Walenty / Slowal

Commit 92fd5c36277e29243591b6c00b97b8ef63c67c05

Authored by Bartłomiej Nitoń
1 parent a2ea5daa

Changed morfeusz based validations to use morfeusz2.

Inline Side-by-side
Showing 7 changed files with 619 additions and 319 deletions
common/_morfeusz2.so 0 → 100755
View file @92fd5c3
No preview for this file type
common/morfeusz.py deleted
View file @a2ea5da
1   -# encoding=UTF-8
2   -
3   -# Copyright © 2007, 2008, 2010, 2011 Jakub Wilk <jwilk@jwilk.net>
4   -#
5   -# Permission is hereby granted, free of charge, to any person obtaining a copy
6   -# of this software and associated documentation files (the “Software”), to deal
7   -# in the Software without restriction, including without limitation the rights
8   -# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
9   -# copies of the Software, and to permit persons to whom the Software is
10   -# furnished to do so, subject to the following conditions:
11   -#
12   -# The above copyright notice and this permission notice shall be included in
13   -# all copies or substantial portions of the Software.
14   -#
15   -# THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16   -# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17   -# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18   -# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19   -# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20   -# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21   -# SOFTWARE.
22   -
23   -'''
24   -Bindings for Morfeusz_, a Polish morphological analyser.
25   -
26   -.. _Morfeusz:
27   - http://sgjp.pl/morfeusz/
28   -'''
29   -
30   -from __future__ import with_statement
31   -
32   -import collections
33   -import ctypes
34   -import sys
35   -
36   -py3k = sys.version_info >= (3, 0)
37   -
38   -if py3k:
39   - import _thread as thread
40   -else:
41   - import thread
42   -if not py3k:
43   - from itertools import izip as zip
44   -
45   -if py3k:
46   - unicode = str
47   -
48   -__author__ = 'Jakub Wilk <jwilk@jwilk.net>'
49   -__version__ = '0.3300'
50   -__all__ = ['analyse', 'about', 'expand_tags', 'ATTRIBUTES', 'VALUES']
51   -
52   -ATTRIBUTES = '''
53   -subst=number case gender
54   -depr=number case gender
55   -adj=number case gender degree
56   -adja=
57   -adjc=
58   -adjp=
59   -adv=degree
60   -num=number case gender accommodability
61   -numcol=number case gender accommodability
62   -ppron12=number case gender person accentability
63   -ppron3=number case gender person accentability post_prepositionality
64   -siebie=case
65   -fin=number person aspect
66   -bedzie=number person aspect
67   -aglt=number person aspect vocalicity
68   -praet=number gender aspect agglutination
69   -impt=number person aspect
70   -imps=aspect
71   -inf=aspect
72   -pcon=aspect
73   -pant=aspect
74   -ger=number case gender aspect negation
75   -pact=number case gender aspect negation
76   -ppas=number case gender aspect negation
77   -winien=number gender aspect
78   -pred=
79   -prep=case vocalicity
80   -conj=
81   -comp=
82   -brev=fullstoppedness
83   -burk=
84   -interj=
85   -qub=vocalicity
86   -xxs=number case gender
87   -xxx=
88   -interp=
89   -ign=
90   -sp=
91   -'''
92   -ATTRIBUTES = \
93   -dict(
94   - (key, tuple(values.split()))
95   - for line in ATTRIBUTES.splitlines() if line
96   - for (key, values) in (line.split('=', 1),)
97   -)
98   -
99   -VALUES = '''
100   -number=sg pl
101   -case=nom gen dat acc inst loc voc
102   -gender=m1 m2 m3 f n1 n2 p1 p2 p3
103   -person=pri sec ter
104   -degree=pos comp sup
105   -aspect=imperf perf
106   -negation=aff neg
107   -accentability=akc nakc
108   -post_prepositionality=npraep praep
109   -accommodability=congr rec
110   -agglutination=agl nagl
111   -vocalicity=nwok wok
112   -fullstoppedness=pun npun
113   -'''
114   -VALUES = \
115   -dict(
116   - (key, tuple(values.split()))
117   - for line in VALUES.splitlines() if line
118   - for (key, values) in (line.split('=', 1),)
119   -)
120   -
121   -libmorfeusz = ctypes.CDLL('libmorfeusz.so.0')
122   -
123   -MORFOPT_ENCODING = 1
124   -MORFEUSZ_UTF_8 = 8
125   -
126   -MORFOPT_WHITESPACE = 2
127   -MORFEUSZ_SKIP_WHITESPACE = 0
128   -MORFEUSZ_KEEP_WHITESPACE = 2
129   -
130   -libmorfeusz.morfeusz_set_option(MORFOPT_ENCODING, MORFEUSZ_UTF_8)
131   -libmorfeusz_lock = thread.allocate_lock()
132   -
133   -class InterpEdge(ctypes.Structure):
134   - _fields_ = \
135   - (
136   - ('i', ctypes.c_int),
137   - ('j', ctypes.c_int),
138   - ('_orth', ctypes.c_char_p),
139   - ('_base', ctypes.c_char_p),
140   - ('_tags', ctypes.c_char_p)
141   - )
142   -
143   - if py3k:
144   - @property
145   - def tags(self):
146   - if self._tags is not None:
147   - return self._tags.decode('UTF-8')
148   - else:
149   - @property
150   - def tags(self):
151   - return self._tags
152   -
153   - @property
154   - def orth(self):
155   - if self._orth is not None:
156   - return self._orth.decode('UTF-8')
157   -
158   - @property
159   - def base(self):
160   - if self._base is not None:
161   - return self._base.decode('UTF-8')
162   -
163   -libmorfeusz_analyse = libmorfeusz.morfeusz_analyse
164   -libmorfeusz_analyse.restype = ctypes.POINTER(InterpEdge)
165   -libmorfeusz_about = libmorfeusz.morfeusz_about
166   -libmorfeusz_about.restype = ctypes.c_char_p
167   -
168   -def expand_tags(tags, expand_dot=True, expand_underscore=True):
169   -
170   - if tags is None:
171   - yield
172   - return
173   - tags = str(tags)
174   - for tag in tags.split('|'):
175   - tag = tag.split(':')
176   - pos = tag.pop(0)
177   - chunks = [(pos,)]
178   - chunks += \
179   - (
180   - VALUES[attribute] if chunk == '_' and expand_underscore
181   - else chunk.split('.')
182   - for chunk, attribute in zip(tag, ATTRIBUTES[pos])
183   - )
184   -
185   - if not expand_dot:
186   - yield ':'.join('.'.join(values) for values in chunks)
187   - continue
188   -
189   - def expand_chunks(i):
190   - if i >= len(chunks):
191   - yield ()
192   - else:
193   - tail = tuple(expand_chunks(i + 1))
194   - for chunk_variant in chunks[i]:
195   - for tail_variant in tail:
196   - yield (chunk_variant,) + tail_variant
197   -
198   - for x in expand_chunks(0):
199   - yield ':'.join(x)
200   -
201   -_expand_tags = expand_tags
202   -
203   -def _dont_expand_tags(s, **kwargs):
204   - return [s]
205   -
206   -def analyse(text, expand_tags=True, expand_dot=True, expand_underscore=True, dag=False, keep_whitespace=False):
207   - '''
208   - Analyse the text.
209   - '''
210   - expand_tags = _expand_tags if expand_tags else _dont_expand_tags
211   - text = unicode(text)
212   - text = text.encode('UTF-8')
213   - analyse = _analyse_as_dag if dag else _analyse_as_list
214   - return analyse(
215   - text=text,
216   - expand_tags=expand_tags,
217   - expand_dot=expand_dot,
218   - expand_underscore=expand_underscore,
219   - keep_whitespace=keep_whitespace
220   - )
221   -
222   -def _analyse_as_dag(text, expand_tags, expand_dot, expand_underscore, keep_whitespace):
223   - result = []
224   - with libmorfeusz_lock:
225   - if keep_whitespace:
226   - if libmorfeusz.morfeusz_set_option(MORFOPT_WHITESPACE, MORFEUSZ_KEEP_WHITESPACE) != 1:
227   - raise NotImplementedError("This version of Morfeusz doesn't support keep_whitespace")
228   - for edge in libmorfeusz_analyse(text):
229   - if edge.i == -1:
230   - break
231   - for tag in expand_tags(edge.tags, expand_dot=expand_dot, expand_underscore=expand_underscore):
232   - result += [(edge.i, edge.j, (edge.orth, edge.base, tag))]
233   - if keep_whitespace:
234   - libmorfeusz.morfeusz_set_option(MORFOPT_WHITESPACE, MORFEUSZ_SKIP_WHITESPACE)
235   - return result
236   -
237   -def _analyse_as_list(text, expand_tags, expand_dot, expand_underscore, keep_whitespace):
238   - dag = collections.defaultdict(list)
239   - with libmorfeusz_lock:
240   - if keep_whitespace:
241   - if libmorfeusz.morfeusz_set_option(MORFOPT_WHITESPACE, MORFEUSZ_KEEP_WHITESPACE) != 1:
242   - raise NotImplementedError("This version of Morfeusz doesn't support keep_whitespace")
243   - for edge in libmorfeusz_analyse(text):
244   - if edge.i == -1:
245   - break
246   - for tag in expand_tags(edge.tags, expand_dot=expand_dot, expand_underscore=expand_underscore):
247   - dag[edge.i] += [((edge.orth, edge.base, tag), edge.j)]
248   - if keep_whitespace:
249   - libmorfeusz.morfeusz_set_option(MORFOPT_WHITESPACE, MORFEUSZ_SKIP_WHITESPACE)
250   - def expand_dag(i):
251   - nexts = dag[i]
252   - if not nexts:
253   - yield []
254   - else:
255   - for head, j in nexts:
256   - for tail in expand_dag(j):
257   - yield [head] + tail
258   - return list(expand_dag(0))
259   -
260   -def about():
261   - '''
262   - Return a string containing information on authors and version of the
263   - underlying library.
264   - '''
265   - about = libmorfeusz_about()
266   - try:
267   - return about.decode('UTF-8')
268   - except UnicodeError:
269   - return about.decode('ISO-8859-2')
270   -
271   -# vim:ts=4 sw=4 et
common/morfeusz2.py 0 → 100644
  View file @92fd5c3
  1 +# This file was automatically generated by SWIG (http://www.swig.org).
  2 +# Version 2.0.4
  3 +#
  4 +# Do not make changes to this file unless you know what you are doing--modify
  5 +# the SWIG interface file instead.
  6 +
  7 +
  8 +
  9 +from sys import version_info
  10 +if version_info >= (2,6,0):
  11 + def swig_import_helper():
  12 + from os.path import dirname
  13 + import imp
  14 + fp = None
  15 + try:
  16 + fp, pathname, description = imp.find_module('_morfeusz2', [dirname(__file__)])
  17 + except ImportError:
  18 + import _morfeusz2
  19 + return _morfeusz2
  20 + if fp is not None:
  21 + try:
  22 + _mod = imp.load_module('_morfeusz2', fp, pathname, description)
  23 + finally:
  24 + fp.close()
  25 + return _mod
  26 + _morfeusz2 = swig_import_helper()
  27 + del swig_import_helper
  28 +else:
  29 + import _morfeusz2
  30 +del version_info
  31 +try:
  32 + _swig_property = property
  33 +except NameError:
  34 + pass # Python < 2.2 doesn't have 'property'.
  35 +def _swig_setattr_nondynamic(self,class_type,name,value,static=1):
  36 + if (name == "thisown"): return self.this.own(value)
  37 + if (name == "this"):
  38 + if type(value).__name__ == 'SwigPyObject':
  39 + self.__dict__[name] = value
  40 + return
  41 + method = class_type.__swig_setmethods__.get(name,None)
  42 + if method: return method(self,value)
  43 + if (not static):
  44 + self.__dict__[name] = value
  45 + else:
  46 + raise AttributeError("You cannot add attributes to %s" % self)
  47 +
  48 +def _swig_setattr(self,class_type,name,value):
  49 + return _swig_setattr_nondynamic(self,class_type,name,value,0)
  50 +
  51 +def _swig_getattr(self,class_type,name):
  52 + if (name == "thisown"): return self.this.own()
  53 + method = class_type.__swig_getmethods__.get(name,None)
  54 + if method: return method(self)
  55 + raise AttributeError(name)
  56 +
  57 +def _swig_repr(self):
  58 + try: strthis = "proxy of " + self.this.__repr__()
  59 + except: strthis = ""
  60 + return "<%s.%s; %s >" % (self.__class__.__module__, self.__class__.__name__, strthis,)
  61 +
  62 +try:
  63 + _object = object
  64 + _newclass = 1
  65 +except AttributeError:
  66 + class _object : pass
  67 + _newclass = 0
  68 +
  69 +
  70 +class SwigPyIterator(_object):
  71 + __swig_setmethods__ = {}
  72 + __setattr__ = lambda self, name, value: _swig_setattr(self, SwigPyIterator, name, value)
  73 + __swig_getmethods__ = {}
  74 + __getattr__ = lambda self, name: _swig_getattr(self, SwigPyIterator, name)
  75 + def __init__(self, *args, **kwargs): raise AttributeError("No constructor defined - class is abstract")
  76 + __repr__ = _swig_repr
  77 + __swig_destroy__ = _morfeusz2.delete_SwigPyIterator
  78 + __del__ = lambda self : None;
  79 + def value(self): return _morfeusz2.SwigPyIterator_value(self)
  80 + def incr(self, n = 1): return _morfeusz2.SwigPyIterator_incr(self, n)
  81 + def decr(self, n = 1): return _morfeusz2.SwigPyIterator_decr(self, n)
  82 + def distance(self, *args): return _morfeusz2.SwigPyIterator_distance(self, *args)
  83 + def equal(self, *args): return _morfeusz2.SwigPyIterator_equal(self, *args)
  84 + def copy(self): return _morfeusz2.SwigPyIterator_copy(self)
  85 + def next(self): return _morfeusz2.SwigPyIterator_next(self)
  86 + def __next__(self): return _morfeusz2.SwigPyIterator___next__(self)
  87 + def previous(self): return _morfeusz2.SwigPyIterator_previous(self)
  88 + def advance(self, *args): return _morfeusz2.SwigPyIterator_advance(self, *args)
  89 + def __eq__(self, *args): return _morfeusz2.SwigPyIterator___eq__(self, *args)
  90 + def __ne__(self, *args): return _morfeusz2.SwigPyIterator___ne__(self, *args)
  91 + def __iadd__(self, *args): return _morfeusz2.SwigPyIterator___iadd__(self, *args)
  92 + def __isub__(self, *args): return _morfeusz2.SwigPyIterator___isub__(self, *args)
  93 + def __add__(self, *args): return _morfeusz2.SwigPyIterator___add__(self, *args)
  94 + def __sub__(self, *args): return _morfeusz2.SwigPyIterator___sub__(self, *args)
  95 + def __iter__(self): return self
  96 +SwigPyIterator_swigregister = _morfeusz2.SwigPyIterator_swigregister
  97 +SwigPyIterator_swigregister(SwigPyIterator)
  98 +
  99 +class InterpsList(_object):
  100 + __swig_setmethods__ = {}
  101 + __setattr__ = lambda self, name, value: _swig_setattr(self, InterpsList, name, value)
  102 + __swig_getmethods__ = {}
  103 + __getattr__ = lambda self, name: _swig_getattr(self, InterpsList, name)
  104 + __repr__ = _swig_repr
  105 + def iterator(self): return _morfeusz2.InterpsList_iterator(self)
  106 + def __iter__(self): return self.iterator()
  107 + def __nonzero__(self): return _morfeusz2.InterpsList___nonzero__(self)
  108 + def __bool__(self): return _morfeusz2.InterpsList___bool__(self)
  109 + def __len__(self): return _morfeusz2.InterpsList___len__(self)
  110 + def pop(self): return _morfeusz2.InterpsList_pop(self)
  111 + def __getslice__(self, *args): return _morfeusz2.InterpsList___getslice__(self, *args)
  112 + def __setslice__(self, *args): return _morfeusz2.InterpsList___setslice__(self, *args)
  113 + def __delslice__(self, *args): return _morfeusz2.InterpsList___delslice__(self, *args)
  114 + def __delitem__(self, *args): return _morfeusz2.InterpsList___delitem__(self, *args)
  115 + def __getitem__(self, *args): return _morfeusz2.InterpsList___getitem__(self, *args)
  116 + def __setitem__(self, *args): return _morfeusz2.InterpsList___setitem__(self, *args)
  117 + def append(self, *args): return _morfeusz2.InterpsList_append(self, *args)
  118 + def empty(self): return _morfeusz2.InterpsList_empty(self)
  119 + def size(self): return _morfeusz2.InterpsList_size(self)
  120 + def clear(self): return _morfeusz2.InterpsList_clear(self)
  121 + def swap(self, *args): return _morfeusz2.InterpsList_swap(self, *args)
  122 + def get_allocator(self): return _morfeusz2.InterpsList_get_allocator(self)
  123 + def begin(self): return _morfeusz2.InterpsList_begin(self)
  124 + def end(self): return _morfeusz2.InterpsList_end(self)
  125 + def rbegin(self): return _morfeusz2.InterpsList_rbegin(self)
  126 + def rend(self): return _morfeusz2.InterpsList_rend(self)
  127 + def pop_back(self): return _morfeusz2.InterpsList_pop_back(self)
  128 + def erase(self, *args): return _morfeusz2.InterpsList_erase(self, *args)
  129 + def __init__(self, *args):
  130 + this = _morfeusz2.new_InterpsList(*args)
  131 + try: self.this.append(this)
  132 + except: self.this = this
  133 + def push_back(self, *args): return _morfeusz2.InterpsList_push_back(self, *args)
  134 + def front(self): return _morfeusz2.InterpsList_front(self)
  135 + def back(self): return _morfeusz2.InterpsList_back(self)
  136 + def assign(self, *args): return _morfeusz2.InterpsList_assign(self, *args)
  137 + def resize(self, *args): return _morfeusz2.InterpsList_resize(self, *args)
  138 + def insert(self, *args): return _morfeusz2.InterpsList_insert(self, *args)
  139 + def reserve(self, *args): return _morfeusz2.InterpsList_reserve(self, *args)
  140 + def capacity(self): return _morfeusz2.InterpsList_capacity(self)
  141 + __swig_destroy__ = _morfeusz2.delete_InterpsList
  142 + __del__ = lambda self : None;
  143 +InterpsList_swigregister = _morfeusz2.InterpsList_swigregister
  144 +InterpsList_swigregister(InterpsList)
  145 +
  146 +class StringsList(_object):
  147 + __swig_setmethods__ = {}
  148 + __setattr__ = lambda self, name, value: _swig_setattr(self, StringsList, name, value)
  149 + __swig_getmethods__ = {}
  150 + __getattr__ = lambda self, name: _swig_getattr(self, StringsList, name)
  151 + __repr__ = _swig_repr
  152 + def iterator(self): return _morfeusz2.StringsList_iterator(self)
  153 + def __iter__(self): return self.iterator()
  154 + def __nonzero__(self): return _morfeusz2.StringsList___nonzero__(self)
  155 + def __bool__(self): return _morfeusz2.StringsList___bool__(self)
  156 + def __len__(self): return _morfeusz2.StringsList___len__(self)
  157 + def pop(self): return _morfeusz2.StringsList_pop(self)
  158 + def __getslice__(self, *args): return _morfeusz2.StringsList___getslice__(self, *args)
  159 + def __setslice__(self, *args): return _morfeusz2.StringsList___setslice__(self, *args)
  160 + def __delslice__(self, *args): return _morfeusz2.StringsList___delslice__(self, *args)
  161 + def __delitem__(self, *args): return _morfeusz2.StringsList___delitem__(self, *args)
  162 + def __getitem__(self, *args): return _morfeusz2.StringsList___getitem__(self, *args)
  163 + def __setitem__(self, *args): return _morfeusz2.StringsList___setitem__(self, *args)
  164 + def append(self, *args): return _morfeusz2.StringsList_append(self, *args)
  165 + def empty(self): return _morfeusz2.StringsList_empty(self)
  166 + def size(self): return _morfeusz2.StringsList_size(self)
  167 + def clear(self): return _morfeusz2.StringsList_clear(self)
  168 + def swap(self, *args): return _morfeusz2.StringsList_swap(self, *args)
  169 + def get_allocator(self): return _morfeusz2.StringsList_get_allocator(self)
  170 + def begin(self): return _morfeusz2.StringsList_begin(self)
  171 + def end(self): return _morfeusz2.StringsList_end(self)
  172 + def rbegin(self): return _morfeusz2.StringsList_rbegin(self)
  173 + def rend(self): return _morfeusz2.StringsList_rend(self)
  174 + def pop_back(self): return _morfeusz2.StringsList_pop_back(self)
  175 + def erase(self, *args): return _morfeusz2.StringsList_erase(self, *args)
  176 + def __init__(self, *args):
  177 + this = _morfeusz2.new_StringsList(*args)
  178 + try: self.this.append(this)
  179 + except: self.this = this
  180 + def push_back(self, *args): return _morfeusz2.StringsList_push_back(self, *args)
  181 + def front(self): return _morfeusz2.StringsList_front(self)
  182 + def back(self): return _morfeusz2.StringsList_back(self)
  183 + def assign(self, *args): return _morfeusz2.StringsList_assign(self, *args)
  184 + def resize(self, *args): return _morfeusz2.StringsList_resize(self, *args)
  185 + def insert(self, *args): return _morfeusz2.StringsList_insert(self, *args)
  186 + def reserve(self, *args): return _morfeusz2.StringsList_reserve(self, *args)
  187 + def capacity(self): return _morfeusz2.StringsList_capacity(self)
  188 + __swig_destroy__ = _morfeusz2.delete_StringsList
  189 + __del__ = lambda self : None;
  190 +StringsList_swigregister = _morfeusz2.StringsList_swigregister
  191 +StringsList_swigregister(StringsList)
  192 +
  193 +class StringsLinkedList(_object):
  194 + __swig_setmethods__ = {}
  195 + __setattr__ = lambda self, name, value: _swig_setattr(self, StringsLinkedList, name, value)
  196 + __swig_getmethods__ = {}
  197 + __getattr__ = lambda self, name: _swig_getattr(self, StringsLinkedList, name)
  198 + __repr__ = _swig_repr
  199 + def iterator(self): return _morfeusz2.StringsLinkedList_iterator(self)
  200 + def __iter__(self): return self.iterator()
  201 + def __nonzero__(self): return _morfeusz2.StringsLinkedList___nonzero__(self)
  202 + def __bool__(self): return _morfeusz2.StringsLinkedList___bool__(self)
  203 + def __len__(self): return _morfeusz2.StringsLinkedList___len__(self)
  204 + def pop(self): return _morfeusz2.StringsLinkedList_pop(self)
  205 + def __getslice__(self, *args): return _morfeusz2.StringsLinkedList___getslice__(self, *args)
  206 + def __setslice__(self, *args): return _morfeusz2.StringsLinkedList___setslice__(self, *args)
  207 + def __delslice__(self, *args): return _morfeusz2.StringsLinkedList___delslice__(self, *args)
  208 + def __delitem__(self, *args): return _morfeusz2.StringsLinkedList___delitem__(self, *args)
  209 + def __getitem__(self, *args): return _morfeusz2.StringsLinkedList___getitem__(self, *args)
  210 + def __setitem__(self, *args): return _morfeusz2.StringsLinkedList___setitem__(self, *args)
  211 + def append(self, *args): return _morfeusz2.StringsLinkedList_append(self, *args)
  212 + def empty(self): return _morfeusz2.StringsLinkedList_empty(self)
  213 + def size(self): return _morfeusz2.StringsLinkedList_size(self)
  214 + def clear(self): return _morfeusz2.StringsLinkedList_clear(self)
  215 + def swap(self, *args): return _morfeusz2.StringsLinkedList_swap(self, *args)
  216 + def get_allocator(self): return _morfeusz2.StringsLinkedList_get_allocator(self)
  217 + def begin(self): return _morfeusz2.StringsLinkedList_begin(self)
  218 + def end(self): return _morfeusz2.StringsLinkedList_end(self)
  219 + def rbegin(self): return _morfeusz2.StringsLinkedList_rbegin(self)
  220 + def rend(self): return _morfeusz2.StringsLinkedList_rend(self)
  221 + def pop_back(self): return _morfeusz2.StringsLinkedList_pop_back(self)
  222 + def erase(self, *args): return _morfeusz2.StringsLinkedList_erase(self, *args)
  223 + def __init__(self, *args):
  224 + this = _morfeusz2.new_StringsLinkedList(*args)
  225 + try: self.this.append(this)
  226 + except: self.this = this
  227 + def push_back(self, *args): return _morfeusz2.StringsLinkedList_push_back(self, *args)
  228 + def front(self): return _morfeusz2.StringsLinkedList_front(self)
  229 + def back(self): return _morfeusz2.StringsLinkedList_back(self)
  230 + def assign(self, *args): return _morfeusz2.StringsLinkedList_assign(self, *args)
  231 + def resize(self, *args): return _morfeusz2.StringsLinkedList_resize(self, *args)
  232 + def insert(self, *args): return _morfeusz2.StringsLinkedList_insert(self, *args)
  233 + def pop_front(self): return _morfeusz2.StringsLinkedList_pop_front(self)
  234 + def push_front(self, *args): return _morfeusz2.StringsLinkedList_push_front(self, *args)
  235 + def reverse(self): return _morfeusz2.StringsLinkedList_reverse(self)
  236 + __swig_destroy__ = _morfeusz2.delete_StringsLinkedList
  237 + __del__ = lambda self : None;
  238 +StringsLinkedList_swigregister = _morfeusz2.StringsLinkedList_swigregister
  239 +StringsLinkedList_swigregister(StringsLinkedList)
  240 +
  241 +class StringsSet(_object):
  242 + __swig_setmethods__ = {}
  243 + __setattr__ = lambda self, name, value: _swig_setattr(self, StringsSet, name, value)
  244 + __swig_getmethods__ = {}
  245 + __getattr__ = lambda self, name: _swig_getattr(self, StringsSet, name)
  246 + __repr__ = _swig_repr
  247 + def iterator(self): return _morfeusz2.StringsSet_iterator(self)
  248 + def __iter__(self): return self.iterator()
  249 + def __nonzero__(self): return _morfeusz2.StringsSet___nonzero__(self)
  250 + def __bool__(self): return _morfeusz2.StringsSet___bool__(self)
  251 + def __len__(self): return _morfeusz2.StringsSet___len__(self)
  252 + def append(self, *args): return _morfeusz2.StringsSet_append(self, *args)
  253 + def __contains__(self, *args): return _morfeusz2.StringsSet___contains__(self, *args)
  254 + def __getitem__(self, *args): return _morfeusz2.StringsSet___getitem__(self, *args)
  255 + def __init__(self, *args):
  256 + this = _morfeusz2.new_StringsSet(*args)
  257 + try: self.this.append(this)
  258 + except: self.this = this
  259 + def empty(self): return _morfeusz2.StringsSet_empty(self)
  260 + def size(self): return _morfeusz2.StringsSet_size(self)
  261 + def clear(self): return _morfeusz2.StringsSet_clear(self)
  262 + def swap(self, *args): return _morfeusz2.StringsSet_swap(self, *args)
  263 + def count(self, *args): return _morfeusz2.StringsSet_count(self, *args)
  264 + def begin(self): return _morfeusz2.StringsSet_begin(self)
  265 + def end(self): return _morfeusz2.StringsSet_end(self)
  266 + def rbegin(self): return _morfeusz2.StringsSet_rbegin(self)
  267 + def rend(self): return _morfeusz2.StringsSet_rend(self)
  268 + def erase(self, *args): return _morfeusz2.StringsSet_erase(self, *args)
  269 + def find(self, *args): return _morfeusz2.StringsSet_find(self, *args)
  270 + def lower_bound(self, *args): return _morfeusz2.StringsSet_lower_bound(self, *args)
  271 + def upper_bound(self, *args): return _morfeusz2.StringsSet_upper_bound(self, *args)
  272 + def equal_range(self, *args): return _morfeusz2.StringsSet_equal_range(self, *args)
  273 + def insert(self, *args): return _morfeusz2.StringsSet_insert(self, *args)
  274 + __swig_destroy__ = _morfeusz2.delete_StringsSet
  275 + __del__ = lambda self : None;
  276 +StringsSet_swigregister = _morfeusz2.StringsSet_swigregister
  277 +StringsSet_swigregister(StringsSet)
  278 +
  279 +SEPARATE_NUMBERING = _morfeusz2.SEPARATE_NUMBERING
  280 +CONTINUOUS_NUMBERING = _morfeusz2.CONTINUOUS_NUMBERING
  281 +CONDITIONALLY_CASE_SENSITIVE = _morfeusz2.CONDITIONALLY_CASE_SENSITIVE
  282 +STRICTLY_CASE_SENSITIVE = _morfeusz2.STRICTLY_CASE_SENSITIVE
  283 +IGNORE_CASE = _morfeusz2.IGNORE_CASE
  284 +SKIP_WHITESPACES = _morfeusz2.SKIP_WHITESPACES
  285 +APPEND_WHITESPACES = _morfeusz2.APPEND_WHITESPACES
  286 +KEEP_WHITESPACES = _morfeusz2.KEEP_WHITESPACES
  287 +ANALYSE_ONLY = _morfeusz2.ANALYSE_ONLY
  288 +GENERATE_ONLY = _morfeusz2.GENERATE_ONLY
  289 +BOTH_ANALYSE_AND_GENERATE = _morfeusz2.BOTH_ANALYSE_AND_GENERATE
  290 +class Morfeusz(_object):
  291 + __swig_setmethods__ = {}
  292 + __setattr__ = lambda self, name, value: _swig_setattr(self, Morfeusz, name, value)
  293 + __swig_getmethods__ = {}
  294 + __getattr__ = lambda self, name: _swig_getattr(self, Morfeusz, name)
  295 + def __init__(self, *args, **kwargs): raise AttributeError("No constructor defined - class is abstract")
  296 + __repr__ = _swig_repr
  297 + __swig_getmethods__["getVersion"] = lambda x: _morfeusz2.Morfeusz_getVersion
  298 + if _newclass:getVersion = staticmethod(_morfeusz2.Morfeusz_getVersion)
  299 + __swig_getmethods__["getDefaultDictName"] = lambda x: _morfeusz2.Morfeusz_getDefaultDictName
  300 + if _newclass:getDefaultDictName = staticmethod(_morfeusz2.Morfeusz_getDefaultDictName)
  301 + __swig_getmethods__["getCopyright"] = lambda x: _morfeusz2.Morfeusz_getCopyright
  302 + if _newclass:getCopyright = staticmethod(_morfeusz2.Morfeusz_getCopyright)
  303 + __swig_getmethods__["createInstance"] = lambda x: _morfeusz2.Morfeusz_createInstance
  304 + if _newclass:createInstance = staticmethod(_morfeusz2.Morfeusz_createInstance)
  305 + def getDictID(self): return _morfeusz2.Morfeusz_getDictID(self)
  306 + def getDictCopyright(self): return _morfeusz2.Morfeusz_getDictCopyright(self)
  307 + def clone(self): return _morfeusz2.Morfeusz_clone(self)
  308 + __swig_destroy__ = _morfeusz2.delete_Morfeusz
  309 + __del__ = lambda self : None;
  310 + def analyse(self, text):
  311 + """
  312 + Analyse given text and return a list of MorphInterpretation objects.
  313 + """
  314 + res = InterpsList()
  315 + _morfeusz2.Morfeusz_analyse(self, text, res)
  316 + return res
  317 +
  318 +
  319 + def generate(self, lemma, tagId=None):
  320 + """
  321 + Perform morphological synthesis on given text and return a list of MorphInterpretation objects.
  322 + """
  323 + if tagId is not None:
  324 + return self._generateByTagId(lemma, tagId)
  325 + else:
  326 + res = InterpsList()
  327 + _morfeusz2.Morfeusz_generate(self, lemma, res)
  328 + return res
  329 +
  330 +
  331 + def setAggl(self, optionString):
  332 + """
  333 + Select agglutination rules option
  334 + """
  335 + _morfeusz2.Morfeusz_setAggl(self, optionString.encode('utf8'))
  336 +
  337 +
  338 + def getAggl(self): return _morfeusz2.Morfeusz_getAggl(self)
  339 + def setPraet(self, optionString):
  340 + """
  341 + Select past tense segmentation
  342 + """
  343 + _morfeusz2.Morfeusz_setPraet(self, optionString.encode('utf8'))
  344 +
  345 +
  346 + def getPraet(self): return _morfeusz2.Morfeusz_getPraet(self)
  347 + def setCaseHandling(self, option):
  348 + """
  349 + Set case handling option (valid options are CONDITIONALLY_CASE_SENSITIVE, STRICTLY_CASE_SENSITIVE, IGNORE_CASE)
  350 + """
  351 + _morfeusz2.Morfeusz_setCaseHandling(self, option)
  352 +
  353 +
  354 + def getCaseHandling(self): return _morfeusz2.Morfeusz_getCaseHandling(self)
  355 + def setTokenNumbering(self, option):
  356 + """
  357 + Set token numbering option (valid options are SEPARATE_NUMBERING, CONTINUOUS_NUMBERING)
  358 + """
  359 + _morfeusz2.Morfeusz_setTokenNumbering(self, option)
  360 +
  361 +
  362 + def getTokenNumbering(self): return _morfeusz2.Morfeusz_getTokenNumbering(self)
  363 + def setWhitespaceHandling(self, option):
  364 + """
  365 + Set whitespace handling handling option (valid options are SKIP_WHITESPACES, KEEP_WHITESPACES, APPEND_WHITESPACES)
  366 + """
  367 + _morfeusz2.Morfeusz_setWhitespaceHandling(self, option)
  368 +
  369 +
  370 + def getWhitespaceHandling(self): return _morfeusz2.Morfeusz_getWhitespaceHandling(self)
  371 + def getIdResolver(self): return _morfeusz2.Morfeusz_getIdResolver(self)
  372 + def setDictionary(self, dictName):
  373 + """
  374 + Set dictionary to be used by this instance (by name)
  375 + """
  376 + _morfeusz2.Morfeusz_setDictionary(self, dictName.encode('utf8'))
  377 +
  378 +
  379 + __swig_setmethods__["dictionarySearchPaths"] = _morfeusz2.Morfeusz_dictionarySearchPaths_set
  380 + __swig_getmethods__["dictionarySearchPaths"] = _morfeusz2.Morfeusz_dictionarySearchPaths_get
  381 + if _newclass:dictionarySearchPaths = _swig_property(_morfeusz2.Morfeusz_dictionarySearchPaths_get, _morfeusz2.Morfeusz_dictionarySearchPaths_set)
  382 + def getAvailableAgglOptions(self): return _morfeusz2.Morfeusz_getAvailableAgglOptions(self)
  383 + def getAvailablePraetOptions(self): return _morfeusz2.Morfeusz_getAvailablePraetOptions(self)
  384 + def _generateByTagId(self, *args): return _morfeusz2.Morfeusz__generateByTagId(self, *args)
  385 + def analyse_iter(self, text):
  386 + """
  387 + Analyse given text and return an iterator over MorphInterpretation objects as a result.
  388 + """
  389 + return _morfeusz2.Morfeusz__analyseAsIterator(self, text)
  390 +
  391 +
  392 +Morfeusz_swigregister = _morfeusz2.Morfeusz_swigregister
  393 +Morfeusz_swigregister(Morfeusz)
  394 +
  395 +def Morfeusz_getVersion():
  396 + return _morfeusz2.Morfeusz_getVersion()
  397 +Morfeusz_getVersion = _morfeusz2.Morfeusz_getVersion
  398 +
  399 +def Morfeusz_getDefaultDictName():
  400 + return _morfeusz2.Morfeusz_getDefaultDictName()
  401 +Morfeusz_getDefaultDictName = _morfeusz2.Morfeusz_getDefaultDictName
  402 +
  403 +def Morfeusz_getCopyright():
  404 + return _morfeusz2.Morfeusz_getCopyright()
  405 +Morfeusz_getCopyright = _morfeusz2.Morfeusz_getCopyright
  406 +
  407 +def Morfeusz_createInstance(*args):
  408 + return _morfeusz2.Morfeusz_createInstance(*args)
  409 +Morfeusz_createInstance = _morfeusz2.Morfeusz_createInstance
  410 +cvar = _morfeusz2.cvar
  411 +
  412 +class ResultsIterator(_object):
  413 + __swig_setmethods__ = {}
  414 + __setattr__ = lambda self, name, value: _swig_setattr(self, ResultsIterator, name, value)
  415 + __swig_getmethods__ = {}
  416 + __getattr__ = lambda self, name: _swig_getattr(self, ResultsIterator, name)
  417 + def __init__(self, *args, **kwargs): raise AttributeError("No constructor defined - class is abstract")
  418 + __repr__ = _swig_repr
  419 + def hasNext(self): return _morfeusz2.ResultsIterator_hasNext(self)
  420 + def peek(self): return _morfeusz2.ResultsIterator_peek(self)
  421 + def next(self):
  422 + if self.hasNext():
  423 + return _morfeusz2.ResultsIterator_next(self)
  424 + else:
  425 + raise StopIteration
  426 +
  427 +
  428 + __swig_destroy__ = _morfeusz2.delete_ResultsIterator
  429 + __del__ = lambda self : None;
  430 + def __iter__(self): return _morfeusz2.ResultsIterator___iter__(self)
  431 +ResultsIterator_swigregister = _morfeusz2.ResultsIterator_swigregister
  432 +ResultsIterator_swigregister(ResultsIterator)
  433 +
  434 +class IdResolver(_object):
  435 + __swig_setmethods__ = {}
  436 + __setattr__ = lambda self, name, value: _swig_setattr(self, IdResolver, name, value)
  437 + __swig_getmethods__ = {}
  438 + __getattr__ = lambda self, name: _swig_getattr(self, IdResolver, name)
  439 + def __init__(self, *args, **kwargs): raise AttributeError("No constructor defined - class is abstract")
  440 + __repr__ = _swig_repr
  441 + def getTagsetId(self): return _morfeusz2.IdResolver_getTagsetId(self)
  442 + def getTag(self, tagId):
  443 + return _morfeusz2.IdResolver_getTag(self, tagId).decode('utf8')
  444 +
  445 +
  446 + def getTagId(self, tag):
  447 + return _morfeusz2.IdResolver_getTagId(self, tag.encode('utf8'))
  448 +
  449 +
  450 + def getName(self, nameId):
  451 + return _morfeusz2.IdResolver_getName(self, nameId).decode('utf8')
  452 +
  453 +
  454 + def getNameId(self, name):
  455 + return _morfeusz2.IdResolver_getNameId(self, name.encode('utf8'))
  456 +
  457 +
  458 + def getLabelsAsUnicode(self, labelsId):
  459 + return _morfeusz2.IdResolver_getLabelsAsString(self, labelsId).decode('utf8')
  460 +
  461 +
  462 + def getLabels(self, labelsId):
  463 + return { l.decode('utf8') for l in _morfeusz2.IdResolver_getLabels(self, labelsId) }
  464 +
  465 +
  466 + def getLabelsId(self, labelsStr):
  467 + return _morfeusz2.IdResolver_getLabelsId(self, labelsStr.encode('utf8'))
  468 +
  469 +
  470 + def getTagsCount(self): return _morfeusz2.IdResolver_getTagsCount(self)
  471 + def getNamesCount(self): return _morfeusz2.IdResolver_getNamesCount(self)
  472 + def getLabelsCount(self): return _morfeusz2.IdResolver_getLabelsCount(self)
  473 + __swig_destroy__ = _morfeusz2.delete_IdResolver
  474 + __del__ = lambda self : None;
  475 +IdResolver_swigregister = _morfeusz2.IdResolver_swigregister
  476 +IdResolver_swigregister(IdResolver)
  477 +
  478 +class MorphInterpretation(_object):
  479 + __swig_setmethods__ = {}
  480 + __setattr__ = lambda self, name, value: _swig_setattr(self, MorphInterpretation, name, value)
  481 + __swig_getmethods__ = {}
  482 + __getattr__ = lambda self, name: _swig_getattr(self, MorphInterpretation, name)
  483 + __repr__ = _swig_repr
  484 + def __init__(self):
  485 + this = _morfeusz2.new_MorphInterpretation()
  486 + try: self.this.append(this)
  487 + except: self.this = this
  488 + __swig_getmethods__["createIgn"] = lambda x: _morfeusz2.MorphInterpretation_createIgn
  489 + if _newclass:createIgn = staticmethod(_morfeusz2.MorphInterpretation_createIgn)
  490 + __swig_getmethods__["createWhitespace"] = lambda x: _morfeusz2.MorphInterpretation_createWhitespace
  491 + if _newclass:createWhitespace = staticmethod(_morfeusz2.MorphInterpretation_createWhitespace)
  492 + def isIgn(self): return _morfeusz2.MorphInterpretation_isIgn(self)
  493 + def isWhitespace(self): return _morfeusz2.MorphInterpretation_isWhitespace(self)
  494 + def getTag(self, morfeusz):
  495 + """
  496 + Returns tag as string.
  497 + """
  498 + return _morfeusz2.MorphInterpretation_getTag(self, morfeusz)
  499 +
  500 +
  501 + def getName(self, morfeusz):
  502 + """
  503 + Returns this interpretation named entity as string
  504 + """
  505 + return _morfeusz2.MorphInterpretation_getName(self, morfeusz)
  506 +
  507 +
  508 + def getLabelsAsUnicode(self, morfeusz):
  509 + """
  510 + Returns this interpretation labels as string
  511 + """
  512 + return _morfeusz2.MorphInterpretation_getLabelsAsString(self, morfeusz).decode('utf8')
  513 +
  514 +
  515 + def getLabels(self, morfeusz):
  516 + """
  517 + Returns this interpretation labels as a set of strings
  518 + """
  519 + return { l.decode('utf8') for l in _morfeusz2.MorphInterpretation_getLabels(self, morfeusz) }
  520 +
  521 +
  522 + __swig_setmethods__["startNode"] = _morfeusz2.MorphInterpretation_startNode_set
  523 + __swig_getmethods__["startNode"] = _morfeusz2.MorphInterpretation_startNode_get
  524 + if _newclass:startNode = _swig_property(_morfeusz2.MorphInterpretation_startNode_get, _morfeusz2.MorphInterpretation_startNode_set)
  525 + __swig_setmethods__["endNode"] = _morfeusz2.MorphInterpretation_endNode_set
  526 + __swig_getmethods__["endNode"] = _morfeusz2.MorphInterpretation_endNode_get
  527 + if _newclass:endNode = _swig_property(_morfeusz2.MorphInterpretation_endNode_get, _morfeusz2.MorphInterpretation_endNode_set)
  528 + __swig_setmethods__["_orth"] = _morfeusz2.MorphInterpretation__orth_set
  529 + __swig_getmethods__["_orth"] = _morfeusz2.MorphInterpretation__orth_get
  530 + if _newclass:_orth = _swig_property(_morfeusz2.MorphInterpretation__orth_get, _morfeusz2.MorphInterpretation__orth_set)
  531 + __swig_setmethods__["_lemma"] = _morfeusz2.MorphInterpretation__lemma_set
  532 + __swig_getmethods__["_lemma"] = _morfeusz2.MorphInterpretation__lemma_get
  533 + if _newclass:_lemma = _swig_property(_morfeusz2.MorphInterpretation__lemma_get, _morfeusz2.MorphInterpretation__lemma_set)
  534 + __swig_setmethods__["tagId"] = _morfeusz2.MorphInterpretation_tagId_set
  535 + __swig_getmethods__["tagId"] = _morfeusz2.MorphInterpretation_tagId_get
  536 + if _newclass:tagId = _swig_property(_morfeusz2.MorphInterpretation_tagId_get, _morfeusz2.MorphInterpretation_tagId_set)
  537 + __swig_setmethods__["nameId"] = _morfeusz2.MorphInterpretation_nameId_set
  538 + __swig_getmethods__["nameId"] = _morfeusz2.MorphInterpretation_nameId_get
  539 + if _newclass:nameId = _swig_property(_morfeusz2.MorphInterpretation_nameId_get, _morfeusz2.MorphInterpretation_nameId_set)
  540 + __swig_setmethods__["labelsId"] = _morfeusz2.MorphInterpretation_labelsId_set
  541 + __swig_getmethods__["labelsId"] = _morfeusz2.MorphInterpretation_labelsId_get
  542 + if _newclass:labelsId = _swig_property(_morfeusz2.MorphInterpretation_labelsId_get, _morfeusz2.MorphInterpretation_labelsId_set)
  543 + @property
  544 + def orth(self):
  545 + return self._orth.decode('utf8')
  546 +
  547 + @orth.setter
  548 + def orth(self, val):
  549 + self._orth = val.encode('utf8')
  550 +
  551 + @property
  552 + def lemma(self):
  553 + return self._lemma.decode('utf8')
  554 +
  555 + @lemma.setter
  556 + def lemma(self, val):
  557 + self._lemma = val.encode('utf8')
  558 +
  559 + __swig_destroy__ = _morfeusz2.delete_MorphInterpretation
  560 + __del__ = lambda self : None;
  561 +MorphInterpretation_swigregister = _morfeusz2.MorphInterpretation_swigregister
  562 +MorphInterpretation_swigregister(MorphInterpretation)
  563 +
  564 +def MorphInterpretation_createIgn(*args):
  565 + return _morfeusz2.MorphInterpretation_createIgn(*args)
  566 +MorphInterpretation_createIgn = _morfeusz2.MorphInterpretation_createIgn
  567 +
  568 +def MorphInterpretation_createWhitespace(*args):
  569 + return _morfeusz2.MorphInterpretation_createWhitespace(*args)
  570 +MorphInterpretation_createWhitespace = _morfeusz2.MorphInterpretation_createWhitespace
  571 +
  572 +# This file is compatible with both classic and new-style classes.
  573 +
  574 +
... ...
dictionary/ajax_argument_form.py
  View file @92fd5c3
... ... @@ -27,7 +27,6 @@ from django.db.models import Count, Q, Sum
27 27 from accounts.models import can_modify_phraseology_only
28 28 from common.decorators import render, ajax, AjaxError
29 29 from common.js_to_obj import jsArgToObj
30   -from common.morfeusz import analyse
31 30 from dictionary.forms import AddArgumentForm, ArgPropositionsForm, Atribute_Model, \
32 31 AtributeChoiceForm, AtributeTextForm, ValueAttrMultiValueForm, \
33 32 TextAttrMultiValueForm, PositionsForm, SelectArgumentForm, \
... ... @@ -38,6 +37,7 @@ from dictionary.models import Argument, Argument_Model, Atribute, Atribute_Value
38 37 sortatributes, sortPositions, sortArguments, get_or_create_attr_parameter, \
39 38 get_attr_models_to_exclude, is_morfeusz_exception, get_or_create_attribute, \
40 39 get_or_create_parameter_attr_value
  40 +from settings import MORFEUSZ2
41 41  
42 42 @render('argument_form.html')
43 43 @ajax(method='post', encode_result=False)
... ... @@ -317,7 +317,6 @@ def create_type_form(pos, arg_model, phraseologic_modification=False,
317 317 def create_attributes_forms(pos, arg_model, subforms_values, form_type='standard'):
318 318 sheets = []
319 319 attribute_models = get_attribute_models(arg_model, subforms_values)
320   - #attr_values_to_exclude = get_attr_values_to_exclude(arg_model)
321 320 param_models_to_exclude = get_parameter_types_to_exclude(arg_model)
322 321 for i in range(len(attribute_models)):
323 322 attr_form_values = []
... ... @@ -488,8 +487,8 @@ def get_parameter_types_to_exclude(arg_model):
488 487 def is_correct_lemma(argument_model, attribute_model, lemma):
489 488 correct_form = False
490 489 possible_pos_tags = argument_model.get_possible_lemma_tags(attribute_model)
491   - for interp in analyse(lemma):
492   - if (is_single_word(interp) and base_form_correct(interp, lemma) and
  490 + for interp in MORFEUSZ2.analyse(lemma.encode('utf8')):
  491 + if (base_form_correct(interp, lemma) and
493 492 pos_tag_correct(interp, possible_pos_tags)):
494 493 correct_form = True
495 494 break
... ... @@ -501,18 +500,13 @@ def is_correct_lemma(argument_model, attribute_model, lemma):
501 500 break
502 501 return correct_form
503 502  
504   -def is_single_word(interp):
505   - if len(interp) == 1:
506   - return True
507   - return False
508   -
509 503 def base_form_correct(interp, lemma):
510   - if interp[0][1] == lemma:
  504 + if interp.lemma == lemma:
511 505 return True
512 506 return False
513 507  
514 508 def pos_tag_correct(interp, possible_pos_tags):
515   - tagstr = interp[0][2]
  509 + tagstr = interp.getTag(MORFEUSZ2)
516 510 pos_tag = tagstr.split(':')[0]
517 511 if possible_pos_tags.filter(name=pos_tag).exists():
518 512 return True
... ... @@ -520,16 +514,18 @@ def pos_tag_correct(interp, possible_pos_tags):
520 514  
521 515 def contains_separator(lemma):
522 516 contains_separator = False
523   - for interp in analyse(lemma):
524   - if len(interp) > 1:
  517 + results_iter = MORFEUSZ2.analyse_iter(lemma.encode('utf8'))
  518 + while results_iter.hasNext():
  519 + if results_iter.peek().orth != lemma:
525 520 contains_separator = True
526 521 break
  522 + results_iter.next()
527 523 return contains_separator
528 524  
529 525 def is_preposition_case_pair_valid(preposition_obj, case_obj):
530 526 # postp is used by prepadjp
531 527 case_str = unicode(case_obj)
532   - prep_str = unicode(preposition_obj)
  528 + prep_str = unicode(preposition_obj).split()[-1]
533 529 if case_str != 'postp':
534 530 # str is used by prepadjp
535 531 if case_str == 'str':
... ... @@ -537,12 +533,14 @@ def is_preposition_case_pair_valid(preposition_obj, case_obj):
537 533 else:
538 534 pcase = [case_str]
539 535 for case in pcase:
540   - for interp in analyse(prep_str):
541   - for token in interp:
542   - tag = token[2].split(':')
543   - if tag[0] == 'prep':
544   - if tag[1] == case:
545   - return True
  536 + for interp in MORFEUSZ2.analyse(prep_str.encode('utf8')):
  537 + tagstr = interp.getTag(MORFEUSZ2)
  538 + tag_parts = tagstr.split(':')
  539 + if len(tag_parts) > 1:
  540 + interp_pos = tag_parts[0]
  541 + interp_case = tag_parts[1]
  542 + if interp_pos == 'prep' and interp_case == case:
  543 + return True
546 544 return False
547 545 return True
548 546  
... ...
dictionary/ajax_lemma_view.py
  View file @92fd5c3
... ... @@ -85,7 +85,6 @@ from accounts.models import UserSettings, RealizedLemma, UserStats, RealizedPhra
85 85 can_modify_phraseology_only, get_anon_profile
86 86  
87 87 from ajax_jqgrid import JqGridAjax, default_sort_rules, default_filter_rules
88   -from common.morfeusz import analyse
89 88  
90 89 import locale
91 90 from functools import cmp_to_key
... ...
dictionary/validation.py
  View file @92fd5c3
... ... @@ -27,13 +27,13 @@ import operator
27 27 from django.db.models import Sum, Q
28 28  
29 29 from common.js_to_obj import frameObjToSerializableDict
30   -from common.morfeusz import analyse
31 30 from dictionary.common_func import subframe_exists
32 31 from dictionary.convert_frames import frame_conversion
33 32 from dictionary.forms import FrameAspectForm, FrameOpinionForm
34 33 from dictionary.models import Argument_Model, AspectRelationsGroup, Frame_Char_Model, \
35 34 Frame_Characteristic, Lemma, Position, PositionCategory, \
36 35 get_attribute_models, sortArguments, sortPositions
  36 +from settings import MORFEUSZ2
37 37  
38 38 def get_wrong_aspect_frames(lemma, frames):
39 39 wrong_aspect_frames = []
... ... @@ -47,18 +47,18 @@ def check_aspect(lemma, frame):
47 47 inf_present = False
48 48 frame_aspect_obj = frame.characteristics.get(type=u'ASPEKT')
49 49 frame_aspect = frame_aspect_obj.value.value
50   - interps = analyse(lemma.entry)
51   - if interps[0][0][1] and frame_aspect != '_':
  50 + interps = MORFEUSZ2.analyse(lemma.entry.encode('utf8'))
  51 + if frame_aspect != '_':
52 52 for interp in interps:
53   - for token in interp:
54   - tag = token[2].split(':')
55   - if tag[0] == 'inf':
56   - inf_present = True
57   - if tag[1] == frame_aspect:
58   - good_aspect = True
59   - break
60   - if good_aspect:
61   - break
  53 + tagstr = interp.getTag(MORFEUSZ2)
  54 + tag_parts = tagstr.split(':')
  55 + pos = tag_parts[0]
  56 + if pos == 'inf':
  57 + aspect = tag_parts[1]
  58 + inf_present = True
  59 + if aspect == frame_aspect:
  60 + good_aspect = True
  61 + break
62 62 if good_aspect or not inf_present:
63 63 good_aspect = True
64 64 return good_aspect
... ... @@ -84,13 +84,13 @@ def get_missing_aspects_msg(lemma):
84 84  
85 85 def get_possible_aspects(lemma):
86 86 possible_aspects = []
87   - interps = analyse(lemma.entry)
88   - if interps[0][0][1]:
89   - for interp in interps:
90   - for token in interp:
91   - tag = token[2].split(':')
92   - if tag[0] == 'inf' and tag[1] not in possible_aspects:
93   - possible_aspects.append(tag[1])
  87 + interps = MORFEUSZ2.analyse(lemma.entry.encode('utf8'))
  88 + for interp in interps:
  89 + tagstr = interp.getTag(MORFEUSZ2)
  90 + tag_parts = tagstr.split(':')
  91 + pos = tag_parts[0]
  92 + if pos == 'inf' and tag_parts[1] not in possible_aspects:
  93 + possible_aspects.append(tag_parts[1])
94 94 return possible_aspects
95 95  
96 96 def match_arg_poss(arg_poss, frame):
... ... @@ -355,7 +355,7 @@ def prep_check(arg):
355 355 #if arg.find('prep') != -1 and not arg.startswith('comprepnp('):
356 356 if arg.startswith('prep'):
357 357 params = arg[arg.find('(')+1:arg.find(')')].split(',')
358   - pform = params[0]
  358 + pform = params[0].split()[-1]
359 359 pcase = params[1]
360 360 # postp is used by prepadjp
361 361 if pcase != 'postp':
... ... @@ -365,12 +365,12 @@ def prep_check(arg):
365 365 else:
366 366 pcase = [pcase]
367 367 for case in pcase:
368   - for interp in analyse(pform):
369   - for token in interp:
370   - tag = token[2].split(':')
371   - if tag[0] == 'prep':
372   - if tag[1] == case:
373   - return True
  368 + for interp in MORFEUSZ2.analyse(pform.encode('utf8')):
  369 + tagstr = interp.getTag(MORFEUSZ2)
  370 + tag_parts = tagstr.split(':')
  371 + pos = tag_parts[0]
  372 + if pos == 'prep' and tag_parts[1] == case:
  373 + return True
374 374 return False
375 375 return True
376 376  
... ...
manage.py
  View file @92fd5c3
... ... @@ -18,4 +18,4 @@ if __name__ == &quot;__main__&quot;:
18 18 #
19 19 # from django.core.management import execute_from_command_line
20 20 #
21 21 -# execute_from_command_line(sys.argv)
  22 +# execute_from_command_line(sys.argv)
22 23 \ No newline at end of file
... ...