diff --git a/csvkit/utilities/csvdiff.py b/csvkit/utilities/csvdiff.py
new file mode 100644
index 0000000000000000000000000000000000000000..4b5fefb450b48cf2d8c3fd2999475e27ad354867
--- /dev/null
+++ b/csvkit/utilities/csvdiff.py
@@ -0,0 +1,78 @@
+#!/usr/bin/env python
+
+import sys
+import daff
+
+from csvkit import CSVKitReader, CSVKitWriter
+from csvkit.cli import CSVKitUtility, match_column_identifier
+
+class CSVDiff(CSVKitUtility):
+ description = 'Compare CSV files. Like unix "diff" command, but for tabular data.'
+ epilog = 'Note that the diff operation requires reading all files into memory. Don\'t try this on very large files.'
+ override_flags = ['f', 'H']
+
+ def add_arguments(self):
+ self.argparser.add_argument(metavar="FILE", nargs='*', dest='input_paths', default=['-'],
+ help='The CSV files to operate on.')
+ self.argparser.add_argument('-c', '--columns', dest='columns',
+ help='The column name(s) to use for comparison. Should be either one name or a comma-separated list of names. May also be left unspecified, in which case we\'ll guess something plausible.')
+ self.argparser.add_argument('--color', dest='color', action='store_true',
+ help='Decorate output with colors and glyphs.')
+
+ def main(self):
+ self.input_files = []
+
+ for path in self.args.input_paths:
+ self.input_files.append(self._open_input_file(path))
+
+ ct = len(self.input_files)
+ if ct < 2:
+ self.argparser.error('You must specify two or three files to compare.\n(If three, the first file should be a common ancestor of the remaining two)')
+ if ct > 3:
+ self.argparser.error('You must specify at most three files to compare.')
+
+ match_column_names = []
+ if self.args.columns:
+ match_column_names = self._parse_match_column_names(self.args.columns)
+
+ tables = []
+
+ for f in self.input_files:
+ tables.append(list(CSVKitReader(f, **self.reader_kwargs)))
+ f.close()
+
+ flags = daff.CompareFlags()
+ for c in match_column_names:
+ flags.addPrimaryKey(c)
+
+ result = []
+ tab = daff.PythonTableView(result)
+ tab1 = daff.PythonTableView(tables[ct-2])
+ tab2 = daff.PythonTableView(tables[ct-1])
+ if ct == 3:
+ tab0 = daff.PythonTableView(tables[0])
+ alignment = daff.Coopy.compareTables3(tab0,tab1,tab2,flags).align()
+ else:
+ alignment = daff.Coopy.compareTables(tab1,tab2,flags).align()
+ daff.TableDiff(alignment,flags).hilite(tab)
+
+ if self.args.color or (self.output_file == sys.stdout and sys.stdout.isatty()):
+ self.output_file.write(daff.TerminalDiffRender().render(tab).encode('utf-8'))
+ else:
+ output = CSVKitWriter(self.output_file, **self.writer_kwargs)
+ for row in result:
+ output.writerow(row)
+
+ def _parse_match_column_names(self, join_string):
+ """
+ Parse a list of match columns.
+ """
+ return list(map(str.strip, join_string.split(',')))
+
+
+def launch_new_instance():
+ utility = CSVDiff()
+ utility.main()
+
+if __name__ == "__main__":
+ launch_new_instance()
diff --git a/docs/cli.rst b/docs/cli.rst
index cf9dca80f0c34a48589352ebc896cc21a7f6bc18..aaf9903b1005b8d9bed945bb213ce276c81f8d3a 100644
--- a/docs/cli.rst
+++ b/docs/cli.rst
@@ -32,6 +32,7 @@ Output (and Analysis)
.. toctree::
:maxdepth: 1
+ scripts/csvdiff
scripts/csvformat
scripts/csvjson
scripts/csvlook
diff --git a/docs/scripts/csvdiff.rst b/docs/scripts/csvdiff.rst
new file mode 100644
index 0000000000000000000000000000000000000000..2c5b253be0baea5e32bb75debea6023b5becdcac
--- /dev/null
+++ b/docs/scripts/csvdiff.rst
@@ -0,0 +1,58 @@
+=======
+csvdiff
+=======
+
+Description
+===========
+
+Compare two CSV tables, and produce another CSV table that summarizes their differences.::
+
+ usage: csvdiff [-h] [-d DELIMITER] [-t] [-q QUOTECHAR] [-u {0,1,2,3}] [-b]
+ [-p ESCAPECHAR] [-z MAXFIELDSIZE] [-e ENCODING] [-S] [-v] [-l]
+ [--zero] [-c COLUMNS] [--color]
+ [FILE [FILE ...]]
+
+ Compare CSV files. Like unix "diff" command, but for tabular data.
+
+ positional arguments:
+ FILE The CSV files to operate on.
+
+ optional arguments:
+ -h, --help show this help message and exit
+ -c COLUMNS, --columns COLUMNS
+ The column name(s) to use for comparison. Should be
+ either one name or a comma-separated list of names.
+ May also be left unspecified, in which case we'll
+ guess something plausible.
+ --color Decorate output with colors and glyphs.
+
+ Note that the diff operation requires reading all files into memory. Don't try
+ this on very large files.
+
+
+See also: :doc:`../common_arguments`.
+
+Examples
+========
+
+::
+
+ csvdiff file1.csv file2.csv
+
+This command says you have two files to compare, file1.csv and file2.csv.
+A typical output will look like this:
+
+.. raw:: html
+
+ <p style="font-family: monospace;">
+ <span style="font-weight:bold">@@</span> ,<span style="font-weight:bold">bridge</span> ,<span style="font-weight:bold">designer</span> ,<span style="font-weight:bold">length</span><br/>
+ ,Brooklyn ,"J. A. Roebling" ,1595<br/>
+ <span style="color:green"></span><span style="color:green;font-weight:bold">+++</span>,<span style="color:green"></span><span style="color:green;font-weight:bold">Manhattan</span> ,"<span style="color:green"></span><span style="color:green;font-weight:bold">G. Lindenthal</span>" ,<span style="color:green"></span><span style="color:green;font-weight:bold">1470</span><br/>
+ <span style="color:red"></span><span style="color:red;font-weight:bold"></span><span style="color:blue;font-weight:bold"></span><span style="color:blue;font-weight:bold">→</span><span style="color:green;font-weight:bold"></span><span style="color:green;font-weight:bold"></span> ,Williamsburg ,"<span style="color:red"></span><span style="color:red;font-weight:bold">D. Duck</span><span style="color:blue;font-weight:bold"></span><span style="color:blue;font-weight:bold">→</span><span style="color:green;font-weight:bold"></span><span style="color:green;font-weight:bold">L. L. Buck</span>" ,1600<br/>
+ ,Queensborough ,"Palmer & Hornbostel",1182<br/>
+ ...,... ,... ,...<br/>
+ ,"George Washington","O. H. Ammann" ,3500<br/>
+ <span style="color:red"></span><span style="color:red;font-weight:bold">---</span>,<span style="color:red"></span><span style="color:red;font-weight:bold">Spamspan</span> ,"<span style="color:red"></span><span style="color:red;font-weight:bold">S. Spamington</span>" ,<span style="color:red"></span><span style="color:red;font-weight:bold">10000</span>
+ </p>
+
+See http://dataprotocols.org/tabular-diff-format/ for information on the diff format. Color highlighting will by default only activate when the result is being shown on a console.
diff --git a/requirements-py2.txt b/requirements-py2.txt
index f3fad914d193f598252bfd7444f350ffa63fff00..2bcec1cca8d7397d0abeb94e739d42c6b3e81621 100644
--- a/requirements-py2.txt
+++ b/requirements-py2.txt
@@ -13,3 +13,4 @@ six>=1.6.1
ordereddict>=1.1
simplejson>=3.6.3
sphinx_rtd_theme
+daff
diff --git a/requirements-py3.txt b/requirements-py3.txt
index 3c26240ef0cb97b6290b1f6c8d93d9c45c7d7939..43985c53a134d7c875e3e6df98cf8a346ea0dd69 100644
--- a/requirements-py3.txt
+++ b/requirements-py3.txt
@@ -8,3 +8,4 @@ openpyxl>=2.0.3
tox>=1.3
six>=1.6.1
sphinx_rtd_theme
+daff
diff --git a/setup.py b/setup.py
index 04ae53f8834854550c6ed16dd04d314f0538fc42..dc3b7015852ef8ebf2d9f0f247b5012c2f670973 100644
--- a/setup.py
+++ b/setup.py
@@ -57,6 +57,7 @@ setup(
'console_scripts': [
'csvclean = csvkit.utilities.csvclean:launch_new_instance',
'csvcut = csvkit.utilities.csvcut:launch_new_instance',
+ 'csvdiff = csvkit.utilities.csvdiff:launch_new_instance',
'csvformat = csvkit.utilities.csvformat:launch_new_instance',
'csvgrep = csvkit.utilities.csvgrep:launch_new_instance',
'csvjoin = csvkit.utilities.csvjoin:launch_new_instance',
diff --git a/tests/test_utilities/test_csvdiff.py b/tests/test_utilities/test_csvdiff.py
new file mode 100644
index 0000000000000000000000000000000000000000..997a60605364d116b169ba90cd91d90dc8e0bf1a
--- /dev/null
+++ b/tests/test_utilities/test_csvdiff.py
@@ -0,0 +1,22 @@
+#!/usr/bin/env python
+
+import six
+
+try:
+ import unittest2 as unittest
+except ImportError:
+ import unittest
+
+from csvkit.utilities.csvdiff import CSVDiff
+
+class TestCSVJoin(unittest.TestCase):
+ def test_basic(self):
+ args = ['examples/join_a.csv', 'examples/join_b.csv']
+ output_file = six.StringIO()
+
+ utility = CSVDiff(args, output_file)
+ utility.main()
+
+ output = six.StringIO(output_file.getvalue())
+
+ self.assertEqual(len(output.readlines()), 6)