| blob | 9ad1b39ac38669d485795f308a0ee3e6fcd32788 |
1 ;;; ediff-diff.el --- diff-related utilities
3 ;; Copyright (C) 1994-2013 Free Software Foundation, Inc.
5 ;; Author: Michael Kifer <kifer@cs.stonybrook.edu>
6 ;; Package: ediff
8 ;; This file is part of GNU Emacs.
10 ;; GNU Emacs is free software: you can redistribute it and/or modify
11 ;; it under the terms of the GNU General Public License as published by
12 ;; the Free Software Foundation, either version 3 of the License, or
13 ;; (at your option) any later version.
15 ;; GNU Emacs is distributed in the hope that it will be useful,
16 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 ;; GNU General Public License for more details.
20 ;; You should have received a copy of the GNU General Public License
21 ;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>.
23 ;;; Commentary:
25 ;;; Code:
36 "Diff related utilities."
41 "Program to use for generating the differential of the two files."
45 "Program to be used for three-way comparison.
46 Must produce output compatible with Unix's diff3 program."
51 ;; The following functions must precede all defcustom-defined variables.
59 "The shell used to run diff and patch.
60 If user's .profile or .cshrc files are set up correctly, any shell
61 will do. However, some people set $prompt or other things
62 incorrectly, which leads to undesirable output messages. These may
63 cause Ediff to fail. In such a case, set `ediff-shell' to a shell that
64 you are not using or, better, fix your shell's startup file."
69 "Utility to use to determine if two files are identical.
70 It must return code 0, if its arguments are identical files."
75 "Options to pass to `ediff-cmp-program'.
76 If GNU diff is used as `ediff-cmp-program', then the most useful options
77 are `-I REGEXP', to ignore changes whose lines match the REGEXP."
87 "Options to pass to `ediff-diff-program'.
88 If Unix diff is used as `ediff-diff-program',
89 then a useful option is `-w', to ignore space.
90 Options `-c', `-u', and `-i' are not allowed. Case sensitivity can be
93 Do not remove the default options. If you need to change this variable, add new
94 options after the default ones.
96 This variable is not for customizing the look of the differences produced by
98 `ediff-custom-diff-options' for that."
104 "If t, skip over difference regions that differ only in letter case.
105 This variable can be set either in .emacs or toggled interactively.
109 "Option that causes the diff program to ignore case of letters."
114 "Option that causes the diff3 program to ignore case of letters.
115 GNU diff3 doesn't have such an option."
119 ;; the actual options used in comparison
123 "Program to use for generating custom diff output for saving it in a file.
124 This output is not used by Ediff internally."
128 "Options to pass to `ediff-custom-diff-program'."
132 ;;; Support for diff3
137 "Options to pass to `ediff-diff3-program'."
142 ;; the actual options used in comparison
147 "Regexp that matches normal output lines from `ediff-diff3-program'.
148 Lines that do not match are assumed to be error messages."
152 ;; keeps the status of the current diff in 3-way jobs.
153 ;; the status can be =diff(A), =diff(B), or =diff(A+B)
157 ;;; Fine differences
160 "If `on', Ediff auto-highlights fine diffs for the current diff region.
161 If `off', auto-highlighting is not used. If `nix', no fine diffs are shown
162 at all, unless the user force-refines the region by hitting `*'.
164 This variable can be set either in .emacs or toggled interactively.
168 "If t, skip over difference regions that differ only in the white space and line breaks.
169 This variable can be set either in .emacs or toggled interactively.
175 ;;; General
180 "[0-9,]+[acd][0-9,]+\C-m?$"
188 "Regexp that matches normal output lines from `ediff-diff-program'.
189 This is mostly lifted from Emerge, except that Ediff also considers
190 warnings and `Missing newline'-type messages to be normal output.
199 "value is a function symbol depending on the kind of job is to be done.
200 For 2-way jobs and for ediff-merge, it should be `ediff-setup-diff-regions'.
201 For jobs requiring diff3, it should be `ediff-setup-diff-regions3'.
203 The function should take three mandatory arguments, file-A, file-B, and
204 file-C. It may ignore file C for diff2 jobs. It should also take
208 ;;; Functions
210 ;; Generate the difference vector and overlays for the two files
211 ;; With optional arg REG-TO-REFINE, refine this region.
212 ;; File-C argument is not used here. It is there just because
213 ;; ediff-setup-diff-regions is called via a funcall to
214 ;; ediff-setup-diff-regions-function, which can also have the value
215 ;; ediff-setup-diff-regions3, which takes 4 arguments.
217 ;; looking for '-c', '-i', '-u', or 'c', 'i', 'u' among clustered non-long options
219 ediff-diff-options)
222 ;; create, if it doesn't exist
230 ediff-diff-buffer ediff-word-mode ediff-narrow-bounds)))
232 ;; Run the diff program on FILE1 and FILE2 and put the output in DIFF-BUFFER
233 ;; Return the size of DIFF-BUFFER
234 ;; The return code isn't used in the program at present.
242 ;; 1 is an error exit code
248 ;; 1 is an error exit code
253 ;; this erases the diff buffer automatically
255 diff-buffer
256 'synchronize
257 ediff-actual-diff-options file1 file2)
264 ;; If file-A/B/C is nil, do 2-way comparison with the non-nil buffers
265 ;; This function works for diff3 and diff2 jobs
276 ediff-actual-diff3-options
277 ediff-actual-diff-options)
279 ediff-diff3-ok-lines-regexp
280 ediff-diff-ok-lines-regexp))
284 diff-options
285 ;; The shuffle below is because we can compare 3-way
286 ;; or in several 2-way fashions, like fA fC, fA fB,
287 ;; or fB fC.
297 ;; "Refining difference region %d ... done" (1+ reg-num))
304 ;; fixup diff-list
321 ))
324 ))
338 ;; If diff reports errors, show them then quit.
347 ;; BOUNDS specifies visibility bounds to use.
348 ;; WORD-MODE tells whether we are in the word-mode or not.
349 ;; If WORD-MODE, also construct vector of diffs using word numbers.
350 ;; Else, use point values.
351 ;; This function handles diff-2 jobs including the case of
352 ;; merging buffers and files without ancestor.
363 diff-list shift-A shift-B
364 )
366 ;; diff list contains word numbers, unless changed later
368 diff-list))
369 ;; we don't use visibility bounds for buffer C when merging
374 shift-B
378 ;; reset point in buffers A/B/C
396 a-begin)))
404 b-begin)))
405 a-begin-pt a-end-pt b-begin-pt b-end-pt
406 c-begin c-end c-begin-pt c-end-pt)
407 ;; fix the beginning and end numbers, because diff is somewhat
408 ;; strange about how it numbers lines
412 a-end a-begin)
416 b-end b-begin)
417 ;; (string-equal diff-type "c")
423 c-end b-end)
425 c-end a-end))
427 ;; compute main diff vector
429 ;; make diff-list contain word numbers
437 nil nil ; dummy ancestor
438 nil ; state of diff
439 nil ; state of merge
440 nil ; state of ancestor
441 )
444 nil nil ; dummy buf C
445 nil nil ; dummy ancestor
446 nil ; state of diff
447 nil ; state of merge
448 nil ; state of ancestor
449 ))
450 ))
451 a-prev a-end
452 b-prev b-end
453 c-prev c-end)
454 ;; else convert lines to points
458 ;; we must disable and then restore longlines-mode
466 a-prev a-end
467 a-prev-pt a-end-pt)
470 ))
481 b-prev b-end
482 b-prev-pt b-end-pt)
485 ))
497 c-prev c-end
498 c-prev-pt c-end-pt)
501 )))
504 diff-list
508 a-begin-pt a-end-pt b-begin-pt b-end-pt
509 c-begin-pt c-end-pt
510 nil nil ; dummy ancestor
511 ;; state of diff
512 ;; shows which buff is different from the other two
514 ediff-default-variant ; state of merge
515 nil ; state of ancestor
516 )
518 b-begin-pt b-end-pt
519 nil nil ; dummy buf C
520 nil nil ; dummy ancestor
521 nil nil ; dummy state of diff & merge
522 nil ; dummy state of ancestor
523 )))
524 )))
527 diff-list
528 ))
538 ;; set up vector showing the status of merge regions
546 ;; state of merge: prefers/default-A/B or combined
548 ;; whether the ancestor region is empty
549 state-of-ancestor)))
550 ;; the first elt designates type of list
552 )))
560 ;; ediff-extract-diffs puts the type of diff-list as the first elt
561 ;; of this list. The type is either 'points or 'words
565 buf-type ediff-narrow-bounds)))
568 buf-type ediff-narrow-bounds)))
569 diff-overlay-list list-element total-diffs
570 begin end pt-saved overlay state-of-diff)
575 ;; shift, if necessary
590 )
598 )
601 ;; Put overlays at appropriate places in buffer
602 ;; convert word numbers to points, if necessary
617 ediff-use-faces ediff-highlight-all-diffs)
623 buf-type current-diff total-diffs))
624 ;; Record all overlays for this difference.
625 ;; The 2-d elt, nil, is a place holder for the fine diff vector.
626 ;; The 3-d elt, nil, is a place holder for no-fine-diffs flag.
627 ;; The 4-th elt says which diff region is different from the other two
628 ;; (3-way jobs only).
631 diff-overlay-list
633 diff-list
639 ))
641 ;; `n' is the diff region to work on. Default is ediff-current-difference.
642 ;; if `flag' is 'noforce then make fine-diffs only if this region's fine
643 ;; diffs have not been computed before.
644 ;; if `flag' is 'skip then don't compute fine diffs for this region.
657 ;; n is within the range
668 cumulative-fine-diff-length)
671 ;; then don't refine.
672 ;; If the region happens to be entirely whitespace or empty then
673 ;; mark as such.
676 ediff-merge-job)
680 "Region %d is empty in all buffers but %S"
686 "Region %d in buffer %S is empty"
691 )
692 ;; if all regions happen to be whitespace
694 ;; mark as space only
696 ;; if some regions are white and others don't, then mark as
697 ;; non-white-space-only
700 ;; don't compute fine diffs if diff vector exists
704 "Only white-space differences in region %d %s"
713 ;; don't compute fine diffs for this region
718 "Region %d exceeds the auto-refinement limit. Type `%s' to refine"
722 )))
724 ;; recompute fine diffs
728 ediff-buffer-A
729 tmp-buffer
730 ediff-control-buffer)
737 ediff-buffer-B
738 tmp-buffer
739 ediff-control-buffer)
748 ediff-buffer-C
749 tmp-buffer
750 ediff-control-buffer)
755 ;; save temp file names.
757 ediff-temp-file-B file-B
758 ediff-temp-file-C file-C)
760 ;; set the new vector of fine diffs, if none exists
766 ;; In merge-jobs, whitespace-C is t, since
767 ;; ediff-empty-diff-region-p returns t in this case
768 whitespace-C)
776 ;; in merge jobs, the merge buffer is never refined
782 ;; all regions are white space
784 ;; none is white space and no fine diffs detected
794 "Only white-space differences in region %d %s"
804 )
807 )))
809 ;; Interface to ediff-make-fine-diffs. Checks for auto-refine limit, etc.
812 ediff-use-faces
816 (> ediff-auto-refine-limit
819 (> ediff-auto-refine-limit
825 ;; highlight if fine diffs already exist
830 ;; if fine diff vector is not set for diff N, then do nothing
835 ;; when faces are supported, set faces and priorities of fine overlays
846 'default
848 buf-type ediff-fine-diff-face-alist)
849 ))
851 0
855 buf-type
856 ediff-current-diff-overlay-alist))
862 fine-diff-vector)))
864 ;; Set overlays over the regions that denote delimiters
878 )
882 ))
885 ;; Convert diff list to overlays for a given DIFF-REGION
886 ;; in buffer of type BUF-TYPE
892 combined-merge-diff-list
893 diff-overlay-list list-element
894 begin end overlay)
900 ;; if it is a combined merge then set overlays in buff C specially
905 combined-merge-diff-list region-num)
906 ;; regular fine diff
918 ;; Put overlays at appropriate places in buffers
919 ;; convert lines to points, if necessary
924 ;; record all overlays for this difference region
929 ;; convert the list of difference information into a vector
930 ;; for fast access
933 )))
941 ))
944 ;; Stolen from emerge.el
946 ;; This save-excursion allows ediff-get-diff3-group to be called for the
947 ;; various groups of lines (1, 2, 3) in any order, and for the lines to
948 ;; appear in any order. The reason this is necessary is that Gnu diff3
949 ;; can produce the groups in the order 1, 2, 3 or 1, 3, 2.
954 ;; treatment depends on whether it is an "a" group or a "c" group
956 ;; it is a "c" group
958 ;; it has two numbers
963 ;; it has one number
967 ;; it is an "a" group
973 ;; If WORD-MODE, construct vector of diffs using word numbers.
974 ;; Else, use point values.
975 ;; WORD-MODE also tells if we are in the word-mode or not.
976 ;; If THREE-WAY-COMP, then it is a 3-way comparison. Else, it is merging
977 ;; with ancestor, in which case buffer-C contents is identical to buffer-A/B,
978 ;; contents (unless buffer-A is narrowed) depending on ediff-default-variant's
979 ;; value.
980 ;; BOUNDS specifies visibility bounds to use.
994 diff-list shift-A shift-B shift-C
995 )
997 ;; diff list contains word numbers or points, depending on word-mode
999 diff-list))
1004 shift-B
1007 shift-C
1012 ;; reset point in buffers A, B, C
1027 ;; leave point after matched line
1030 ;; if the files A and B are the same and not 3way-comparison,
1031 ;; ignore the difference
1049 state-of-ancestor
1050 c-begin c-end
1051 a-begin-pt a-end-pt
1052 b-begin-pt b-end-pt
1053 c-begin-pt c-end-pt
1054 anc-begin-pt anc-end-pt)
1061 c-end c-or-anc-end))
1064 c-end b-end))
1067 c-end a-end)))
1069 ;; compute main diff vector
1071 ;; make diff-list contain word numbers
1078 nil nil ; dummy ancestor
1079 nil ; state of diff
1080 nil ; state of merge
1081 nil ; state of ancestor
1082 )))
1083 a-prev a-end
1084 b-prev b-end
1085 c-prev c-end)
1086 ;; else convert lines to points
1090 ;; we must disable and then restore longlines-mode
1098 a-prev a-end
1099 a-prev-pt a-end-pt)
1102 ))
1113 b-prev b-end
1114 b-prev-pt b-end-pt)
1117 ))
1128 c-prev c-end
1129 c-prev-pt c-end-pt)
1132 ))
1143 anc-prev c-or-anc-end)
1146 )))
1149 diff-list
1150 ;; if comparing with ancestor, then there also is a
1151 ;; state-of-difference marker
1154 a-begin-pt a-end-pt
1155 b-begin-pt b-end-pt
1156 c-begin-pt c-end-pt
1157 nil nil ; ancestor begin/end
1158 state-of-diff-comparison
1159 nil ; state of merge
1160 nil ; state of ancestor
1161 ))
1163 b-begin-pt b-end-pt
1164 c-begin-pt c-end-pt
1165 anc-begin-pt anc-end-pt
1166 state-of-diff-merge
1167 state-of-merge
1168 state-of-ancestor
1169 )))
1170 )))
1171 ))
1174 diff-list
1175 ))
1177 ;; Generate the difference vector and overlays for three files
1178 ;; File-C is either the third file to compare (in case of 3-way comparison)
1179 ;; or it is the ancestor file.
1181 ;; looking for '-i' or a 'i' among clustered non-long options
1191 ediff-actual-diff3-options file-A file-B file-C)
1194 ;;(message "Computing differences ... done")
1197 ediff-diff-buffer
1198 ediff-word-mode ediff-3way-comparison-job ediff-narrow-bounds)
1199 ))
1202 ;; Execute PROGRAM asynchronously, unless OS/2, Windows-*, or DOS, or unless
1203 ;; SYNCH is non-nil. BUFFER must be a buffer object, and must be alive. The
1204 ;; OPTIONS arg is a list of options to pass to PROGRAM. It may be a blank
1205 ;; string. All elements in FILES must be strings. We also delete nil from
1206 ;; args.
1209 ;; If this is a buffer job, we are diffing temporary files
1210 ;; produced by Emacs with ediff-coding-system-for-write, so
1211 ;; use the same encoding to read the results.
1214 ediff-coding-system-for-write
1215 ediff-coding-system-for-read))
1216 args)
1219 ;; the --binary option, if present, should be used only for buffer jobs
1220 ;; or for refining the differences
1226 proc)
1231 synch)
1232 ;; In Windows do it synchronously, since Windows doesn't let us
1233 ;; delete files used by other processes. Thus, in ediff-buffers
1234 ;; and similar functions, we can't delete temp files because
1235 ;; they might be used by the asynch process that computes
1236 ;; custom diffs. So, we have to wait till custom diff
1237 ;; subprocess is done.
1238 ;; In DOS, must synchronize because DOS doesn't have
1239 ;; asynchronous processes.
1241 ;; On other systems, do it asynchronously.
1249 )))
1252 ;; This is shell-command-filter from simple.el in Emacs.
1253 ;; Copied here because XEmacs doesn't have it.
1255 ;; Do save-excursion by hand so that we can leave point numerically unchanged
1256 ;; despite an insertion immediately after it.
1259 opoint
1269 ;; insert-before-markers moved this marker: set it back.
1271 ;; Finish our save-excursion.
1276 ;; like shell-command-sentinel but doesn't print an exit status message
1277 ;; we do this because diff always exits with status 1, if diffs are found
1278 ;; so shell-command-sentinel displays a confusing message to the user
1288 ;;; Word functions used to refine the current diff
1291 "Function to call to move to the next word.
1295 ;; \240 is Unicode symbol for nonbreakable whitespace
1297 "Characters constituting white space.
1303 "Characters that constitute words of type 1.
1304 More precisely, [ediff-word-1] is a regexp that matches type 1 words.
1309 "Characters that constitute words of type 2.
1310 More precisely, [ediff-word-2] is a regexp that matches type 2 words.
1315 "Characters that constitute words of type 3.
1316 More precisely, [ediff-word-3] is a regexp that matches type 3 words.
1322 "Characters that constitute words of type 4.
1323 More precisely, [ediff-word-4] is a regexp that matches type 4 words.
1327 ;; Split region along word boundaries. Each word will be on its own line.
1328 ;; Output to buffer out-buffer.
1330 "Move point one word forward.
1331 There are four types of words, each of which consists entirely of
1332 characters in `ediff-word-1', `ediff-word-2', `ediff-word-3', or
1333 `ediff-word-4'. Words are recognized by passing these one after another as
1334 arguments to `skip-chars-forward'."
1341 ))
1346 ;; eval in control buf to let user create local versions for
1347 ;; different invocations
1350 ediff-forward-word-function)
1351 ediff-forward-word-function))
1352 inbuf-syntax-tbl sv-point diff-string)
1357 ediff-syntax-table)
1362 ;; Make sure that temp buff syntax table is the same as the original buf
1363 ;; syntax tbl, because we use ediff-forward-word in both and
1364 ;; ediff-forward-word depends on the syntax classes of characters.
1379 ;; copy string specified as BEG END from IN-BUF to OUT-BUF
1387 ;; goto word #n starting at current position in buffer `buf'
1388 ;; For ediff, a word is determined by ediff-forward-word-function
1389 ;; If `flag' is non-nil, goto the end of the n-th word.
1391 ;; remember val ediff-forward-word-function has in ctl buf
1406 "Return t if files F1 and F2 have identical contents."
1411 ;; In the remote case, this works only if F1 and F2 are
1412 ;; located on the same remote host.
1419 ))
1422 ;; F1 and F2 are not located on the same host.
1429 )))
1433 "Return t if D1 and D2 have the same content.
1434 D1 and D2 can either be both directories or both regular files.
1435 Symlinks and the likes are not handled.
1436 If FILTER-RE is non-nil, recursive checking in directories
1437 affects only files whose names match the expression."
1438 ;; Normalize empty filter RE to nil.
1440 ;; Indicate progress
1443 ;; D1 & D2 directories => recurse
1455 )
1458 ))
1460 ;; D1 & D2 are both files => compare directly
1464 ;; Otherwise => false: unequal contents
1465 )
1466 )
1468 ;; If lists have the same length and names of files are pairwise equal
1469 ;; (removing the directories) then compare contents pairwise.
1470 ;; True if all contents are the same; false otherwise
1472 ;; First, check only the names (works quickly and ensures a
1473 ;; precondition for subsequent code)
1477 ;; With name equality established, compare the entries
1478 ;; through recursion.
1486 )
1487 ;; if reached the end then lists are equal
1489 )
1490 )
1493 ;; ARG1 is a regexp, ARG2 is a list of full-filenames
1494 ;; Delete all entries that match the regexp
1509 ediff-actual-diff3-options
1512 ediff-actual-diff3-options ediff-diff3-options)
1513 )
1515 ediff-actual-diff3-options ediff-actual-diff3-options)
1516 )
1519 ;; Ignore case handling - some ideas from drew.adams@@oracle.com
1537 )
1541 ;; Local Variables:
1542 ;; eval: (put 'ediff-defvar-local 'lisp-indent-hook 'defun)
1543 ;; eval: (put 'ediff-with-current-buffer 'lisp-indent-hook 1)
1544 ;; eval: (put 'ediff-with-current-buffer 'edebug-form-spec '(form body))
1545 ;; End:
1547 ;;; ediff-diff.el ends here