• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1:mod:`difflib` --- Helpers for computing deltas
2===============================================
3
4.. module:: difflib
5   :synopsis: Helpers for computing differences between objects.
6.. moduleauthor:: Tim Peters <tim_one@users.sourceforge.net>
7.. sectionauthor:: Tim Peters <tim_one@users.sourceforge.net>
8.. Markup by Fred L. Drake, Jr. <fdrake@acm.org>
9
10.. testsetup::
11
12   import sys
13   from difflib import *
14
15.. versionadded:: 2.1
16
17This module provides classes and functions for comparing sequences. It
18can be used for example, for comparing files, and can produce difference
19information in various formats, including HTML and context and unified
20diffs. For comparing directories and files, see also, the :mod:`filecmp` module.
21
22.. class:: SequenceMatcher
23
24   This is a flexible class for comparing pairs of sequences of any type, so long
25   as the sequence elements are :term:`hashable`.  The basic algorithm predates, and is a
26   little fancier than, an algorithm published in the late 1980's by Ratcliff and
27   Obershelp under the hyperbolic name "gestalt pattern matching."  The idea is to
28   find the longest contiguous matching subsequence that contains no "junk"
29   elements (the Ratcliff and Obershelp algorithm doesn't address junk).  The same
30   idea is then applied recursively to the pieces of the sequences to the left and
31   to the right of the matching subsequence.  This does not yield minimal edit
32   sequences, but does tend to yield matches that "look right" to people.
33
34   **Timing:** The basic Ratcliff-Obershelp algorithm is cubic time in the worst
35   case and quadratic time in the expected case. :class:`SequenceMatcher` is
36   quadratic time for the worst case and has expected-case behavior dependent in a
37   complicated way on how many elements the sequences have in common; best case
38   time is linear.
39
40   **Automatic junk heuristic:** :class:`SequenceMatcher` supports a heuristic that
41   automatically treats certain sequence items as junk. The heuristic counts how many
42   times each individual item appears in the sequence. If an item's duplicates (after
43   the first one) account for more than 1% of the sequence and the sequence is at least
44   200 items long, this item is marked as "popular" and is treated as junk for
45   the purpose of sequence matching. This heuristic can be turned off by setting
46   the ``autojunk`` argument to ``False`` when creating the :class:`SequenceMatcher`.
47
48   .. versionadded:: 2.7.1
49      The *autojunk* parameter.
50
51.. class:: Differ
52
53   This is a class for comparing sequences of lines of text, and producing
54   human-readable differences or deltas.  Differ uses :class:`SequenceMatcher`
55   both to compare sequences of lines, and to compare sequences of characters
56   within similar (near-matching) lines.
57
58   Each line of a :class:`Differ` delta begins with a two-letter code:
59
60   +----------+-------------------------------------------+
61   | Code     | Meaning                                   |
62   +==========+===========================================+
63   | ``'- '`` | line unique to sequence 1                 |
64   +----------+-------------------------------------------+
65   | ``'+ '`` | line unique to sequence 2                 |
66   +----------+-------------------------------------------+
67   | ``'  '`` | line common to both sequences             |
68   +----------+-------------------------------------------+
69   | ``'? '`` | line not present in either input sequence |
70   +----------+-------------------------------------------+
71
72   Lines beginning with '``?``' attempt to guide the eye to intraline differences,
73   and were not present in either input sequence. These lines can be confusing if
74   the sequences contain tab characters.
75
76
77.. class:: HtmlDiff
78
79   This class can be used to create an HTML table (or a complete HTML file
80   containing the table) showing a side by side, line by line comparison of text
81   with inter-line and intra-line change highlights.  The table can be generated in
82   either full or contextual difference mode.
83
84   The constructor for this class is:
85
86
87   .. function:: __init__(tabsize=8, wrapcolumn=None, linejunk=None, charjunk=IS_CHARACTER_JUNK)
88
89      Initializes instance of :class:`HtmlDiff`.
90
91      *tabsize* is an optional keyword argument to specify tab stop spacing and
92      defaults to ``8``.
93
94      *wrapcolumn* is an optional keyword to specify column number where lines are
95      broken and wrapped, defaults to ``None`` where lines are not wrapped.
96
97      *linejunk* and *charjunk* are optional keyword arguments passed into :func:`ndiff`
98      (used by :class:`HtmlDiff` to generate the side by side HTML differences).  See
99      :func:`ndiff` documentation for argument default values and descriptions.
100
101   The following methods are public:
102
103
104   .. function:: make_file(fromlines, tolines [, fromdesc][, todesc][, context][, numlines])
105
106      Compares *fromlines* and *tolines* (lists of strings) and returns a string which
107      is a complete HTML file containing a table showing line by line differences with
108      inter-line and intra-line changes highlighted.
109
110      *fromdesc* and *todesc* are optional keyword arguments to specify from/to file
111      column header strings (both default to an empty string).
112
113      *context* and *numlines* are both optional keyword arguments. Set *context* to
114      ``True`` when contextual differences are to be shown, else the default is
115      ``False`` to show the full files. *numlines* defaults to ``5``.  When *context*
116      is ``True`` *numlines* controls the number of context lines which surround the
117      difference highlights.  When *context* is ``False`` *numlines* controls the
118      number of lines which are shown before a difference highlight when using the
119      "next" hyperlinks (setting to zero would cause the "next" hyperlinks to place
120      the next difference highlight at the top of the browser without any leading
121      context).
122
123
124   .. function:: make_table(fromlines, tolines [, fromdesc][, todesc][, context][, numlines])
125
126      Compares *fromlines* and *tolines* (lists of strings) and returns a string which
127      is a complete HTML table showing line by line differences with inter-line and
128      intra-line changes highlighted.
129
130      The arguments for this method are the same as those for the :meth:`make_file`
131      method.
132
133   :file:`Tools/scripts/diff.py` is a command-line front-end to this class and
134   contains a good example of its use.
135
136   .. versionadded:: 2.4
137
138
139.. function:: context_diff(a, b[, fromfile][, tofile][, fromfiledate][, tofiledate][, n][, lineterm])
140
141   Compare *a* and *b* (lists of strings); return a delta (a :term:`generator`
142   generating the delta lines) in context diff format.
143
144   Context diffs are a compact way of showing just the lines that have changed plus
145   a few lines of context.  The changes are shown in a before/after style.  The
146   number of context lines is set by *n* which defaults to three.
147
148   By default, the diff control lines (those with ``***`` or ``---``) are created
149   with a trailing newline.  This is helpful so that inputs created from
150   :func:`file.readlines` result in diffs that are suitable for use with
151   :func:`file.writelines` since both the inputs and outputs have trailing
152   newlines.
153
154   For inputs that do not have trailing newlines, set the *lineterm* argument to
155   ``""`` so that the output will be uniformly newline free.
156
157   The context diff format normally has a header for filenames and modification
158   times.  Any or all of these may be specified using strings for *fromfile*,
159   *tofile*, *fromfiledate*, and *tofiledate*.  The modification times are normally
160   expressed in the ISO 8601 format. If not specified, the
161   strings default to blanks.
162
163      >>> s1 = ['bacon\n', 'eggs\n', 'ham\n', 'guido\n']
164      >>> s2 = ['python\n', 'eggy\n', 'hamster\n', 'guido\n']
165      >>> for line in context_diff(s1, s2, fromfile='before.py', tofile='after.py'):
166      ...     sys.stdout.write(line)  # doctest: +NORMALIZE_WHITESPACE
167      *** before.py
168      --- after.py
169      ***************
170      *** 1,4 ****
171      ! bacon
172      ! eggs
173      ! ham
174        guido
175      --- 1,4 ----
176      ! python
177      ! eggy
178      ! hamster
179        guido
180
181   See :ref:`difflib-interface` for a more detailed example.
182
183   .. versionadded:: 2.3
184
185
186.. function:: get_close_matches(word, possibilities[, n][, cutoff])
187
188   Return a list of the best "good enough" matches.  *word* is a sequence for which
189   close matches are desired (typically a string), and *possibilities* is a list of
190   sequences against which to match *word* (typically a list of strings).
191
192   Optional argument *n* (default ``3``) is the maximum number of close matches to
193   return; *n* must be greater than ``0``.
194
195   Optional argument *cutoff* (default ``0.6``) is a float in the range [0, 1].
196   Possibilities that don't score at least that similar to *word* are ignored.
197
198   The best (no more than *n*) matches among the possibilities are returned in a
199   list, sorted by similarity score, most similar first.
200
201      >>> get_close_matches('appel', ['ape', 'apple', 'peach', 'puppy'])
202      ['apple', 'ape']
203      >>> import keyword
204      >>> get_close_matches('wheel', keyword.kwlist)
205      ['while']
206      >>> get_close_matches('apple', keyword.kwlist)
207      []
208      >>> get_close_matches('accept', keyword.kwlist)
209      ['except']
210
211
212.. function:: ndiff(a, b[, linejunk][, charjunk])
213
214   Compare *a* and *b* (lists of strings); return a :class:`Differ`\ -style
215   delta (a :term:`generator` generating the delta lines).
216
217   Optional keyword parameters *linejunk* and *charjunk* are for filter functions
218   (or ``None``):
219
220   *linejunk*: A function that accepts a single string argument, and returns true
221   if the string is junk, or false if not. The default is (``None``), starting with
222   Python 2.3.  Before then, the default was the module-level function
223   :func:`IS_LINE_JUNK`, which filters out lines without visible characters, except
224   for at most one pound character (``'#'``). As of Python 2.3, the underlying
225   :class:`SequenceMatcher` class does a dynamic analysis of which lines are so
226   frequent as to constitute noise, and this usually works better than the pre-2.3
227   default.
228
229   *charjunk*: A function that accepts a character (a string of length 1), and
230   returns if the character is junk, or false if not. The default is module-level
231   function :func:`IS_CHARACTER_JUNK`, which filters out whitespace characters (a
232   blank or tab; note: bad idea to include newline in this!).
233
234   :file:`Tools/scripts/ndiff.py` is a command-line front-end to this function.
235
236      >>> diff = ndiff('one\ntwo\nthree\n'.splitlines(1),
237      ...              'ore\ntree\nemu\n'.splitlines(1))
238      >>> print ''.join(diff),
239      - one
240      ?  ^
241      + ore
242      ?  ^
243      - two
244      - three
245      ?  -
246      + tree
247      + emu
248
249
250.. function:: restore(sequence, which)
251
252   Return one of the two sequences that generated a delta.
253
254   Given a *sequence* produced by :meth:`Differ.compare` or :func:`ndiff`, extract
255   lines originating from file 1 or 2 (parameter *which*), stripping off line
256   prefixes.
257
258   Example:
259
260      >>> diff = ndiff('one\ntwo\nthree\n'.splitlines(1),
261      ...              'ore\ntree\nemu\n'.splitlines(1))
262      >>> diff = list(diff) # materialize the generated delta into a list
263      >>> print ''.join(restore(diff, 1)),
264      one
265      two
266      three
267      >>> print ''.join(restore(diff, 2)),
268      ore
269      tree
270      emu
271
272
273.. function:: unified_diff(a, b[, fromfile][, tofile][, fromfiledate][, tofiledate][, n][, lineterm])
274
275   Compare *a* and *b* (lists of strings); return a delta (a :term:`generator`
276   generating the delta lines) in unified diff format.
277
278   Unified diffs are a compact way of showing just the lines that have changed plus
279   a few lines of context.  The changes are shown in an inline style (instead of
280   separate before/after blocks).  The number of context lines is set by *n* which
281   defaults to three.
282
283   By default, the diff control lines (those with ``---``, ``+++``, or ``@@``) are
284   created with a trailing newline.  This is helpful so that inputs created from
285   :func:`file.readlines` result in diffs that are suitable for use with
286   :func:`file.writelines` since both the inputs and outputs have trailing
287   newlines.
288
289   For inputs that do not have trailing newlines, set the *lineterm* argument to
290   ``""`` so that the output will be uniformly newline free.
291
292   The context diff format normally has a header for filenames and modification
293   times.  Any or all of these may be specified using strings for *fromfile*,
294   *tofile*, *fromfiledate*, and *tofiledate*.  The modification times are normally
295   expressed in the ISO 8601 format. If not specified, the
296   strings default to blanks.
297
298      >>> s1 = ['bacon\n', 'eggs\n', 'ham\n', 'guido\n']
299      >>> s2 = ['python\n', 'eggy\n', 'hamster\n', 'guido\n']
300      >>> for line in unified_diff(s1, s2, fromfile='before.py', tofile='after.py'):
301      ...     sys.stdout.write(line)   # doctest: +NORMALIZE_WHITESPACE
302      --- before.py
303      +++ after.py
304      @@ -1,4 +1,4 @@
305      -bacon
306      -eggs
307      -ham
308      +python
309      +eggy
310      +hamster
311       guido
312
313   See :ref:`difflib-interface` for a more detailed example.
314
315   .. versionadded:: 2.3
316
317
318.. function:: IS_LINE_JUNK(line)
319
320   Return true for ignorable lines.  The line *line* is ignorable if *line* is
321   blank or contains a single ``'#'``, otherwise it is not ignorable.  Used as a
322   default for parameter *linejunk* in :func:`ndiff` before Python 2.3.
323
324
325.. function:: IS_CHARACTER_JUNK(ch)
326
327   Return true for ignorable characters.  The character *ch* is ignorable if *ch*
328   is a space or tab, otherwise it is not ignorable.  Used as a default for
329   parameter *charjunk* in :func:`ndiff`.
330
331
332.. seealso::
333
334   `Pattern Matching: The Gestalt Approach <http://www.drdobbs.com/database/pattern-matching-the-gestalt-approach/184407970>`_
335      Discussion of a similar algorithm by John W. Ratcliff and D. E. Metzener. This
336      was published in `Dr. Dobb's Journal <http://www.drdobbs.com/>`_ in July, 1988.
337
338
339.. _sequence-matcher:
340
341SequenceMatcher Objects
342-----------------------
343
344The :class:`SequenceMatcher` class has this constructor:
345
346
347.. class:: SequenceMatcher(isjunk=None, a='', b='', autojunk=True)
348
349   Optional argument *isjunk* must be ``None`` (the default) or a one-argument
350   function that takes a sequence element and returns true if and only if the
351   element is "junk" and should be ignored. Passing ``None`` for *isjunk* is
352   equivalent to passing ``lambda x: 0``; in other words, no elements are ignored.
353   For example, pass::
354
355      lambda x: x in " \t"
356
357   if you're comparing lines as sequences of characters, and don't want to synch up
358   on blanks or hard tabs.
359
360   The optional arguments *a* and *b* are sequences to be compared; both default to
361   empty strings.  The elements of both sequences must be :term:`hashable`.
362
363   The optional argument *autojunk* can be used to disable the automatic junk
364   heuristic.
365
366   .. versionadded:: 2.7.1
367      The *autojunk* parameter.
368
369   :class:`SequenceMatcher` objects have the following methods:
370
371   .. method:: set_seqs(a, b)
372
373      Set the two sequences to be compared.
374
375   :class:`SequenceMatcher` computes and caches detailed information about the
376   second sequence, so if you want to compare one sequence against many
377   sequences, use :meth:`set_seq2` to set the commonly used sequence once and
378   call :meth:`set_seq1` repeatedly, once for each of the other sequences.
379
380
381   .. method:: set_seq1(a)
382
383      Set the first sequence to be compared.  The second sequence to be compared
384      is not changed.
385
386
387   .. method:: set_seq2(b)
388
389      Set the second sequence to be compared.  The first sequence to be compared
390      is not changed.
391
392
393   .. method:: find_longest_match(alo, ahi, blo, bhi)
394
395      Find longest matching block in ``a[alo:ahi]`` and ``b[blo:bhi]``.
396
397      If *isjunk* was omitted or ``None``, :meth:`find_longest_match` returns
398      ``(i, j, k)`` such that ``a[i:i+k]`` is equal to ``b[j:j+k]``, where ``alo
399      <= i <= i+k <= ahi`` and ``blo <= j <= j+k <= bhi``. For all ``(i', j',
400      k')`` meeting those conditions, the additional conditions ``k >= k'``, ``i
401      <= i'``, and if ``i == i'``, ``j <= j'`` are also met. In other words, of
402      all maximal matching blocks, return one that starts earliest in *a*, and
403      of all those maximal matching blocks that start earliest in *a*, return
404      the one that starts earliest in *b*.
405
406         >>> s = SequenceMatcher(None, " abcd", "abcd abcd")
407         >>> s.find_longest_match(0, 5, 0, 9)
408         Match(a=0, b=4, size=5)
409
410      If *isjunk* was provided, first the longest matching block is determined
411      as above, but with the additional restriction that no junk element appears
412      in the block.  Then that block is extended as far as possible by matching
413      (only) junk elements on both sides. So the resulting block never matches
414      on junk except as identical junk happens to be adjacent to an interesting
415      match.
416
417      Here's the same example as before, but considering blanks to be junk. That
418      prevents ``' abcd'`` from matching the ``' abcd'`` at the tail end of the
419      second sequence directly.  Instead only the ``'abcd'`` can match, and
420      matches the leftmost ``'abcd'`` in the second sequence:
421
422         >>> s = SequenceMatcher(lambda x: x==" ", " abcd", "abcd abcd")
423         >>> s.find_longest_match(0, 5, 0, 9)
424         Match(a=1, b=0, size=4)
425
426      If no blocks match, this returns ``(alo, blo, 0)``.
427
428      .. versionchanged:: 2.6
429         This method returns a :term:`named tuple` ``Match(a, b, size)``.
430
431
432   .. method:: get_matching_blocks()
433
434      Return list of triples describing matching subsequences. Each triple is of
435      the form ``(i, j, n)``, and means that ``a[i:i+n] == b[j:j+n]``.  The
436      triples are monotonically increasing in *i* and *j*.
437
438      The last triple is a dummy, and has the value ``(len(a), len(b), 0)``.  It
439      is the only triple with ``n == 0``.  If ``(i, j, n)`` and ``(i', j', n')``
440      are adjacent triples in the list, and the second is not the last triple in
441      the list, then ``i+n != i'`` or ``j+n != j'``; in other words, adjacent
442      triples always describe non-adjacent equal blocks.
443
444      .. XXX Explain why a dummy is used!
445
446      .. versionchanged:: 2.5
447         The guarantee that adjacent triples always describe non-adjacent blocks
448         was implemented.
449
450      .. doctest::
451
452         >>> s = SequenceMatcher(None, "abxcd", "abcd")
453         >>> s.get_matching_blocks()
454         [Match(a=0, b=0, size=2), Match(a=3, b=2, size=2), Match(a=5, b=4, size=0)]
455
456
457   .. method:: get_opcodes()
458
459      Return list of 5-tuples describing how to turn *a* into *b*. Each tuple is
460      of the form ``(tag, i1, i2, j1, j2)``.  The first tuple has ``i1 == j1 ==
461      0``, and remaining tuples have *i1* equal to the *i2* from the preceding
462      tuple, and, likewise, *j1* equal to the previous *j2*.
463
464      The *tag* values are strings, with these meanings:
465
466      +---------------+---------------------------------------------+
467      | Value         | Meaning                                     |
468      +===============+=============================================+
469      | ``'replace'`` | ``a[i1:i2]`` should be replaced by          |
470      |               | ``b[j1:j2]``.                               |
471      +---------------+---------------------------------------------+
472      | ``'delete'``  | ``a[i1:i2]`` should be deleted.  Note that  |
473      |               | ``j1 == j2`` in this case.                  |
474      +---------------+---------------------------------------------+
475      | ``'insert'``  | ``b[j1:j2]`` should be inserted at          |
476      |               | ``a[i1:i1]``. Note that ``i1 == i2`` in     |
477      |               | this case.                                  |
478      +---------------+---------------------------------------------+
479      | ``'equal'``   | ``a[i1:i2] == b[j1:j2]`` (the sub-sequences |
480      |               | are equal).                                 |
481      +---------------+---------------------------------------------+
482
483      For example:
484
485         >>> a = "qabxcd"
486         >>> b = "abycdf"
487         >>> s = SequenceMatcher(None, a, b)
488         >>> for tag, i1, i2, j1, j2 in s.get_opcodes():
489         ...    print ("%7s a[%d:%d] (%s) b[%d:%d] (%s)" %
490         ...           (tag, i1, i2, a[i1:i2], j1, j2, b[j1:j2]))
491          delete a[0:1] (q) b[0:0] ()
492           equal a[1:3] (ab) b[0:2] (ab)
493         replace a[3:4] (x) b[2:3] (y)
494           equal a[4:6] (cd) b[3:5] (cd)
495          insert a[6:6] () b[5:6] (f)
496
497
498   .. method:: get_grouped_opcodes([n])
499
500      Return a :term:`generator` of groups with up to *n* lines of context.
501
502      Starting with the groups returned by :meth:`get_opcodes`, this method
503      splits out smaller change clusters and eliminates intervening ranges which
504      have no changes.
505
506      The groups are returned in the same format as :meth:`get_opcodes`.
507
508      .. versionadded:: 2.3
509
510
511   .. method:: ratio()
512
513      Return a measure of the sequences' similarity as a float in the range [0,
514      1].
515
516      Where T is the total number of elements in both sequences, and M is the
517      number of matches, this is 2.0\*M / T. Note that this is ``1.0`` if the
518      sequences are identical, and ``0.0`` if they have nothing in common.
519
520      This is expensive to compute if :meth:`get_matching_blocks` or
521      :meth:`get_opcodes` hasn't already been called, in which case you may want
522      to try :meth:`quick_ratio` or :meth:`real_quick_ratio` first to get an
523      upper bound.
524
525
526   .. method:: quick_ratio()
527
528      Return an upper bound on :meth:`ratio` relatively quickly.
529
530
531   .. method:: real_quick_ratio()
532
533      Return an upper bound on :meth:`ratio` very quickly.
534
535
536The three methods that return the ratio of matching to total characters can give
537different results due to differing levels of approximation, although
538:meth:`quick_ratio` and :meth:`real_quick_ratio` are always at least as large as
539:meth:`ratio`:
540
541   >>> s = SequenceMatcher(None, "abcd", "bcde")
542   >>> s.ratio()
543   0.75
544   >>> s.quick_ratio()
545   0.75
546   >>> s.real_quick_ratio()
547   1.0
548
549
550.. _sequencematcher-examples:
551
552SequenceMatcher Examples
553------------------------
554
555This example compares two strings, considering blanks to be "junk:"
556
557   >>> s = SequenceMatcher(lambda x: x == " ",
558   ...                     "private Thread currentThread;",
559   ...                     "private volatile Thread currentThread;")
560
561:meth:`ratio` returns a float in [0, 1], measuring the similarity of the
562sequences.  As a rule of thumb, a :meth:`ratio` value over 0.6 means the
563sequences are close matches:
564
565   >>> print round(s.ratio(), 3)
566   0.866
567
568If you're only interested in where the sequences match,
569:meth:`get_matching_blocks` is handy:
570
571   >>> for block in s.get_matching_blocks():
572   ...     print "a[%d] and b[%d] match for %d elements" % block
573   a[0] and b[0] match for 8 elements
574   a[8] and b[17] match for 21 elements
575   a[29] and b[38] match for 0 elements
576
577Note that the last tuple returned by :meth:`get_matching_blocks` is always a
578dummy, ``(len(a), len(b), 0)``, and this is the only case in which the last
579tuple element (number of elements matched) is ``0``.
580
581If you want to know how to change the first sequence into the second, use
582:meth:`get_opcodes`:
583
584   >>> for opcode in s.get_opcodes():
585   ...     print "%6s a[%d:%d] b[%d:%d]" % opcode
586    equal a[0:8] b[0:8]
587   insert a[8:8] b[8:17]
588    equal a[8:29] b[17:38]
589
590.. seealso::
591
592   * The :func:`get_close_matches` function in this module which shows how
593     simple code building on :class:`SequenceMatcher` can be used to do useful
594     work.
595
596   * `Simple version control recipe
597     <https://code.activestate.com/recipes/576729/>`_ for a small application
598     built with :class:`SequenceMatcher`.
599
600
601.. _differ-objects:
602
603Differ Objects
604--------------
605
606Note that :class:`Differ`\ -generated deltas make no claim to be **minimal**
607diffs. To the contrary, minimal diffs are often counter-intuitive, because they
608synch up anywhere possible, sometimes accidental matches 100 pages apart.
609Restricting synch points to contiguous matches preserves some notion of
610locality, at the occasional cost of producing a longer diff.
611
612The :class:`Differ` class has this constructor:
613
614
615.. class:: Differ([linejunk[, charjunk]])
616
617   Optional keyword parameters *linejunk* and *charjunk* are for filter functions
618   (or ``None``):
619
620   *linejunk*: A function that accepts a single string argument, and returns true
621   if the string is junk.  The default is ``None``, meaning that no line is
622   considered junk.
623
624   *charjunk*: A function that accepts a single character argument (a string of
625   length 1), and returns true if the character is junk. The default is ``None``,
626   meaning that no character is considered junk.
627
628   :class:`Differ` objects are used (deltas generated) via a single method:
629
630
631   .. method:: Differ.compare(a, b)
632
633      Compare two sequences of lines, and generate the delta (a sequence of lines).
634
635      Each sequence must contain individual single-line strings ending with
636      newlines.  Such sequences can be obtained from the
637      :meth:`~file.readlines` method of file-like objects.  The delta
638      generated also consists of newline-terminated strings, ready to be
639      printed as-is via the :meth:`~file.writelines` method of a
640      file-like object.
641
642
643.. _differ-examples:
644
645Differ Example
646--------------
647
648This example compares two texts. First we set up the texts, sequences of
649individual single-line strings ending with newlines (such sequences can also be
650obtained from the :meth:`~file.readlines` method of file-like objects):
651
652   >>> text1 = '''  1. Beautiful is better than ugly.
653   ...   2. Explicit is better than implicit.
654   ...   3. Simple is better than complex.
655   ...   4. Complex is better than complicated.
656   ... '''.splitlines(1)
657   >>> len(text1)
658   4
659   >>> text1[0][-1]
660   '\n'
661   >>> text2 = '''  1. Beautiful is better than ugly.
662   ...   3.   Simple is better than complex.
663   ...   4. Complicated is better than complex.
664   ...   5. Flat is better than nested.
665   ... '''.splitlines(1)
666
667Next we instantiate a Differ object:
668
669   >>> d = Differ()
670
671Note that when instantiating a :class:`Differ` object we may pass functions to
672filter out line and character "junk."  See the :meth:`Differ` constructor for
673details.
674
675Finally, we compare the two:
676
677   >>> result = list(d.compare(text1, text2))
678
679``result`` is a list of strings, so let's pretty-print it:
680
681   >>> from pprint import pprint
682   >>> pprint(result)
683   ['    1. Beautiful is better than ugly.\n',
684    '-   2. Explicit is better than implicit.\n',
685    '-   3. Simple is better than complex.\n',
686    '+   3.   Simple is better than complex.\n',
687    '?     ++\n',
688    '-   4. Complex is better than complicated.\n',
689    '?            ^                     ---- ^\n',
690    '+   4. Complicated is better than complex.\n',
691    '?           ++++ ^                      ^\n',
692    '+   5. Flat is better than nested.\n']
693
694As a single multi-line string it looks like this:
695
696   >>> import sys
697   >>> sys.stdout.writelines(result)
698       1. Beautiful is better than ugly.
699   -   2. Explicit is better than implicit.
700   -   3. Simple is better than complex.
701   +   3.   Simple is better than complex.
702   ?     ++
703   -   4. Complex is better than complicated.
704   ?            ^                     ---- ^
705   +   4. Complicated is better than complex.
706   ?           ++++ ^                      ^
707   +   5. Flat is better than nested.
708
709
710.. _difflib-interface:
711
712A command-line interface to difflib
713-----------------------------------
714
715This example shows how to use difflib to create a ``diff``-like utility.
716It is also contained in the Python source distribution, as
717:file:`Tools/scripts/diff.py`.
718
719.. testcode::
720
721   """ Command line interface to difflib.py providing diffs in four formats:
722
723   * ndiff:    lists every line and highlights interline changes.
724   * context:  highlights clusters of changes in a before/after format.
725   * unified:  highlights clusters of changes in an inline format.
726   * html:     generates side by side comparison with change highlights.
727
728   """
729
730   import sys, os, time, difflib, optparse
731
732   def main():
733        # Configure the option parser
734       usage = "usage: %prog [options] fromfile tofile"
735       parser = optparse.OptionParser(usage)
736       parser.add_option("-c", action="store_true", default=False,
737                         help='Produce a context format diff (default)')
738       parser.add_option("-u", action="store_true", default=False,
739                         help='Produce a unified format diff')
740       hlp = 'Produce HTML side by side diff (can use -c and -l in conjunction)'
741       parser.add_option("-m", action="store_true", default=False, help=hlp)
742       parser.add_option("-n", action="store_true", default=False,
743                         help='Produce a ndiff format diff')
744       parser.add_option("-l", "--lines", type="int", default=3,
745                         help='Set number of context lines (default 3)')
746       (options, args) = parser.parse_args()
747
748       if len(args) == 0:
749           parser.print_help()
750           sys.exit(1)
751       if len(args) != 2:
752           parser.error("need to specify both a fromfile and tofile")
753
754       n = options.lines
755       fromfile, tofile = args # as specified in the usage string
756
757       # we're passing these as arguments to the diff function
758       fromdate = time.ctime(os.stat(fromfile).st_mtime)
759       todate = time.ctime(os.stat(tofile).st_mtime)
760       with open(fromfile, 'U') as f:
761           fromlines = f.readlines()
762       with open(tofile, 'U') as f:
763           tolines = f.readlines()
764
765       if options.u:
766           diff = difflib.unified_diff(fromlines, tolines, fromfile, tofile,
767                                       fromdate, todate, n=n)
768       elif options.n:
769           diff = difflib.ndiff(fromlines, tolines)
770       elif options.m:
771           diff = difflib.HtmlDiff().make_file(fromlines, tolines, fromfile,
772                                               tofile, context=options.c,
773                                               numlines=n)
774       else:
775           diff = difflib.context_diff(fromlines, tolines, fromfile, tofile,
776                                       fromdate, todate, n=n)
777
778       # we're using writelines because diff is a generator
779       sys.stdout.writelines(diff)
780
781   if __name__ == '__main__':
782       main()
783