1 """Text wrapping and filling.

     2 """

     3 

     4 # Copyright (C) 1999-2001 Gregory P. Ward.

     5 # Copyright (C) 2002, 2003 Python Software Foundation.

     6 # Written by Greg Ward <gward@python.net>

     7 

     8 __revision__ = "$Id: textwrap.py,v 1.1 2009/02/05 23:03:30 stechong Exp $"

     9 

    10 import string, re

    11 import types

    12 

    13 # Do the right thing with boolean values for all known Python versions

    14 # (so this module can be copied to projects that don't depend on Python

    15 # 2.3, e.g. Optik and Docutils).

    16 try:

    17     True, False

    18 except NameError:

    19     (True, False) = (1, 0)

    20 

    21 # For Python 1.5, just ignore unicode (try it as str)

    22 try:

    23     unicode

    24 except NameError:

    25     unicode=str

    26 

    27 __all__ = ['TextWrapper', 'wrap', 'fill']

    28 

    29 # Hardcode the recognized whitespace characters to the US-ASCII

    30 # whitespace characters.  The main reason for doing this is that in

    31 # ISO-8859-1, 0xa0 is non-breaking whitespace, so in certain locales

    32 # that character winds up in string.whitespace.  Respecting

    33 # string.whitespace in those cases would 1) make textwrap treat 0xa0 the

    34 # same as any other whitespace char, which is clearly wrong (it's a

    35 # *non-breaking* space), 2) possibly cause problems with Unicode,

    36 # since 0xa0 is not in range(128).

    37 _whitespace = '\t\n\x0b\x0c\r '

    38 

    39 class TextWrapper:

    40     """

    41     Object for wrapping/filling text.  The public interface consists of

    42     the wrap() and fill() methods; the other methods are just there for

    43     subclasses to override in order to tweak the default behaviour.

    44     If you want to completely replace the main wrapping algorithm,

    45     you'll probably have to override _wrap_chunks().

    46 

    47     Several instance attributes control various aspects of wrapping:

    48       width (default: 70)

    49         the maximum width of wrapped lines (unless break_long_words

    50         is false)

    51       initial_indent (default: "")

    52         string that will be prepended to the first line of wrapped

    53         output.  Counts towards the line's width.

    54       subsequent_indent (default: "")

    55         string that will be prepended to all lines save the first

    56         of wrapped output; also counts towards each line's width.

    57       expand_tabs (default: true)

    58         Expand tabs in input text to spaces before further processing.

    59         Each tab will become 1 .. 8 spaces, depending on its position in

    60         its line.  If false, each tab is treated as a single character.

    61       replace_whitespace (default: true)

    62         Replace all whitespace characters in the input text by spaces

    63         after tab expansion.  Note that if expand_tabs is false and

    64         replace_whitespace is true, every tab will be converted to a

    65         single space!

    66       fix_sentence_endings (default: false)

    67         Ensure that sentence-ending punctuation is always followed

    68         by two spaces.  Off by default because the algorithm is

    69         (unavoidably) imperfect.

    70       break_long_words (default: true)

    71         Break words longer than 'width'.  If false, those words will not

    72         be broken, and some lines might be longer than 'width'.

    73     """

    74 

    75     whitespace_trans = string.maketrans(_whitespace, ' ' * len(_whitespace))

    76 

    77     unicode_whitespace_trans = {}

    78     uspace = ord(unicode(' '))

    79     for x in map(ord, _whitespace):

    80         unicode_whitespace_trans[x] = uspace

    81 

    82     # This funky little regex is just the trick for splitting

    83     # text up into word-wrappable chunks.  E.g.

    84     #   "Hello there -- you goof-ball, use the -b option!"

    85     # splits into

    86     #   Hello/ /there/ /--/ /you/ /goof-/ball,/ /use/ /the/ /-b/ /option!

    87     # (after stripping out empty strings).

    88     try:

    89         wordsep_re = re.compile(r'(\s+|'                  # any whitespace

    90                                 r'[^\s\w]*\w{2,}-(?=\w{2,})|' # hyphenated words

    91                                 r'(?<=[\w\!\"\'\&\.\,\?])-{2,}(?=\w))')   # em-dash

    92     except:

    93         # Under python 1.5, the above regular expression does not compile because

    94         # of positive look-behind assertions (?<=). This stripped down version

    95         # does but it causes some regressions in the testsuite. Better than

    96         # nothing...

    97         wordsep_re = re.compile(r'(\s+|'                  # any whitespace

    98                                 r'[^\s\w]*\w{2,}-(?=\w{2,})|' # hyphenated words

    99                                 r')')   # em-dash

   100 

   101     # XXX this is not locale- or charset-aware -- string.lowercase

   102     # is US-ASCII only (and therefore English-only)

   103     sentence_end_re = re.compile(r'[%s]'              # lowercase letter

   104                                  r'[\.\!\?]'          # sentence-ending punct.

   105                                  r'[\"\']?'           # optional end-of-quote

   106                                  % string.lowercase)

   107 

   108 

   109     def __init__(self,

   110                  width=70,

   111                  initial_indent="",

   112                  subsequent_indent="",

   113                  expand_tabs=True,

   114                  replace_whitespace=True,

   115                  fix_sentence_endings=False,

   116                  break_long_words=True):

   117         self.width = width

   118         self.initial_indent = initial_indent

   119         self.subsequent_indent = subsequent_indent

   120         self.expand_tabs = expand_tabs

   121         self.replace_whitespace = replace_whitespace

   122         self.fix_sentence_endings = fix_sentence_endings

   123         self.break_long_words = break_long_words

   124 

   125 

   126     # -- Private methods -----------------------------------------------

   127     # (possibly useful for subclasses to override)

   128 

   129     def _munge_whitespace(self, text):

   130         """_munge_whitespace(text : string) -> string

   131 

   132         Munge whitespace in text: expand tabs and convert all other

   133         whitespace characters to spaces.  Eg. " foo\tbar\n\nbaz"

   134         becomes " foo    bar  baz".

   135         """

   136         if self.expand_tabs:

   137             text = string.expandtabs(text)

   138         if self.replace_whitespace:

   139             if isinstance(text, types.StringType):

   140                 text = string.translate(text, self.whitespace_trans)

   141             elif isinstance(text, types.UnicodeType):

   142                 # This has to be Python 2.0+ (no unicode before), so

   143                 # use directly string methods (the string module does not

   144                 # support translate() with dictionary for unicode).

   145                 text = text.translate(self.unicode_whitespace_trans)

   146         return text

   147 

   148 

   149     def _split(self, text):

   150         """_split(text : string) -> [string]

   151 

   152         Split the text to wrap into indivisible chunks.  Chunks are

   153         not quite the same as words; see wrap_chunks() for full

   154         details.  As an example, the text

   155           Look, goof-ball -- use the -b option!

   156         breaks into the following chunks:

   157           'Look,', ' ', 'goof-', 'ball', ' ', '--', ' ',

   158           'use', ' ', 'the', ' ', '-b', ' ', 'option!'

   159         """

   160         chunks = self.wordsep_re.split(text)

   161         chunks = filter(None, chunks)

   162         return chunks

   163 

   164     def _fix_sentence_endings(self, chunks):

   165         """_fix_sentence_endings(chunks : [string])

   166 

   167         Correct for sentence endings buried in 'chunks'.  Eg. when the

   168         original text contains "... foo.\nBar ...", munge_whitespace()

   169         and split() will convert that to [..., "foo.", " ", "Bar", ...]

   170         which has one too few spaces; this method simply changes the one

   171         space to two.

   172         """

   173         i = 0

   174         pat = self.sentence_end_re

   175         while i < len(chunks)-1:

   176             if chunks[i+1] == " " and pat.search(chunks[i]):

   177                 chunks[i+1] = "  "

   178                 i = i+2

   179             else:

   180                 i = i+1

   181 

   182     def _handle_long_word(self, chunks, cur_line, cur_len, width):

   183         """_handle_long_word(chunks : [string],

   184                              cur_line : [string],

   185                              cur_len : int, width : int)

   186 

   187         Handle a chunk of text (most likely a word, not whitespace) that

   188         is too long to fit in any line.

   189         """

   190         space_left = max(width - cur_len, 1)

   191 

   192         # If we're allowed to break long words, then do so: put as much

   193         # of the next chunk onto the current line as will fit.

   194         if self.break_long_words:

   195             cur_line.append(chunks[0][0:space_left])

   196             chunks[0] = chunks[0][space_left:]

   197 

   198         # Otherwise, we have to preserve the long word intact.  Only add

   199         # it to the current line if there's nothing already there --

   200         # that minimizes how much we violate the width constraint.

   201         elif not cur_line:

   202             cur_line.append(chunks.pop(0))

   203 

   204         # If we're not allowed to break long words, and there's already

   205         # text on the current line, do nothing.  Next time through the

   206         # main loop of _wrap_chunks(), we'll wind up here again, but

   207         # cur_len will be zero, so the next line will be entirely

   208         # devoted to the long word that we can't handle right now.

   209 

   210     def _wrap_chunks(self, chunks):

   211         """_wrap_chunks(chunks : [string]) -> [string]

   212 

   213         Wrap a sequence of text chunks and return a list of lines of

   214         length 'self.width' or less.  (If 'break_long_words' is false,

   215         some lines may be longer than this.)  Chunks correspond roughly

   216         to words and the whitespace between them: each chunk is

   217         indivisible (modulo 'break_long_words'), but a line break can

   218         come between any two chunks.  Chunks should not have internal

   219         whitespace; ie. a chunk is either all whitespace or a "word".

   220         Whitespace chunks will be removed from the beginning and end of

   221         lines, but apart from that whitespace is preserved.

   222         """

   223         lines = []

   224         if self.width <= 0:

   225             raise ValueError("invalid width %r (must be > 0)" % self.width)

   226 

   227         while chunks:

   228 

   229             # Start the list of chunks that will make up the current line.

   230             # cur_len is just the length of all the chunks in cur_line.

   231             cur_line = []

   232             cur_len = 0

   233 

   234             # Figure out which static string will prefix this line.

   235             if lines:

   236                 indent = self.subsequent_indent

   237             else:

   238                 indent = self.initial_indent

   239 

   240             # Maximum width for this line.

   241             width = self.width - len(indent)

   242 

   243             # First chunk on line is whitespace -- drop it, unless this

   244             # is the very beginning of the text (ie. no lines started yet).

   245             if string.strip(chunks[0]) == '' and lines:

   246                 del chunks[0]

   247 

   248             while chunks:

   249                 l = len(chunks[0])

   250 

   251                 # Can at least squeeze this chunk onto the current line.

   252                 if cur_len + l <= width:

   253                     cur_line.append(chunks.pop(0))

   254                     cur_len = cur_len + l

   255 

   256                 # Nope, this line is full.

   257                 else:

   258                     break

   259 

   260             # The current line is full, and the next chunk is too big to

   261             # fit on *any* line (not just this one).

   262             if chunks and len(chunks[0]) > width:

   263                 self._handle_long_word(chunks, cur_line, cur_len, width)

   264 

   265             # If the last chunk on this line is all whitespace, drop it.

   266             if cur_line and string.strip(cur_line[-1]) == '':

   267                 del cur_line[-1]

   268 

   269             # Convert current line back to a string and store it in list

   270             # of all lines (return value).

   271             if cur_line:

   272                 lines.append(indent + string.join(cur_line, ''))

   273 

   274         return lines

   275 

   276 

   277     # -- Public interface ----------------------------------------------

   278 

   279     def wrap(self, text):

   280         """wrap(text : string) -> [string]

   281 

   282         Reformat the single paragraph in 'text' so it fits in lines of

   283         no more than 'self.width' columns, and return a list of wrapped

   284         lines.  Tabs in 'text' are expanded with string.expandtabs(),

   285         and all other whitespace characters (including newline) are

   286         converted to space.

   287         """

   288         text = self._munge_whitespace(text)

   289         indent = self.initial_indent

   290         chunks = self._split(text)

   291         if self.fix_sentence_endings:

   292             self._fix_sentence_endings(chunks)

   293         return self._wrap_chunks(chunks)

   294 

   295     def fill(self, text):

   296         """fill(text : string) -> string

   297 

   298         Reformat the single paragraph in 'text' to fit in lines of no

   299         more than 'self.width' columns, and return a new string

   300         containing the entire wrapped paragraph.

   301         """

   302         return string.join(self.wrap(text), "\n")

   303 

   304 

   305 # -- Convenience interface ---------------------------------------------

   306 

   307 def wrap(text, width=70, **kwargs):

   308     """Wrap a single paragraph of text, returning a list of wrapped lines.

   309 

   310     Reformat the single paragraph in 'text' so it fits in lines of no

   311     more than 'width' columns, and return a list of wrapped lines.  By

   312     default, tabs in 'text' are expanded with string.expandtabs(), and

   313     all other whitespace characters (including newline) are converted to

   314     space.  See TextWrapper class for available keyword args to customize

   315     wrapping behaviour.

   316     """

   317     kwargs["width"] = width

   318     w = apply(TextWrapper, (), kwargs)

   319     return w.wrap(text)

   320 

   321 def fill(text, width=70, **kwargs):

   322     """Fill a single paragraph of text, returning a new string.

   323 

   324     Reformat the single paragraph in 'text' to fit in lines of no more

   325     than 'width' columns, and return a new string containing the entire

   326     wrapped paragraph.  As with wrap(), tabs are expanded and other

   327     whitespace characters converted to space.  See TextWrapper class for

   328     available keyword args to customize wrapping behaviour.

   329     """

   330     kwargs["width"] = width

   331     w = apply(TextWrapper, (), kwargs)

   332     return w.fill(text)

   333 

   334 

   335 # -- Loosely related functionality -------------------------------------

   336 

   337 def dedent(text):

   338     """dedent(text : string) -> string

   339 

   340     Remove any whitespace than can be uniformly removed from the left

   341     of every line in `text`.

   342 

   343     This can be used e.g. to make triple-quoted strings line up with

   344     the left edge of screen/whatever, while still presenting it in the

   345     source code in indented form.

   346 

   347     For example:

   348 

   349         def test():

   350             # end first line with \ to avoid the empty line!

   351             s = '''\

   352             hello

   353               world

   354             '''

   355             print repr(s)          # prints '    hello\n      world\n    '

   356             print repr(dedent(s))  # prints 'hello\n  world\n'

   357     """

   358     lines = string.split(string.expandtabs(text), '\n')

   359     margin = None

   360     for line in lines:

   361         content = string.lstrip(line)

   362         if not content:

   363             continue

   364         indent = len(line) - len(content)

   365         if margin is None:

   366             margin = indent

   367         else:

   368             margin = min(margin, indent)

   369 

   370     if margin is not None and margin > 0:

   371         for i in range(len(lines)):

   372             lines[i] = lines[i][margin:]

   373 

   374     return string.join(lines, "\n")