3 This is a Pure Python module to hyphenate text.
5 It is inspired by Ruby's Text::Hyphen, but currently reads standard *.dic files,
6 that must be installed separately.
8 In the future it's maybe nice if dictionaries could be distributed together with
9 this module, in a slightly prepared form, like in Ruby's Text::Hyphen.
11 Wilbert Berendsen, March 2008
12 info@wilbertberendsen.nl
20 __all__ = ("Hyphenator")
22 # cache of per-file Hyph_dict objects
25 # precompile some stuff
26 parse_hex = re.compile(r'\^{2}([0-9a-f]{2})').sub
27 parse = re.compile(r'(\d?)(\D?)').findall
29 def hexrepl(matchObj):
30 return unichr(int(matchObj.group(1), 16))
33 class parse_alt(object):
35 Parse nonstandard hyphen pattern alternative.
36 The instance returns a special int with data about the current position
37 in the pattern when called with an odd value.
39 def __init__(self, pat, alt):
43 self.index = int(alt[1])
44 self.cut = int(alt[2]) + 1
47 self.cut = len(re.sub(r'[\d\.]', '', pat)) + 1
48 if pat.startswith('.'):
51 def __call__(self, val):
55 return dint(val, (self.change, self.index, self.cut))
62 Just an int some other data can be stuck to in a data attribute.
63 Call with ref=other to use the data from the other dint.
65 def __new__(cls, value, data=None, ref=None):
66 obj = int.__new__(cls, value)
67 if ref and type(ref) == dint:
74 class Hyph_dict(object):
76 Reads a hyph_*.dic file and stores the hyphenation patterns.
78 -filename : filename of hyph_*.dic to read
80 def __init__(self, filename):
82 f = open(filename, 'rb')
83 charset = f.readline().strip()
84 if charset.startswith(b'charset '):
85 charset = charset[8:].strip()
88 pat = pat.decode(charset.decode('latin1')).strip()
89 if not pat or pat[0] == '%': continue
90 # replace ^^hh with the real character
91 pat = parse_hex(hexrepl, pat)
92 # read nonstandard hyphen alternatives
94 pat, alt = pat.split('/', 1)
95 factory = parse_alt(pat, alt)
98 tag, value = zip(*[(s, factory(i or "0")) for i, s in parse(pat)])
99 # if only zeros, skip this pattern
100 if max(value) == 0: continue
101 # chop zeros from beginning and end, and store start offset.
102 start, end = 0, len(value)
103 while not value[start]: start += 1
104 while not value[end-1]: end -= 1
105 self.patterns[''.join(tag)] = start, value[start:end]
108 self.maxlen = max(map(len, self.patterns.keys()))
110 def positions(self, word):
112 Returns a list of positions where the word can be hyphenated.
113 E.g. for the dutch word 'lettergrepen' this method returns
116 Each position is a 'data int' (dint) with a data attribute.
117 If the data attribute is not None, it contains a tuple with
118 information about nonstandard hyphenation at that point:
121 change: is a string like 'ff=f', that describes how hyphenation
123 index: where to substitute the change, counting from the current
125 cut: how many characters to remove while substituting the nonstandard
129 points = self.cache.get(word)
131 prepWord = '.%s.' % word
132 res = [0] * (len(prepWord) + 1)
133 for i in range(len(prepWord) - 1):
134 for j in range(i + 1, min(i + self.maxlen, len(prepWord)) + 1):
135 p = self.patterns.get(prepWord[i:j])
138 s = slice(i + offset, i + offset + len(value))
139 res[s] = map(max, value, res[s])
141 points = [dint(i - 1, ref=r) for i, r in enumerate(res) if r % 2]
142 self.cache[word] = points
146 class Hyphenator(object):
148 Reads a hyph_*.dic file and stores the hyphenation patterns.
149 Provides methods to hyphenate strings in various ways.
151 -filename : filename of hyph_*.dic to read
152 -left: make the first syllabe not shorter than this
153 -right: make the last syllabe not shorter than this
154 -cache: if true (default), use a cached copy of the dic file, if possible
156 left and right may also later be changed:
160 def __init__(self, filename, left=2, right=2, cache=True):
163 if not cache or filename not in hdcache:
164 hdcache[filename] = Hyph_dict(filename)
165 self.hd = hdcache[filename]
167 def positions(self, word):
169 Returns a list of positions where the word can be hyphenated.
170 See also Hyph_dict.positions. The points that are too far to
171 the left or right are removed.
173 right = len(word) - self.right
174 return [i for i in self.hd.positions(word) if self.left <= i <= right]
176 def iterate(self, word):
178 Iterate over all hyphenation possibilities, the longest first.
180 if isinstance(word, str):
181 word = word.decode('latin1')
182 for p in reversed(self.positions(word)):
184 # get the nonstandard hyphenation data
185 change, index, cut = p.data
187 change = change.upper()
188 c1, c2 = change.split('=')
189 yield word[:p+index] + c1, c2 + word[p+index+cut:]
191 yield word[:p], word[p:]
193 def wrap(self, word, width, hyphen='-'):
195 Return the longest possible first part and the last part of the
196 hyphenated word. The first part has the hyphen already attached.
197 Returns None, if there is no hyphenation point before width, or
198 if the word could not be hyphenated.
201 for w1, w2 in self.iterate(word):
203 return w1 + hyphen, w2
205 def inserted(self, word, hyphen='-'):
207 Returns the word as a string with all the possible hyphens inserted.
208 E.g. for the dutch word 'lettergrepen' this method returns
209 the string 'let-ter-gre-pen'. The hyphen string to use can be
210 given as the second parameter, that defaults to '-'.
212 if isinstance(word, bytes):
213 word = word.decode('latin1')
215 for p in reversed(self.positions(word)):
217 # get the nonstandard hyphenation data
218 change, index, cut = p.data
220 change = change.upper()
221 l[p + index : p + index + cut] = change.replace('=', hyphen)
229 if __name__ == "__main__":
231 dict_file = sys.argv[1]
232 word = sys.argv[2].decode('latin1')
234 h = Hyphenator(dict_file, left=1, right=1)