NAME

dwdiff - a delimited word diff program

SYNOPSIS

dwdiff [OPTIONS] OLD FILE NEW FILE
dwdiff [OPTIONS] --diff-input [DIFF FILE]

DESCRIPTION

dwdiff is a diff program that operates at the word level instead of the line level. It is different from wdiff in that it allows the user to specify what should be considered whitespace, and in that it takes an optional list of characters that should be considered delimiters. Delimiters are single characters that are treated as if they are words, even when there is no whitespace separating them from preceding words or delimiters. dwdiff is mostly command-line compatible with wdiff. Only the --autopager, --terminal and --avoid-wraps options are not supported.

The default output from dwdiff is the new text, with the deleted and inserted parts annotated with markers. Command line options are available to change both what is printed, and the markers.

OPTIONS

dwdiff accepts the following options (Note that all strings will first be escape expanded. All standard \-escapes are supported, as well as \u and \U Unicode escapes):

-h, --help: Display a short help message.
-v, --version: Print version and copyright information.
-d delimiters, --delimiters=delimiters: Specify a list of characters to be used as delimiters. The default is empty.
-P, --punctuation: Use punctuation characters as delimiters. The exact set of punctuation characters depends on the current locale.
-W whitespace, --white-space=whitespace: Specify a list of characters to be used as whitespace. The default set of whitespace characters depends on the current locale.
-u, --diff-input: Interpret the input as the output from diff in the Unified Diff format (usually produced by diff -u). In this case only one input file is allowed. This option allows reformating diff output with dwdiff, and is useful for example to post-process the output of svn diff.
-1, --no-deleted: Suppress printing of words deleted from the first file.
-2, --no-inserted: Suppress printing of words inserted in the second file.
-3, --no-common: Suppress printing of words common to both files.
-L[width], --line-numbers[=width]: Show line numbers at the start of each line. The line numbers displayed are the line number in the old file and the line number in the new file respectively. The optional width argument is the minimum number of positions per line number.
-Cnum, --context=num: Show num lines of context before and after each changes. A line with only -- is printed between blocks of changes.
-s, --statistics: Print statistics when done. The numbers printed include the number of words from in both files, the number of deleted words, the number of inserted words, and the number of changed words. The number of changed words is counted as the number of words that are removed from the first file, and the number of words that replace them from the second file. All of these numbers are also expressed as a percentage of the total number of words in the file the words came from.
-i, --ignore-case: Ignore differences in case when comparing words. This option is only available if the diff program that is called provides it.
-I, --ignore-formatting: Ignore differences in formatting of characters. This option switches to using the Unicode compatibility decomposition instead of the canonical decomposition. The compatibility decomposition discards formatting information. For example, the ligature fi will be decomposed into two separate characters for the purposes of comparison. However, also super- and subscript will be regarded equal as well as different rotations of the same character.
-c[spec], --color[=spec]: Color mode. The optional spec can be used to customize the colors. spec consists of [delete],[insert]. If either is omited it will be set to its default color (bright red or bright green respectively). Both parts of the spec consist of [foreground][:background]. To obtain a list of permissible color names, use the word ``list'' as spec. Alternatively, you can specify any escape sequence to set attributes as a color by prepending e:.

: The standard markers for the begin and end of deleted and inserted text are suppressed, but any markers specified on the command line will still be printed.

-l, --less-mode: As -p but also overstrike deleted whitespace.
-p, --printer: Use overstriking with an underscore and bold text to emphasize changes. This is implemented by first printing the underscore or a duplicate of the character to be printed, followed by a backspace, followed by the character. On regular terminals you won't see any effect. The less(1) command will however show underlined and bold text.

: The standard markers for the begin and end of deleted and inserted text are suppressed, but any markers specified on the command line will still be printed.

-mnum, --match-context=num: Use num words of context before and after words for matching. Words in the old text will then only match words in the new text if words surrounding them are also equal. This improves the output for dwdiff for large changes with frequently occurring words. However, using context requires more disk space and more processing time. The default value is 1. Set this option to 0 to revert to the pre 1.5 behavior.
--aggregate-changes: Allow multiple close changes to be treated as one change, if context words are used (see --match-context). This option reduces the processing time as the changes reported by the diff program are not post-processed to give more precise results.
-A algorithm, --algorithm=algorithm: Select the algorithm to be used for determining differences. There are three possible values for algorithm: best, which tries to find the minimal set of changes, normal, which trades some optimality for speed, and fast, which assumes that the input is large and contains few changes. By default the normal algorithm is used.
-S[marker], --paragraph-separator[=marker]: Show insertion or deletion of blocks of lines with only whitespace characters. A special marker is inserted into the output to indicate these blocks. The default marker is --.
--wdiff-output: Create wdiff compatible output. The dwdiff program uses a different output algorithm, which provides a more intuitive output.
-w string, --start-delete=string: Specify a string to mark begin of deleted text. The default is [-.
-x string, --stop-delete=string: Specify a string to mark end of deleted text. The default is -].
-y string, --start-insert=string: Specify a string to mark begin of inserted text. The default is {+.
-z string, --stop-insert=string: Specify a string to mark end of inserted text. The default is +}.
-R, --repeat-markers: Repeat the begin and end markers at the start and end of line if a change crosses a newline.
--profile=name: Load the profile with name name from the .dwdiffrc file in the user's home directory. Only the last profile named is actually read, unless it is followed by a --no-profile option. By default the profile named default is read.
--no-profile: Disable reading of profiles, unless followed by another --profile option. This also disables reading the default profile.

A single dash (-) as a file can be used to denote standard input. Only one file can be read from standard input. To stop dwdiff from interpreting file names that start with a dash as options, one can specify a double dash (--) after which dwdiff will interpret any following arguments as files to read.

The exit status of dwdiff indicates the result of the comparison: 0 if the files are the same, 1 if the files are different. Should an error occur, dwdiff will exit with status 2.

PROFILES

Since version 2.1.0, dwdiff allows users to create profiles. A profile is a set of options that can be loaded using the --profile option. Profiles are specified in the .dwdiffrc in the user's home directory. The format of the .dwdiffrc file is as follows:

A line contains either a profile header (a text enclosed in []) or a long option name without leading dashes, optionally followed by it's argument. Arguments are stripped of whitespace surrounding it. If an argument starts and ends with matching double or single quotes, these are removed as well. Comments can be included by starting a line with a hash mark (#).

If no profile is specified, the profile named default will be loaded, unless the --no-profile option is specified. Here is an example .dwdiffrc:

[default]
color red,blue
punctuation

[html]
start-delete <span style="color=red">
stop-delete </span>
start-insert <span style="color=green">
stop-insert </span>

[space-only]
# Use only a space as whitespace.
white-space " "

BUGS

If you think you have found a bug, please check that you are using the latest version of dwdiff <http://os.ghalkes.nl/dwdiff.html>. When reporting bugs, please include a minimal example that demonstrates the problem.

AUTHOR

G.P. Halkes <dwdiff@ghalkes.nl>

COPYRIGHT

Copyright © 2006-2017 G.P. Halkes and others
dwdiff is licensed under the GNU General Public License version 3.
For more details on the license, see the file COPYING in the documentation directory. On Un*x systems this is usually /usr/share/doc/dwdiff-$VERSION$.