Messed up the man page in my previous commit. Now all is well again.
[dragonfly.git] / usr.bin / patch / patch.1
CommitLineData
48d201a5 1.\" $OpenBSD: patch.1,v 1.17 2003/10/31 20:20:45 millert Exp $
88147647 2.\" $DragonFly: src/usr.bin/patch/patch.1,v 1.4 2004/12/26 18:58:59 swildner Exp $
48d201a5
JS
3.\"
4.\" Copyright 1986, Larry Wall
5.\"
6.\" Redistribution and use in source and binary forms, with or without
7.\" modification, are permitted provided that the following condition
8.\" is met:
9.\" 1. Redistributions of source code must retain the above copyright
10.\" notice, this condition and the following disclaimer.
11.\"
12.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
13.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
14.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
15.\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
16.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
17.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
18.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
19.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
20.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
21.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
22.\" SUCH DAMAGE.
23.\"
24.Dd July 23, 2003
25.Dt PATCH 1
26.Os
27.Sh NAME
28.Nm patch
29.Nd apply a diff file to an original
30.Sh SYNOPSIS
31.Nm patch
32.Op Cm options
33.Op Ar origfile Op Ar patchfile
34.Nm patch
35.Pf \*(Lt Ar patchfile
36.Sh DESCRIPTION
37.Nm
38will take a patch file containing any of the four forms of difference
39listing produced by the
40.Xr diff 1
41program and apply those differences to an original file,
42producing a patched version.
43If
44.Ar patchfile
45is omitted, or is a hyphen, the patch will be read from the standard input.
46.Pp
47.Nm
48will attempt to determine the type of the diff listing, unless over-ruled by a
49.Fl c ,
50.Fl e ,
51.Fl n ,
52or
53.Fl u
54option.
55Context diffs (old-style, new-style, and unified) and
56normal diffs are applied directly by the
57.Nm
58program itself, whereas ed diffs are simply fed to the
59.Xr ed 1
60editor via a pipe.
61.Pp
62If the
63.Ar patchfile
64contains more than one patch,
65.Nm
66will try to apply each of them as if they came from separate patch files.
67This means, among other things, that it is assumed that the name of the file
68to patch must be determined for each diff listing, and that the garbage before
69each diff listing will be examined for interesting things such as file names
70and revision level (see the section on
71.Sx Filename Determination
72below).
73.Pp
74The options are as follows:
75.Bl -tag -width Ds
88147647 76.It Fl b , Fl Fl backup
48d201a5
JS
77Save a backup copy of the file before it is modified.
78By default the original file is saved with a backup extension of
79.Qq .orig
80unless the file already has a numbered backup, in which case a numbered
81backup is made.
82This is equivalent to specifying
83.Qo Fl V Ar existing Qc .
84This option is currently the default but that will change in a future release.
85.It Fl B , Fl Fl prefix
86Causes the next argument to be interpreted as a prefix to the backup file
87name.
88If this argument is specified, any argument to
89.Fl z
90will be ignored.
91.It Fl c , Fl Fl context
92Forces
93.Nm
94to interpret the patch file as a context diff.
95.It Fl C , Fl Fl check
96Checks that the patch would apply cleanly, but does not modify anything.
97.It Fl d , Fl Fl directory
98Causes
99.Nm
100to interpret the next argument as a directory, and
101.Xr cd 1
102to it before doing anything else.
103.It Fl D , Fl Fl ifdef
104Causes
105.Nm
106to use the
107.Qq #ifdef...#endif
108construct to mark changes.
109The argument following will be used as the differentiating symbol.
110Note that, unlike the C compiler, there must be a space between the
111.Fl D
112and the argument.
113.It Fl e , Fl Fl ed
114Forces
115.Nm
116to interpret the patch file as an
117.Xr ed 1
118script.
119.It Fl E , Fl Fl remove-empty-files
120Causes
121.Nm
122to remove output files that are empty after the patches have been applied.
123This option is useful when applying patches that create or remove files.
124.It Fl f , Fl Fl force
125Forces
126.Nm
127to assume that the user knows exactly what he or she is doing, and to not
128ask any questions.
129It assumes the following:
130skip patches for which a file to patch can't be found;
131patch files even though they have the wrong version for the
132.Qq Prereq:
133line in the patch;
134and assume that patches are not reversed even if they look like they are.
135This option does not suppress commentary; use
136.Fl s
137for that.
138.It Xo
139.Fl F Ns Aq Ar number ,
140.Fl Fl fuzz Aq Ar number
141.Xc
142Sets the maximum fuzz factor.
143This option only applies to context diffs, and causes
144.Nm
145to ignore up to that many lines in looking for places to install a hunk.
146Note that a larger fuzz factor increases the odds of a faulty patch.
147The default fuzz factor is 2, and it may not be set to more than
148the number of lines of context in the context diff, ordinarily 3.
149.It Fl i , Fl Fl input
150Causes the next argument to be interpreted as the input file name
151(i.e. a patchfile).
152This option may be specified multiple times.
153.It Fl l , Fl Fl ignore-whitespace
154Causes the pattern matching to be done loosely, in case the tabs and
155spaces have been munged in your input file.
156Any sequence of whitespace in the pattern line will match any sequence
157in the input file.
158Normal characters must still match exactly.
159Each line of the context must still match a line in the input file.
160.It Fl n , Fl Fl normal
161Forces
162.Nm
163to interpret the patch file as a normal diff.
164.It Fl N , Fl Fl forward
165Causes
166.Nm
167to ignore patches that it thinks are reversed or already applied.
168See also
169.Fl R .
170.It Fl o , Fl Fl output
171Causes the next argument to be interpreted as the output file name.
172.It Xo
173.Fl p Ns Aq Ar number ,
174.Fl Fl strip Aq Ar number
175.Xc
176Sets the pathname strip count,
177which controls how pathnames found in the patch file are treated,
178in case you keep your files in a different directory than the person who sent
179out the patch.
180The strip count specifies how many slashes are to be stripped from
181the front of the pathname.
182(Any intervening directory names also go away.)
183For example, supposing the file name in the patch file was
184.Pa /u/howard/src/blurfl/blurfl.c :
185.Pp
186Setting
187.Fl p Ns Ar 0
188gives the entire pathname unmodified.
189.Pp
190.Fl p Ns Ar 1
191gives
192.Pp
193.D1 Pa u/howard/src/blurfl/blurfl.c
194.Pp
195without the leading slash.
196.Pp
197.Fl p Ns Ar 4
198gives
199.Pp
200.D1 Pa blurfl/blurfl.c
201.Pp
202Not specifying
203.Fl p
204at all just gives you
205.Pa blurfl.c ,
206unless all of the directories in the leading path
207.Pq Pa u/howard/src/blurfl
208exist and that path is relative,
209in which case you get the entire pathname unmodified.
210Whatever you end up with is looked for either in the current directory,
211or the directory specified by the
212.Fl d
213option.
214.It Fl r , Fl Fl reject-file
215Causes the next argument to be interpreted as the reject file name.
216.It Fl R , Fl Fl reverse
217Tells
218.Nm
219that this patch was created with the old and new files swapped.
220(Yes, I'm afraid that does happen occasionally, human nature being what it
221is.)
222.Nm
223will attempt to swap each hunk around before applying it.
224Rejects will come out in the swapped format.
225The
226.Fl R
227option will not work with ed diff scripts because there is too little
228information to reconstruct the reverse operation.
229.Pp
230If the first hunk of a patch fails,
231.Nm
232will reverse the hunk to see if it can be applied that way.
233If it can, you will be asked if you want to have the
234.Fl R
235option set.
236If it can't, the patch will continue to be applied normally.
237(Note: this method cannot detect a reversed patch if it is a normal diff
238and if the first command is an append (i.e. it should have been a delete)
239since appends always succeed, due to the fact that a null context will match
240anywhere.
241Luckily, most patches add or change lines rather than delete them, so most
242reversed normal diffs will begin with a delete, which will fail, triggering
243the heuristic.)
244.It Xo
245.Fl s , Fl Fl quiet ,
246.Fl Fl silent
247.Xc
248Makes
249.Nm
250do its work silently, unless an error occurs.
251.It Fl t , Fl Fl batch
252Similar to
253.Fl f ,
254in that it suppresses questions, but makes some different assumptions:
255skip patches for which a file to patch can't be found (the same as
256.Fl f ) ;
257skip patches for which the file has the wrong version for the
258.Qq Prereq:
259line in the patch;
260and assume that patches are reversed if they look like they are.
261.It Fl u , Fl Fl unified
262Forces
263.Nm
264to interpret the patch file as a unified context diff (a unidiff).
265.It Fl v , Fl Fl version
266Causes
267.Nm
268to print out its revision header and patch level.
269.It Fl V , Fl Fl version-control
270Causes the next argument to be interpreted as a method for creating
271backup file names.
272The type of backups made can also be given in the
273.Ev PATCH_VERSION_CONTROL
274or
275.Ev VERSION_CONTROL
276environment variables, which are overridden by this option.
277The
278.Fl B
279option overrides this option, causing the prefix to always be used for
280making backup file names.
281The values of the
282.Ev PATCH_VERSION_CONTROL
283and
284.Ev VERSION_CONTROL
285environment variables and the argument to the
286.Fl V
287option are like the GNU Emacs
288.Dq version-control
289variable; they also recognize synonyms that are more descriptive.
290The valid values are (unique abbreviations are accepted):
291.Bl -tag -width Ds -offset indent
292.It t , numbered
293Always make numbered backups.
294.It nil , existing
295Make numbered backups of files that already have them,
296simple backups of the others.
297.It never , simple
298Always make simple backups.
299.El
300.It Xo
301.Fl x Ns Aq Ar number ,
302.Fl Fl debug Aq Ar number
303.Xc
304Sets internal debugging flags, and is of interest only to
305.Nm
306patchers.
f7e25d55 307.It Fl z , Fl Fl suffix
48d201a5
JS
308Causes the next argument to be interpreted as the backup extension, to be
309used in place of
310.Qq .orig .
311.It Fl Fl posix
312Enables strict
313.St -p1003.2
314conformance, specifically:
315.Bl -enum
316.It
317Backup files are not created unless the
88147647 318.Fl b
48d201a5
JS
319option is specified.
320.It
321If unspecified, the file name used is the first of the old, new and
322index files that exists.
323.El
324.El
325.Ss Patch Application
326.Nm
327will try to skip any leading garbage, apply the diff,
328and then skip any trailing garbage.
329Thus you could feed an article or message containing a
330diff listing to
331.Nm patch ,
332and it should work.
333If the entire diff is indented by a consistent amount,
334this will be taken into account.
335.Pp
336With context diffs, and to a lesser extent with normal diffs,
337.Nm
338can detect when the line numbers mentioned in the patch are incorrect,
339and will attempt to find the correct place to apply each hunk of the patch.
340As a first guess, it takes the line number mentioned for the hunk, plus or
341minus any offset used in applying the previous hunk.
342If that is not the correct place,
343.Nm
344will scan both forwards and backwards for a set of lines matching the context
345given in the hunk.
346First
347.Nm
348looks for a place where all lines of the context match.
349If no such place is found, and it's a context diff, and the maximum fuzz factor
350is set to 1 or more, then another scan takes place ignoring the first and last
351line of context.
352If that fails, and the maximum fuzz factor is set to 2 or more,
353the first two and last two lines of context are ignored,
354and another scan is made.
355.Pq The default maximum fuzz factor is 2.
356.Pp
357If
358.Nm
359cannot find a place to install that hunk of the patch, it will put the hunk
360out to a reject file, which normally is the name of the output file plus
361.Qq .rej .
362(Note that the rejected hunk will come out in context diff form whether the
363input patch was a context diff or a normal diff.
364If the input was a normal diff, many of the contexts will simply be null.)
365The line numbers on the hunks in the reject file may be different than
366in the patch file: they reflect the approximate location patch thinks the
367failed hunks belong in the new file rather than the old one.
368.Pp
369As each hunk is completed, you will be told whether the hunk succeeded or
370failed, and which line (in the new file)
371.Nm
372thought the hunk should go on.
373If this is different from the line number specified in the diff,
374you will be told the offset.
375A single large offset MAY be an indication that a hunk was installed in the
376wrong place.
377You will also be told if a fuzz factor was used to make the match, in which
378case you should also be slightly suspicious.
379.Ss Filename Determination
380If no original file is specified on the command line,
381.Nm
382will try to figure out from the leading garbage what the name of the file
383to edit is.
384When checking a prospective file name, pathname components are stripped
385as specified by the
386.Fl p
387option and the file's existence and writability are checked relative
388to the current working directory (or the directory specified by the
389.Fl d
390option).
391.Pp
392If the diff is a context or unified diff,
393.Nm
394is able to determine the old and new file names from the diff header.
395For context diffs, the
396.Dq old
397file is specified in the line beginning with
398.Qq ***
399and the
400.Dq new
401file is specified in the line beginning with
402.Qq --- .
403For a unified diff, the
404.Dq old
405file is specified in the line beginning with
406.Qq ---
407and the
408.Dq new
409file is specified in the line beginning with
410.Qq +++ .
411If there is an
412.Qq Index:
413line in the leading garbage (regardless of the diff type),
414.Nm
415will use the file name from that line as the
416.Dq index
417file.
418.Pp
419.Nm
420will choose the file name by performing the following steps, with the first
421match used:
422.Bl -enum
423.It
424If
425.Nm
426is operating in strict
427.St -p1003.2
428mode, the first of the
429.Dq old ,
430.Dq new
431and
432.Dq index
433file names that exist is used.
434Otherwise,
435.Nm
436will examine either the
437.Dq old
438and
439.Dq new
440file names or, for a non-context diff, the
441.Dq index
442file name, and choose the file name with the fewest path components,
443the shortest basename, and the shortest total file name length (in that order).
444.It
445If no file exists,
446.Nm
447checks for the existence of the files in an SCCS or RCS directory
448(using the appropriate prefix or suffix) using the criteria specified
449above.
450If found,
451.Nm
452will attempt to get or check out the file.
453.It
454If no suitable file was found to patch, the patch file is a context or
455unified diff, and the old file was zero length, the new file name is
456created and used.
457.It
458If the file name still cannot be determined,
459.Nm
460will prompt the user for the file name to use.
461.El
462.Pp
463Additionally, if the leading garbage contains a
464.Qq Prereq:\ \&
465line,
466.Nm
467will take the first word from the prerequisites line (normally a version
468number) and check the input file to see if that word can be found.
469If not,
470.Nm
471will ask for confirmation before proceeding.
472.Pp
473The upshot of all this is that you should be able to say, while in a news
474interface, the following:
475.Pp
476.Dl | patch -d /usr/src/local/blurfl
477.Pp
478and patch a file in the blurfl directory directly from the article containing
479the patch.
480.Ss Backup Files
481By default, the patched version is put in place of the original, with
482the original file backed up to the same name with the extension
483.Qq .orig ,
484or as specified by the
485.Fl B ,
486.Fl V ,
487or
488.Fl z
489options.
490The extension used for making backup files may also be specified in the
491.Ev SIMPLE_BACKUP_SUFFIX
492environment variable, which is overridden by the options above.
493.Pp
494If the backup file is a symbolic or hard link to the original file,
495.Nm
496creates a new backup file name by changing the first lowercase letter
497in the last component of the file's name into uppercase.
498If there are no more lowercase letters in the name,
499it removes the first character from the name.
500It repeats this process until it comes up with a
501backup file that does not already exist or is not linked to the original file.
502.Pp
503You may also specify where you want the output to go with the
504.Fl o
505option; if that file already exists, it is backed up first.
506.Ss Notes For Patch Senders
507There are several things you should bear in mind if you are going to
508be sending out patches:
509.Pp
510First, you can save people a lot of grief by keeping a
511.Pa patchlevel.h
512file which is patched to increment the patch level as the first diff in the
513patch file you send out.
514If you put a
515.Qq Prereq:
516line in with the patch, it won't let them apply
517patches out of order without some warning.
518.Pp
519Second, make sure you've specified the file names right, either in a
520context diff header, or with an
521.Qq Index:
522line.
523If you are patching something in a subdirectory, be sure to tell the patch
524user to specify a
525.Fl p
526option as needed.
527.Pp
528Third, you can create a file by sending out a diff that compares a
529null file to the file you want to create.
530This will only work if the file you want to create doesn't exist already in
531the target directory.
532.Pp
533Fourth, take care not to send out reversed patches, since it makes people wonder
534whether they already applied the patch.
535.Pp
536Fifth, while you may be able to get away with putting 582 diff listings into
537one file, it is probably wiser to group related patches into separate files in
538case something goes haywire.
539.Sh ENVIRONMENT
540.Bl -tag -width "PATCH_VERSION_CONTROL" -compact
541.It Ev POSIXLY_CORRECT
542When set,
543.Nm
544behaves as if the
545.Fl Fl posix
546option has been specified.
547.It Ev SIMPLE_BACKUP_SUFFIX
548Extension to use for backup file names instead of
549.Qq .orig .
550.It Ev TMPDIR
551Directory to put temporary files in; default is
552.Pa /tmp .
553.It Ev PATCH_VERSION_CONTROL
554Selects when numbered backup files are made.
555.It Ev VERSION_CONTROL
556Same as
557.Ev PATCH_VERSION_CONTROL .
558.El
559.Sh FILES
560.Bl -tag -width "$TMPDIR/patch*" -compact
561.It Pa $TMPDIR/patch*
562.Nm
563temporary files
564.It Pa /dev/tty
565used to read input when
566.Nm
567prompts the user
568.El
569.Sh DIAGNOSTICS
570Too many to list here, but generally indicative that
571.Nm
572couldn't parse your patch file.
573.Pp
574The message
575.Qq Hmm...
576indicates that there is unprocessed text in the patch file and that
577.Nm
578is attempting to intuit whether there is a patch in that text and, if so,
579what kind of patch it is.
580.Pp
581The
582.Nm
583utility exits with one of the following values:
584.Pp
585.Bl -tag -width Ds -compact -offset indent
586.It \&0
587Successful completion.
588.It \&1
589One or more lines were written to a reject file.
590.It \*[Gt]\&1
591An error occurred.
592.El
593.Pp
594When applying a set of patches in a loop it behooves you to check this
595exit status so you don't apply a later patch to a partially patched file.
596.Sh SEE ALSO
597.Xr diff 1
598.Sh AUTHORS
599.An Larry Wall
600with many other contributors.
601.Sh CAVEATS
602.Nm
603cannot tell if the line numbers are off in an ed script, and can only detect
604bad line numbers in a normal diff when it finds a
605.Qq change
606or a
607.Qq delete
608command.
609A context diff using fuzz factor 3 may have the same problem.
610Until a suitable interactive interface is added, you should probably do
611a context diff in these cases to see if the changes made sense.
612Of course, compiling without errors is a pretty good indication that the patch
613worked, but not always.
614.Pp
615.Nm
616usually produces the correct results, even when it has to do a lot of
617guessing.
618However, the results are guaranteed to be correct only when the patch is
619applied to exactly the same version of the file that the patch was
620generated from.
621.Sh BUGS
622Could be smarter about partial matches, excessively deviant offsets and
623swapped code, but that would take an extra pass.
624.Pp
625Check patch mode
626.Pq Fl C
627will fail if you try to check several patches in succession that build on
628each other.
629The entire
630.Nm
631code would have to be restructured to keep temporary files around so that it
632can handle this situation.
633.Pp
634If code has been duplicated (for instance with #ifdef OLDCODE ... #else ...
635#endif),
636.Nm
637is incapable of patching both versions, and, if it works at all, will likely
638patch the wrong one, and tell you that it succeeded to boot.
639.Pp
640If you apply a patch you've already applied,
641.Nm
642will think it is a reversed patch, and offer to un-apply the patch.
643This could be construed as a feature.