Lib/textwrap.py

   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__ = "$Id$"
   9
  10 import string, re
  11
  12 # Do the right thing with boolean values for all known Python versions
  13 # (so this module can be copied to projects that don't depend on Python
  14 # 2.3, e.g. Optik and Docutils) by uncommenting the block of code below.
  15 #try:
  16 #    True, False
  17 #except NameError:
  18 #    (True, False) = (1, 0)
  19
  20 __all__ = ['TextWrapper', 'wrap', 'fill']
  21
  22 # Hardcode the recognized whitespace characters to the US-ASCII
  23 # whitespace characters.  The main reason for doing this is that in
  24 # ISO-8859-1, 0xa0 is non-breaking whitespace, so in certain locales
  25 # that character winds up in string.whitespace.  Respecting
  26 # string.whitespace in those cases would 1) make textwrap treat 0xa0 the
  27 # same as any other whitespace char, which is clearly wrong (it's a
  28 # *non-breaking* space), 2) possibly cause problems with Unicode,
  29 # since 0xa0 is not in range(128).
  30 _whitespace = '\t\n\x0b\x0c\r '
  31
  32 class TextWrapper:
  33     """
  34     Object for wrapping/filling text.  The public interface consists of
  35     the wrap() and fill() methods; the other methods are just there for
  36     subclasses to override in order to tweak the default behaviour.
  37     If you want to completely replace the main wrapping algorithm,
  38     you'll probably have to override _wrap_chunks().
  39
  40     Several instance attributes control various aspects of wrapping:
  41       width (default: 70)
  42         the maximum width of wrapped lines (unless break_long_words
  43         is false)
  44       initial_indent (default: "")
  45         string that will be prepended to the first line of wrapped
  46         output.  Counts towards the line's width.
  47       subsequent_indent (default: "")
  48         string that will be prepended to all lines save the first
  49         of wrapped output; also counts towards each line's width.
  50       expand_tabs (default: true)
  51         Expand tabs in input text to spaces before further processing.
  52         Each tab will become 1 .. 8 spaces, depending on its position in
  53         its line.  If false, each tab is treated as a single character.
  54       replace_whitespace (default: true)
  55         Replace all whitespace characters in the input text by spaces
  56         after tab expansion.  Note that if expand_tabs is false and
  57         replace_whitespace is true, every tab will be converted to a
  58         single space!
  59       fix_sentence_endings (default: false)
  60         Ensure that sentence-ending punctuation is always followed
  61         by two spaces.  Off by default because the algorithm is
  62         (unavoidably) imperfect.
  63       break_long_words (default: true)
  64         Break words longer than 'width'.  If false, those words will not
  65         be broken, and some lines might be longer than 'width'.
  66       break_on_hyphens (default: true)
  67         Allow breaking hyphenated words. If true, wrapping will occur
  68         preferably on whitespaces and right after hyphens part of
  69         compound words.
  70       drop_whitespace (default: true)
  71         Drop leading and trailing whitespace from lines.
  72     """
  73
  74     whitespace_trans = string.maketrans(_whitespace, ' ' * len(_whitespace))
  75
  76     unicode_whitespace_trans = {}
  77     uspace = ord(u' ')
  78     for x in map(ord, _whitespace):
  79         unicode_whitespace_trans[x] = uspace
  80
  81     # This funky little regex is just the trick for splitting
  82     # text up into word-wrappable chunks.  E.g.
  83     #   "Hello there -- you goof-ball, use the -b option!"
  84     # splits into
  85     #   Hello/ /there/ /--/ /you/ /goof-/ball,/ /use/ /the/ /-b/ /option!
  86     # (after stripping out empty strings).
  87     wordsep_re = re.compile(
  88         r'(\s+|'                                  # any whitespace
  89         r'[^\s\w]*\w+[a-zA-Z]-(?=\w+[a-zA-Z])|'   # hyphenated words
  90         r'(?<=[\w\!\"\'\&\.\,\?])-{2,}(?=\w))')   # em-dash
  91
  92     # This less funky little regex just split on recognized spaces. E.g.
  93     #   "Hello there -- you goof-ball, use the -b option!"
  94     # splits into
  95     #   Hello/ /there/ /--/ /you/ /goof-ball,/ /use/ /the/ /-b/ /option!/
  96     wordsep_simple_re = re.compile(r'(\s+)')
  97
  98     # XXX this is not locale- or charset-aware -- string.lowercase
  99     # is US-ASCII only (and therefore English-only)
 100     sentence_end_re = re.compile(r'[%s]'              # lowercase letter
 101                                  r'[\.\!\?]'          # sentence-ending punct.
 102                                  r'[\"\']?'           # optional end-of-quote
 103                                  r'\Z'                # end of chunk
 104                                  % string.lowercase)
 105
 106
 107     def __init__(self,
 108                  width=70,
 109                  initial_indent="",
 110                  subsequent_indent="",
 111                  expand_tabs=True,
 112                  replace_whitespace=True,
 113                  fix_sentence_endings=False,
 114                  break_long_words=True,
 115                  drop_whitespace=True,
 116                  break_on_hyphens=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         self.drop_whitespace = drop_whitespace
 125         self.break_on_hyphens = break_on_hyphens
 126
 127
 128     # -- Private methods -----------------------------------------------
 129     # (possibly useful for subclasses to override)
 130
 131     def _munge_whitespace(self, text):
 132         """_munge_whitespace(text : string) -> string
 133
 134         Munge whitespace in text: expand tabs and convert all other
 135         whitespace characters to spaces.  Eg. " foo\tbar\n\nbaz"
 136         becomes " foo    bar  baz".
 137         """
 138         if self.expand_tabs:
 139             text = text.expandtabs()
 140         if self.replace_whitespace:
 141             if isinstance(text, str):
 142                 text = text.translate(self.whitespace_trans)
 143             elif isinstance(text, unicode):
 144                 text = text.translate(self.unicode_whitespace_trans)
 145         return text
 146
 147
 148     def _split(self, text):
 149         """_split(text : string) -> [string]
 150
 151         Split the text to wrap into indivisible chunks.  Chunks are
 152         not quite the same as words; see wrap_chunks() for full
 153         details.  As an example, the text
 154           Look, goof-ball -- use the -b option!
 155         breaks into the following chunks:
 156           'Look,', ' ', 'goof-', 'ball', ' ', '--', ' ',
 157           'use', ' ', 'the', ' ', '-b', ' ', 'option!'
 158         if break_on_hyphens is True, or in:
 159           'Look,', ' ', 'goof-ball', ' ', '--', ' ',
 160           'use', ' ', 'the', ' ', '-b', ' ', option!'
 161         otherwise.
 162         """
 163         if self.break_on_hyphens is True:
 164             chunks = self.wordsep_re.split(text)
 165         else:
 166             chunks = self.wordsep_simple_re.split(text)
 167         chunks = filter(None, chunks)  # remove empty chunks
 168         return chunks
 169
 170     def _fix_sentence_endings(self, chunks):
 171         """_fix_sentence_endings(chunks : [string])
 172
 173         Correct for sentence endings buried in 'chunks'.  Eg. when the
 174         original text contains "... foo.\nBar ...", munge_whitespace()
 175         and split() will convert that to [..., "foo.", " ", "Bar", ...]
 176         which has one too few spaces; this method simply changes the one
 177         space to two.
 178         """
 179         i = 0
 180         pat = self.sentence_end_re
 181         while i < len(chunks)-1:
 182             if chunks[i+1] == " " and pat.search(chunks[i]):
 183                 chunks[i+1] = "  "
 184                 i += 2
 185             else:
 186                 i += 1
 187
 188     def _handle_long_word(self, reversed_chunks, cur_line, cur_len, width):
 189         """_handle_long_word(chunks : [string],
 190                              cur_line : [string],
 191                              cur_len : int, width : int)
 192
 193         Handle a chunk of text (most likely a word, not whitespace) that
 194         is too long to fit in any line.
 195         """
 196         # Figure out when indent is larger than the specified width, and make
 197         # sure at least one character is stripped off on every pass
 198         if width < 1:
 199             space_left = 1
 200         else:
 201             space_left = width - cur_len
 202
 203         # If we're allowed to break long words, then do so: put as much
 204         # of the next chunk onto the current line as will fit.
 205         if self.break_long_words:
 206             cur_line.append(reversed_chunks[-1][:space_left])
 207             reversed_chunks[-1] = reversed_chunks[-1][space_left:]
 208
 209         # Otherwise, we have to preserve the long word intact.  Only add
 210         # it to the current line if there's nothing already there --
 211         # that minimizes how much we violate the width constraint.
 212         elif not cur_line:
 213             cur_line.append(reversed_chunks.pop())
 214
 215         # If we're not allowed to break long words, and there's already
 216         # text on the current line, do nothing.  Next time through the
 217         # main loop of _wrap_chunks(), we'll wind up here again, but
 218         # cur_len will be zero, so the next line will be entirely
 219         # devoted to the long word that we can't handle right now.
 220
 221     def _wrap_chunks(self, chunks):
 222         """_wrap_chunks(chunks : [string]) -> [string]
 223
 224         Wrap a sequence of text chunks and return a list of lines of
 225         length 'self.width' or less.  (If 'break_long_words' is false,
 226         some lines may be longer than this.)  Chunks correspond roughly
 227         to words and the whitespace between them: each chunk is
 228         indivisible (modulo 'break_long_words'), but a line break can
 229         come between any two chunks.  Chunks should not have internal
 230         whitespace; ie. a chunk is either all whitespace or a "word".
 231         Whitespace chunks will be removed from the beginning and end of
 232         lines, but apart from that whitespace is preserved.
 233         """
 234         lines = []
 235         if self.width <= 0:
 236             raise ValueError("invalid width %r (must be > 0)" % self.width)
 237
 238         # Arrange in reverse order so items can be efficiently popped
 239         # from a stack of chucks.
 240         chunks.reverse()
 241
 242         while chunks:
 243
 244             # Start the list of chunks that will make up the current line.
 245             # cur_len is just the length of all the chunks in cur_line.
 246             cur_line = []
 247             cur_len = 0
 248
 249             # Figure out which static string will prefix this line.
 250             if lines:
 251                 indent = self.subsequent_indent
 252             else:
 253                 indent = self.initial_indent
 254
 255             # Maximum width for this line.
 256             width = self.width - len(indent)
 257
 258             # First chunk on line is whitespace -- drop it, unless this
 259             # is the very beginning of the text (ie. no lines started yet).
 260             if self.drop_whitespace and chunks[-1].strip() == '' and lines:
 261                 del chunks[-1]
 262
 263             while chunks:
 264                 l = len(chunks[-1])
 265
 266                 # Can at least squeeze this chunk onto the current line.
 267                 if cur_len + l <= width:
 268                     cur_line.append(chunks.pop())
 269                     cur_len += l
 270
 271                 # Nope, this line is full.
 272                 else:
 273                     break
 274
 275             # The current line is full, and the next chunk is too big to
 276             # fit on *any* line (not just this one).
 277             if chunks and len(chunks[-1]) > width:
 278                 self._handle_long_word(chunks, cur_line, cur_len, width)
 279
 280             # If the last chunk on this line is all whitespace, drop it.
 281             if self.drop_whitespace and cur_line and cur_line[-1].strip() == '':
 282                 del cur_line[-1]
 283
 284             # Convert current line back to a string and store it in list
 285             # of all lines (return value).
 286             if cur_line:
 287                 lines.append(indent + ''.join(cur_line))
 288
 289         return lines
 290
 291
 292     # -- Public interface ----------------------------------------------
 293
 294     def wrap(self, text):
 295         """wrap(text : string) -> [string]
 296
 297         Reformat the single paragraph in 'text' so it fits in lines of
 298         no more than 'self.width' columns, and return a list of wrapped
 299         lines.  Tabs in 'text' are expanded with string.expandtabs(),
 300         and all other whitespace characters (including newline) are
 301         converted to space.
 302         """
 303         text = self._munge_whitespace(text)
 304         chunks = self._split(text)
 305         if self.fix_sentence_endings:
 306             self._fix_sentence_endings(chunks)
 307         return self._wrap_chunks(chunks)
 308
 309     def fill(self, text):
 310         """fill(text : string) -> string
 311
 312         Reformat the single paragraph in 'text' to fit in lines of no
 313         more than 'self.width' columns, and return a new string
 314         containing the entire wrapped paragraph.
 315         """
 316         return "\n".join(self.wrap(text))
 317
 318
 319 # -- Convenience interface ---------------------------------------------
 320
 321 def wrap(text, width=70, **kwargs):
 322     """Wrap a single paragraph of text, returning a list of wrapped lines.
 323
 324     Reformat the single paragraph in 'text' so it fits in lines of no
 325     more than 'width' columns, and return a list of wrapped lines.  By
 326     default, tabs in 'text' are expanded with string.expandtabs(), and
 327     all other whitespace characters (including newline) are converted to
 328     space.  See TextWrapper class for available keyword args to customize
 329     wrapping behaviour.
 330     """
 331     w = TextWrapper(width=width, **kwargs)
 332     return w.wrap(text)
 333
 334 def fill(text, width=70, **kwargs):
 335     """Fill a single paragraph of text, returning a new string.
 336
 337     Reformat the single paragraph in 'text' to fit in lines of no more
 338     than 'width' columns, and return a new string containing the entire
 339     wrapped paragraph.  As with wrap(), tabs are expanded and other
 340     whitespace characters converted to space.  See TextWrapper class for
 341     available keyword args to customize wrapping behaviour.
 342     """
 343     w = TextWrapper(width=width, **kwargs)
 344     return w.fill(text)
 345
 346
 347 # -- Loosely related functionality -------------------------------------
 348
 349 _whitespace_only_re = re.compile('^[ \t]+$', re.MULTILINE)
 350 _leading_whitespace_re = re.compile('(^[ \t]*)(?:[^ \t\n])', re.MULTILINE)
 351
 352 def dedent(text):
 353     """Remove any common leading whitespace from every line in `text`.
 354
 355     This can be used to make triple-quoted strings line up with the left
 356     edge of the display, while still presenting them in the source code
 357     in indented form.
 358
 359     Note that tabs and spaces are both treated as whitespace, but they
 360     are not equal: the lines "  hello" and "\thello" are
 361     considered to have no common leading whitespace.  (This behaviour is
 362     new in Python 2.5; older versions of this module incorrectly
 363     expanded tabs before searching for common leading whitespace.)
 364     """
 365     # Look for the longest leading string of spaces and tabs common to
 366     # all lines.
 367     margin = None
 368     text = _whitespace_only_re.sub('', text)
 369     indents = _leading_whitespace_re.findall(text)
 370     for indent in indents:
 371         if margin is None:
 372             margin = indent
 373
 374         # Current line more deeply indented than previous winner:
 375         # no change (previous winner is still on top).
 376         elif indent.startswith(margin):
 377             pass
 378
 379         # Current line consistent with and no deeper than previous winner:
 380         # it's the new winner.
 381         elif margin.startswith(indent):
 382             margin = indent
 383
 384         # Current line and previous winner have no common whitespace:
 385         # there is no margin.
 386         else:
 387             margin = ""
 388             break
 389
 390     # sanity check (testing/debugging only)
 391     if 0 and margin:
 392         for line in text.split("\n"):
 393             assert not line or line.startswith(margin), \
 394                    "line = %r, margin = %r" % (line, margin)
 395
 396     if margin:
 397         text = re.sub(r'(?m)^' + margin, '', text)
 398     return text
 399
 400 if __name__ == "__main__":
 401     #print dedent("\tfoo\n\tbar")
 402     #print dedent("  \thello there\n  \t  how are you?")
 403     print dedent("Hello there.\n  This is indented.")