDONLY, 0);
106: /* File was opened successfully. */
107: if (ifd != -1) {
108: if (fstat(ifd, &sb) == -1)
109: err(2, "error getting file status from %s", source_file);
110:
111: /* Regular file. */
1.25 ray 112: if (S_ISREG(sb.st_mode)) {
113: close(ifd);
1.17 otto 114: return (NULL);
1.25 ray 115: }
1.17 otto 116: } else {
117: /* If ``-'' does not exist the user meant stdin. */
118: if (errno == ENOENT && strcmp(source_file, "-") == 0)
119: ifd = STDIN_FILENO;
120: else
121: err(2, "error opening %s", source_file);
122: }
123:
124: /* Not a regular file, so copy input into temporary file. */
1.19 ray 125: if (asprintf(&target_file, "%s/sdiff.XXXXXXXXXX", tmpdir) == -1)
126: err(2, "asprintf");
127: if ((ofd = mkstemp(target_file)) == -1) {
1.17 otto 128: warn("error opening %s", target_file);
129: goto FAIL;
130: }
131: while ((rcount = read(ifd, buf, sizeof(buf))) != -1 &&
132: rcount != 0) {
133: ssize_t wcount;
134:
135: wcount = write(ofd, buf, (size_t)rcount);
136: if (-1 == wcount || rcount != wcount) {
137: warn("error writing to %s", target_file);
138: goto FAIL;
139: }
140: }
141: if (rcount == -1) {
142: warn("error reading from %s", source_file);
143: goto FAIL;
144: }
145:
146: close(ifd);
147: close(ofd);
148:
149: return (target_file);
150:
151: FAIL:
152: unlink(target_file);
153: exit(2);
154: }
155:
1.1 tedu 156: int
157: main(int argc, char **argv)
158: {
1.13 otto 159: FILE *diffpipe, *file1, *file2;
1.10 deraadt 160: size_t diffargc = 0, wflag = WIDTH;
1.1 tedu 161: int ch, fd[2], status;
162: pid_t pid;
1.17 otto 163: char **diffargv, *diffprog = "diff", *filename1, *filename2,
164: *tmp1, *tmp2, *s1, *s2;
1.1 tedu 165:
166: /*
167: * Process diff flags.
168: */
169: /*
170: * Allocate memory for diff arguments and NULL.
171: * Each flag has at most one argument, so doubling argc gives an
172: * upper limit of how many diff args can be passed. argv[0],
173: * file1, and file2 won't have arguments so doubling them will
174: * waste some memory; however we need an extra space for the
175: * NULL at the end, so it sort of works out.
176: */
1.23 deraadt 177: if (!(diffargv = calloc(argc, sizeof(char **) * 2)))
1.6 tedu 178: err(2, "main");
1.1 tedu 179:
180: /* Add first argument, the program name. */
181: diffargv[diffargc++] = diffprog;
182:
1.8 tedu 183: while ((ch = getopt_long(argc, argv, "aBbdEHI:ilo:stWw:",
1.1 tedu 184: longopts, NULL)) != -1) {
185: const char *errstr;
186:
187: switch (ch) {
188: case 'a':
189: diffargv[diffargc++] = "-a";
190: break;
191: case 'B':
192: diffargv[diffargc++] = "-B";
193: break;
194: case 'b':
195: diffargv[diffargc++] = "-b";
196: break;
197: case 'd':
198: diffargv[diffargc++] = "-d";
199: break;
200: case 'E':
201: diffargv[diffargc++] = "-E";
202: break;
203: case 'F':
1.4 tedu 204: diffargv[0] = diffprog = optarg;
1.1 tedu 205: break;
206: case 'H':
207: diffargv[diffargc++] = "-H";
208: break;
209: case 'I':
1.13 otto 210: Iflag = 1;
1.1 tedu 211: diffargv[diffargc++] = "-I";
212: diffargv[diffargc++] = optarg;
213: break;
214: case 'i':
215: diffargv[diffargc++] = "-i";
216: break;
217: case 'l':
218: lflag = 1;
219: break;
220: case 'o':
221: if ((outfile = fopen(optarg, "w")) == NULL)
222: err(2, "could not open: %s", optarg);
223: break;
224: case 'S':
225: diffargv[diffargc++] = "--strip-trailing-cr";
226: break;
227: case 's':
228: sflag = 1;
229: break;
230: case 't':
231: diffargv[diffargc++] = "-t";
232: break;
233: case 'W':
234: diffargv[diffargc++] = "-w";
235: break;
236: case 'w':
237: wflag = strtonum(optarg, WIDTH_MIN,
1.9 tedu 238: INT_MAX, &errstr);
1.1 tedu 239: if (errstr)
240: errx(2, "width is %s: %s", errstr, optarg);
241: break;
242: default:
243: usage();
244: }
245:
246: }
247: argc -= optind;
248: argv += optind;
249:
1.18 ray 250: if (argc != 2)
1.4 tedu 251: usage();
1.19 ray 252:
1.26 ray 253: if ((tmpdir = getenv("TMPDIR")) == NULL || *tmpdir == '\0')
1.19 ray 254: tmpdir = _PATH_TMP;
1.4 tedu 255:
1.17 otto 256: filename1 = argv[0];
257: filename2 = argv[1];
258:
259: /*
260: * Create temporary files for diff and sdiff to share if file1
261: * or file2 are not regular files. This allows sdiff and diff
262: * to read the same inputs if one or both inputs are stdin.
263: *
264: * If any temporary files were created, their names would be
265: * saved in tmp1 or tmp2. tmp1 should never equal tmp2.
266: */
267: tmp1 = tmp2 = NULL;
268: /* file1 and file2 are the same, so copy to same temp file. */
269: if (strcmp(filename1, filename2) == 0) {
270: if ((tmp1 = mktmpcpy(filename1)))
271: filename1 = filename2 = tmp1;
272: /* Copy file1 and file2 into separate temp files. */
273: } else {
274: if ((tmp1 = mktmpcpy(filename1)))
275: filename1 = tmp1;
276: if ((tmp2 = mktmpcpy(filename2)))
277: filename2 = tmp2;
278: }
279:
280: diffargv[diffargc++] = filename1;
281: diffargv[diffargc++] = filename2;
1.1 tedu 282: /* Add NULL to end of array to indicate end of array. */
283: diffargv[diffargc++] = NULL;
284:
285: /* Subtract column divider and divide by two. */
286: width = (wflag - 3) / 2;
287: /* Make sure line_width can fit in size_t. */
1.27 ! ray 288: if (width > (SIZE_MAX - 3) / 2)
1.1 tedu 289: errx(2, "width is too large: %zu", width);
290: line_width = width * 2 + 3;
291:
292: if (pipe(fd))
293: err(2, "pipe");
294:
295: switch(pid = fork()) {
296: case 0:
297: /* child */
298: /* We don't read from the pipe. */
1.5 tedu 299: close(fd[0]);
1.1 tedu 300: if (dup2(fd[1], STDOUT_FILENO) == -1)
301: err(2, "child could not duplicate descriptor");
302: /* Free unused descriptor. */
1.5 tedu 303: close(fd[1]);
1.1 tedu 304:
1.14 otto 305: execvp(diffprog, diffargv);
1.1 tedu 306: err(2, "could not execute diff: %s", diffprog);
307: case -1:
308: err(2, "could not fork");
309: }
310:
311: /* parent */
312: /* We don't write to the pipe. */
1.5 tedu 313: close(fd[1]);
1.1 tedu 314:
315: /* Open pipe to diff command. */
1.13 otto 316: if ((diffpipe = fdopen(fd[0], "r")) == NULL)
1.1 tedu 317: err(2, "could not open diff pipe");
1.17 otto 318: if ((file1 = fopen(filename1, "r")) == NULL)
319: err(2, "could not open %s", filename1);
320: if ((file2 = fopen(filename2, "r")) == NULL)
321: err(2, "could not open %s", filename2);
322:
1.1 tedu 323: /* Line numbers start at one. */
324: file1ln = file2ln = 1;
325:
326: /* Read and parse diff output. */
1.13 otto 327: while (parsecmd(diffpipe, file1, file2) != EOF)
1.1 tedu 328: ;
1.13 otto 329: fclose(diffpipe);
1.1 tedu 330:
331: /* Wait for diff to exit. */
332: if (waitpid(pid, &status, 0) == -1 || !WIFEXITED(status) ||
333: WEXITSTATUS(status) >= 2)
334: err(2, "diff exited abnormally");
1.17 otto 335:
336: /* Delete and free unneeded temporary files. */
337: if (tmp1)
338: if (unlink(tmp1))
339: warn("error deleting %s", tmp1);
340: if (tmp2)
341: if (unlink(tmp2))
342: warn("error deleting %s", tmp2);
343: free(tmp1);
344: free(tmp2);
345: filename1 = filename2 = tmp1 = tmp2 = NULL;
1.1 tedu 346:
347: /* No more diffs, so print common lines. */
1.13 otto 348: if (lflag)
349: while ((s1 = xfgets(file1)))
350: enqueue(s1, ' ', NULL);
351: else
352: for (;;) {
353: s1 = xfgets(file1);
354: s2 = xfgets(file2);
355: if (s1 || s2)
356: enqueue(s1, ' ', s2);
357: else
358: break;
359: }
360: fclose(file1);
361: fclose(file2);
1.1 tedu 362: /* Process unmodified lines. */
363: processq();
364:
365: /* Return diff exit status. */
366: return (WEXITSTATUS(status));
367: }
368:
369: /*
370: * Prints an individual column (left or right), taking into account
371: * that tabs are variable-width. Takes a string, the current column
1.12 deraadt 372: * the cursor is on the screen, and the maximum value of the column.
1.1 tedu 373: * The column value is updated as we go along.
374: */
375: static void
376: printcol(const char *s, size_t *col, const size_t col_max)
377: {
378:
379: for (; *s && *col < col_max; ++s) {
380: size_t new_col;
381:
382: switch (*s) {
383: case '\t':
384: /*
385: * If rounding to next multiple of eight causes
386: * an integer overflow, just return.
387: */
1.27 ! ray 388: if (*col > SIZE_MAX - 8)
1.1 tedu 389: return;
390:
391: /* Round to next multiple of eight. */
392: new_col = (*col / 8 + 1) * 8;
393:
394: /*
395: * If printing the tab goes past the column
396: * width, don't print it and just quit.
397: */
398: if (new_col > col_max)
399: return;
400: *col = new_col;
401: break;
402:
403: default:
404: ++(*col);
405: }
406:
407: putchar(*s);
408: }
409: }
410:
411: /*
412: * Prompts user to either choose between two strings or edit one, both,
413: * or neither.
414: */
415: static void
416: prompt(const char *s1, const char *s2)
417: {
1.14 otto 418: char *cmd;
1.1 tedu 419:
420: /* Print command prompt. */
421: putchar('%');
422:
423: /* Get user input. */
1.14 otto 424: for (; (cmd = xfgets(stdin)); free(cmd)) {
1.1 tedu 425: const char *p;
426:
427: /* Skip leading whitespace. */
428: for (p = cmd; isspace(*p); ++p)
429: ;
430:
431: switch (*p) {
432: case 'e':
433: /* Skip `e'. */
434: ++p;
435:
436: if (eparse(p, s1, s2) == -1)
437: goto USAGE;
438: break;
439:
440: case 'l':
1.22 ray 441: case '1':
1.1 tedu 442: /* Choose left column as-is. */
443: if (s1 != NULL)
444: fprintf(outfile, "%s\n", s1);
445:
446: /* End of command parsing. */
447: break;
448:
449: case 'q':
450: goto QUIT;
451:
452: case 'r':
1.22 ray 453: case '2':
1.1 tedu 454: /* Choose right column as-is. */
455: if (s2 != NULL)
456: fprintf(outfile, "%s\n", s2);
457:
458: /* End of command parsing. */
459: break;
460:
461: case 's':
462: sflag = 1;
463: goto PROMPT;
464:
465: case 'v':
466: sflag = 0;
467: /* FALLTHROUGH */
468:
469: default:
470: /* Interactive usage help. */
471: USAGE:
472: int_usage();
473: PROMPT:
474: putchar('%');
475:
476: /* Prompt user again. */
477: continue;
478: }
479:
1.14 otto 480: free(cmd);
1.1 tedu 481: return;
482: }
483:
484: /*
485: * If there was no error, we received an EOF from stdin, so we
486: * should quit.
487: */
488: QUIT:
1.5 tedu 489: fclose(outfile);
1.1 tedu 490: exit(0);
491: }
492:
493: /*
494: * Takes two strings, separated by a column divider. NULL strings are
495: * treated as empty columns. If the divider is the ` ' character, the
496: * second column is not printed (-l flag). In this case, the second
497: * string must be NULL. When the second column is NULL, the divider
498: * does not print the trailing space following the divider character.
499: *
500: * Takes into account that tabs can take multiple columns.
501: */
502: static void
503: println(const char *s1, const char div, const char *s2)
504: {
505: size_t col;
506:
507: /* Print first column. Skips if s1 == NULL. */
508: col = 0;
509: if (s1) {
510: /* Skip angle bracket and space. */
511: printcol(s1, &col, width);
512:
513: }
514:
515: /* Only print left column. */
516: if (div == ' ' && !s2) {
517: putchar('\n');
518: return;
519: }
520:
521: /* Otherwise, we pad this column up to width. */
522: for (; col < width; ++col)
523: putchar(' ');
524:
525: /*
526: * Print column divider. If there is no second column, we don't
527: * need to add the space for padding.
528: */
529: if (!s2) {
530: printf(" %c\n", div);
531: return;
532: }
533: printf(" %c ", div);
534: col += 3;
535:
536: /* Skip angle bracket and space. */
537: printcol(s2, &col, line_width);
538:
539: putchar('\n');
540: }
541:
542: /*
543: * Reads a line from file and returns as a string. If EOF is reached,
544: * NULL is returned. The returned string must be freed afterwards.
545: */
546: static char *
547: xfgets(FILE *file)
548: {
549: const char delim[3] = {'\0', '\0', '\0'};
550: char *s;
551:
552: /* XXX - Is this necessary? */
553: clearerr(file);
554:
555: if (!(s = fparseln(file, NULL, NULL, delim, 0)) &&
556: ferror(file))
557: err(2, "error reading file");
558:
559: if (!s) {
560: return (NULL);
561: }
562:
563: return (s);
564: }
565:
566: /*
1.13 otto 567: * Parse ed commands from diffpipe and print lines from file1 (lines
568: * to change or delete) or file2 (lines to add or change).
569: * Returns EOF or 0.
1.1 tedu 570: */
571: static int
1.13 otto 572: parsecmd(FILE *diffpipe, FILE *file1, FILE *file2)
1.1 tedu 573: {
1.13 otto 574: size_t file1start, file1end, file2start, file2end, n;
1.1 tedu 575: /* ed command line and pointer to characters in line */
1.11 tedu 576: char *line, *p, *q;
577: const char *errstr;
578: char c, cmd;
1.1 tedu 579:
580: /* Read ed command. */
1.13 otto 581: if (!(line = xfgets(diffpipe)))
1.1 tedu 582: return (EOF);
583:
584: p = line;
585: /* Go to character after line number. */
586: while (isdigit(*p))
587: ++p;
1.11 tedu 588: c = *p;
589: *p++ = 0;
590: file1start = strtonum(line, 0, INT_MAX, &errstr);
591: if (errstr)
592: errx(2, "file1 start is %s: %s", errstr, line);
1.1 tedu 593:
594: /* A range is specified for file1. */
1.11 tedu 595: if (c == ',') {
1.1 tedu 596:
1.11 tedu 597: q = p;
598: /* Go to character after file2end. */
599: while (isdigit(*p))
600: ++p;
601: c = *p;
602: *p++ = 0;
603: file1end = strtonum(q, 0, INT_MAX, &errstr);
604: if (errstr)
605: errx(2, "file1 end is %s: %s", errstr, line);
1.1 tedu 606: if (file1start > file1end)
607: errx(2, "invalid line range in file1: %s", line);
608:
609: } else
610: file1end = file1start;
611:
1.11 tedu 612: cmd = c;
1.1 tedu 613: /* Check that cmd is valid. */
614: if (!(cmd == 'a' || cmd == 'c' || cmd == 'd'))
615: errx(2, "ed command not recognized: %c: %s", cmd, line);
616:
1.11 tedu 617: q = p;
1.1 tedu 618: /* Go to character after line number. */
619: while (isdigit(*p))
620: ++p;
1.11 tedu 621: c = *p;
622: *p++ = 0;
623: file2start = strtonum(q, 0, INT_MAX, &errstr);
624: if (errstr)
625: errx(2, "file2 start is %s: %s", errstr, line);
1.1 tedu 626:
627: /*
628: * There should either be a comma signifying a second line
629: * number or the line should just end here.
630: */
1.11 tedu 631: if (c != ',' && c != '\0')
632: errx(2, "invalid line range in file2: %c: %s", c, line);
1.1 tedu 633:
1.11 tedu 634: if (c == ',') {
1.1 tedu 635:
1.11 tedu 636: file2end = strtonum(p, 0, INT_MAX, &errstr);
637: if (errstr)
638: errx(2, "file2 end is %s: %s", errstr, line);
1.1 tedu 639: if (file2start >= file2end)
640: errx(2, "invalid line range in file2: %s", line);
641: } else
642: file2end = file2start;
643:
644: /* Appends happen _after_ stated line. */
645: if (cmd == 'a') {
646: if (file1start != file1end)
647: errx(2, "append cannot have a file1 range: %s",
648: line);
1.27 ! ray 649: if (file1start == SIZE_MAX)
1.1 tedu 650: errx(2, "file1 line range too high: %s", line);
651: file1start = ++file1end;
652: }
653: /*
654: * I'm not sure what the deal is with the line numbers for
655: * deletes, though.
656: */
657: else if (cmd == 'd') {
658: if (file2start != file2end)
659: errx(2, "delete cannot have a file2 range: %s",
660: line);
1.27 ! ray 661: if (file2start == SIZE_MAX)
1.1 tedu 662: errx(2, "file2 line range too high: %s", line);
663: file2start = ++file2end;
664: }
665:
1.13 otto 666: /*
667: * Continue reading file1 and file2 until we reach line numbers
668: * specified by diff. Should only happen with -I flag.
669: */
670: for (; file1ln < file1start && file2ln < file2start;
671: ++file1ln, ++file2ln) {
1.14 otto 672: char *s1, *s2;
1.1 tedu 673:
1.13 otto 674: if (!(s1 = xfgets(file1)))
1.1 tedu 675: errx(2, "file1 shorter than expected");
1.13 otto 676: if (!(s2 = xfgets(file2)))
677: errx(2, "file2 shorter than expected");
1.1 tedu 678:
679: /* If the -l flag was specified, print only left column. */
1.13 otto 680: if (lflag) {
1.14 otto 681: free(s2);
1.13 otto 682: /*
683: * XXX - If -l and -I are both specified, all
684: * unchanged or ignored lines are shown with a
685: * `(' divider. This matches GNU sdiff, but I
686: * believe it is a bug. Just check out:
687: * gsdiff -l -I '^$' samefile samefile.
688: */
689: if (Iflag)
690: enqueue(s1, '(', NULL);
691: else
692: enqueue(s1, ' ', NULL);
693: } else
694: enqueue(s1, ' ', s2);
695: }
696: /* Ignore deleted lines. */
697: for (; file1ln < file1start; ++file1ln) {
1.14 otto 698: char *s;
1.13 otto 699:
700: if (!(s = xfgets(file1)))
701: errx(2, "file1 shorter than expected");
702:
703: enqueue(s, '(', NULL);
1.1 tedu 704: }
1.13 otto 705: /* Ignore added lines. */
706: for (; file2ln < file2start; ++file2ln) {
1.14 otto 707: char *s;
1.13 otto 708:
709: if (!(s = xfgets(file2)))
710: errx(2, "file2 shorter than expected");
711:
712: /* If -l flag was given, don't print right column. */
713: if (lflag)
1.14 otto 714: free(s);
1.13 otto 715: else
716: enqueue(NULL, ')', s);
717: }
718:
719: /* Process unmodified or skipped lines. */
1.1 tedu 720: processq();
721:
722: switch (cmd) {
723: case 'a':
1.13 otto 724: printa(file2, file2end);
725: n = file2end - file2start + 1;
1.1 tedu 726: break;
727:
728: case 'c':
1.13 otto 729: printc(file1, file1end, file2, file2end);
730: n = file1end - file1start + 1 + 1 + file2end - file2start + 1;
1.1 tedu 731: break;
732:
733: case 'd':
1.13 otto 734: printd(file1, file1end);
735: n = file1end - file1start + 1;
1.1 tedu 736: break;
737:
738: default:
739: errx(2, "invalid diff command: %c: %s", cmd, line);
740: }
741:
1.13 otto 742: /* Skip to next ed line. */
743: while (n--)
744: if (!xfgets(diffpipe))
745: errx(2, "diff ended early");
746:
1.3 tedu 747: return (0);
1.1 tedu 748: }
749:
750: /*
751: * Queues up a diff line.
752: */
753: static void
1.14 otto 754: enqueue(char *left, char div, char *right)
1.1 tedu 755: {
756: struct diffline *diffp;
757:
758: if (!(diffp = malloc(sizeof(struct diffline))))
1.6 tedu 759: err(2, "enqueue");
1.1 tedu 760: diffp->left = left;
761: diffp->div = div;
762: diffp->right = right;
763: SIMPLEQ_INSERT_TAIL(&diffhead, diffp, diffentries);
764: }
765:
766: /*
767: * Free a diffline structure and its elements.
768: */
769: static void
1.14 otto 770: freediff(struct diffline *diffp)
1.1 tedu 771: {
1.14 otto 772: free(diffp->left);
773: free(diffp->right);
774: free(diffp);
1.1 tedu 775: }
776:
777: /*
778: * Append second string into first. Repeated appends to the same string
779: * are cached, making this an O(n) function, where n = strlen(append).
780: */
781: static void
782: astrcat(char **s, const char *append)
783: {
784: /* Length of string in previous run. */
785: static size_t offset = 0;
1.15 otto 786: size_t newsiz;
1.1 tedu 787: /*
788: * String from previous run. Compared to *s to see if we are
789: * dealing with the same string. If so, we can use offset.
790: */
1.14 otto 791: static const char *oldstr = NULL;
1.1 tedu 792: char *newstr;
793:
794:
795: /*
796: * First string is NULL, so just copy append.
797: */
798: if (!*s) {
799: if (!(*s = strdup(append)))
1.6 tedu 800: err(2, "astrcat");
1.1 tedu 801:
802: /* Keep track of string. */
803: offset = strlen(*s);
804: oldstr = *s;
805:
806: return;
807: }
808:
809: /*
810: * *s is a string so concatenate.
811: */
812:
813: /* Did we process the same string in the last run? */
814: /*
815: * If this is a different string from the one we just processed
816: * cache new string.
817: */
818: if (oldstr != *s) {
819: offset = strlen(*s);
820: oldstr = *s;
821: }
822:
1.15 otto 823: /* Size = strlen(*s) + \n + strlen(append) + '\0'. */
824: newsiz = offset + 1 + strlen(append) + 1;
1.1 tedu 825:
826: /* Resize *s to fit new string. */
1.15 otto 827: newstr = realloc(*s, newsiz);
1.1 tedu 828: if (newstr == NULL)
1.6 tedu 829: err(2, "astrcat");
1.1 tedu 830: *s = newstr;
831:
1.15 otto 832: /* *s + offset should be end of string. */
1.1 tedu 833: /* Concatenate. */
1.15 otto 834: strlcpy(*s + offset, "\n", newsiz - offset);
835: strlcat(*s + offset, append, newsiz - offset);
1.1 tedu 836:
1.15 otto 837: /* New string length should be exactly newsiz - 1 characters. */
1.1 tedu 838: /* Store generated string's values. */
1.15 otto 839: offset = newsiz - 1;
1.1 tedu 840: oldstr = *s;
841: }
842:
843: /*
844: * Process diff set queue, printing, prompting, and saving each diff
845: * line stored in queue.
846: */
847: static void
848: processq(void)
849: {
850: struct diffline *diffp;
1.14 otto 851: char divc, *left, *right;
1.1 tedu 852:
853: /* Don't process empty queue. */
854: if (SIMPLEQ_EMPTY(&diffhead))
855: return;
856:
1.16 otto 857: /* Remember the divider. */
858: divc = SIMPLEQ_FIRST(&diffhead)->div;
859:
1.1 tedu 860: left = NULL;
861: right = NULL;
862: /*
863: * Go through set of diffs, concatenating each line in left or
864: * right column into two long strings, `left' and `right'.
865: */
866: SIMPLEQ_FOREACH(diffp, &diffhead, diffentries) {
867: /*
1.13 otto 868: * Print changed lines if -s was given,
869: * print all lines if -s was not given.
1.1 tedu 870: */
1.16 otto 871: if (!sflag || diffp->div == '|' || diffp->div == '<' ||
872: diffp->div == '>')
1.1 tedu 873: println(diffp->left, diffp->div, diffp->right);
874:
875: /* Append new lines to diff set. */
876: if (diffp->left)
877: astrcat(&left, diffp->left);
878: if (diffp->right)
879: astrcat(&right, diffp->right);
880: }
881:
882: /* Empty queue and free each diff line and its elements. */
883: while (!SIMPLEQ_EMPTY(&diffhead)) {
884: diffp = SIMPLEQ_FIRST(&diffhead);
1.13 otto 885: SIMPLEQ_REMOVE_HEAD(&diffhead, diffentries);
1.1 tedu 886: freediff(diffp);
887: }
888:
889: /* Write to outfile, prompting user if lines are different. */
1.13 otto 890: if (outfile)
1.14 otto 891: switch (divc) {
1.13 otto 892: case ' ': case '(': case ')':
1.1 tedu 893: fprintf(outfile, "%s\n", left);
1.13 otto 894: break;
895: case '|': case '<': case '>':
1.1 tedu 896: prompt(left, right);
1.13 otto 897: break;
898: default:
1.14 otto 899: errx(2, "invalid divider: %c", divc);
1.13 otto 900: }
1.1 tedu 901:
902: /* Free left and right. */
1.14 otto 903: free(left);
904: free(right);
1.1 tedu 905: }
906:
907: /*
908: * Print lines following an (a)ppend command.
909: */
910: static void
911: printa(FILE *file, size_t line2)
912: {
913: char *line;
914:
915: for (; file2ln <= line2; ++file2ln) {
916: if (!(line = xfgets(file)))
917: errx(2, "append ended early");
918: enqueue(NULL, '>', line);
919: }
920:
921: processq();
922: }
923:
924: /*
925: * Print lines following a (c)hange command, from file1ln to file1end
926: * and from file2ln to file2end.
927: */
928: static void
929: printc(FILE *file1, size_t file1end, FILE *file2, size_t file2end)
930: {
931: struct fileline {
1.14 otto 932: SIMPLEQ_ENTRY(fileline) fileentries;
933: char *line;
1.1 tedu 934: };
935: SIMPLEQ_HEAD(, fileline) delqhead = SIMPLEQ_HEAD_INITIALIZER(delqhead);
936:
937: /* Read lines to be deleted. */
938: for (; file1ln <= file1end; ++file1ln) {
939: struct fileline *linep;
1.14 otto 940: char *line1;
1.1 tedu 941:
942: /* Read lines from both. */
943: if (!(line1 = xfgets(file1)))
944: errx(2, "error reading file1 in delete in change");
945:
946: /* Add to delete queue. */
947: if (!(linep = malloc(sizeof(struct fileline))))
1.6 tedu 948: err(2, "printc");
1.1 tedu 949: linep->line = line1;
950: SIMPLEQ_INSERT_TAIL(&delqhead, linep, fileentries);
951: }
952:
953: /* Process changed lines.. */
954: for (; !SIMPLEQ_EMPTY(&delqhead) && file2ln <= file2end;
955: ++file2ln) {
956: struct fileline *del;
957: char *add;
958:
959: /* Get add line. */
1.13 otto 960: if (!(add = xfgets(file2)))
961: errx(2, "error reading add in change");
1.1 tedu 962:
963: del = SIMPLEQ_FIRST(&delqhead);
964: enqueue(del->line, '|', add);
965: SIMPLEQ_REMOVE_HEAD(&delqhead, fileentries);
966: /*
967: * Free fileline structure but not its elements since
968: * they are queued up.
969: */
970: free(del);
971: }
972: processq();
973:
974: /* Process remaining lines to add. */
975: for (; file2ln <= file2end; ++file2ln) {
976: char *add;
977:
978: /* Get add line. */
1.13 otto 979: if (!(add = xfgets(file2)))
980: errx(2, "error reading add in change");
1.1 tedu 981:
982: enqueue(NULL, '>', add);
983: }
984: processq();
985:
986: /* Process remaining lines to delete. */
987: while (!SIMPLEQ_EMPTY(&delqhead)) {
988: struct fileline *filep;
989:
990: filep = SIMPLEQ_FIRST(&delqhead);
991: enqueue(filep->line, '<', NULL);
992: SIMPLEQ_REMOVE_HEAD(&delqhead, fileentries);
993: free(filep);
994: }
995: processq();
996: }
997:
998: /*
999: * Print deleted lines from file, from file1ln to file1end.
1000: */
1001: static void
1.13 otto 1002: printd(FILE *file1, size_t file1end)
1.1 tedu 1003: {
1.14 otto 1004: char *line1;
1.1 tedu 1005:
1006: /* Print out lines file1ln to line2. */
1007: for (; file1ln <= file1end; ++file1ln) {
1008: if (!(line1 = xfgets(file1)))
1009: errx(2, "file1 ended early in delete");
1010: enqueue(line1, '<', NULL);
1011: }
1012: processq();
1013: }
1014:
1015: /*
1016: * Interactive mode usage.
1017: */
1018: static void
1019: int_usage(void)
1020: {
1021: puts("e:\tedit blank diff\n"
1022: "eb:\tedit both diffs concatenated\n"
1023: "el:\tedit left diff\n"
1024: "er:\tedit right diff\n"
1.22 ray 1025: "l | 1:\tchoose left diff\n"
1026: "r | 2:\tchoose right diff\n"
1.1 tedu 1027: "s:\tsilent mode--don't print identical lines\n"
1028: "v:\tverbose mode--print identical lines\n"
1029: "q:\tquit");
1030: }
1031:
1032: static void
1033: usage(void)
1034: {
1035: extern char *__progname;
1036:
1037: fprintf(stderr,
1.8 tedu 1038: "usage: %s [-abdilstW] [-I regexp] [-o outfile] [-w width] file1 file2\n",
1.1 tedu 1039: __progname);
1040: exit(2);
1041: }