Commit 05086d573805674d1d187bbd149cb08fb08f8384
1 parent
96182ff0
dodanie obsługi nazw własnych i kwalifikatorów w segmentacji
git-svn-id: svn://svn.nlp.ipipan.waw.pl/morfeusz/trunk@277 ff4e3ee1-f430-4e82-ade0-24591c43f1fd
Showing
26 changed files
with
2069 additions
and
351 deletions
fsabuilder/morfeusz_builder
@@ -180,15 +180,15 @@ def _concatFiles(inputFiles): | @@ -180,15 +180,15 @@ def _concatFiles(inputFiles): | ||
180 | 180 | ||
181 | def _readNamesAndQualifiers(inputFiles): | 181 | def _readNamesAndQualifiers(inputFiles): |
182 | names = set([u'']) | 182 | names = set([u'']) |
183 | - qualifiers = set([tuple(frozenset([u'']))]) | 183 | + qualifiers = set([frozenset()]) |
184 | for line in _concatFiles(inputFiles): | 184 | for line in _concatFiles(inputFiles): |
185 | line = line.strip().decode('utf8') | 185 | line = line.strip().decode('utf8') |
186 | if line: | 186 | if line: |
187 | _, _, _, name, qualifier = convertinput.parseLine(line) | 187 | _, _, _, name, qualifier = convertinput.parseLine(line) |
188 | names.add(name) | 188 | names.add(name) |
189 | - qualifiers.add(tuple(sorted(qualifier.split(u'|')))) | 189 | + qualifiers.add(convertinput.parseQualifiers(qualifier)) |
190 | namesMap = dict([(name, idx) for idx, name in enumerate(sorted(list(names)))]) | 190 | namesMap = dict([(name, idx) for idx, name in enumerate(sorted(list(names)))]) |
191 | - qualifiersMap = dict([(frozenset(quals), idx) for idx, quals in enumerate(sorted(list(qualifiers)))]) | 191 | + qualifiersMap = dict([(quals, idx) for idx, quals in enumerate(sorted(qualifiers, key=lambda q: tuple(sorted(q))))]) |
192 | exceptions.validate( | 192 | exceptions.validate( |
193 | len(qualifiersMap) <= limits.MAX_QUALIFIERS_COMBINATIONS, | 193 | len(qualifiersMap) <= limits.MAX_QUALIFIERS_COMBINATIONS, |
194 | u'Too many qualifiers combinations. The limit is %d' % limits.MAX_QUALIFIERS_COMBINATIONS) | 194 | u'Too many qualifiers combinations. The limit is %d' % limits.MAX_QUALIFIERS_COMBINATIONS) |
@@ -259,7 +259,7 @@ def _doBuildDictionaryPart(opts, tagset, namesMap, qualifiersMap, isGenerator): | @@ -259,7 +259,7 @@ def _doBuildDictionaryPart(opts, tagset, namesMap, qualifiersMap, isGenerator): | ||
259 | 259 | ||
260 | logging.info('reading segmentation rules') | 260 | logging.info('reading segmentation rules') |
261 | rulesParserVersion = rulesParser.RulesParser.PARSE4ANALYZER if not isGenerator else rulesParser.RulesParser.PARSE4GENERATOR | 261 | rulesParserVersion = rulesParser.RulesParser.PARSE4ANALYZER if not isGenerator else rulesParser.RulesParser.PARSE4GENERATOR |
262 | - segmentRulesManager = rulesParser.RulesParser(tagset, rulesParserVersion).parse(opts.segmentsFile) | 262 | + segmentRulesManager = rulesParser.RulesParser(tagset, namesMap, qualifiersMap, rulesParserVersion).parse(opts.segmentsFile) |
263 | segmentationRulesData = segmentRulesManager.serialize() | 263 | segmentationRulesData = segmentRulesManager.serialize() |
264 | logging.info('done reading segmentation rules') | 264 | logging.info('done reading segmentation rules') |
265 | 265 |
fsabuilder/morfeuszbuilder/fsa/convertinput.py
@@ -43,6 +43,12 @@ def parseLine(line): | @@ -43,6 +43,12 @@ def parseLine(line): | ||
43 | raise ValueError('input line "%s" does not have 3, 4 or 5 tab-separated fields' % line) | 43 | raise ValueError('input line "%s" does not have 3, 4 or 5 tab-separated fields' % line) |
44 | return orth, base, tag, name, qualifier | 44 | return orth, base, tag, name, qualifier |
45 | 45 | ||
46 | +def parseQualifiers(string): | ||
47 | + if string: | ||
48 | + return frozenset(string.split(u'|')) | ||
49 | + else: | ||
50 | + return frozenset() | ||
51 | + | ||
46 | class PolimorfConverter4Analyzer(object): | 52 | class PolimorfConverter4Analyzer(object): |
47 | 53 | ||
48 | def __init__(self, tagset, namesMap, qualifiersMap, encoder, inputEncoding, segmentRulesManager): | 54 | def __init__(self, tagset, namesMap, qualifiersMap, encoder, inputEncoding, segmentRulesManager): |
@@ -62,9 +68,9 @@ class PolimorfConverter4Analyzer(object): | @@ -62,9 +68,9 @@ class PolimorfConverter4Analyzer(object): | ||
62 | 68 | ||
63 | tagnum = self.tagset.getTagnum4Tag(tag) | 69 | tagnum = self.tagset.getTagnum4Tag(tag) |
64 | namenum = self.namesMap[name] | 70 | namenum = self.namesMap[name] |
65 | - typenum = self.segmentRulesManager.lexeme2SegmentTypeNum(base, tagnum) | ||
66 | - qualifiers = qualifier.split('|') if qualifier else frozenset([u'']) | ||
67 | - qualsnum = self.qualifiersMap[frozenset(qualifiers)] | 71 | + qualifiers = parseQualifiers(qualifier) |
72 | + qualsnum = self.qualifiersMap[qualifiers] | ||
73 | + typenum = self.segmentRulesManager.lexeme2SegmentTypeNum(base, tagnum, namenum, qualsnum) | ||
68 | 74 | ||
69 | assert not ( | 75 | assert not ( |
70 | self.segmentRulesManager.shiftOrthMagic.shouldReplaceLemmaWithOrth(typenum) | 76 | self.segmentRulesManager.shiftOrthMagic.shouldReplaceLemmaWithOrth(typenum) |
@@ -138,9 +144,9 @@ class PolimorfConverter4Generator(object): | @@ -138,9 +144,9 @@ class PolimorfConverter4Generator(object): | ||
138 | base, homonymId = assumedBase, assumedHomonymId | 144 | base, homonymId = assumedBase, assumedHomonymId |
139 | tagnum = self.tagset.getTagnum4Tag(tag) | 145 | tagnum = self.tagset.getTagnum4Tag(tag) |
140 | namenum = self.namesMap[name] | 146 | namenum = self.namesMap[name] |
141 | - qualifiers = qualifier.split('|') if qualifier else frozenset([u'']) | ||
142 | - qualsnum = self.qualifiersMap[frozenset(qualifiers)] | ||
143 | - typenum = self.segmentRulesManager.lexeme2SegmentTypeNum(base, tagnum) | 147 | + qualifiers = parseQualifiers(qualifier) |
148 | + qualsnum = self.qualifiersMap[qualifiers] | ||
149 | + typenum = self.segmentRulesManager.lexeme2SegmentTypeNum(base, tagnum, namenum, qualsnum) | ||
144 | 150 | ||
145 | if self.segmentRulesManager.shiftOrthMagic.shouldReplaceLemmaWithOrth(typenum): | 151 | if self.segmentRulesManager.shiftOrthMagic.shouldReplaceLemmaWithOrth(typenum): |
146 | # print 'replace %s %s %s %d with %s %s %s %d' % (orth, base, tag, typenum, orth, orth, tag, typenum) | 152 | # print 'replace %s %s %s %d with %s %s %s %d' % (orth, base, tag, typenum, orth, orth, tag, typenum) |
fsabuilder/morfeuszbuilder/segrules/rulesManager.py
@@ -33,8 +33,8 @@ class RulesManager(object): | @@ -33,8 +33,8 @@ class RulesManager(object): | ||
33 | def addDFA(self, optionsMap, dfa): | 33 | def addDFA(self, optionsMap, dfa): |
34 | self.options2DFA[self._options2Key(optionsMap)] = dfa | 34 | self.options2DFA[self._options2Key(optionsMap)] = dfa |
35 | 35 | ||
36 | - def lexeme2SegmentTypeNum(self, lemma, tagnum): | ||
37 | - res = self.segtypes.lexeme2Segnum(lemma, tagnum) | 36 | + def lexeme2SegmentTypeNum(self, lemma, tagnum, namenum, labelsnum): |
37 | + res = self.segtypes.lexeme2Segnum(lemma, tagnum, namenum, labelsnum) | ||
38 | if res is None: | 38 | if res is None: |
39 | raise ValueError() | 39 | raise ValueError() |
40 | else: | 40 | else: |
fsabuilder/morfeuszbuilder/segrules/rulesParser.py
@@ -13,8 +13,10 @@ class RulesParser(object): | @@ -13,8 +13,10 @@ class RulesParser(object): | ||
13 | PARSE4GENERATOR = 1 | 13 | PARSE4GENERATOR = 1 |
14 | PARSE4ANALYZER = 2 | 14 | PARSE4ANALYZER = 2 |
15 | 15 | ||
16 | - def __init__(self, tagset, rulesType): | 16 | + def __init__(self, tagset, namesMap, labelsMap, rulesType): |
17 | self.tagset = tagset | 17 | self.tagset = tagset |
18 | + self.namesMap = namesMap | ||
19 | + self.labelsMap = labelsMap | ||
18 | assert rulesType in (RulesParser.PARSE4GENERATOR, RulesParser.PARSE4ANALYZER) | 20 | assert rulesType in (RulesParser.PARSE4GENERATOR, RulesParser.PARSE4ANALYZER) |
19 | self.rulesType = rulesType | 21 | self.rulesType = rulesType |
20 | 22 | ||
@@ -43,7 +45,7 @@ class RulesParser(object): | @@ -43,7 +45,7 @@ class RulesParser(object): | ||
43 | 'segment types', | 45 | 'segment types', |
44 | 'separator chars']) | 46 | 'separator chars']) |
45 | key2Defs = self._getKey2Defs(segtypesConfigFile) | 47 | key2Defs = self._getKey2Defs(segtypesConfigFile) |
46 | - segtypesHelper = segtypes.Segtypes(self.tagset, segtypesConfigFile) | 48 | + segtypesHelper = segtypes.Segtypes(self.tagset, self.namesMap, self.labelsMap, segtypesConfigFile) |
47 | separatorsList = separatorChars.parseSeparatorChars(segtypesConfigFile) \ | 49 | separatorsList = separatorChars.parseSeparatorChars(segtypesConfigFile) \ |
48 | if self.rulesType == RulesParser.PARSE4ANALYZER \ | 50 | if self.rulesType == RulesParser.PARSE4ANALYZER \ |
49 | else [] | 51 | else [] |
fsabuilder/morfeuszbuilder/tagset/segtypes.py
@@ -5,7 +5,7 @@ Created on 17 lut 2014 | @@ -5,7 +5,7 @@ Created on 17 lut 2014 | ||
5 | ''' | 5 | ''' |
6 | import re | 6 | import re |
7 | import logging | 7 | import logging |
8 | -import sys | 8 | +import itertools |
9 | from morfeuszbuilder.utils import exceptions | 9 | from morfeuszbuilder.utils import exceptions |
10 | 10 | ||
11 | def _cutHomonymFromLemma(lemma): | 11 | def _cutHomonymFromLemma(lemma): |
@@ -15,19 +15,25 @@ def _cutHomonymFromLemma(lemma): | @@ -15,19 +15,25 @@ def _cutHomonymFromLemma(lemma): | ||
15 | 15 | ||
16 | class Segtypes(object): | 16 | class Segtypes(object): |
17 | 17 | ||
18 | - def __init__(self, tagset, segrulesConfigFile): | 18 | + def __init__(self, tagset, namesMap, labelsMap, segrulesConfigFile): |
19 | 19 | ||
20 | self.tagset = tagset | 20 | self.tagset = tagset |
21 | + self.namesMap = namesMap | ||
22 | + self.labelsMap = labelsMap | ||
23 | + self._reverseLabelsMap = dict([(v, k) for (k, v) in labelsMap.iteritems()]) | ||
21 | 24 | ||
22 | self.filename = segrulesConfigFile.filename | 25 | self.filename = segrulesConfigFile.filename |
23 | 26 | ||
24 | - self.segtypes = set() | ||
25 | - self.segtype2Segnum = {} | ||
26 | - self.segnum2Segtype = {} | 27 | + self.segtypes = [] |
28 | + # self.segtype2Segnum = {} | ||
29 | + # self.segnum2Segtype = {} | ||
27 | self.patternsList = [] | 30 | self.patternsList = [] |
28 | - | ||
29 | - self._tagnum2Segnum = {} | ||
30 | - self._lemmaTagnum2Segnum = {} | 31 | + |
32 | + # (lemma, tagnum) -> [namenum, labelsnum, segnum] | ||
33 | + self._segnumsMap = {} | ||
34 | + | ||
35 | + # self._tagnum2Segnum = {} | ||
36 | + # self._lemmaTagnum2Segnum = {} | ||
31 | 37 | ||
32 | self._readSegtypes(segrulesConfigFile) | 38 | self._readSegtypes(segrulesConfigFile) |
33 | self._readLexemes(segrulesConfigFile) | 39 | self._readLexemes(segrulesConfigFile) |
@@ -38,11 +44,9 @@ class Segtypes(object): | @@ -38,11 +44,9 @@ class Segtypes(object): | ||
38 | # print self._tagnum2Segnum | 44 | # print self._tagnum2Segnum |
39 | logging.info('segment number -> segment type') | 45 | logging.info('segment number -> segment type') |
40 | logging.info('------------------------------') | 46 | logging.info('------------------------------') |
41 | - logging.info(str(self.segnum2Segtype)) | 47 | + logging.info(dict(enumerate(self.segtypes))) |
42 | logging.info('------------------------------') | 48 | logging.info('------------------------------') |
43 | 49 | ||
44 | -# self._debugSegnums() | ||
45 | - | ||
46 | def _validate(self, msg, lineNum, cond): | 50 | def _validate(self, msg, lineNum, cond): |
47 | if not cond: | 51 | if not cond: |
48 | raise exceptions.ConfigFileException(self.filename, lineNum, msg) | 52 | raise exceptions.ConfigFileException(self.filename, lineNum, msg) |
@@ -58,155 +62,171 @@ class Segtypes(object): | @@ -58,155 +62,171 @@ class Segtypes(object): | ||
58 | u'Segment type already defined: "%s"' % line, | 62 | u'Segment type already defined: "%s"' % line, |
59 | lineNum, | 63 | lineNum, |
60 | line not in self.segtypes) | 64 | line not in self.segtypes) |
61 | - self.segtypes.add(line) | ||
62 | - | ||
63 | - | 65 | + self.segtypes.append(line) |
66 | + | ||
64 | def _readTags(self, segrulesConfigFile): | 67 | def _readTags(self, segrulesConfigFile): |
65 | gotWildcardPattern = False | 68 | gotWildcardPattern = False |
66 | for lineNum, line in segrulesConfigFile.enumerateLinesInSection('tags'): | 69 | for lineNum, line in segrulesConfigFile.enumerateLinesInSection('tags'): |
67 | - splitLine = re.split(r'\s+', line.strip()) | ||
68 | - self._validate( | ||
69 | - u'Line in [tags] section must contain exactly two fields - segment type and tag pattern', | ||
70 | - lineNum, | ||
71 | - len(splitLine) == 2) | ||
72 | - segtype, pattern = splitLine | ||
73 | - self._validate( | ||
74 | - u'Undeclared segment type: "%s"' % segtype, | ||
75 | - lineNum, | ||
76 | - segtype in self.segtypes) | ||
77 | - self._validate( | ||
78 | - u'Segment type must be a lowercase alphanumeric with optional underscores', | ||
79 | - lineNum, | ||
80 | - re.match(r'[a-z_]+', segtype)) | ||
81 | - self._validate( | ||
82 | - u'Pattern must contain only ":", "%", "." and lowercase alphanumeric letters', | ||
83 | - lineNum, | ||
84 | - re.match(r'[a-z_\.\:\%]+', pattern)) | ||
85 | - | ||
86 | - self._validate( | ||
87 | - u'Pattern that matches everything must be the last one', | ||
88 | - lineNum - 1, | ||
89 | - not gotWildcardPattern) | ||
90 | - | ||
91 | - if segtype in self.segtype2Segnum: | ||
92 | - segnum = self.segtype2Segnum[segtype] | ||
93 | - else: | ||
94 | - segnum = len(self.segtype2Segnum) | ||
95 | - self.segtype2Segnum[segtype] = segnum | ||
96 | - | ||
97 | - segtypePattern = SegtypePattern(None, pattern, segnum) | ||
98 | - | 70 | + self._parsePattern(lineNum, line, withLemma=False) |
99 | self._validate( | 71 | self._validate( |
100 | - u'There is no tag that matches pattern "%s".' % pattern, | ||
101 | - lineNum, | ||
102 | - any([segtypePattern.tryToMatch(None, tag) != -1 for tag in self.tagset.getAllTags()])) | ||
103 | - | ||
104 | - self.patternsList.append(segtypePattern) | ||
105 | - | ||
106 | - gotWildcardPattern = gotWildcardPattern or pattern == '%' | ||
107 | - | ||
108 | - self.segnum2Segtype = dict([(v, k) for (k, v) in self.segtype2Segnum.iteritems()]) | 72 | + u'Pattern that matches everything must be the last one', |
73 | + lineNum - 1, | ||
74 | + not gotWildcardPattern) | ||
75 | + gotWildcardPattern = gotWildcardPattern or self.patternsList[-1].isWildcardPattern() | ||
76 | + | ||
77 | + self._validate( | ||
78 | + u'There must be a pattern that matches everything at the end of [tags] section', | ||
79 | + lineNum, | ||
80 | + self.patternsList[-1].isWildcardPattern()) | ||
109 | 81 | ||
110 | def _readLexemes(self, segrulesConfigFile): | 82 | def _readLexemes(self, segrulesConfigFile): |
111 | for lineNum, line in segrulesConfigFile.enumerateLinesInSection('lexemes'): | 83 | for lineNum, line in segrulesConfigFile.enumerateLinesInSection('lexemes'): |
112 | - split = re.split(r'\s+', line.strip()) | ||
113 | - self._validate( | ||
114 | - u'Line in [lexemes] section must contain exactly two fields - segment type and lexeme pattern', | ||
115 | - lineNum, | ||
116 | - len(split) == 2) | ||
117 | - segtype, pattern = split | 84 | + self._parsePattern(lineNum, line, withLemma=True) |
85 | + | ||
86 | + def _parseAdditionalConstraints(self, lineNum, fields): | ||
87 | + res = {} | ||
88 | + for f in fields: | ||
89 | + match = re.match(r'(name|labels)=([\w_]+)', f, re.U) | ||
118 | self._validate( | 90 | self._validate( |
119 | - u'Undeclared segment type: "%s"' % segtype, | ||
120 | - lineNum, | ||
121 | - segtype in self.segtypes) | 91 | + u'invalid name or labels constraint: "%s"' % f, |
92 | + lineNum, | ||
93 | + match) | ||
94 | + key = match.group(1) | ||
95 | + value = match.group(2) | ||
122 | self._validate( | 96 | self._validate( |
123 | - u'Segment type must be a lowercase alphanumeric with optional underscores', | ||
124 | - lineNum, | ||
125 | - re.match(r'[a-z_]+', segtype)) | 97 | + u'%s already specified' % key, |
98 | + lineNum, | ||
99 | + key not in res) | ||
100 | + if key == 'labels': | ||
101 | + if value: | ||
102 | + value = frozenset(value.split(u'|')) | ||
103 | + else: | ||
104 | + value = frozenset() | ||
105 | + res[key] = value | ||
106 | + return res | ||
107 | + | ||
108 | + def _parsePattern(self, lineNum, line, withLemma): | ||
109 | + split = re.split(r'\s+', line.strip()) | ||
110 | + if withLemma: | ||
126 | self._validate( | 111 | self._validate( |
127 | - u'Pattern must contain encodedForm and part-of-speech fields', | ||
128 | - lineNum, | ||
129 | - re.match(r'.+?\:[a-z_]+', pattern, re.U)) | ||
130 | - | ||
131 | - if segtype in self.segtype2Segnum: | ||
132 | - segnum = self.segtype2Segnum[segtype] | ||
133 | - else: | ||
134 | - segnum = len(self.segtype2Segnum) | ||
135 | - self.segtype2Segnum[segtype] = segnum | ||
136 | - | ||
137 | - lemma, pos = pattern.split(':', 1) | ||
138 | - | ||
139 | - segtypePattern = SegtypePattern(lemma, pos + ':%', segnum) | ||
140 | - | 112 | + u'Line in [lexemes] section must contain 3 to 5 fields - segment type, lemma, tag pattern and optional constraints on name and labels', |
113 | + lineNum, | ||
114 | + len(split) in [3, 4, 5]) | ||
115 | + segtype = split[0] | ||
116 | + lemma = split[1] | ||
117 | + pattern = split[2] | ||
118 | + additionalConstraints = self._parseAdditionalConstraints(lineNum, split[3:]) | ||
119 | + else: | ||
141 | self._validate( | 120 | self._validate( |
142 | - u'There is no tag that matches pattern "%s".' % (pos + ':%'), | ||
143 | - lineNum, | ||
144 | - any([segtypePattern.tryToMatch(lemma, tag) != -1 for tag in self.tagset.getAllTags()])) | ||
145 | - | ||
146 | - self.patternsList.append(segtypePattern) | ||
147 | - | ||
148 | - def _debugSegnums(self): | ||
149 | - for tagnum, segnum in self._tagnum2Segnum.items(): | ||
150 | - print self.tagset.getTag4Tagnum(tagnum), '-->', self.segnum2Segtype[segnum] | ||
151 | - | ||
152 | - for (base, tagnum), segnum in self._lemmaTagnum2Segnum.items(): | ||
153 | - print base, self.tagset.getTag4Tagnum(tagnum), '-->', self.segnum2Segtype[segnum] | ||
154 | - | ||
155 | - def _indexSegnums(self): | ||
156 | -# logging.info('indexing segment type numbers...') | ||
157 | - # index tags | 121 | + u'Line in [tags] section must contain 2 to 4 fields - segment type, tag pattern and optional constraints on name and labels', |
122 | + lineNum, | ||
123 | + len(split) in [2, 3, 4]) | ||
124 | + segtype = split[0] | ||
125 | + lemma = None | ||
126 | + pattern = split[1] | ||
127 | + additionalConstraints = self._parseAdditionalConstraints(lineNum, split[2:]) | ||
128 | + self._validate( | ||
129 | + u'Undeclared segment type: "%s"' % segtype, | ||
130 | + lineNum, | ||
131 | + segtype in self.segtypes) | ||
132 | + segnum = self.segtypes.index(segtype) | ||
133 | + | ||
134 | + self._validate( | ||
135 | + u'Pattern must contain only ":", "%", "." and lowercase alphanumeric letters', | ||
136 | + lineNum, | ||
137 | + re.match(r'[a-z_\.\:\%]+', pattern)) | ||
138 | + | ||
139 | + segtypePattern = SegtypePattern( | ||
140 | + lemma, | ||
141 | + pattern, | ||
142 | + additionalConstraints.get('name', u''), | ||
143 | + additionalConstraints.get('labels', frozenset()), | ||
144 | + segnum) | ||
145 | + # print 'segtypePattern', repr(str(segtypePattern)) | ||
146 | + self._validate( | ||
147 | + u'There is no tag that matches pattern "%s".' % (pattern), | ||
148 | + lineNum, | ||
149 | + any([segtypePattern.tryToMatch(lemma, tag) != -1 for tag in self.tagset.getAllTags()])) | ||
150 | + self.patternsList.append(segtypePattern) | ||
151 | + | ||
152 | + def _getAllExistingLabelsnumCombinations(self, labels): | ||
153 | + if labels: | ||
154 | + for labelsCombination, labelsnum in self.labelsMap.iteritems(): | ||
155 | + if labels <= labelsCombination: | ||
156 | + yield labelsnum | ||
157 | + else: | ||
158 | + yield 0 | ||
159 | + | ||
160 | + def _indexOnePattern(self, p): | ||
161 | + | ||
158 | for tag in self.tagset.getAllTags(): | 162 | for tag in self.tagset.getAllTags(): |
159 | - tagnum = self.tagset.getTagnum4Tag(tag) | ||
160 | - for p in self.patternsList: | ||
161 | - segnum = p.tryToMatch(None, tag) | ||
162 | - if segnum >= 0 and tagnum not in self._tagnum2Segnum: | ||
163 | - self._tagnum2Segnum[tagnum] = segnum | ||
164 | - | 163 | + segnum = p.tryToMatch(p.lemma, tag) |
164 | + if segnum != -1: | ||
165 | + tagnum = self.tagset.getTagnum4Tag(tag) | ||
166 | + self._segnumsMap.setdefault((p.lemma, tagnum), []) | ||
167 | + namenum = self.namesMap[p.name] | ||
168 | + for labelsnum in self._getAllExistingLabelsnumCombinations(p.labels): | ||
169 | + self._segnumsMap[(p.lemma, tagnum)].append((namenum, labelsnum, segnum)) | ||
170 | + | ||
171 | + def _indexSegnums(self): | ||
172 | + logging.info('indexing segment type numbers...') | ||
173 | + | ||
165 | # index lexemes | 174 | # index lexemes |
166 | for p in self.patternsList: | 175 | for p in self.patternsList: |
167 | - if p.lemma: | ||
168 | - for tag in self.tagset.getAllTags(): | ||
169 | - tagnum = self.tagset.getTagnum4Tag(tag) | ||
170 | - if not (p.lemma, tagnum) in self._lemmaTagnum2Segnum: | ||
171 | - segnum = p.tryToMatch(p.lemma, tag) | ||
172 | - if segnum != -1: | ||
173 | - self._lemmaTagnum2Segnum[(p.lemma, tagnum)] = segnum | ||
174 | -# logging.info('indexing segment type numbers - done') | ||
175 | -# self._debugSegnums() | 176 | + self._indexOnePattern(p) |
177 | + | ||
178 | + # logging.info(self._segnumsMap) | ||
176 | 179 | ||
177 | def hasSegtype(self, segTypeString): | 180 | def hasSegtype(self, segTypeString): |
178 | - return segTypeString in self.segtype2Segnum | 181 | + # return segTypeString in self.segtype2Segnum |
182 | + return segTypeString in self.segtypes | ||
179 | 183 | ||
180 | def getSegnum4Segtype(self, segTypeString): | 184 | def getSegnum4Segtype(self, segTypeString): |
181 | - return self.segtype2Segnum[segTypeString] | 185 | + return self.segtypes.index(segTypeString) |
186 | + # return self.segtype2Segnum[segTypeString] | ||
182 | 187 | ||
183 | - def lexeme2Segnum(self, lemma, tagnum): | ||
184 | - lemma = _cutHomonymFromLemma(lemma) | ||
185 | - res = self._lemmaTagnum2Segnum.get((lemma, tagnum), None) | ||
186 | - if res is None: | ||
187 | - res = self._tagnum2Segnum.get(tagnum, None) | ||
188 | - return res | 188 | + def lexeme2Segnum(self, lemma, tagnum, namenum, labelsnum): |
189 | + | ||
190 | + if (lemma, tagnum) in self._segnumsMap: | ||
191 | + for (n, l, segnum) in self._segnumsMap[(lemma, tagnum)]: | ||
192 | + if (n, l) == (namenum, labelsnum) \ | ||
193 | + or (n, l) == (0, 0)\ | ||
194 | + or (n == 0 and l == labelsnum)\ | ||
195 | + or (l == 0 and n == namenum): | ||
196 | + return segnum | ||
197 | + | ||
198 | + if not lemma is None: | ||
199 | + return self.lexeme2Segnum(None, tagnum, namenum, labelsnum) | ||
200 | + else: | ||
201 | + assert False | ||
189 | 202 | ||
190 | def getMaxSegnum(self): | 203 | def getMaxSegnum(self): |
191 | - return max(self.segnum2Segtype.keys()) | 204 | + return len(self.segtypes) - 1 |
192 | 205 | ||
193 | class SegtypePattern(object): | 206 | class SegtypePattern(object): |
194 | 207 | ||
195 | - def __init__(self, lemma, pattern, segnum): | 208 | + def __init__(self, lemma, pattern, name, labels, segnum): |
196 | self.lemma = _cutHomonymFromLemma(lemma) | 209 | self.lemma = _cutHomonymFromLemma(lemma) |
197 | self.pattern = pattern | 210 | self.pattern = pattern |
211 | + self.name = name | ||
212 | + self.labels = labels | ||
198 | self.segnum = segnum | 213 | self.segnum = segnum |
199 | - | 214 | + |
200 | def tryToMatch(self, lemma, tag): | 215 | def tryToMatch(self, lemma, tag): |
201 | -# tag2Match = tag + ':' if not tag.endswith(':') else tag | ||
202 | -# print tag2Match | ||
203 | patterns2Match = [] | 216 | patterns2Match = [] |
204 | patterns2Match.append(self.pattern.replace('%', '.*')) | 217 | patterns2Match.append(self.pattern.replace('%', '.*')) |
205 | patterns2Match.append(re.sub(r'\:\%$', '', self.pattern).replace('%', '.*')) | 218 | patterns2Match.append(re.sub(r'\:\%$', '', self.pattern).replace('%', '.*')) |
206 | - lemma = _cutHomonymFromLemma(lemma) | ||
207 | - if (self.lemma is None or self.lemma == lemma) \ | ||
208 | - and any([re.match(p, tag) for p in patterns2Match]): | 219 | + # patterns2Match.append(re.sub(r'$', ':%', self.pattern).replace('%', '.*')) |
220 | + if self.lemma is None: | ||
221 | + lemma = None | ||
222 | + if any([re.match('^'+p+'$', tag) for p in patterns2Match]) \ | ||
223 | + and self.lemma == lemma: | ||
209 | return self.segnum | 224 | return self.segnum |
210 | else: | 225 | else: |
211 | -# print 'NOT match', lemma.encode('utf8') if lemma else '%', tag, self.segnum | ||
212 | return -1 | 226 | return -1 |
227 | + | ||
228 | + def isWildcardPattern(self): | ||
229 | + return (self.lemma, self.pattern, self.name, self.labels) == (None, '%', u'', frozenset()) | ||
230 | + | ||
231 | + def __str__(self): | ||
232 | + return u'%s %s %s %s -> %d' % (self.lemma, self.pattern, self.name, self.labels, self.segnum) |
input/segmenty.dat
@@ -506,181 +506,181 @@ pred pred | @@ -506,181 +506,181 @@ pred pred | ||
506 | samodz % | 506 | samodz % |
507 | 507 | ||
508 | [lexemes] | 508 | [lexemes] |
509 | -tym tym:adv | ||
510 | -bardziej bardziej:adv | ||
511 | -niemniej niemniej:qub | ||
512 | -pewno pewno:qub | ||
513 | -pol_zloz pół:num:comp | ||
514 | -i i:conj | ||
515 | -by by:qub | ||
516 | -li li:qub | ||
517 | -ze +że:qub | ||
518 | -z +ż:qub | ||
519 | -killfile +ć:qub | ||
520 | -killfile +anka:subst:% | ||
521 | -killfile +owa:subst:% | ||
522 | -killfile +anka:subst:% | ||
523 | -killfile +ina:subst:% | ||
524 | -killfile +yna:subst:% | ||
525 | -killfile +ówna:subst:% | ||
526 | -z_aglt_by aby:comp | ||
527 | -z_aglt bowiem:comp | ||
528 | -z_aglt_by by:comp | ||
529 | -z_aglt cóż:subst | ||
530 | -z_aglt_nwok czemu:adv | ||
531 | -z_aglt_by czyżby:qub | ||
532 | -z_aglt_by choćby:comp | ||
533 | -z_aglt_by chociażby:comp | ||
534 | -z_aglt_nwok dlaczego:adv | ||
535 | -z_aglt_nwok dopóki:comp | ||
536 | -z_aglt_nwok dopóty:conj | ||
537 | -z_aglt_by gdyby:comp | ||
538 | -z_aglt_nwok gdy:adv | ||
539 | -z_aglt_nwok gdzie:qub | ||
540 | -z_aglt_nwok gdzie:adv | ||
541 | -z_aglt_nwok kto:subst:% | ||
542 | -z_aglt_by jakby:comp | ||
543 | -z_aglt_by jakoby:comp | ||
544 | -z_aglt_nwok kiedy:adv | ||
545 | -z_aglt_nwok kiedy:comp | ||
546 | -z_aglt_nwok tylko:qub | ||
547 | -z_aglt_by żeby:comp | ||
548 | -z_aglt_nwok to:conj | ||
549 | -z_aglt_nwok chyba:qub | ||
550 | -z_aglt_nwok że:qub | ||
551 | -z_aglt_nwok czy:conj | ||
552 | -z_aglt_by oby:qub | ||
553 | -z_aglt_by bodajby:qub | ||
554 | -z_aglt_nwok co:comp | ||
555 | -z_aglt_nwok bo:comp | ||
556 | -z_aglt_nwok byle:comp | ||
557 | -z_aglt_by byleby:comp | ||
558 | -dywiz -:interp | ||
559 | -pauza —:interp | ||
560 | -polpauza –:interp | ||
561 | -kropka .:interp | ||
562 | -przecinek ,:interp | ||
563 | -n on:ppron3:sg:gen.acc:m1.m2.m3:ter:nakc:praep | ||
564 | -adj_anty_zloz ten:adj:% | ||
565 | -adj_anty_zloz tenże:adj:% | ||
566 | -adj_anty_zloz ck:adj:% | ||
567 | -adj_anty_zloz c.k.:adj:% | ||
568 | -adj_anty_zloz ki:adj:% | ||
569 | -adj_anty_zloz si:adj:% | ||
570 | -adj_anty_zloz ow:adj:% | ||
571 | -adj_anty_zloz ów:adj:% | ||
572 | -adj_anty_zloz ówże:adj:% | ||
573 | -adj_anty_zloz mój:adj:% | ||
574 | -adj_anty_zloz a-z:adj:% | ||
575 | -adj_anty_zloz a-ż:adj:% | ||
576 | -adj_anty_zloz kiż:adj:% | ||
577 | -adj_anty_zloz be:adj:% | ||
578 | -adj_anty_zloz caca:adj:% | ||
579 | -adj_anty_zloz czyj:adj:% | ||
580 | -adj_anty_zloz oboj:adj:% | ||
581 | -adj_anty_zloz on:adj:% | ||
582 | -adj_anty_zloz tyli:adj:% | ||
583 | -adj_anty_zloz rany:adj:% | ||
584 | -adj_anty_zloz taki:adj:% | ||
585 | -adj_anty_zloz pop:adj:% | ||
586 | -subst_anty_zloz co:subst:% | ||
587 | -subst_anty_zloz ar:subst:%:m3 | ||
588 | -subst_anty_zloz as:subst:%:m2 | ||
589 | -subst_anty_zloz as:subst:%:m1 | ||
590 | -subst_anty_zloz as:subst:%:m2 | ||
591 | -subst_anty_zloz as:subst:%:m3 | ||
592 | -subst_anty_zloz as:subst:%:n2 | ||
593 | -subst_anty_zloz at:subst:%:n2 | ||
594 | -subst_anty_zloz at:subst:%:m2 | ||
595 | -subst_anty_zloz az:subst:%:m1 | ||
596 | -subst_anty_zloz be:subst:%:n2 | ||
597 | -subst_anty_zloz bp:subst:%:m1 | ||
598 | -subst_anty_zloz ce:subst:%:n2 | ||
599 | -subst_anty_zloz de:subst:%:n2 | ||
600 | -subst_anty_zloz do:subst:%:n2 | ||
601 | -subst_anty_zloz dr:subst:%:m1 | ||
602 | -subst_anty_zloz ef:subst:%:n2 | ||
603 | -subst_anty_zloz el:subst:%:n2 | ||
604 | -subst_anty_zloz em:subst:%:n2 | ||
605 | -subst_anty_zloz en:subst:%:n2 | ||
606 | -subst_anty_zloz er:subst:%:n2 | ||
607 | -subst_anty_zloz es:subst:%:m3 | ||
608 | -subst_anty_zloz es:subst:%:n2 | ||
609 | -subst_anty_zloz fa:subst:%:n2 | ||
610 | -subst_anty_zloz go:subst:%:n2 | ||
611 | -subst_anty_zloz ha:subst:%:n2 | ||
612 | -subst_anty_zloz id:subst:%:m3 | ||
613 | -subst_anty_zloz id:subst:%:n2 | ||
614 | -subst_anty_zloz in:subst:%:n2 | ||
615 | -subst_anty_zloz ka:subst:%:n2 | ||
616 | -subst_anty_zloz kb:subst:%:m3 | ||
617 | -subst_anty_zloz km:subst:%:m3 | ||
618 | -subst_anty_zloz ku:subst:%:n2 | ||
619 | -subst_anty_zloz la:subst:%:n2 | ||
620 | -subst_anty_zloz li:subst:%:n2 | ||
621 | -subst_anty_zloz li:subst:%:f | ||
622 | -subst_anty_zloz mi:subst:%:n2 | ||
623 | -subst_anty_zloz ni:subst:%:n2 | ||
624 | -subst_anty_zloz no:subst:%:n2 | ||
625 | -subst_anty_zloz nr:subst:%:m3 | ||
626 | -subst_anty_zloz om:subst:%:m3 | ||
627 | -subst_anty_zloz oz:subst:%:m3 | ||
628 | -subst_anty_zloz pe:subst:%:n2 | ||
629 | -subst_anty_zloz pi:subst:%:n2 | ||
630 | -subst_anty_zloz re:subst:%:n2 | ||
631 | -subst_anty_zloz rh:subst:%:n2 | ||
632 | -subst_anty_zloz ro:subst:%:n2 | ||
633 | -subst_anty_zloz si:subst:%:n2 | ||
634 | -subst_anty_zloz su:subst:%:n2 | ||
635 | -subst_anty_zloz te:subst:%:n2 | ||
636 | -subst_anty_zloz tv:subst:%:f | ||
637 | -subst_anty_zloz ud:subst:%:m3 | ||
638 | -subst_anty_zloz uf:subst:%:m3 | ||
639 | -subst_anty_zloz ul:subst:%:m3 | ||
640 | -subst_anty_zloz um:subst:%:m3 | ||
641 | -subst_anty_zloz ut:subst:%:n2 | ||
642 | -subst_anty_zloz vw:subst:%:m3 | ||
643 | -subst_anty_zloz vw:subst:%:n2 | ||
644 | -subst_anty_zloz wc:subst:%:n2 | ||
645 | -subst_anty_zloz we:subst:%:n2 | ||
646 | -subst_anty_zloz wf:subst:%:m3 | ||
647 | -subst_anty_zloz wu:subst:%:n2 | ||
648 | -prep_na na:prep:% | ||
649 | -prep_n do:prep:% | ||
650 | -prep_n dla:prep:% | ||
651 | -prep_n koło:prep:% | ||
652 | -prep_n o:prep:% | ||
653 | -prep_n po:prep:% | ||
654 | -prep_n poza:prep:% | ||
655 | -prep_n spoza:prep:% | ||
656 | -prep_n za:prep:% | ||
657 | -prep_n zza:prep:% | ||
658 | -sufs +znawca:subst:% | ||
659 | -sufs +dawca:subst:% | ||
660 | -sufs +biorca:subst:% | ||
661 | -sufs +żerca:subst:% | ||
662 | -sufs +maniak:subst:% | ||
663 | -sufs +logia:subst:% | ||
664 | -sufs +log:subst:% | ||
665 | -suf_num +latek:subst:% | ||
666 | -suf_num +latka:subst:% | ||
667 | -suf_num +lecie:subst:% | ||
668 | -suf_num +krotność:subst:% | ||
669 | -suf_ord ty:adj:% | ||
670 | -suf_ord y:adj:% | ||
671 | -rom_i I:romandig | ||
672 | -rom_v V:romandig | ||
673 | -rom_x X:romandig | ||
674 | -rom_l L:romandig | ||
675 | -rom_d D:romandig | ||
676 | -rom_c C:romandig | ||
677 | -rom_m M:romandig | ||
678 | -ja ja:ppron12:sg:nom:% | ||
679 | -ty ty:ppron12:sg:nom:% | ||
680 | -my my:ppron12:pl:nom:% | ||
681 | -wy wy:ppron12:pl:nom:% | ||
682 | -pref_dyw e-+:prefa | ||
683 | -pref_dyw e-+:prefs | 509 | +tym tym adv |
510 | +bardziej bardziej adv | ||
511 | +niemniej niemniej qub | ||
512 | +pewno pewno qub | ||
513 | +pol_zloz pół num:comp | ||
514 | +i i conj | ||
515 | +by by qub | ||
516 | +li li qub | ||
517 | +ze +że qub | ||
518 | +z +ż qub | ||
519 | +killfile +ć qub | ||
520 | +killfile +anka subst:% | ||
521 | +killfile +owa subst:% | ||
522 | +killfile +anka subst:% | ||
523 | +killfile +ina subst:% | ||
524 | +killfile +yna subst:% | ||
525 | +killfile +ówna subst:% | ||
526 | +z_aglt_by aby comp | ||
527 | +z_aglt bowiem comp | ||
528 | +z_aglt_by by comp | ||
529 | +z_aglt cóż subst:% | ||
530 | +z_aglt_nwok czemu adv | ||
531 | +z_aglt_by czyżby qub | ||
532 | +z_aglt_by choćby comp | ||
533 | +z_aglt_by chociażby comp | ||
534 | +z_aglt_nwok dlaczego adv | ||
535 | +z_aglt_nwok dopóki comp | ||
536 | +z_aglt_nwok dopóty conj | ||
537 | +z_aglt_by gdyby comp | ||
538 | +z_aglt_nwok gdy adv | ||
539 | +z_aglt_nwok gdzie qub | ||
540 | +z_aglt_nwok gdzie adv | ||
541 | +z_aglt_nwok kto subst:% | ||
542 | +z_aglt_by jakby comp | ||
543 | +z_aglt_by jakoby comp | ||
544 | +z_aglt_nwok kiedy adv | ||
545 | +z_aglt_nwok kiedy comp | ||
546 | +z_aglt_nwok tylko qub | ||
547 | +z_aglt_by żeby comp | ||
548 | +z_aglt_nwok to conj | ||
549 | +z_aglt_nwok chyba qub | ||
550 | +z_aglt_nwok że qub | ||
551 | +z_aglt_nwok czy conj | ||
552 | +z_aglt_by oby qub | ||
553 | +z_aglt_by bodajby qub | ||
554 | +z_aglt_nwok co comp | ||
555 | +z_aglt_nwok bo comp | ||
556 | +z_aglt_nwok byle comp | ||
557 | +z_aglt_by byleby comp | ||
558 | +dywiz - interp | ||
559 | +pauza — interp | ||
560 | +polpauza – interp | ||
561 | +kropka . interp | ||
562 | +przecinek , interp | ||
563 | +n on ppron3:sg:gen.acc:m1.m2.m3:ter:nakc:praep | ||
564 | +adj_anty_zloz ten adj:% | ||
565 | +adj_anty_zloz tenże adj:% | ||
566 | +adj_anty_zloz ck adj:% | ||
567 | +adj_anty_zloz c.k. adj:% | ||
568 | +adj_anty_zloz ki adj:% | ||
569 | +adj_anty_zloz si adj:% | ||
570 | +adj_anty_zloz ow adj:% | ||
571 | +adj_anty_zloz ów adj:% | ||
572 | +adj_anty_zloz ówże adj:% | ||
573 | +adj_anty_zloz mój adj:% | ||
574 | +adj_anty_zloz a-z adj:% | ||
575 | +adj_anty_zloz a-ż adj:% | ||
576 | +adj_anty_zloz kiż adj:% | ||
577 | +adj_anty_zloz be adj:% | ||
578 | +adj_anty_zloz caca adj:% | ||
579 | +adj_anty_zloz czyj adj:% | ||
580 | +adj_anty_zloz oboj adj:% | ||
581 | +adj_anty_zloz on adj:% | ||
582 | +adj_anty_zloz tyli adj:% | ||
583 | +adj_anty_zloz rany adj:% | ||
584 | +adj_anty_zloz taki adj:% | ||
585 | +adj_anty_zloz pop adj:% | ||
586 | +subst_anty_zloz co subst:% | ||
587 | +subst_anty_zloz ar subst:%:m3 | ||
588 | +subst_anty_zloz as subst:%:m2 | ||
589 | +subst_anty_zloz as subst:%:m1 | ||
590 | +subst_anty_zloz as subst:%:m2 | ||
591 | +subst_anty_zloz as subst:%:m3 | ||
592 | +subst_anty_zloz as subst:%:n2 | ||
593 | +subst_anty_zloz at subst:%:n2 | ||
594 | +subst_anty_zloz at subst:%:m2 | ||
595 | +subst_anty_zloz az subst:%:m1 | ||
596 | +subst_anty_zloz be subst:%:n2 | ||
597 | +subst_anty_zloz bp subst:%:m1 | ||
598 | +subst_anty_zloz ce subst:%:n2 | ||
599 | +subst_anty_zloz de subst:%:n2 | ||
600 | +subst_anty_zloz do subst:%:n2 | ||
601 | +subst_anty_zloz dr subst:%:m1 | ||
602 | +subst_anty_zloz ef subst:%:n2 | ||
603 | +subst_anty_zloz el subst:%:n2 | ||
604 | +subst_anty_zloz em subst:%:n2 | ||
605 | +subst_anty_zloz en subst:%:n2 | ||
606 | +subst_anty_zloz er subst:%:n2 | ||
607 | +subst_anty_zloz es subst:%:m3 | ||
608 | +subst_anty_zloz es subst:%:n2 | ||
609 | +subst_anty_zloz fa subst:%:n2 | ||
610 | +subst_anty_zloz go subst:%:n2 | ||
611 | +subst_anty_zloz ha subst:%:n2 | ||
612 | +subst_anty_zloz id subst:%:m3 | ||
613 | +subst_anty_zloz id subst:%:n2 | ||
614 | +subst_anty_zloz in subst:%:n2 | ||
615 | +subst_anty_zloz ka subst:%:n2 | ||
616 | +subst_anty_zloz kb subst:%:m3 | ||
617 | +subst_anty_zloz km subst:%:m3 | ||
618 | +subst_anty_zloz ku subst:%:n2 | ||
619 | +subst_anty_zloz la subst:%:n2 | ||
620 | +subst_anty_zloz li subst:%:n2 | ||
621 | +subst_anty_zloz li subst:%:f | ||
622 | +subst_anty_zloz mi subst:%:n2 | ||
623 | +subst_anty_zloz ni subst:%:n2 | ||
624 | +subst_anty_zloz no subst:%:n2 | ||
625 | +subst_anty_zloz nr subst:%:m3 | ||
626 | +subst_anty_zloz om subst:%:m3 | ||
627 | +subst_anty_zloz oz subst:%:m3 | ||
628 | +subst_anty_zloz pe subst:%:n2 | ||
629 | +subst_anty_zloz pi subst:%:n2 | ||
630 | +subst_anty_zloz re subst:%:n2 | ||
631 | +subst_anty_zloz rh subst:%:n2 | ||
632 | +subst_anty_zloz ro subst:%:n2 | ||
633 | +subst_anty_zloz si subst:%:n2 | ||
634 | +subst_anty_zloz su subst:%:n2 | ||
635 | +subst_anty_zloz te subst:%:n2 | ||
636 | +subst_anty_zloz tv subst:%:f | ||
637 | +subst_anty_zloz ud subst:%:m3 | ||
638 | +subst_anty_zloz uf subst:%:m3 | ||
639 | +subst_anty_zloz ul subst:%:m3 | ||
640 | +subst_anty_zloz um subst:%:m3 | ||
641 | +subst_anty_zloz ut subst:%:n2 | ||
642 | +subst_anty_zloz vw subst:%:m3 | ||
643 | +subst_anty_zloz vw subst:%:n2 | ||
644 | +subst_anty_zloz wc subst:%:n2 | ||
645 | +subst_anty_zloz we subst:%:n2 | ||
646 | +subst_anty_zloz wf subst:%:m3 | ||
647 | +subst_anty_zloz wu subst:%:n2 | ||
648 | +prep_na na prep:% | ||
649 | +prep_n do prep:% | ||
650 | +prep_n dla prep:% | ||
651 | +prep_n koło prep:% | ||
652 | +prep_n o prep:% | ||
653 | +prep_n po prep:% | ||
654 | +prep_n poza prep:% | ||
655 | +prep_n spoza prep:% | ||
656 | +prep_n za prep:% | ||
657 | +prep_n zza prep:% | ||
658 | +sufs +znawca subst:% | ||
659 | +sufs +dawca subst:% | ||
660 | +sufs +biorca subst:% | ||
661 | +sufs +żerca subst:% | ||
662 | +sufs +maniak subst:% | ||
663 | +sufs +logia subst:% | ||
664 | +sufs +log subst:% | ||
665 | +suf_num +latek subst:% | ||
666 | +suf_num +latka subst:% | ||
667 | +suf_num +lecie subst:% | ||
668 | +suf_num +krotność subst:% | ||
669 | +suf_ord ty adj:% | ||
670 | +suf_ord y adj:% | ||
671 | +rom_i I romandig | ||
672 | +rom_v V romandig | ||
673 | +rom_x X romandig | ||
674 | +rom_l L romandig | ||
675 | +rom_d D romandig | ||
676 | +rom_c C romandig | ||
677 | +rom_m M romandig | ||
678 | +ja ja ppron12:sg:nom:% | ||
679 | +ty ty ppron12:sg:nom:% | ||
680 | +my my ppron12:pl:nom:% | ||
681 | +wy wy ppron12:pl:nom:% | ||
682 | +pref_dyw e-+ prefa | ||
683 | +pref_dyw e-+ prefs | ||
684 | 684 | ||
685 | [separator chars] | 685 | [separator chars] |
686 | # , | 686 | # , |
nbproject/configurations.xml
@@ -339,8 +339,6 @@ | @@ -339,8 +339,6 @@ | ||
339 | ex="false" | 339 | ex="false" |
340 | tool="1" | 340 | tool="1" |
341 | flavor2="4"> | 341 | flavor2="4"> |
342 | - <ccTool flags="2"> | ||
343 | - </ccTool> | ||
344 | </item> | 342 | </item> |
345 | <item path="build/morfeusz/wrappers/morfeuszPERL_wrap.cxx" | 343 | <item path="build/morfeusz/wrappers/morfeuszPERL_wrap.cxx" |
346 | ex="false" | 344 | ex="false" |
@@ -969,22 +967,16 @@ | @@ -969,22 +967,16 @@ | ||
969 | ex="false" | 967 | ex="false" |
970 | tool="1" | 968 | tool="1" |
971 | flavor2="4"> | 969 | flavor2="4"> |
972 | - <ccTool flags="2"> | ||
973 | - </ccTool> | ||
974 | </item> | 970 | </item> |
975 | <item path="morfeusz/deserialization/morphInterps/InterpretedChunksDecoder4Analyzer.cpp" | 971 | <item path="morfeusz/deserialization/morphInterps/InterpretedChunksDecoder4Analyzer.cpp" |
976 | ex="false" | 972 | ex="false" |
977 | tool="1" | 973 | tool="1" |
978 | flavor2="4"> | 974 | flavor2="4"> |
979 | - <ccTool flags="2"> | ||
980 | - </ccTool> | ||
981 | </item> | 975 | </item> |
982 | <item path="morfeusz/deserialization/morphInterps/InterpretedChunksDecoder4Generator.cpp" | 976 | <item path="morfeusz/deserialization/morphInterps/InterpretedChunksDecoder4Generator.cpp" |
983 | ex="false" | 977 | ex="false" |
984 | tool="1" | 978 | tool="1" |
985 | flavor2="4"> | 979 | flavor2="4"> |
986 | - <ccTool flags="2"> | ||
987 | - </ccTool> | ||
988 | </item> | 980 | </item> |
989 | <item path="morfeusz/fsa/const.cpp" ex="false" tool="1" flavor2="4"> | 981 | <item path="morfeusz/fsa/const.cpp" ex="false" tool="1" flavor2="4"> |
990 | </item> | 982 | </item> |
@@ -1035,12 +1027,8 @@ | @@ -1035,12 +1027,8 @@ | ||
1035 | </ccTool> | 1027 | </ccTool> |
1036 | </item> | 1028 | </item> |
1037 | <item path="morfeusz/segrules/SegrulesFSA.cpp" ex="false" tool="1" flavor2="4"> | 1029 | <item path="morfeusz/segrules/SegrulesFSA.cpp" ex="false" tool="1" flavor2="4"> |
1038 | - <ccTool flags="2"> | ||
1039 | - </ccTool> | ||
1040 | </item> | 1030 | </item> |
1041 | <item path="morfeusz/segrules/segrules.cpp" ex="false" tool="1" flavor2="4"> | 1031 | <item path="morfeusz/segrules/segrules.cpp" ex="false" tool="1" flavor2="4"> |
1042 | - <ccTool flags="2"> | ||
1043 | - </ccTool> | ||
1044 | </item> | 1032 | </item> |
1045 | <item path="morfeusz/test_runner.cpp" ex="false" tool="1" flavor2="4"> | 1033 | <item path="morfeusz/test_runner.cpp" ex="false" tool="1" flavor2="4"> |
1046 | <ccTool flags="0"> | 1034 | <ccTool flags="0"> |
tests/analyzer/test_digits/segmentation.dat
tests/analyzer/test_names/segmentation.dat
tests/analyzer/test_qualifiers/segmentation.dat
tests/analyzer/test_segtypes/ARGS
0 → 100644
1 | +--aggl permissive |
tests/analyzer/test_segtypes/dictionary.tab
0 → 100644
1 | +mikro mikro prefs | ||
2 | +euro euro prefs | ||
3 | +super super prefs | ||
4 | +kot kot subst:sg:nom:m1 imie mikro|zzz | ||
5 | +kot kot subst:sg:nom:m1 nazwisko euro | ||
6 | +kot kot subst:sg:nom:m1 xxx|yyy|super | ||
7 | +kot kot subst:sg:nom:m1 samodzielnie|zzz | ||
8 | +kot kot subst:sg:nom:m2 imie samodzielnie | ||
9 | +kot kot adj:sg:acc:m3:pos nigdy | ||
10 | +pies pies subst:sg:acc:m3 aaa|zzz|euro | ||
11 | +pies pies subst:sg:nom:m1 nazwisko yyy|zzz|samodzielnie |
tests/analyzer/test_segtypes/input.txt
0 → 100644
tests/analyzer/test_segtypes/output.txt
0 → 100644
1 | +[0,1,kot,kot,subst:sg:nom:m1,_,samodzielnie|zzz | ||
2 | + 0,1,kot,kot,subst:sg:nom:m2,imie,samodzielnie] | ||
3 | +[0,1,eurokot,eurokot,subst:sg:nom:m1,nazwisko,euro] | ||
4 | +[0,1,superkot,superkot,subst:sg:nom:m1,_,super|xxx|yyy] | ||
5 | +[0,1,mikrokot,mikrokot,subst:sg:nom:m1,imie,mikro|zzz] | ||
6 | +[0,1,kotkot,kotkot,ign,_,_] | ||
7 | +[0,1,pies,pies,subst:sg:nom:m1,nazwisko,samodzielnie|yyy|zzz] | ||
8 | +[0,1,europies,europies,subst:sg:acc:m3,_,aaa|zzz|euro] | ||
9 | +[0,1,superpies,superpies,ign,_,_] | ||
10 | +[0,1,mikropies,mikropies,ign,_,_] | ||
11 | + |
tests/analyzer/test_segtypes/segmentation.dat
0 → 100644
1 | +[options] | ||
2 | +aggl=strict permissive isolated | ||
3 | +praet=split composite | ||
4 | + | ||
5 | +[combinations] | ||
6 | + | ||
7 | +samodzielnie | ||
8 | +mikro> tylko_z_mikro | ||
9 | +euro> tylko_z_euro | ||
10 | +super> tylko_z_super | ||
11 | + | ||
12 | +[segment types] | ||
13 | +tylko_z_mikro | ||
14 | +tylko_z_euro | ||
15 | +tylko_z_super | ||
16 | +samodzielnie | ||
17 | +nigdy | ||
18 | +mikro | ||
19 | +euro | ||
20 | +super | ||
21 | + | ||
22 | +[lexemes] | ||
23 | +tylko_z_mikro kot subst:% name=imie labels=zzz | ||
24 | +tylko_z_euro kot subst:% name=nazwisko | ||
25 | +tylko_z_super kot subst:% labels=xxx|yyy | ||
26 | +samodzielnie kot subst:% | ||
27 | + | ||
28 | +mikro mikro % | ||
29 | +euro euro % | ||
30 | +super super % | ||
31 | + | ||
32 | +[tags] | ||
33 | +tylko_z_euro subst:% labels=aaa | ||
34 | +samodzielnie subst:% | ||
35 | +nigdy % | ||
36 | + | ||
37 | +[separator chars] | ||
38 | +# , | ||
39 | +44 | ||
40 | + | ||
41 | +# . | ||
42 | +46 | ||
43 | + | ||
44 | +# ; | ||
45 | +59 |
tests/analyzer/test_segtypes/tagset.dat
0 → 100644
1 | +#!TAGSET-ID pl.sgjp.morfeusz-0.5.0 | ||
2 | + | ||
3 | +[TAGS] | ||
4 | +# special: unknown word (ignotum): | ||
5 | +0 ign | ||
6 | +# special: space/blank: | ||
7 | +1 sp | ||
8 | +# NOUNS | ||
9 | +694 subst:sg:nom:m1 | ||
10 | +695 subst:sg:nom:m2 | ||
11 | +696 subst:sg:nom:m3 | ||
12 | +697 subst:sg:nom:n1 | ||
13 | +698 subst:sg:nom:n2 | ||
14 | +693 subst:sg:nom:f | ||
15 | +676 subst:sg:gen:m1 | ||
16 | +677 subst:sg:gen:m2 | ||
17 | +678 subst:sg:gen:m3 | ||
18 | +679 subst:sg:gen:n1 | ||
19 | +680 subst:sg:gen:n2 | ||
20 | +675 subst:sg:gen:f | ||
21 | +670 subst:sg:dat:m1 | ||
22 | +671 subst:sg:dat:m2 | ||
23 | +672 subst:sg:dat:m3 | ||
24 | +673 subst:sg:dat:n1 | ||
25 | +674 subst:sg:dat:n2 | ||
26 | +669 subst:sg:dat:f | ||
27 | +664 subst:sg:acc:m1 | ||
28 | +665 subst:sg:acc:m2 | ||
29 | +666 subst:sg:acc:m3 | ||
30 | +667 subst:sg:acc:n1 | ||
31 | +668 subst:sg:acc:n2 | ||
32 | +663 subst:sg:acc:f | ||
33 | +682 subst:sg:inst:m1 | ||
34 | +683 subst:sg:inst:m2 | ||
35 | +684 subst:sg:inst:m3 | ||
36 | +685 subst:sg:inst:n1 | ||
37 | +686 subst:sg:inst:n2 | ||
38 | +681 subst:sg:inst:f | ||
39 | +688 subst:sg:loc:m1 | ||
40 | +689 subst:sg:loc:m2 | ||
41 | +690 subst:sg:loc:m3 | ||
42 | +691 subst:sg:loc:n1 | ||
43 | +692 subst:sg:loc:n2 | ||
44 | +687 subst:sg:loc:f | ||
45 | +700 subst:sg:voc:m1 | ||
46 | +701 subst:sg:voc:m2 | ||
47 | +702 subst:sg:voc:m3 | ||
48 | +703 subst:sg:voc:n1 | ||
49 | +704 subst:sg:voc:n2 | ||
50 | +699 subst:sg:voc:f | ||
51 | +646 subst:pl:nom:m1 | ||
52 | +647 subst:pl:nom:m2 | ||
53 | +648 subst:pl:nom:m3 | ||
54 | +649 subst:pl:nom:n1 | ||
55 | +650 subst:pl:nom:n2 | ||
56 | +651 subst:pl:nom:p1 | ||
57 | +652 subst:pl:nom:p2 | ||
58 | +653 subst:pl:nom:p3 | ||
59 | +645 subst:pl:nom:f | ||
60 | +619 subst:pl:gen:m1 | ||
61 | +620 subst:pl:gen:m2 | ||
62 | +621 subst:pl:gen:m3 | ||
63 | +622 subst:pl:gen:n1 | ||
64 | +623 subst:pl:gen:n2 | ||
65 | +624 subst:pl:gen:p1 | ||
66 | +625 subst:pl:gen:p2 | ||
67 | +626 subst:pl:gen:p3 | ||
68 | +618 subst:pl:gen:f | ||
69 | +610 subst:pl:dat:m1 | ||
70 | +611 subst:pl:dat:m2 | ||
71 | +612 subst:pl:dat:m3 | ||
72 | +613 subst:pl:dat:n1 | ||
73 | +614 subst:pl:dat:n2 | ||
74 | +615 subst:pl:dat:p1 | ||
75 | +616 subst:pl:dat:p2 | ||
76 | +617 subst:pl:dat:p3 | ||
77 | +609 subst:pl:dat:f | ||
78 | +601 subst:pl:acc:m1 | ||
79 | +602 subst:pl:acc:m2 | ||
80 | +603 subst:pl:acc:m3 | ||
81 | +604 subst:pl:acc:n1 | ||
82 | +605 subst:pl:acc:n2 | ||
83 | +606 subst:pl:acc:p1 | ||
84 | +607 subst:pl:acc:p2 | ||
85 | +608 subst:pl:acc:p3 | ||
86 | +600 subst:pl:acc:f | ||
87 | +628 subst:pl:inst:m1 | ||
88 | +629 subst:pl:inst:m2 | ||
89 | +630 subst:pl:inst:m3 | ||
90 | +631 subst:pl:inst:n1 | ||
91 | +632 subst:pl:inst:n2 | ||
92 | +633 subst:pl:inst:p1 | ||
93 | +634 subst:pl:inst:p2 | ||
94 | +635 subst:pl:inst:p3 | ||
95 | +627 subst:pl:inst:f | ||
96 | +637 subst:pl:loc:m1 | ||
97 | +638 subst:pl:loc:m2 | ||
98 | +639 subst:pl:loc:m3 | ||
99 | +640 subst:pl:loc:n1 | ||
100 | +641 subst:pl:loc:n2 | ||
101 | +642 subst:pl:loc:p1 | ||
102 | +643 subst:pl:loc:p2 | ||
103 | +644 subst:pl:loc:p3 | ||
104 | +636 subst:pl:loc:f | ||
105 | +654 subst:pl:voc:f | ||
106 | +655 subst:pl:voc:m1 | ||
107 | +656 subst:pl:voc:m2 | ||
108 | +657 subst:pl:voc:m3 | ||
109 | +658 subst:pl:voc:n1 | ||
110 | +659 subst:pl:voc:n2 | ||
111 | +660 subst:pl:voc:p1 | ||
112 | +661 subst:pl:voc:p2 | ||
113 | +662 subst:pl:voc:p3 | ||
114 | +# depreciative nominal flexeme: | ||
115 | +149 depr:pl:nom:m2 | ||
116 | +150 depr:pl:voc:m2 | ||
117 | +# nominal compounds forming form: | ||
118 | +599 substa | ||
119 | +# PERSONAL PRONOUNS | ||
120 | +443 ppron12:sg:acc:m1.m2.m3.f.n1.n2:pri:akc | ||
121 | +444 ppron12:sg:acc:m1.m2.m3.f.n1.n2:pri:nakc | ||
122 | +445 ppron12:sg:acc:m1.m2.m3.f.n1.n2:sec:akc | ||
123 | +446 ppron12:sg:acc:m1.m2.m3.f.n1.n2:sec:nakc | ||
124 | +447 ppron12:sg:dat:m1.m2.m3.f.n1.n2:pri:akc | ||
125 | +448 ppron12:sg:dat:m1.m2.m3.f.n1.n2:pri:nakc | ||
126 | +449 ppron12:sg:dat:m1.m2.m3.f.n1.n2:sec:akc | ||
127 | +450 ppron12:sg:dat:m1.m2.m3.f.n1.n2:sec:nakc | ||
128 | +451 ppron12:sg:gen:m1.m2.m3.f.n1.n2:pri:akc | ||
129 | +452 ppron12:sg:gen:m1.m2.m3.f.n1.n2:pri:nakc | ||
130 | +453 ppron12:sg:gen:m1.m2.m3.f.n1.n2:sec:akc | ||
131 | +454 ppron12:sg:gen:m1.m2.m3.f.n1.n2:sec:nakc | ||
132 | +455 ppron12:sg:inst:m1.m2.m3.f.n1.n2:pri | ||
133 | +456 ppron12:sg:inst:m1.m2.m3.f.n1.n2:sec | ||
134 | +457 ppron12:sg:loc:m1.m2.m3.f.n1.n2:pri | ||
135 | +458 ppron12:sg:loc:m1.m2.m3.f.n1.n2:sec | ||
136 | +459 ppron12:sg:nom:m1.m2.m3.f.n1.n2:pri | ||
137 | +460 ppron12:sg:nom:m1.m2.m3.f.n1.n2:sec | ||
138 | +461 ppron12:sg:voc:m1.m2.m3.f.n1.n2:sec | ||
139 | +429 ppron12:pl:acc:_:pri | ||
140 | +430 ppron12:pl:acc:_:sec | ||
141 | +431 ppron12:pl:dat:_:pri | ||
142 | +432 ppron12:pl:dat:_:sec | ||
143 | +433 ppron12:pl:gen:_:pri | ||
144 | +434 ppron12:pl:gen:_:sec | ||
145 | +435 ppron12:pl:inst:_:pri | ||
146 | +436 ppron12:pl:inst:_:sec | ||
147 | +437 ppron12:pl:loc:_:pri | ||
148 | +438 ppron12:pl:loc:_:sec | ||
149 | +439 ppron12:pl:nom:_:pri | ||
150 | +440 ppron12:pl:nom:_:sec | ||
151 | +441 ppron12:pl:voc:_:pri | ||
152 | +442 ppron12:pl:voc:_:sec | ||
153 | +474 ppron3:sg:acc:f:ter:_:npraep | ||
154 | +475 ppron3:sg:acc:f:ter:_:praep | ||
155 | +476 ppron3:sg:acc:m1.m2.m3:ter:akc:npraep | ||
156 | +477 ppron3:sg:acc:m1.m2.m3:ter:akc:praep | ||
157 | +478 ppron3:sg:acc:m1.m2.m3:ter:nakc:npraep | ||
158 | +479 ppron3:sg:acc:m1.m2.m3:ter:nakc:praep | ||
159 | +480 ppron3:sg:acc:n1.n2:ter:_:npraep | ||
160 | +481 ppron3:sg:acc:n1.n2:ter:_:praep | ||
161 | +482 ppron3:sg:dat:f:ter:_:npraep | ||
162 | +483 ppron3:sg:dat:f:ter:_:praep | ||
163 | +484 ppron3:sg:dat:m1.m2.m3:ter:akc:npraep | ||
164 | +485 ppron3:sg:dat:m1.m2.m3:ter:nakc:npraep | ||
165 | +486 ppron3:sg:dat:m1.m2.m3:ter:_:praep | ||
166 | +487 ppron3:sg:dat:n1.n2:ter:akc:npraep | ||
167 | +488 ppron3:sg:dat:n1.n2:ter:nakc:npraep | ||
168 | +489 ppron3:sg:dat:n1.n2:ter:_:praep | ||
169 | +490 ppron3:sg:gen.acc:m1.m2.m3:ter:nakc:praep | ||
170 | +491 ppron3:sg:gen:f:ter:_:npraep | ||
171 | +492 ppron3:sg:gen:f:ter:_:praep | ||
172 | +493 ppron3:sg:gen:m1.m2.m3:ter:akc:npraep | ||
173 | +494 ppron3:sg:gen:m1.m2.m3:ter:akc:praep | ||
174 | +495 ppron3:sg:gen:m1.m2.m3:ter:nakc:npraep | ||
175 | +496 ppron3:sg:gen:m1.m2.m3:ter:nakc:praep | ||
176 | +497 ppron3:sg:gen:n1.n2:ter:akc:npraep | ||
177 | +498 ppron3:sg:gen:n1.n2:ter:nakc:npraep | ||
178 | +499 ppron3:sg:gen:n1.n2:ter:_:praep | ||
179 | +500 ppron3:sg:inst:f:ter:_:praep | ||
180 | +501 ppron3:sg:inst:m1.m2.m3:ter:_:_ | ||
181 | +502 ppron3:sg:inst:n1.n2:ter:_:_ | ||
182 | +503 ppron3:sg:loc:f:ter:_:_ | ||
183 | +504 ppron3:sg:loc:m1.m2.m3:ter:_:_ | ||
184 | +505 ppron3:sg:loc:n1.n2:ter:_:_ | ||
185 | +506 ppron3:sg:nom:f:ter:_:_ | ||
186 | +507 ppron3:sg:nom:m1.m2.m3:ter:_:_ | ||
187 | +508 ppron3:sg:nom:n1.n2:ter:_:_ | ||
188 | +462 ppron3:pl:acc:m1.p1:ter:_:npraep | ||
189 | +463 ppron3:pl:acc:m1.p1:ter:_:praep | ||
190 | +464 ppron3:pl:acc:m2.m3.f.n1.n2.p2.p3:ter:_:npraep | ||
191 | +465 ppron3:pl:acc:m2.m3.f.n1.n2.p2.p3:ter:_:praep | ||
192 | +466 ppron3:pl:dat:_:ter:_:npraep | ||
193 | +467 ppron3:pl:dat:_:ter:_:praep | ||
194 | +468 ppron3:pl:gen:_:ter:_:npraep | ||
195 | +469 ppron3:pl:gen:_:ter:_:praep | ||
196 | +470 ppron3:pl:inst:_:ter:_:_ | ||
197 | +471 ppron3:pl:loc:_:ter:_:_ | ||
198 | +472 ppron3:pl:nom:m1.p1:ter:_:_ | ||
199 | +473 ppron3:pl:nom:m2.m3.f.n1.n2.p2.p3:ter:_:_ | ||
200 | +# PRONOUN ‘SIEBIE’ | ||
201 | +594 siebie:acc | ||
202 | +595 siebie:dat | ||
203 | +596 siebie:gen | ||
204 | +597 siebie:inst | ||
205 | +598 siebie:loc | ||
206 | +# ADJECTIVES | ||
207 | +5 adj:pl:acc:m1.p1:com | ||
208 | +6 adj:pl:acc:m1.p1:pos | ||
209 | +7 adj:pl:acc:m1.p1:sup | ||
210 | +8 adj:pl:acc:m2.m3.f.n1.n2.p2.p3:com | ||
211 | +9 adj:pl:acc:m2.m3.f.n1.n2.p2.p3:pos | ||
212 | +10 adj:pl:acc:m2.m3.f.n1.n2.p2.p3:sup | ||
213 | +11 adj:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:com | ||
214 | +12 adj:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:pos | ||
215 | +13 adj:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:sup | ||
216 | +14 adj:pl:gen:m1.m2.m3.f.n1.n2.p1.p2.p3:com | ||
217 | +15 adj:pl:gen:m1.m2.m3.f.n1.n2.p1.p2.p3:pos | ||
218 | +16 adj:pl:gen:m1.m2.m3.f.n1.n2.p1.p2.p3:sup | ||
219 | +17 adj:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:com | ||
220 | +18 adj:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:pos | ||
221 | +19 adj:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:sup | ||
222 | +20 adj:pl:loc:m1.m2.m3.f.n1.n2.p1.p2.p3:com | ||
223 | +21 adj:pl:loc:m1.m2.m3.f.n1.n2.p1.p2.p3:pos | ||
224 | +22 adj:pl:loc:m1.m2.m3.f.n1.n2.p1.p2.p3:sup | ||
225 | +23 adj:pl:nom:m1.p1:pos | ||
226 | +24 adj:pl:nom:m2.m3.f.n1.n2.p2.p3:pos | ||
227 | +25 adj:pl:nom.voc:m1.p1:com | ||
228 | +26 adj:pl:nom.voc:m1.p1:pos | ||
229 | +27 adj:pl:nom.voc:m1.p1:sup | ||
230 | +28 adj:pl:nom.voc:m2.m3.f.n1.n2.p2.p3:com | ||
231 | +29 adj:pl:nom.voc:m2.m3.f.n1.n2.p2.p3:pos | ||
232 | +30 adj:pl:nom.voc:m2.m3.f.n1.n2.p2.p3:sup | ||
233 | +31 adj:sg:acc:f:com | ||
234 | +32 adj:sg:acc:f:pos | ||
235 | +33 adj:sg:acc:f:sup | ||
236 | +34 adj:sg:acc:m1.m2:com | ||
237 | +35 adj:sg:acc:m1.m2:pos | ||
238 | +36 adj:sg:acc:m1.m2:sup | ||
239 | +37 adj:sg:acc:m3:com | ||
240 | +38 adj:sg:acc:m3:pos | ||
241 | +39 adj:sg:acc:m3:sup | ||
242 | +40 adj:sg:acc:n1.n2:com | ||
243 | +41 adj:sg:acc:n1.n2:pos | ||
244 | +42 adj:sg:acc:n1.n2:sup | ||
245 | +43 adj:sg:dat:f:com | ||
246 | +44 adj:sg:dat:f:pos | ||
247 | +45 adj:sg:dat:f:sup | ||
248 | +46 adj:sg:dat:m1.m2.m3.n1.n2:com | ||
249 | +47 adj:sg:dat:m1.m2.m3.n1.n2:pos | ||
250 | +48 adj:sg:dat:m1.m2.m3.n1.n2:sup | ||
251 | +49 adj:sg:gen:f:com | ||
252 | +50 adj:sg:gen:f:pos | ||
253 | +51 adj:sg:gen:f:sup | ||
254 | +52 adj:sg:gen:m1.m2.m3.n1.n2:com | ||
255 | +53 adj:sg:gen:m1.m2.m3.n1.n2:pos | ||
256 | +54 adj:sg:gen:m1.m2.m3.n1.n2:sup | ||
257 | +55 adj:sg:inst:f:com | ||
258 | +56 adj:sg:inst:f:pos | ||
259 | +57 adj:sg:inst:f:sup | ||
260 | +58 adj:sg:inst:m1.m2.m3.n1.n2:com | ||
261 | +59 adj:sg:inst:m1.m2.m3.n1.n2:pos | ||
262 | +60 adj:sg:inst:m1.m2.m3.n1.n2:sup | ||
263 | +61 adj:sg:loc:f:com | ||
264 | +62 adj:sg:loc:f:pos | ||
265 | +63 adj:sg:loc:f:sup | ||
266 | +64 adj:sg:loc:m1.m2.m3.n1.n2:com | ||
267 | +65 adj:sg:loc:m1.m2.m3.n1.n2:pos | ||
268 | +66 adj:sg:loc:m1.m2.m3.n1.n2:sup | ||
269 | +67 adj:sg:nom:f:pos | ||
270 | +68 adj:sg:nom:m1.m2.m3:pos | ||
271 | +69 adj:sg:nom:n1.n2:pos | ||
272 | +70 adj:sg:nom.voc:f:com | ||
273 | +71 adj:sg:nom.voc:f:pos | ||
274 | +72 adj:sg:nom.voc:f:sup | ||
275 | +73 adj:sg:nom.voc:m1.m2.m3:com | ||
276 | +74 adj:sg:nom.voc:m1.m2.m3:pos | ||
277 | +75 adj:sg:nom.voc:m1.m2.m3:sup | ||
278 | +76 adj:sg:nom.voc:n1.n2:com | ||
279 | +77 adj:sg:nom.voc:n1.n2:pos | ||
280 | +78 adj:sg:nom.voc:n1.n2:sup | ||
281 | +# adjectival compounds forming form: | ||
282 | +2 adja | ||
283 | +# predicative adjective: | ||
284 | +3 adjc | ||
285 | +# post-prepositional adjective: | ||
286 | +4 adjp | ||
287 | +# VERBS | ||
288 | +# finitive (present/future) flexeme: | ||
289 | +153 fin:pl:pri:imperf | ||
290 | +154 fin:pl:pri:imperf.perf | ||
291 | +155 fin:pl:pri:perf | ||
292 | +156 fin:pl:sec:imperf | ||
293 | +157 fin:pl:sec:imperf.perf | ||
294 | +158 fin:pl:sec:perf | ||
295 | +159 fin:pl:ter:imperf | ||
296 | +160 fin:pl:ter:imperf.perf | ||
297 | +161 fin:pl:ter:perf | ||
298 | +162 fin:sg:pri:imperf | ||
299 | +163 fin:sg:pri:imperf.perf | ||
300 | +164 fin:sg:pri:perf | ||
301 | +165 fin:sg:sec:imperf | ||
302 | +166 fin:sg:sec:imperf.perf | ||
303 | +167 fin:sg:sec:perf | ||
304 | +168 fin:sg:ter:imperf | ||
305 | +169 fin:sg:ter:imperf.perf | ||
306 | +170 fin:sg:ter:perf | ||
307 | +# past flexeme: | ||
308 | +# praet=split (unsued otherwise): | ||
309 | +509 praet:pl:m1.p1:imperf | ||
310 | +510 praet:pl:m1.p1:imperf.perf | ||
311 | +511 praet:pl:m1.p1:perf | ||
312 | +521 praet:pl:m2.m3.f.n1.n2.p2.p3:imperf | ||
313 | +522 praet:pl:m2.m3.f.n1.n2.p2.p3:imperf.perf | ||
314 | +523 praet:pl:m2.m3.f.n1.n2.p2.p3:perf | ||
315 | +533 praet:sg:f:imperf | ||
316 | +534 praet:sg:f:imperf.perf | ||
317 | +535 praet:sg:f:perf | ||
318 | +545 praet:sg:m1.m2.m3:imperf | ||
319 | +546 praet:sg:m1.m2.m3:imperf:agl | ||
320 | +547 praet:sg:m1.m2.m3:imperf:nagl | ||
321 | +548 praet:sg:m1.m2.m3:imperf.perf | ||
322 | +549 praet:sg:m1.m2.m3:perf | ||
323 | +550 praet:sg:m1.m2.m3:perf:agl | ||
324 | +551 praet:sg:m1.m2.m3:perf:nagl | ||
325 | +561 praet:sg:n1.n2:imperf | ||
326 | +562 praet:sg:n1.n2:imperf.perf | ||
327 | +563 praet:sg:n1.n2:perf | ||
328 | +# praet=composite (unsued otherwise): | ||
329 | +512 praet:pl:m1.p1:pri:imperf | ||
330 | +513 praet:pl:m1.p1:pri:imperf.perf | ||
331 | +514 praet:pl:m1.p1:pri:perf | ||
332 | +515 praet:pl:m1.p1:sec:imperf | ||
333 | +516 praet:pl:m1.p1:sec:imperf.perf | ||
334 | +517 praet:pl:m1.p1:sec:perf | ||
335 | +518 praet:pl:m1.p1:ter:imperf | ||
336 | +519 praet:pl:m1.p1:ter:imperf.perf | ||
337 | +520 praet:pl:m1.p1:ter:perf | ||
338 | +524 praet:pl:m2.m3.f.n1.n2.p2.p3:pri:imperf | ||
339 | +525 praet:pl:m2.m3.f.n1.n2.p2.p3:pri:imperf.perf | ||
340 | +526 praet:pl:m2.m3.f.n1.n2.p2.p3:pri:perf | ||
341 | +527 praet:pl:m2.m3.f.n1.n2.p2.p3:sec:imperf | ||
342 | +528 praet:pl:m2.m3.f.n1.n2.p2.p3:sec:imperf.perf | ||
343 | +529 praet:pl:m2.m3.f.n1.n2.p2.p3:sec:perf | ||
344 | +530 praet:pl:m2.m3.f.n1.n2.p2.p3:ter:imperf | ||
345 | +531 praet:pl:m2.m3.f.n1.n2.p2.p3:ter:imperf.perf | ||
346 | +532 praet:pl:m2.m3.f.n1.n2.p2.p3:ter:perf | ||
347 | +536 praet:sg:f:pri:imperf | ||
348 | +537 praet:sg:f:pri:imperf.perf | ||
349 | +538 praet:sg:f:pri:perf | ||
350 | +539 praet:sg:f:sec:imperf | ||
351 | +540 praet:sg:f:sec:imperf.perf | ||
352 | +541 praet:sg:f:sec:perf | ||
353 | +542 praet:sg:f:ter:imperf | ||
354 | +543 praet:sg:f:ter:imperf.perf | ||
355 | +544 praet:sg:f:ter:perf | ||
356 | +552 praet:sg:m1.m2.m3:pri:imperf | ||
357 | +553 praet:sg:m1.m2.m3:pri:imperf.perf | ||
358 | +554 praet:sg:m1.m2.m3:pri:perf | ||
359 | +555 praet:sg:m1.m2.m3:sec:imperf | ||
360 | +556 praet:sg:m1.m2.m3:sec:imperf.perf | ||
361 | +557 praet:sg:m1.m2.m3:sec:perf | ||
362 | +558 praet:sg:m1.m2.m3:ter:imperf | ||
363 | +559 praet:sg:m1.m2.m3:ter:imperf.perf | ||
364 | +560 praet:sg:m1.m2.m3:ter:perf | ||
365 | +564 praet:sg:n1.n2:pri:imperf | ||
366 | +565 praet:sg:n1.n2:pri:imperf.perf | ||
367 | +566 praet:sg:n1.n2:pri:perf | ||
368 | +567 praet:sg:n1.n2:sec:imperf | ||
369 | +568 praet:sg:n1.n2:sec:imperf.perf | ||
370 | +569 praet:sg:n1.n2:sec:perf | ||
371 | +570 praet:sg:n1.n2:ter:imperf | ||
372 | +571 praet:sg:n1.n2:ter:imperf.perf | ||
373 | +572 praet:sg:n1.n2:ter:perf | ||
374 | +# conditional mood (used only with praet=composite) | ||
375 | +100 cond:pl:m1.p1:pri:imperf | ||
376 | +101 cond:pl:m1.p1:pri:imperf.perf | ||
377 | +102 cond:pl:m1.p1:pri:perf | ||
378 | +103 cond:pl:m1.p1:sec:imperf | ||
379 | +104 cond:pl:m1.p1:sec:imperf.perf | ||
380 | +105 cond:pl:m1.p1:sec:perf | ||
381 | +106 cond:pl:m1.p1:ter:imperf | ||
382 | +107 cond:pl:m1.p1:ter:imperf.perf | ||
383 | +108 cond:pl:m1.p1:ter:perf | ||
384 | +109 cond:pl:m2.m3.f.n1.n2.p2.p3:pri:imperf | ||
385 | +110 cond:pl:m2.m3.f.n1.n2.p2.p3:pri:imperf.perf | ||
386 | +111 cond:pl:m2.m3.f.n1.n2.p2.p3:pri:perf | ||
387 | +112 cond:pl:m2.m3.f.n1.n2.p2.p3:sec:imperf | ||
388 | +113 cond:pl:m2.m3.f.n1.n2.p2.p3:sec:imperf.perf | ||
389 | +114 cond:pl:m2.m3.f.n1.n2.p2.p3:sec:perf | ||
390 | +115 cond:pl:m2.m3.f.n1.n2.p2.p3:ter:imperf | ||
391 | +116 cond:pl:m2.m3.f.n1.n2.p2.p3:ter:imperf.perf | ||
392 | +117 cond:pl:m2.m3.f.n1.n2.p2.p3:ter:perf | ||
393 | +118 cond:sg:f:pri:imperf | ||
394 | +119 cond:sg:f:pri:imperf.perf | ||
395 | +120 cond:sg:f:pri:perf | ||
396 | +121 cond:sg:f:sec:imperf | ||
397 | +122 cond:sg:f:sec:imperf.perf | ||
398 | +123 cond:sg:f:sec:perf | ||
399 | +124 cond:sg:f:ter:imperf | ||
400 | +125 cond:sg:f:ter:imperf.perf | ||
401 | +126 cond:sg:f:ter:perf | ||
402 | +127 cond:sg:m1.m2.m3:pri:imperf | ||
403 | +128 cond:sg:m1.m2.m3:pri:imperf.perf | ||
404 | +129 cond:sg:m1.m2.m3:pri:perf | ||
405 | +130 cond:sg:m1.m2.m3:sec:imperf | ||
406 | +131 cond:sg:m1.m2.m3:sec:imperf.perf | ||
407 | +132 cond:sg:m1.m2.m3:sec:perf | ||
408 | +133 cond:sg:m1.m2.m3:ter:imperf | ||
409 | +134 cond:sg:m1.m2.m3:ter:imperf.perf | ||
410 | +135 cond:sg:m1.m2.m3:ter:perf | ||
411 | +136 cond:sg:n1.n2:imperf | ||
412 | +137 cond:sg:n1.n2:imperf.perf | ||
413 | +138 cond:sg:n1.n2:perf | ||
414 | +139 cond:sg:n1.n2:pri:imperf | ||
415 | +140 cond:sg:n1.n2:pri:imperf.perf | ||
416 | +141 cond:sg:n1.n2:pri:perf | ||
417 | +142 cond:sg:n1.n2:sec:imperf | ||
418 | +143 cond:sg:n1.n2:sec:imperf.perf | ||
419 | +144 cond:sg:n1.n2:sec:perf | ||
420 | +145 cond:sg:n1.n2:ter:imperf | ||
421 | +146 cond:sg:n1.n2:ter:imperf.perf | ||
422 | +147 cond:sg:n1.n2:ter:perf | ||
423 | +# impersonal flexeme: | ||
424 | +219 imps:imperf | ||
425 | +220 imps:imperf.perf | ||
426 | +221 imps:perf | ||
427 | +# imperative flexeme: | ||
428 | +222 impt:pl:pri:imperf | ||
429 | +223 impt:pl:pri:imperf.perf | ||
430 | +224 impt:pl:pri:perf | ||
431 | +225 impt:pl:sec:imperf | ||
432 | +226 impt:pl:sec:imperf.perf | ||
433 | +227 impt:pl:sec:perf | ||
434 | +228 impt:sg:sec:imperf | ||
435 | +229 impt:sg:sec:imperf.perf | ||
436 | +230 impt:sg:sec:perf | ||
437 | +# infinitival flexeme: | ||
438 | +231 inf:imperf | ||
439 | +232 inf:imperf.perf | ||
440 | +233 inf:perf | ||
441 | +# agglutinative forms of ‘być’: | ||
442 | +83 aglt:pl:pri:imperf:nwok | ||
443 | +84 aglt:pl:pri:imperf:wok | ||
444 | +85 aglt:pl:sec:imperf:nwok | ||
445 | +86 aglt:pl:sec:imperf:wok | ||
446 | +87 aglt:sg:pri:imperf:nwok | ||
447 | +88 aglt:sg:pri:imperf:wok | ||
448 | +89 aglt:sg:sec:imperf:nwok | ||
449 | +90 aglt:sg:sec:imperf:wok | ||
450 | +# future forms of ‘być’: | ||
451 | +91 bedzie:pl:pri:imperf | ||
452 | +92 bedzie:pl:sec:imperf | ||
453 | +93 bedzie:pl:ter:imperf | ||
454 | +94 bedzie:sg:pri:imperf | ||
455 | +95 bedzie:sg:sec:imperf | ||
456 | +96 bedzie:sg:ter:imperf | ||
457 | +# ‘winien’ type verbs: | ||
458 | +705 winien:pl:m1.p1:imperf | ||
459 | +706 winien:pl:m1.p1:pri:imperf | ||
460 | +707 winien:pl:m1.p1:sec:imperf | ||
461 | +708 winien:pl:m1.p1:ter:imperf | ||
462 | +709 winien:pl:m2.m3.f.n1.n2.p2.p3:imperf | ||
463 | +710 winien:pl:m2.m3.f.n1.n2.p2.p3:sec:imperf | ||
464 | +711 winien:pl:m2.m3.f.n1.n2.p2.p3:ter:imperf | ||
465 | +712 winien:sg:f:imperf | ||
466 | +713 winien:sg:f:pri:imperf | ||
467 | +714 winien:sg:f:sec:imperf | ||
468 | +715 winien:sg:f:ter:imperf | ||
469 | +716 winien:sg:m1.m2.m3:imperf | ||
470 | +717 winien:sg:m1.m2.m3:pri:imperf | ||
471 | +718 winien:sg:m1.m2.m3:sec:imperf | ||
472 | +719 winien:sg:m1.m2.m3:ter:imperf | ||
473 | +720 winien:sg:n1.n2:imperf | ||
474 | +721 winien:sg:n1.n2:pri:imperf | ||
475 | +722 winien:sg:n1.n2:sec:imperf | ||
476 | +723 winien:sg:n1.n2:ter:imperf | ||
477 | +# predicative flexeme: | ||
478 | +573 pred | ||
479 | +# gerunds | ||
480 | +171 ger:pl:dat.loc:n2:imperf:aff | ||
481 | +172 ger:pl:dat.loc:n2:imperf:neg | ||
482 | +173 ger:pl:dat.loc:n2:imperf.perf:aff | ||
483 | +174 ger:pl:dat.loc:n2:imperf.perf:neg | ||
484 | +175 ger:pl:dat.loc:n2:perf:aff | ||
485 | +176 ger:pl:dat.loc:n2:perf:neg | ||
486 | +177 ger:pl:gen:n2:imperf:aff | ||
487 | +178 ger:pl:gen:n2:imperf:neg | ||
488 | +179 ger:pl:gen:n2:imperf.perf:aff | ||
489 | +180 ger:pl:gen:n2:imperf.perf:neg | ||
490 | +181 ger:pl:gen:n2:perf:aff | ||
491 | +182 ger:pl:gen:n2:perf:neg | ||
492 | +183 ger:pl:inst:n2:imperf:aff | ||
493 | +184 ger:pl:inst:n2:imperf:neg | ||
494 | +185 ger:pl:inst:n2:imperf.perf:aff | ||
495 | +186 ger:pl:inst:n2:imperf.perf:neg | ||
496 | +187 ger:pl:inst:n2:perf:aff | ||
497 | +188 ger:pl:inst:n2:perf:neg | ||
498 | +189 ger:pl:nom.acc:n2:imperf:aff | ||
499 | +190 ger:pl:nom.acc:n2:imperf:neg | ||
500 | +191 ger:pl:nom.acc:n2:imperf.perf:aff | ||
501 | +192 ger:pl:nom.acc:n2:imperf.perf:neg | ||
502 | +193 ger:pl:nom.acc:n2:perf:aff | ||
503 | +194 ger:pl:nom.acc:n2:perf:neg | ||
504 | +195 ger:sg:dat.loc:n2:imperf:aff | ||
505 | +196 ger:sg:dat.loc:n2:imperf:neg | ||
506 | +197 ger:sg:dat.loc:n2:imperf.perf:aff | ||
507 | +198 ger:sg:dat.loc:n2:imperf.perf:neg | ||
508 | +199 ger:sg:dat.loc:n2:perf:aff | ||
509 | +200 ger:sg:dat.loc:n2:perf:neg | ||
510 | +201 ger:sg:gen:n2:imperf:aff | ||
511 | +202 ger:sg:gen:n2:imperf:neg | ||
512 | +203 ger:sg:gen:n2:imperf.perf:aff | ||
513 | +204 ger:sg:gen:n2:imperf.perf:neg | ||
514 | +205 ger:sg:gen:n2:perf:aff | ||
515 | +206 ger:sg:gen:n2:perf:neg | ||
516 | +207 ger:sg:inst:n2:imperf:aff | ||
517 | +208 ger:sg:inst:n2:imperf:neg | ||
518 | +209 ger:sg:inst:n2:imperf.perf:aff | ||
519 | +210 ger:sg:inst:n2:imperf.perf:neg | ||
520 | +211 ger:sg:inst:n2:perf:aff | ||
521 | +212 ger:sg:inst:n2:perf:neg | ||
522 | +213 ger:sg:nom.acc:n2:imperf:aff | ||
523 | +214 ger:sg:nom.acc:n2:imperf:neg | ||
524 | +215 ger:sg:nom.acc:n2:imperf.perf:aff | ||
525 | +216 ger:sg:nom.acc:n2:imperf.perf:neg | ||
526 | +217 ger:sg:nom.acc:n2:perf:aff | ||
527 | +218 ger:sg:nom.acc:n2:perf:neg | ||
528 | +# participles | ||
529 | +# adverbial participles: | ||
530 | +332 pcon:imperf | ||
531 | +331 pant:perf | ||
532 | +# adjectival active participle: | ||
533 | +267 pact:pl:acc:m1.p1:imperf:aff | ||
534 | +268 pact:pl:acc:m1.p1:imperf:neg | ||
535 | +269 pact:pl:acc:m1.p1:imperf.perf:aff | ||
536 | +270 pact:pl:acc:m1.p1:imperf.perf:neg | ||
537 | +271 pact:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:aff | ||
538 | +272 pact:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:neg | ||
539 | +273 pact:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:aff | ||
540 | +274 pact:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:neg | ||
541 | +275 pact:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:aff | ||
542 | +276 pact:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:neg | ||
543 | +277 pact:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:aff | ||
544 | +278 pact:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:neg | ||
545 | +279 pact:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:aff | ||
546 | +280 pact:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:neg | ||
547 | +281 pact:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:aff | ||
548 | +282 pact:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:neg | ||
549 | +283 pact:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf:aff | ||
550 | +284 pact:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf:neg | ||
551 | +285 pact:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf.perf:aff | ||
552 | +286 pact:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf.perf:neg | ||
553 | +287 pact:pl:nom.voc:m1.p1:imperf:aff | ||
554 | +288 pact:pl:nom.voc:m1.p1:imperf:neg | ||
555 | +289 pact:pl:nom.voc:m1.p1:imperf.perf:aff | ||
556 | +290 pact:pl:nom.voc:m1.p1:imperf.perf:neg | ||
557 | +291 pact:sg:acc.inst:f:imperf:aff | ||
558 | +292 pact:sg:acc.inst:f:imperf:neg | ||
559 | +293 pact:sg:acc.inst:f:imperf.perf:aff | ||
560 | +294 pact:sg:acc.inst:f:imperf.perf:neg | ||
561 | +295 pact:sg:acc:m1.m2:imperf:aff | ||
562 | +296 pact:sg:acc:m1.m2:imperf:neg | ||
563 | +297 pact:sg:acc:m1.m2:imperf.perf:aff | ||
564 | +298 pact:sg:acc:m1.m2:imperf.perf:neg | ||
565 | +299 pact:sg:acc:m3:imperf:aff | ||
566 | +300 pact:sg:acc:m3:imperf:neg | ||
567 | +301 pact:sg:acc:m3:imperf.perf:aff | ||
568 | +302 pact:sg:acc:m3:imperf.perf:neg | ||
569 | +303 pact:sg:dat:m1.m2.m3.n1.n2:imperf:aff | ||
570 | +304 pact:sg:dat:m1.m2.m3.n1.n2:imperf:neg | ||
571 | +305 pact:sg:dat:m1.m2.m3.n1.n2:imperf.perf:aff | ||
572 | +306 pact:sg:dat:m1.m2.m3.n1.n2:imperf.perf:neg | ||
573 | +307 pact:sg:gen.dat.loc:f:imperf:aff | ||
574 | +308 pact:sg:gen.dat.loc:f:imperf:neg | ||
575 | +309 pact:sg:gen.dat.loc:f:imperf.perf:aff | ||
576 | +310 pact:sg:gen.dat.loc:f:imperf.perf:neg | ||
577 | +311 pact:sg:gen:m1.m2.m3.n1.n2:imperf:aff | ||
578 | +312 pact:sg:gen:m1.m2.m3.n1.n2:imperf:neg | ||
579 | +313 pact:sg:gen:m1.m2.m3.n1.n2:imperf.perf:aff | ||
580 | +314 pact:sg:gen:m1.m2.m3.n1.n2:imperf.perf:neg | ||
581 | +315 pact:sg:inst.loc:m1.m2.m3.n1.n2:imperf:aff | ||
582 | +316 pact:sg:inst.loc:m1.m2.m3.n1.n2:imperf:neg | ||
583 | +317 pact:sg:inst.loc:m1.m2.m3.n1.n2:imperf.perf:aff | ||
584 | +318 pact:sg:inst.loc:m1.m2.m3.n1.n2:imperf.perf:neg | ||
585 | +319 pact:sg:nom.acc.voc:n1.n2:imperf:aff | ||
586 | +320 pact:sg:nom.acc.voc:n1.n2:imperf:neg | ||
587 | +321 pact:sg:nom.acc.voc:n1.n2:imperf.perf:aff | ||
588 | +322 pact:sg:nom.acc.voc:n1.n2:imperf.perf:neg | ||
589 | +323 pact:sg:nom.voc:f:imperf:aff | ||
590 | +324 pact:sg:nom.voc:f:imperf:neg | ||
591 | +325 pact:sg:nom.voc:f:imperf.perf:aff | ||
592 | +326 pact:sg:nom.voc:f:imperf.perf:neg | ||
593 | +327 pact:sg:nom.voc:m1.m2.m3:imperf:aff | ||
594 | +328 pact:sg:nom.voc:m1.m2.m3:imperf:neg | ||
595 | +329 pact:sg:nom.voc:m1.m2.m3:imperf.perf:aff | ||
596 | +330 pact:sg:nom.voc:m1.m2.m3:imperf.perf:neg | ||
597 | +# adjectival passive participle: | ||
598 | +333 ppas:pl:acc:m1.p1:imperf:aff | ||
599 | +334 ppas:pl:acc:m1.p1:imperf:neg | ||
600 | +335 ppas:pl:acc:m1.p1:imperf.perf:aff | ||
601 | +336 ppas:pl:acc:m1.p1:imperf.perf:neg | ||
602 | +337 ppas:pl:acc:m1.p1:perf:aff | ||
603 | +338 ppas:pl:acc:m1.p1:perf:neg | ||
604 | +339 ppas:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:aff | ||
605 | +340 ppas:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:neg | ||
606 | +341 ppas:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:aff | ||
607 | +342 ppas:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:neg | ||
608 | +343 ppas:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:perf:aff | ||
609 | +344 ppas:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:perf:neg | ||
610 | +345 ppas:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:aff | ||
611 | +346 ppas:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:neg | ||
612 | +347 ppas:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:aff | ||
613 | +348 ppas:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:neg | ||
614 | +349 ppas:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:perf:aff | ||
615 | +350 ppas:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:perf:neg | ||
616 | +351 ppas:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:aff | ||
617 | +352 ppas:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:neg | ||
618 | +353 ppas:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:aff | ||
619 | +354 ppas:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:neg | ||
620 | +355 ppas:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:perf:aff | ||
621 | +356 ppas:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:perf:neg | ||
622 | +357 ppas:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf:aff | ||
623 | +358 ppas:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf:neg | ||
624 | +359 ppas:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf.perf:aff | ||
625 | +360 ppas:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf.perf:neg | ||
626 | +361 ppas:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:perf:aff | ||
627 | +362 ppas:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:perf:neg | ||
628 | +363 ppas:pl:nom.voc:m1.p1:imperf:aff | ||
629 | +364 ppas:pl:nom.voc:m1.p1:imperf:neg | ||
630 | +365 ppas:pl:nom.voc:m1.p1:imperf.perf:aff | ||
631 | +366 ppas:pl:nom.voc:m1.p1:imperf.perf:neg | ||
632 | +367 ppas:pl:nom.voc:m1.p1:perf:aff | ||
633 | +368 ppas:pl:nom.voc:m1.p1:perf:neg | ||
634 | +369 ppas:sg:acc.inst:f:imperf:aff | ||
635 | +370 ppas:sg:acc.inst:f:imperf:neg | ||
636 | +371 ppas:sg:acc.inst:f:imperf.perf:aff | ||
637 | +372 ppas:sg:acc.inst:f:imperf.perf:neg | ||
638 | +373 ppas:sg:acc.inst:f:perf:aff | ||
639 | +374 ppas:sg:acc.inst:f:perf:neg | ||
640 | +375 ppas:sg:acc:m1.m2:imperf:aff | ||
641 | +376 ppas:sg:acc:m1.m2:imperf:neg | ||
642 | +377 ppas:sg:acc:m1.m2:imperf.perf:aff | ||
643 | +378 ppas:sg:acc:m1.m2:imperf.perf:neg | ||
644 | +379 ppas:sg:acc:m1.m2:perf:aff | ||
645 | +380 ppas:sg:acc:m1.m2:perf:neg | ||
646 | +381 ppas:sg:acc:m3:imperf:aff | ||
647 | +382 ppas:sg:acc:m3:imperf:neg | ||
648 | +383 ppas:sg:acc:m3:imperf.perf:aff | ||
649 | +384 ppas:sg:acc:m3:imperf.perf:neg | ||
650 | +385 ppas:sg:acc:m3:perf:aff | ||
651 | +386 ppas:sg:acc:m3:perf:neg | ||
652 | +387 ppas:sg:dat:m1.m2.m3.n1.n2:imperf:aff | ||
653 | +388 ppas:sg:dat:m1.m2.m3.n1.n2:imperf:neg | ||
654 | +389 ppas:sg:dat:m1.m2.m3.n1.n2:imperf.perf:aff | ||
655 | +390 ppas:sg:dat:m1.m2.m3.n1.n2:imperf.perf:neg | ||
656 | +391 ppas:sg:dat:m1.m2.m3.n1.n2:perf:aff | ||
657 | +392 ppas:sg:dat:m1.m2.m3.n1.n2:perf:neg | ||
658 | +393 ppas:sg:gen.dat.loc:f:imperf:aff | ||
659 | +394 ppas:sg:gen.dat.loc:f:imperf:neg | ||
660 | +395 ppas:sg:gen.dat.loc:f:imperf.perf:aff | ||
661 | +396 ppas:sg:gen.dat.loc:f:imperf.perf:neg | ||
662 | +397 ppas:sg:gen.dat.loc:f:perf:aff | ||
663 | +398 ppas:sg:gen.dat.loc:f:perf:neg | ||
664 | +399 ppas:sg:gen:m1.m2.m3.n1.n2:imperf:aff | ||
665 | +400 ppas:sg:gen:m1.m2.m3.n1.n2:imperf:neg | ||
666 | +401 ppas:sg:gen:m1.m2.m3.n1.n2:imperf.perf:aff | ||
667 | +402 ppas:sg:gen:m1.m2.m3.n1.n2:imperf.perf:neg | ||
668 | +403 ppas:sg:gen:m1.m2.m3.n1.n2:perf:aff | ||
669 | +404 ppas:sg:gen:m1.m2.m3.n1.n2:perf:neg | ||
670 | +405 ppas:sg:inst.loc:m1.m2.m3.n1.n2:imperf:aff | ||
671 | +406 ppas:sg:inst.loc:m1.m2.m3.n1.n2:imperf:neg | ||
672 | +407 ppas:sg:inst.loc:m1.m2.m3.n1.n2:imperf.perf:aff | ||
673 | +408 ppas:sg:inst.loc:m1.m2.m3.n1.n2:imperf.perf:neg | ||
674 | +409 ppas:sg:inst.loc:m1.m2.m3.n1.n2:perf:aff | ||
675 | +410 ppas:sg:inst.loc:m1.m2.m3.n1.n2:perf:neg | ||
676 | +411 ppas:sg:nom.acc.voc:n1.n2:imperf:aff | ||
677 | +412 ppas:sg:nom.acc.voc:n1.n2:imperf:neg | ||
678 | +413 ppas:sg:nom.acc.voc:n1.n2:imperf.perf:aff | ||
679 | +414 ppas:sg:nom.acc.voc:n1.n2:imperf.perf:neg | ||
680 | +415 ppas:sg:nom.acc.voc:n1.n2:perf:aff | ||
681 | +416 ppas:sg:nom.acc.voc:n1.n2:perf:neg | ||
682 | +417 ppas:sg:nom.voc:f:imperf:aff | ||
683 | +418 ppas:sg:nom.voc:f:imperf:neg | ||
684 | +419 ppas:sg:nom.voc:f:imperf.perf:aff | ||
685 | +420 ppas:sg:nom.voc:f:imperf.perf:neg | ||
686 | +421 ppas:sg:nom.voc:f:perf:aff | ||
687 | +422 ppas:sg:nom.voc:f:perf:neg | ||
688 | +423 ppas:sg:nom.voc:m1.m2.m3:imperf:aff | ||
689 | +424 ppas:sg:nom.voc:m1.m2.m3:imperf:neg | ||
690 | +425 ppas:sg:nom.voc:m1.m2.m3:imperf.perf:aff | ||
691 | +426 ppas:sg:nom.voc:m1.m2.m3:imperf.perf:neg | ||
692 | +427 ppas:sg:nom.voc:m1.m2.m3:perf:aff | ||
693 | +428 ppas:sg:nom.voc:m1.m2.m3:perf:neg | ||
694 | +# NUMERALS | ||
695 | +239 num:pl:acc:m1:rec | ||
696 | +240 num:pl:dat.loc:n1.p1.p2:congr.rec | ||
697 | +241 num:pl:dat:m1.m2.m3.n2.f:congr | ||
698 | +242 num:pl:gen.dat.inst.loc:m1.m2.m3.f.n1.n2.p1.p2:congr | ||
699 | +243 num:pl:gen.dat.inst.loc:m1.m2.m3.f.n2:congr | ||
700 | +244 num:pl:gen.dat.loc:m1.m2.m3.n2.f:congr | ||
701 | +245 num:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2:congr | ||
702 | +246 num:pl:gen.loc:m1.m2.m3.n2.f:congr | ||
703 | +247 num:pl:gen:n1.p1.p2:rec | ||
704 | +248 num:pl:inst:f:congr | ||
705 | +249 num:pl:inst:m1.m2.m3.f.n1.n2.p1.p2:congr | ||
706 | +250 num:pl:inst:m1.m2.m3.f.n2:congr | ||
707 | +251 num:pl:inst:m1.m2.m3.n2:congr | ||
708 | +252 num:pl:inst:m1.m2.m3.n2.f:congr | ||
709 | +253 num:pl:inst:n1.p1.p2:rec | ||
710 | +254 num:pl:nom.acc:m1.m2.m3.f.n1.n2.p1.p2:rec | ||
711 | +255 num:pl:nom.acc.voc:f:congr | ||
712 | +256 num:pl:nom.acc.voc:m1:rec | ||
713 | +257 num:pl:nom.acc.voc:m2.m3.f.n1.n2.p1.p2:rec | ||
714 | +258 num:pl:nom.acc.voc:m2.m3.f.n2:rec | ||
715 | +259 num:pl:nom.acc.voc:m2.m3.n2:congr | ||
716 | +260 num:pl:nom.acc.voc:m2.m3.n2.f:congr | ||
717 | +261 num:pl:nom.acc.voc:n1.p1.p2:rec | ||
718 | +262 num:pl:nom.gen.dat.inst.acc.loc.voc:m1.m2.m3.f.n1.n2.p1.p2:rec | ||
719 | +263 num:pl:nom.voc:m1:congr | ||
720 | +264 num:pl:nom.voc:m1:rec | ||
721 | +265 num:sg:nom.gen.dat.inst.acc.loc.voc:f:rec | ||
722 | +266 num:sg:nom.gen.dat.inst.acc.loc.voc:m1.m2.m3.n1.n2:rec | ||
723 | +# numeral compounds forming form: | ||
724 | +238 num:comp | ||
725 | +# PREPOSITIONS | ||
726 | +578 prep:acc | ||
727 | +579 prep:acc:nwok | ||
728 | +580 prep:acc:wok | ||
729 | +581 prep:dat | ||
730 | +582 prep:gen | ||
731 | +583 prep:gen:nwok | ||
732 | +584 prep:gen:wok | ||
733 | +585 prep:inst | ||
734 | +586 prep:inst:nwok | ||
735 | +587 prep:inst:wok | ||
736 | +588 prep:loc | ||
737 | +589 prep:loc:nwok | ||
738 | +590 prep:loc:wok | ||
739 | +591 prep:nom | ||
740 | +# ADVERBS | ||
741 | +79 adv | ||
742 | +80 adv:com | ||
743 | +81 adv:pos | ||
744 | +82 adv:sup | ||
745 | +# OTHER | ||
746 | +# kubliki (particles): | ||
747 | +592 qub | ||
748 | +# conjunctions: | ||
749 | +148 conj | ||
750 | +# complementizers: | ||
751 | +99 comp | ||
752 | +# interjections: | ||
753 | +234 interj | ||
754 | +# burkinostki (bound words): | ||
755 | +98 burk | ||
756 | +# abbreviations: | ||
757 | +97 brev:pun | ||
758 | +97 brev:npun | ||
759 | +# punctuation: | ||
760 | +235 interp | ||
761 | +# digits: | ||
762 | +151 dig | ||
763 | +# Roman digits: | ||
764 | +593 romandig | ||
765 | +# emoticons: | ||
766 | +152 emoticon | ||
767 | +# prefixes: | ||
768 | +574 prefa | ||
769 | +575 prefppas | ||
770 | +576 prefs | ||
771 | +577 prefv | ||
772 | +# (special) | ||
773 | +236 naj | ||
774 | +237 nie |
tests/generator/test_additional_atomic_segments/segmentation.dat
tests/generator/test_digits/segmentation.dat
tests/generator/test_names/segmentation.dat
tests/generator/test_qualifiers/segmentation.dat
tests/generator/test_segtypes/ARGS
0 → 100644
1 | +--aggl permissive |
tests/generator/test_segtypes/dictionary.tab
0 → 100644
1 | +mikro mikro prefs | ||
2 | +euro euro prefs | ||
3 | +super super prefs | ||
4 | +kot kot subst:sg:nom:m1 imie mikro|zzz | ||
5 | +kot kot subst:sg:nom:m1 nazwisko euro | ||
6 | +kot kot subst:sg:nom:m1 xxx|yyy|super | ||
7 | +kot kot subst:sg:nom:m1 samodzielnie|zzz | ||
8 | +kot kot subst:sg:nom:m2 imie samodzielnie | ||
9 | +kot kot adj:sg:acc:m3:pos nigdy | ||
10 | +pies pies subst:sg:acc:m3 aaa|zzz|euro | ||
11 | +pies pies subst:sg:nom:m1 nazwisko yyy|zzz|samodzielnie |
tests/generator/test_segtypes/input.txt
0 → 100644
tests/generator/test_segtypes/output.txt
0 → 100644
1 | +[kot,kot,subst:sg:nom:m1,_,samodzielnie|zzz | ||
2 | + kot,kot,subst:sg:nom:m2,imie,samodzielnie] | ||
3 | +[eurokot,eurokot,subst:sg:nom:m1,nazwisko,euro] | ||
4 | +[superkot,superkot,subst:sg:nom:m1,_,super|xxx|yyy] | ||
5 | +[mikrokot,mikrokot,subst:sg:nom:m1,imie,mikro|zzz] | ||
6 | +[kotkot,kotkot,ign,_,_] | ||
7 | +[pies,pies,subst:sg:nom:m1,nazwisko,samodzielnie|yyy|zzz] | ||
8 | +[europies,europies,subst:sg:acc:m3,_,aaa|zzz|euro] | ||
9 | +[superpies,superpies,ign,_,_] | ||
10 | +[mikropies,mikropies,ign,_,_] | ||
11 | + |
tests/generator/test_segtypes/segmentation.dat
0 → 100644
1 | +[options] | ||
2 | +aggl=strict permissive isolated | ||
3 | +praet=split composite | ||
4 | + | ||
5 | +[combinations] | ||
6 | + | ||
7 | +samodzielnie | ||
8 | +mikro> tylko_z_mikro | ||
9 | +euro> tylko_z_euro | ||
10 | +super> tylko_z_super | ||
11 | + | ||
12 | +[segment types] | ||
13 | +tylko_z_mikro | ||
14 | +tylko_z_euro | ||
15 | +tylko_z_super | ||
16 | +samodzielnie | ||
17 | +nigdy | ||
18 | +mikro | ||
19 | +euro | ||
20 | +super | ||
21 | + | ||
22 | +[lexemes] | ||
23 | +tylko_z_mikro kot subst:% name=imie labels=zzz | ||
24 | +tylko_z_euro kot subst:% name=nazwisko | ||
25 | +tylko_z_super kot subst:% labels=xxx|yyy | ||
26 | +samodzielnie kot subst:% | ||
27 | + | ||
28 | +mikro mikro % | ||
29 | +euro euro % | ||
30 | +super super % | ||
31 | + | ||
32 | +[tags] | ||
33 | +tylko_z_euro subst:% labels=aaa | ||
34 | +samodzielnie subst:% | ||
35 | +nigdy % | ||
36 | + | ||
37 | +[separator chars] | ||
38 | +# , | ||
39 | +44 | ||
40 | + | ||
41 | +# . | ||
42 | +46 | ||
43 | + | ||
44 | +# ; | ||
45 | +59 |
tests/generator/test_segtypes/tagset.dat
0 → 100644
1 | +#!TAGSET-ID pl.sgjp.morfeusz-0.5.0 | ||
2 | + | ||
3 | +[TAGS] | ||
4 | +# special: unknown word (ignotum): | ||
5 | +0 ign | ||
6 | +# special: space/blank: | ||
7 | +1 sp | ||
8 | +# NOUNS | ||
9 | +694 subst:sg:nom:m1 | ||
10 | +695 subst:sg:nom:m2 | ||
11 | +696 subst:sg:nom:m3 | ||
12 | +697 subst:sg:nom:n1 | ||
13 | +698 subst:sg:nom:n2 | ||
14 | +693 subst:sg:nom:f | ||
15 | +676 subst:sg:gen:m1 | ||
16 | +677 subst:sg:gen:m2 | ||
17 | +678 subst:sg:gen:m3 | ||
18 | +679 subst:sg:gen:n1 | ||
19 | +680 subst:sg:gen:n2 | ||
20 | +675 subst:sg:gen:f | ||
21 | +670 subst:sg:dat:m1 | ||
22 | +671 subst:sg:dat:m2 | ||
23 | +672 subst:sg:dat:m3 | ||
24 | +673 subst:sg:dat:n1 | ||
25 | +674 subst:sg:dat:n2 | ||
26 | +669 subst:sg:dat:f | ||
27 | +664 subst:sg:acc:m1 | ||
28 | +665 subst:sg:acc:m2 | ||
29 | +666 subst:sg:acc:m3 | ||
30 | +667 subst:sg:acc:n1 | ||
31 | +668 subst:sg:acc:n2 | ||
32 | +663 subst:sg:acc:f | ||
33 | +682 subst:sg:inst:m1 | ||
34 | +683 subst:sg:inst:m2 | ||
35 | +684 subst:sg:inst:m3 | ||
36 | +685 subst:sg:inst:n1 | ||
37 | +686 subst:sg:inst:n2 | ||
38 | +681 subst:sg:inst:f | ||
39 | +688 subst:sg:loc:m1 | ||
40 | +689 subst:sg:loc:m2 | ||
41 | +690 subst:sg:loc:m3 | ||
42 | +691 subst:sg:loc:n1 | ||
43 | +692 subst:sg:loc:n2 | ||
44 | +687 subst:sg:loc:f | ||
45 | +700 subst:sg:voc:m1 | ||
46 | +701 subst:sg:voc:m2 | ||
47 | +702 subst:sg:voc:m3 | ||
48 | +703 subst:sg:voc:n1 | ||
49 | +704 subst:sg:voc:n2 | ||
50 | +699 subst:sg:voc:f | ||
51 | +646 subst:pl:nom:m1 | ||
52 | +647 subst:pl:nom:m2 | ||
53 | +648 subst:pl:nom:m3 | ||
54 | +649 subst:pl:nom:n1 | ||
55 | +650 subst:pl:nom:n2 | ||
56 | +651 subst:pl:nom:p1 | ||
57 | +652 subst:pl:nom:p2 | ||
58 | +653 subst:pl:nom:p3 | ||
59 | +645 subst:pl:nom:f | ||
60 | +619 subst:pl:gen:m1 | ||
61 | +620 subst:pl:gen:m2 | ||
62 | +621 subst:pl:gen:m3 | ||
63 | +622 subst:pl:gen:n1 | ||
64 | +623 subst:pl:gen:n2 | ||
65 | +624 subst:pl:gen:p1 | ||
66 | +625 subst:pl:gen:p2 | ||
67 | +626 subst:pl:gen:p3 | ||
68 | +618 subst:pl:gen:f | ||
69 | +610 subst:pl:dat:m1 | ||
70 | +611 subst:pl:dat:m2 | ||
71 | +612 subst:pl:dat:m3 | ||
72 | +613 subst:pl:dat:n1 | ||
73 | +614 subst:pl:dat:n2 | ||
74 | +615 subst:pl:dat:p1 | ||
75 | +616 subst:pl:dat:p2 | ||
76 | +617 subst:pl:dat:p3 | ||
77 | +609 subst:pl:dat:f | ||
78 | +601 subst:pl:acc:m1 | ||
79 | +602 subst:pl:acc:m2 | ||
80 | +603 subst:pl:acc:m3 | ||
81 | +604 subst:pl:acc:n1 | ||
82 | +605 subst:pl:acc:n2 | ||
83 | +606 subst:pl:acc:p1 | ||
84 | +607 subst:pl:acc:p2 | ||
85 | +608 subst:pl:acc:p3 | ||
86 | +600 subst:pl:acc:f | ||
87 | +628 subst:pl:inst:m1 | ||
88 | +629 subst:pl:inst:m2 | ||
89 | +630 subst:pl:inst:m3 | ||
90 | +631 subst:pl:inst:n1 | ||
91 | +632 subst:pl:inst:n2 | ||
92 | +633 subst:pl:inst:p1 | ||
93 | +634 subst:pl:inst:p2 | ||
94 | +635 subst:pl:inst:p3 | ||
95 | +627 subst:pl:inst:f | ||
96 | +637 subst:pl:loc:m1 | ||
97 | +638 subst:pl:loc:m2 | ||
98 | +639 subst:pl:loc:m3 | ||
99 | +640 subst:pl:loc:n1 | ||
100 | +641 subst:pl:loc:n2 | ||
101 | +642 subst:pl:loc:p1 | ||
102 | +643 subst:pl:loc:p2 | ||
103 | +644 subst:pl:loc:p3 | ||
104 | +636 subst:pl:loc:f | ||
105 | +654 subst:pl:voc:f | ||
106 | +655 subst:pl:voc:m1 | ||
107 | +656 subst:pl:voc:m2 | ||
108 | +657 subst:pl:voc:m3 | ||
109 | +658 subst:pl:voc:n1 | ||
110 | +659 subst:pl:voc:n2 | ||
111 | +660 subst:pl:voc:p1 | ||
112 | +661 subst:pl:voc:p2 | ||
113 | +662 subst:pl:voc:p3 | ||
114 | +# depreciative nominal flexeme: | ||
115 | +149 depr:pl:nom:m2 | ||
116 | +150 depr:pl:voc:m2 | ||
117 | +# nominal compounds forming form: | ||
118 | +599 substa | ||
119 | +# PERSONAL PRONOUNS | ||
120 | +443 ppron12:sg:acc:m1.m2.m3.f.n1.n2:pri:akc | ||
121 | +444 ppron12:sg:acc:m1.m2.m3.f.n1.n2:pri:nakc | ||
122 | +445 ppron12:sg:acc:m1.m2.m3.f.n1.n2:sec:akc | ||
123 | +446 ppron12:sg:acc:m1.m2.m3.f.n1.n2:sec:nakc | ||
124 | +447 ppron12:sg:dat:m1.m2.m3.f.n1.n2:pri:akc | ||
125 | +448 ppron12:sg:dat:m1.m2.m3.f.n1.n2:pri:nakc | ||
126 | +449 ppron12:sg:dat:m1.m2.m3.f.n1.n2:sec:akc | ||
127 | +450 ppron12:sg:dat:m1.m2.m3.f.n1.n2:sec:nakc | ||
128 | +451 ppron12:sg:gen:m1.m2.m3.f.n1.n2:pri:akc | ||
129 | +452 ppron12:sg:gen:m1.m2.m3.f.n1.n2:pri:nakc | ||
130 | +453 ppron12:sg:gen:m1.m2.m3.f.n1.n2:sec:akc | ||
131 | +454 ppron12:sg:gen:m1.m2.m3.f.n1.n2:sec:nakc | ||
132 | +455 ppron12:sg:inst:m1.m2.m3.f.n1.n2:pri | ||
133 | +456 ppron12:sg:inst:m1.m2.m3.f.n1.n2:sec | ||
134 | +457 ppron12:sg:loc:m1.m2.m3.f.n1.n2:pri | ||
135 | +458 ppron12:sg:loc:m1.m2.m3.f.n1.n2:sec | ||
136 | +459 ppron12:sg:nom:m1.m2.m3.f.n1.n2:pri | ||
137 | +460 ppron12:sg:nom:m1.m2.m3.f.n1.n2:sec | ||
138 | +461 ppron12:sg:voc:m1.m2.m3.f.n1.n2:sec | ||
139 | +429 ppron12:pl:acc:_:pri | ||
140 | +430 ppron12:pl:acc:_:sec | ||
141 | +431 ppron12:pl:dat:_:pri | ||
142 | +432 ppron12:pl:dat:_:sec | ||
143 | +433 ppron12:pl:gen:_:pri | ||
144 | +434 ppron12:pl:gen:_:sec | ||
145 | +435 ppron12:pl:inst:_:pri | ||
146 | +436 ppron12:pl:inst:_:sec | ||
147 | +437 ppron12:pl:loc:_:pri | ||
148 | +438 ppron12:pl:loc:_:sec | ||
149 | +439 ppron12:pl:nom:_:pri | ||
150 | +440 ppron12:pl:nom:_:sec | ||
151 | +441 ppron12:pl:voc:_:pri | ||
152 | +442 ppron12:pl:voc:_:sec | ||
153 | +474 ppron3:sg:acc:f:ter:_:npraep | ||
154 | +475 ppron3:sg:acc:f:ter:_:praep | ||
155 | +476 ppron3:sg:acc:m1.m2.m3:ter:akc:npraep | ||
156 | +477 ppron3:sg:acc:m1.m2.m3:ter:akc:praep | ||
157 | +478 ppron3:sg:acc:m1.m2.m3:ter:nakc:npraep | ||
158 | +479 ppron3:sg:acc:m1.m2.m3:ter:nakc:praep | ||
159 | +480 ppron3:sg:acc:n1.n2:ter:_:npraep | ||
160 | +481 ppron3:sg:acc:n1.n2:ter:_:praep | ||
161 | +482 ppron3:sg:dat:f:ter:_:npraep | ||
162 | +483 ppron3:sg:dat:f:ter:_:praep | ||
163 | +484 ppron3:sg:dat:m1.m2.m3:ter:akc:npraep | ||
164 | +485 ppron3:sg:dat:m1.m2.m3:ter:nakc:npraep | ||
165 | +486 ppron3:sg:dat:m1.m2.m3:ter:_:praep | ||
166 | +487 ppron3:sg:dat:n1.n2:ter:akc:npraep | ||
167 | +488 ppron3:sg:dat:n1.n2:ter:nakc:npraep | ||
168 | +489 ppron3:sg:dat:n1.n2:ter:_:praep | ||
169 | +490 ppron3:sg:gen.acc:m1.m2.m3:ter:nakc:praep | ||
170 | +491 ppron3:sg:gen:f:ter:_:npraep | ||
171 | +492 ppron3:sg:gen:f:ter:_:praep | ||
172 | +493 ppron3:sg:gen:m1.m2.m3:ter:akc:npraep | ||
173 | +494 ppron3:sg:gen:m1.m2.m3:ter:akc:praep | ||
174 | +495 ppron3:sg:gen:m1.m2.m3:ter:nakc:npraep | ||
175 | +496 ppron3:sg:gen:m1.m2.m3:ter:nakc:praep | ||
176 | +497 ppron3:sg:gen:n1.n2:ter:akc:npraep | ||
177 | +498 ppron3:sg:gen:n1.n2:ter:nakc:npraep | ||
178 | +499 ppron3:sg:gen:n1.n2:ter:_:praep | ||
179 | +500 ppron3:sg:inst:f:ter:_:praep | ||
180 | +501 ppron3:sg:inst:m1.m2.m3:ter:_:_ | ||
181 | +502 ppron3:sg:inst:n1.n2:ter:_:_ | ||
182 | +503 ppron3:sg:loc:f:ter:_:_ | ||
183 | +504 ppron3:sg:loc:m1.m2.m3:ter:_:_ | ||
184 | +505 ppron3:sg:loc:n1.n2:ter:_:_ | ||
185 | +506 ppron3:sg:nom:f:ter:_:_ | ||
186 | +507 ppron3:sg:nom:m1.m2.m3:ter:_:_ | ||
187 | +508 ppron3:sg:nom:n1.n2:ter:_:_ | ||
188 | +462 ppron3:pl:acc:m1.p1:ter:_:npraep | ||
189 | +463 ppron3:pl:acc:m1.p1:ter:_:praep | ||
190 | +464 ppron3:pl:acc:m2.m3.f.n1.n2.p2.p3:ter:_:npraep | ||
191 | +465 ppron3:pl:acc:m2.m3.f.n1.n2.p2.p3:ter:_:praep | ||
192 | +466 ppron3:pl:dat:_:ter:_:npraep | ||
193 | +467 ppron3:pl:dat:_:ter:_:praep | ||
194 | +468 ppron3:pl:gen:_:ter:_:npraep | ||
195 | +469 ppron3:pl:gen:_:ter:_:praep | ||
196 | +470 ppron3:pl:inst:_:ter:_:_ | ||
197 | +471 ppron3:pl:loc:_:ter:_:_ | ||
198 | +472 ppron3:pl:nom:m1.p1:ter:_:_ | ||
199 | +473 ppron3:pl:nom:m2.m3.f.n1.n2.p2.p3:ter:_:_ | ||
200 | +# PRONOUN ‘SIEBIE’ | ||
201 | +594 siebie:acc | ||
202 | +595 siebie:dat | ||
203 | +596 siebie:gen | ||
204 | +597 siebie:inst | ||
205 | +598 siebie:loc | ||
206 | +# ADJECTIVES | ||
207 | +5 adj:pl:acc:m1.p1:com | ||
208 | +6 adj:pl:acc:m1.p1:pos | ||
209 | +7 adj:pl:acc:m1.p1:sup | ||
210 | +8 adj:pl:acc:m2.m3.f.n1.n2.p2.p3:com | ||
211 | +9 adj:pl:acc:m2.m3.f.n1.n2.p2.p3:pos | ||
212 | +10 adj:pl:acc:m2.m3.f.n1.n2.p2.p3:sup | ||
213 | +11 adj:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:com | ||
214 | +12 adj:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:pos | ||
215 | +13 adj:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:sup | ||
216 | +14 adj:pl:gen:m1.m2.m3.f.n1.n2.p1.p2.p3:com | ||
217 | +15 adj:pl:gen:m1.m2.m3.f.n1.n2.p1.p2.p3:pos | ||
218 | +16 adj:pl:gen:m1.m2.m3.f.n1.n2.p1.p2.p3:sup | ||
219 | +17 adj:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:com | ||
220 | +18 adj:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:pos | ||
221 | +19 adj:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:sup | ||
222 | +20 adj:pl:loc:m1.m2.m3.f.n1.n2.p1.p2.p3:com | ||
223 | +21 adj:pl:loc:m1.m2.m3.f.n1.n2.p1.p2.p3:pos | ||
224 | +22 adj:pl:loc:m1.m2.m3.f.n1.n2.p1.p2.p3:sup | ||
225 | +23 adj:pl:nom:m1.p1:pos | ||
226 | +24 adj:pl:nom:m2.m3.f.n1.n2.p2.p3:pos | ||
227 | +25 adj:pl:nom.voc:m1.p1:com | ||
228 | +26 adj:pl:nom.voc:m1.p1:pos | ||
229 | +27 adj:pl:nom.voc:m1.p1:sup | ||
230 | +28 adj:pl:nom.voc:m2.m3.f.n1.n2.p2.p3:com | ||
231 | +29 adj:pl:nom.voc:m2.m3.f.n1.n2.p2.p3:pos | ||
232 | +30 adj:pl:nom.voc:m2.m3.f.n1.n2.p2.p3:sup | ||
233 | +31 adj:sg:acc:f:com | ||
234 | +32 adj:sg:acc:f:pos | ||
235 | +33 adj:sg:acc:f:sup | ||
236 | +34 adj:sg:acc:m1.m2:com | ||
237 | +35 adj:sg:acc:m1.m2:pos | ||
238 | +36 adj:sg:acc:m1.m2:sup | ||
239 | +37 adj:sg:acc:m3:com | ||
240 | +38 adj:sg:acc:m3:pos | ||
241 | +39 adj:sg:acc:m3:sup | ||
242 | +40 adj:sg:acc:n1.n2:com | ||
243 | +41 adj:sg:acc:n1.n2:pos | ||
244 | +42 adj:sg:acc:n1.n2:sup | ||
245 | +43 adj:sg:dat:f:com | ||
246 | +44 adj:sg:dat:f:pos | ||
247 | +45 adj:sg:dat:f:sup | ||
248 | +46 adj:sg:dat:m1.m2.m3.n1.n2:com | ||
249 | +47 adj:sg:dat:m1.m2.m3.n1.n2:pos | ||
250 | +48 adj:sg:dat:m1.m2.m3.n1.n2:sup | ||
251 | +49 adj:sg:gen:f:com | ||
252 | +50 adj:sg:gen:f:pos | ||
253 | +51 adj:sg:gen:f:sup | ||
254 | +52 adj:sg:gen:m1.m2.m3.n1.n2:com | ||
255 | +53 adj:sg:gen:m1.m2.m3.n1.n2:pos | ||
256 | +54 adj:sg:gen:m1.m2.m3.n1.n2:sup | ||
257 | +55 adj:sg:inst:f:com | ||
258 | +56 adj:sg:inst:f:pos | ||
259 | +57 adj:sg:inst:f:sup | ||
260 | +58 adj:sg:inst:m1.m2.m3.n1.n2:com | ||
261 | +59 adj:sg:inst:m1.m2.m3.n1.n2:pos | ||
262 | +60 adj:sg:inst:m1.m2.m3.n1.n2:sup | ||
263 | +61 adj:sg:loc:f:com | ||
264 | +62 adj:sg:loc:f:pos | ||
265 | +63 adj:sg:loc:f:sup | ||
266 | +64 adj:sg:loc:m1.m2.m3.n1.n2:com | ||
267 | +65 adj:sg:loc:m1.m2.m3.n1.n2:pos | ||
268 | +66 adj:sg:loc:m1.m2.m3.n1.n2:sup | ||
269 | +67 adj:sg:nom:f:pos | ||
270 | +68 adj:sg:nom:m1.m2.m3:pos | ||
271 | +69 adj:sg:nom:n1.n2:pos | ||
272 | +70 adj:sg:nom.voc:f:com | ||
273 | +71 adj:sg:nom.voc:f:pos | ||
274 | +72 adj:sg:nom.voc:f:sup | ||
275 | +73 adj:sg:nom.voc:m1.m2.m3:com | ||
276 | +74 adj:sg:nom.voc:m1.m2.m3:pos | ||
277 | +75 adj:sg:nom.voc:m1.m2.m3:sup | ||
278 | +76 adj:sg:nom.voc:n1.n2:com | ||
279 | +77 adj:sg:nom.voc:n1.n2:pos | ||
280 | +78 adj:sg:nom.voc:n1.n2:sup | ||
281 | +# adjectival compounds forming form: | ||
282 | +2 adja | ||
283 | +# predicative adjective: | ||
284 | +3 adjc | ||
285 | +# post-prepositional adjective: | ||
286 | +4 adjp | ||
287 | +# VERBS | ||
288 | +# finitive (present/future) flexeme: | ||
289 | +153 fin:pl:pri:imperf | ||
290 | +154 fin:pl:pri:imperf.perf | ||
291 | +155 fin:pl:pri:perf | ||
292 | +156 fin:pl:sec:imperf | ||
293 | +157 fin:pl:sec:imperf.perf | ||
294 | +158 fin:pl:sec:perf | ||
295 | +159 fin:pl:ter:imperf | ||
296 | +160 fin:pl:ter:imperf.perf | ||
297 | +161 fin:pl:ter:perf | ||
298 | +162 fin:sg:pri:imperf | ||
299 | +163 fin:sg:pri:imperf.perf | ||
300 | +164 fin:sg:pri:perf | ||
301 | +165 fin:sg:sec:imperf | ||
302 | +166 fin:sg:sec:imperf.perf | ||
303 | +167 fin:sg:sec:perf | ||
304 | +168 fin:sg:ter:imperf | ||
305 | +169 fin:sg:ter:imperf.perf | ||
306 | +170 fin:sg:ter:perf | ||
307 | +# past flexeme: | ||
308 | +# praet=split (unsued otherwise): | ||
309 | +509 praet:pl:m1.p1:imperf | ||
310 | +510 praet:pl:m1.p1:imperf.perf | ||
311 | +511 praet:pl:m1.p1:perf | ||
312 | +521 praet:pl:m2.m3.f.n1.n2.p2.p3:imperf | ||
313 | +522 praet:pl:m2.m3.f.n1.n2.p2.p3:imperf.perf | ||
314 | +523 praet:pl:m2.m3.f.n1.n2.p2.p3:perf | ||
315 | +533 praet:sg:f:imperf | ||
316 | +534 praet:sg:f:imperf.perf | ||
317 | +535 praet:sg:f:perf | ||
318 | +545 praet:sg:m1.m2.m3:imperf | ||
319 | +546 praet:sg:m1.m2.m3:imperf:agl | ||
320 | +547 praet:sg:m1.m2.m3:imperf:nagl | ||
321 | +548 praet:sg:m1.m2.m3:imperf.perf | ||
322 | +549 praet:sg:m1.m2.m3:perf | ||
323 | +550 praet:sg:m1.m2.m3:perf:agl | ||
324 | +551 praet:sg:m1.m2.m3:perf:nagl | ||
325 | +561 praet:sg:n1.n2:imperf | ||
326 | +562 praet:sg:n1.n2:imperf.perf | ||
327 | +563 praet:sg:n1.n2:perf | ||
328 | +# praet=composite (unsued otherwise): | ||
329 | +512 praet:pl:m1.p1:pri:imperf | ||
330 | +513 praet:pl:m1.p1:pri:imperf.perf | ||
331 | +514 praet:pl:m1.p1:pri:perf | ||
332 | +515 praet:pl:m1.p1:sec:imperf | ||
333 | +516 praet:pl:m1.p1:sec:imperf.perf | ||
334 | +517 praet:pl:m1.p1:sec:perf | ||
335 | +518 praet:pl:m1.p1:ter:imperf | ||
336 | +519 praet:pl:m1.p1:ter:imperf.perf | ||
337 | +520 praet:pl:m1.p1:ter:perf | ||
338 | +524 praet:pl:m2.m3.f.n1.n2.p2.p3:pri:imperf | ||
339 | +525 praet:pl:m2.m3.f.n1.n2.p2.p3:pri:imperf.perf | ||
340 | +526 praet:pl:m2.m3.f.n1.n2.p2.p3:pri:perf | ||
341 | +527 praet:pl:m2.m3.f.n1.n2.p2.p3:sec:imperf | ||
342 | +528 praet:pl:m2.m3.f.n1.n2.p2.p3:sec:imperf.perf | ||
343 | +529 praet:pl:m2.m3.f.n1.n2.p2.p3:sec:perf | ||
344 | +530 praet:pl:m2.m3.f.n1.n2.p2.p3:ter:imperf | ||
345 | +531 praet:pl:m2.m3.f.n1.n2.p2.p3:ter:imperf.perf | ||
346 | +532 praet:pl:m2.m3.f.n1.n2.p2.p3:ter:perf | ||
347 | +536 praet:sg:f:pri:imperf | ||
348 | +537 praet:sg:f:pri:imperf.perf | ||
349 | +538 praet:sg:f:pri:perf | ||
350 | +539 praet:sg:f:sec:imperf | ||
351 | +540 praet:sg:f:sec:imperf.perf | ||
352 | +541 praet:sg:f:sec:perf | ||
353 | +542 praet:sg:f:ter:imperf | ||
354 | +543 praet:sg:f:ter:imperf.perf | ||
355 | +544 praet:sg:f:ter:perf | ||
356 | +552 praet:sg:m1.m2.m3:pri:imperf | ||
357 | +553 praet:sg:m1.m2.m3:pri:imperf.perf | ||
358 | +554 praet:sg:m1.m2.m3:pri:perf | ||
359 | +555 praet:sg:m1.m2.m3:sec:imperf | ||
360 | +556 praet:sg:m1.m2.m3:sec:imperf.perf | ||
361 | +557 praet:sg:m1.m2.m3:sec:perf | ||
362 | +558 praet:sg:m1.m2.m3:ter:imperf | ||
363 | +559 praet:sg:m1.m2.m3:ter:imperf.perf | ||
364 | +560 praet:sg:m1.m2.m3:ter:perf | ||
365 | +564 praet:sg:n1.n2:pri:imperf | ||
366 | +565 praet:sg:n1.n2:pri:imperf.perf | ||
367 | +566 praet:sg:n1.n2:pri:perf | ||
368 | +567 praet:sg:n1.n2:sec:imperf | ||
369 | +568 praet:sg:n1.n2:sec:imperf.perf | ||
370 | +569 praet:sg:n1.n2:sec:perf | ||
371 | +570 praet:sg:n1.n2:ter:imperf | ||
372 | +571 praet:sg:n1.n2:ter:imperf.perf | ||
373 | +572 praet:sg:n1.n2:ter:perf | ||
374 | +# conditional mood (used only with praet=composite) | ||
375 | +100 cond:pl:m1.p1:pri:imperf | ||
376 | +101 cond:pl:m1.p1:pri:imperf.perf | ||
377 | +102 cond:pl:m1.p1:pri:perf | ||
378 | +103 cond:pl:m1.p1:sec:imperf | ||
379 | +104 cond:pl:m1.p1:sec:imperf.perf | ||
380 | +105 cond:pl:m1.p1:sec:perf | ||
381 | +106 cond:pl:m1.p1:ter:imperf | ||
382 | +107 cond:pl:m1.p1:ter:imperf.perf | ||
383 | +108 cond:pl:m1.p1:ter:perf | ||
384 | +109 cond:pl:m2.m3.f.n1.n2.p2.p3:pri:imperf | ||
385 | +110 cond:pl:m2.m3.f.n1.n2.p2.p3:pri:imperf.perf | ||
386 | +111 cond:pl:m2.m3.f.n1.n2.p2.p3:pri:perf | ||
387 | +112 cond:pl:m2.m3.f.n1.n2.p2.p3:sec:imperf | ||
388 | +113 cond:pl:m2.m3.f.n1.n2.p2.p3:sec:imperf.perf | ||
389 | +114 cond:pl:m2.m3.f.n1.n2.p2.p3:sec:perf | ||
390 | +115 cond:pl:m2.m3.f.n1.n2.p2.p3:ter:imperf | ||
391 | +116 cond:pl:m2.m3.f.n1.n2.p2.p3:ter:imperf.perf | ||
392 | +117 cond:pl:m2.m3.f.n1.n2.p2.p3:ter:perf | ||
393 | +118 cond:sg:f:pri:imperf | ||
394 | +119 cond:sg:f:pri:imperf.perf | ||
395 | +120 cond:sg:f:pri:perf | ||
396 | +121 cond:sg:f:sec:imperf | ||
397 | +122 cond:sg:f:sec:imperf.perf | ||
398 | +123 cond:sg:f:sec:perf | ||
399 | +124 cond:sg:f:ter:imperf | ||
400 | +125 cond:sg:f:ter:imperf.perf | ||
401 | +126 cond:sg:f:ter:perf | ||
402 | +127 cond:sg:m1.m2.m3:pri:imperf | ||
403 | +128 cond:sg:m1.m2.m3:pri:imperf.perf | ||
404 | +129 cond:sg:m1.m2.m3:pri:perf | ||
405 | +130 cond:sg:m1.m2.m3:sec:imperf | ||
406 | +131 cond:sg:m1.m2.m3:sec:imperf.perf | ||
407 | +132 cond:sg:m1.m2.m3:sec:perf | ||
408 | +133 cond:sg:m1.m2.m3:ter:imperf | ||
409 | +134 cond:sg:m1.m2.m3:ter:imperf.perf | ||
410 | +135 cond:sg:m1.m2.m3:ter:perf | ||
411 | +136 cond:sg:n1.n2:imperf | ||
412 | +137 cond:sg:n1.n2:imperf.perf | ||
413 | +138 cond:sg:n1.n2:perf | ||
414 | +139 cond:sg:n1.n2:pri:imperf | ||
415 | +140 cond:sg:n1.n2:pri:imperf.perf | ||
416 | +141 cond:sg:n1.n2:pri:perf | ||
417 | +142 cond:sg:n1.n2:sec:imperf | ||
418 | +143 cond:sg:n1.n2:sec:imperf.perf | ||
419 | +144 cond:sg:n1.n2:sec:perf | ||
420 | +145 cond:sg:n1.n2:ter:imperf | ||
421 | +146 cond:sg:n1.n2:ter:imperf.perf | ||
422 | +147 cond:sg:n1.n2:ter:perf | ||
423 | +# impersonal flexeme: | ||
424 | +219 imps:imperf | ||
425 | +220 imps:imperf.perf | ||
426 | +221 imps:perf | ||
427 | +# imperative flexeme: | ||
428 | +222 impt:pl:pri:imperf | ||
429 | +223 impt:pl:pri:imperf.perf | ||
430 | +224 impt:pl:pri:perf | ||
431 | +225 impt:pl:sec:imperf | ||
432 | +226 impt:pl:sec:imperf.perf | ||
433 | +227 impt:pl:sec:perf | ||
434 | +228 impt:sg:sec:imperf | ||
435 | +229 impt:sg:sec:imperf.perf | ||
436 | +230 impt:sg:sec:perf | ||
437 | +# infinitival flexeme: | ||
438 | +231 inf:imperf | ||
439 | +232 inf:imperf.perf | ||
440 | +233 inf:perf | ||
441 | +# agglutinative forms of ‘być’: | ||
442 | +83 aglt:pl:pri:imperf:nwok | ||
443 | +84 aglt:pl:pri:imperf:wok | ||
444 | +85 aglt:pl:sec:imperf:nwok | ||
445 | +86 aglt:pl:sec:imperf:wok | ||
446 | +87 aglt:sg:pri:imperf:nwok | ||
447 | +88 aglt:sg:pri:imperf:wok | ||
448 | +89 aglt:sg:sec:imperf:nwok | ||
449 | +90 aglt:sg:sec:imperf:wok | ||
450 | +# future forms of ‘być’: | ||
451 | +91 bedzie:pl:pri:imperf | ||
452 | +92 bedzie:pl:sec:imperf | ||
453 | +93 bedzie:pl:ter:imperf | ||
454 | +94 bedzie:sg:pri:imperf | ||
455 | +95 bedzie:sg:sec:imperf | ||
456 | +96 bedzie:sg:ter:imperf | ||
457 | +# ‘winien’ type verbs: | ||
458 | +705 winien:pl:m1.p1:imperf | ||
459 | +706 winien:pl:m1.p1:pri:imperf | ||
460 | +707 winien:pl:m1.p1:sec:imperf | ||
461 | +708 winien:pl:m1.p1:ter:imperf | ||
462 | +709 winien:pl:m2.m3.f.n1.n2.p2.p3:imperf | ||
463 | +710 winien:pl:m2.m3.f.n1.n2.p2.p3:sec:imperf | ||
464 | +711 winien:pl:m2.m3.f.n1.n2.p2.p3:ter:imperf | ||
465 | +712 winien:sg:f:imperf | ||
466 | +713 winien:sg:f:pri:imperf | ||
467 | +714 winien:sg:f:sec:imperf | ||
468 | +715 winien:sg:f:ter:imperf | ||
469 | +716 winien:sg:m1.m2.m3:imperf | ||
470 | +717 winien:sg:m1.m2.m3:pri:imperf | ||
471 | +718 winien:sg:m1.m2.m3:sec:imperf | ||
472 | +719 winien:sg:m1.m2.m3:ter:imperf | ||
473 | +720 winien:sg:n1.n2:imperf | ||
474 | +721 winien:sg:n1.n2:pri:imperf | ||
475 | +722 winien:sg:n1.n2:sec:imperf | ||
476 | +723 winien:sg:n1.n2:ter:imperf | ||
477 | +# predicative flexeme: | ||
478 | +573 pred | ||
479 | +# gerunds | ||
480 | +171 ger:pl:dat.loc:n2:imperf:aff | ||
481 | +172 ger:pl:dat.loc:n2:imperf:neg | ||
482 | +173 ger:pl:dat.loc:n2:imperf.perf:aff | ||
483 | +174 ger:pl:dat.loc:n2:imperf.perf:neg | ||
484 | +175 ger:pl:dat.loc:n2:perf:aff | ||
485 | +176 ger:pl:dat.loc:n2:perf:neg | ||
486 | +177 ger:pl:gen:n2:imperf:aff | ||
487 | +178 ger:pl:gen:n2:imperf:neg | ||
488 | +179 ger:pl:gen:n2:imperf.perf:aff | ||
489 | +180 ger:pl:gen:n2:imperf.perf:neg | ||
490 | +181 ger:pl:gen:n2:perf:aff | ||
491 | +182 ger:pl:gen:n2:perf:neg | ||
492 | +183 ger:pl:inst:n2:imperf:aff | ||
493 | +184 ger:pl:inst:n2:imperf:neg | ||
494 | +185 ger:pl:inst:n2:imperf.perf:aff | ||
495 | +186 ger:pl:inst:n2:imperf.perf:neg | ||
496 | +187 ger:pl:inst:n2:perf:aff | ||
497 | +188 ger:pl:inst:n2:perf:neg | ||
498 | +189 ger:pl:nom.acc:n2:imperf:aff | ||
499 | +190 ger:pl:nom.acc:n2:imperf:neg | ||
500 | +191 ger:pl:nom.acc:n2:imperf.perf:aff | ||
501 | +192 ger:pl:nom.acc:n2:imperf.perf:neg | ||
502 | +193 ger:pl:nom.acc:n2:perf:aff | ||
503 | +194 ger:pl:nom.acc:n2:perf:neg | ||
504 | +195 ger:sg:dat.loc:n2:imperf:aff | ||
505 | +196 ger:sg:dat.loc:n2:imperf:neg | ||
506 | +197 ger:sg:dat.loc:n2:imperf.perf:aff | ||
507 | +198 ger:sg:dat.loc:n2:imperf.perf:neg | ||
508 | +199 ger:sg:dat.loc:n2:perf:aff | ||
509 | +200 ger:sg:dat.loc:n2:perf:neg | ||
510 | +201 ger:sg:gen:n2:imperf:aff | ||
511 | +202 ger:sg:gen:n2:imperf:neg | ||
512 | +203 ger:sg:gen:n2:imperf.perf:aff | ||
513 | +204 ger:sg:gen:n2:imperf.perf:neg | ||
514 | +205 ger:sg:gen:n2:perf:aff | ||
515 | +206 ger:sg:gen:n2:perf:neg | ||
516 | +207 ger:sg:inst:n2:imperf:aff | ||
517 | +208 ger:sg:inst:n2:imperf:neg | ||
518 | +209 ger:sg:inst:n2:imperf.perf:aff | ||
519 | +210 ger:sg:inst:n2:imperf.perf:neg | ||
520 | +211 ger:sg:inst:n2:perf:aff | ||
521 | +212 ger:sg:inst:n2:perf:neg | ||
522 | +213 ger:sg:nom.acc:n2:imperf:aff | ||
523 | +214 ger:sg:nom.acc:n2:imperf:neg | ||
524 | +215 ger:sg:nom.acc:n2:imperf.perf:aff | ||
525 | +216 ger:sg:nom.acc:n2:imperf.perf:neg | ||
526 | +217 ger:sg:nom.acc:n2:perf:aff | ||
527 | +218 ger:sg:nom.acc:n2:perf:neg | ||
528 | +# participles | ||
529 | +# adverbial participles: | ||
530 | +332 pcon:imperf | ||
531 | +331 pant:perf | ||
532 | +# adjectival active participle: | ||
533 | +267 pact:pl:acc:m1.p1:imperf:aff | ||
534 | +268 pact:pl:acc:m1.p1:imperf:neg | ||
535 | +269 pact:pl:acc:m1.p1:imperf.perf:aff | ||
536 | +270 pact:pl:acc:m1.p1:imperf.perf:neg | ||
537 | +271 pact:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:aff | ||
538 | +272 pact:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:neg | ||
539 | +273 pact:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:aff | ||
540 | +274 pact:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:neg | ||
541 | +275 pact:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:aff | ||
542 | +276 pact:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:neg | ||
543 | +277 pact:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:aff | ||
544 | +278 pact:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:neg | ||
545 | +279 pact:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:aff | ||
546 | +280 pact:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:neg | ||
547 | +281 pact:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:aff | ||
548 | +282 pact:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:neg | ||
549 | +283 pact:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf:aff | ||
550 | +284 pact:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf:neg | ||
551 | +285 pact:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf.perf:aff | ||
552 | +286 pact:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf.perf:neg | ||
553 | +287 pact:pl:nom.voc:m1.p1:imperf:aff | ||
554 | +288 pact:pl:nom.voc:m1.p1:imperf:neg | ||
555 | +289 pact:pl:nom.voc:m1.p1:imperf.perf:aff | ||
556 | +290 pact:pl:nom.voc:m1.p1:imperf.perf:neg | ||
557 | +291 pact:sg:acc.inst:f:imperf:aff | ||
558 | +292 pact:sg:acc.inst:f:imperf:neg | ||
559 | +293 pact:sg:acc.inst:f:imperf.perf:aff | ||
560 | +294 pact:sg:acc.inst:f:imperf.perf:neg | ||
561 | +295 pact:sg:acc:m1.m2:imperf:aff | ||
562 | +296 pact:sg:acc:m1.m2:imperf:neg | ||
563 | +297 pact:sg:acc:m1.m2:imperf.perf:aff | ||
564 | +298 pact:sg:acc:m1.m2:imperf.perf:neg | ||
565 | +299 pact:sg:acc:m3:imperf:aff | ||
566 | +300 pact:sg:acc:m3:imperf:neg | ||
567 | +301 pact:sg:acc:m3:imperf.perf:aff | ||
568 | +302 pact:sg:acc:m3:imperf.perf:neg | ||
569 | +303 pact:sg:dat:m1.m2.m3.n1.n2:imperf:aff | ||
570 | +304 pact:sg:dat:m1.m2.m3.n1.n2:imperf:neg | ||
571 | +305 pact:sg:dat:m1.m2.m3.n1.n2:imperf.perf:aff | ||
572 | +306 pact:sg:dat:m1.m2.m3.n1.n2:imperf.perf:neg | ||
573 | +307 pact:sg:gen.dat.loc:f:imperf:aff | ||
574 | +308 pact:sg:gen.dat.loc:f:imperf:neg | ||
575 | +309 pact:sg:gen.dat.loc:f:imperf.perf:aff | ||
576 | +310 pact:sg:gen.dat.loc:f:imperf.perf:neg | ||
577 | +311 pact:sg:gen:m1.m2.m3.n1.n2:imperf:aff | ||
578 | +312 pact:sg:gen:m1.m2.m3.n1.n2:imperf:neg | ||
579 | +313 pact:sg:gen:m1.m2.m3.n1.n2:imperf.perf:aff | ||
580 | +314 pact:sg:gen:m1.m2.m3.n1.n2:imperf.perf:neg | ||
581 | +315 pact:sg:inst.loc:m1.m2.m3.n1.n2:imperf:aff | ||
582 | +316 pact:sg:inst.loc:m1.m2.m3.n1.n2:imperf:neg | ||
583 | +317 pact:sg:inst.loc:m1.m2.m3.n1.n2:imperf.perf:aff | ||
584 | +318 pact:sg:inst.loc:m1.m2.m3.n1.n2:imperf.perf:neg | ||
585 | +319 pact:sg:nom.acc.voc:n1.n2:imperf:aff | ||
586 | +320 pact:sg:nom.acc.voc:n1.n2:imperf:neg | ||
587 | +321 pact:sg:nom.acc.voc:n1.n2:imperf.perf:aff | ||
588 | +322 pact:sg:nom.acc.voc:n1.n2:imperf.perf:neg | ||
589 | +323 pact:sg:nom.voc:f:imperf:aff | ||
590 | +324 pact:sg:nom.voc:f:imperf:neg | ||
591 | +325 pact:sg:nom.voc:f:imperf.perf:aff | ||
592 | +326 pact:sg:nom.voc:f:imperf.perf:neg | ||
593 | +327 pact:sg:nom.voc:m1.m2.m3:imperf:aff | ||
594 | +328 pact:sg:nom.voc:m1.m2.m3:imperf:neg | ||
595 | +329 pact:sg:nom.voc:m1.m2.m3:imperf.perf:aff | ||
596 | +330 pact:sg:nom.voc:m1.m2.m3:imperf.perf:neg | ||
597 | +# adjectival passive participle: | ||
598 | +333 ppas:pl:acc:m1.p1:imperf:aff | ||
599 | +334 ppas:pl:acc:m1.p1:imperf:neg | ||
600 | +335 ppas:pl:acc:m1.p1:imperf.perf:aff | ||
601 | +336 ppas:pl:acc:m1.p1:imperf.perf:neg | ||
602 | +337 ppas:pl:acc:m1.p1:perf:aff | ||
603 | +338 ppas:pl:acc:m1.p1:perf:neg | ||
604 | +339 ppas:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:aff | ||
605 | +340 ppas:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:neg | ||
606 | +341 ppas:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:aff | ||
607 | +342 ppas:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:neg | ||
608 | +343 ppas:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:perf:aff | ||
609 | +344 ppas:pl:dat:m1.m2.m3.f.n1.n2.p1.p2.p3:perf:neg | ||
610 | +345 ppas:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:aff | ||
611 | +346 ppas:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:neg | ||
612 | +347 ppas:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:aff | ||
613 | +348 ppas:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:neg | ||
614 | +349 ppas:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:perf:aff | ||
615 | +350 ppas:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2.p3:perf:neg | ||
616 | +351 ppas:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:aff | ||
617 | +352 ppas:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf:neg | ||
618 | +353 ppas:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:aff | ||
619 | +354 ppas:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:imperf.perf:neg | ||
620 | +355 ppas:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:perf:aff | ||
621 | +356 ppas:pl:inst:m1.m2.m3.f.n1.n2.p1.p2.p3:perf:neg | ||
622 | +357 ppas:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf:aff | ||
623 | +358 ppas:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf:neg | ||
624 | +359 ppas:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf.perf:aff | ||
625 | +360 ppas:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:imperf.perf:neg | ||
626 | +361 ppas:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:perf:aff | ||
627 | +362 ppas:pl:nom.acc.voc:m2.m3.f.n1.n2.p2.p3:perf:neg | ||
628 | +363 ppas:pl:nom.voc:m1.p1:imperf:aff | ||
629 | +364 ppas:pl:nom.voc:m1.p1:imperf:neg | ||
630 | +365 ppas:pl:nom.voc:m1.p1:imperf.perf:aff | ||
631 | +366 ppas:pl:nom.voc:m1.p1:imperf.perf:neg | ||
632 | +367 ppas:pl:nom.voc:m1.p1:perf:aff | ||
633 | +368 ppas:pl:nom.voc:m1.p1:perf:neg | ||
634 | +369 ppas:sg:acc.inst:f:imperf:aff | ||
635 | +370 ppas:sg:acc.inst:f:imperf:neg | ||
636 | +371 ppas:sg:acc.inst:f:imperf.perf:aff | ||
637 | +372 ppas:sg:acc.inst:f:imperf.perf:neg | ||
638 | +373 ppas:sg:acc.inst:f:perf:aff | ||
639 | +374 ppas:sg:acc.inst:f:perf:neg | ||
640 | +375 ppas:sg:acc:m1.m2:imperf:aff | ||
641 | +376 ppas:sg:acc:m1.m2:imperf:neg | ||
642 | +377 ppas:sg:acc:m1.m2:imperf.perf:aff | ||
643 | +378 ppas:sg:acc:m1.m2:imperf.perf:neg | ||
644 | +379 ppas:sg:acc:m1.m2:perf:aff | ||
645 | +380 ppas:sg:acc:m1.m2:perf:neg | ||
646 | +381 ppas:sg:acc:m3:imperf:aff | ||
647 | +382 ppas:sg:acc:m3:imperf:neg | ||
648 | +383 ppas:sg:acc:m3:imperf.perf:aff | ||
649 | +384 ppas:sg:acc:m3:imperf.perf:neg | ||
650 | +385 ppas:sg:acc:m3:perf:aff | ||
651 | +386 ppas:sg:acc:m3:perf:neg | ||
652 | +387 ppas:sg:dat:m1.m2.m3.n1.n2:imperf:aff | ||
653 | +388 ppas:sg:dat:m1.m2.m3.n1.n2:imperf:neg | ||
654 | +389 ppas:sg:dat:m1.m2.m3.n1.n2:imperf.perf:aff | ||
655 | +390 ppas:sg:dat:m1.m2.m3.n1.n2:imperf.perf:neg | ||
656 | +391 ppas:sg:dat:m1.m2.m3.n1.n2:perf:aff | ||
657 | +392 ppas:sg:dat:m1.m2.m3.n1.n2:perf:neg | ||
658 | +393 ppas:sg:gen.dat.loc:f:imperf:aff | ||
659 | +394 ppas:sg:gen.dat.loc:f:imperf:neg | ||
660 | +395 ppas:sg:gen.dat.loc:f:imperf.perf:aff | ||
661 | +396 ppas:sg:gen.dat.loc:f:imperf.perf:neg | ||
662 | +397 ppas:sg:gen.dat.loc:f:perf:aff | ||
663 | +398 ppas:sg:gen.dat.loc:f:perf:neg | ||
664 | +399 ppas:sg:gen:m1.m2.m3.n1.n2:imperf:aff | ||
665 | +400 ppas:sg:gen:m1.m2.m3.n1.n2:imperf:neg | ||
666 | +401 ppas:sg:gen:m1.m2.m3.n1.n2:imperf.perf:aff | ||
667 | +402 ppas:sg:gen:m1.m2.m3.n1.n2:imperf.perf:neg | ||
668 | +403 ppas:sg:gen:m1.m2.m3.n1.n2:perf:aff | ||
669 | +404 ppas:sg:gen:m1.m2.m3.n1.n2:perf:neg | ||
670 | +405 ppas:sg:inst.loc:m1.m2.m3.n1.n2:imperf:aff | ||
671 | +406 ppas:sg:inst.loc:m1.m2.m3.n1.n2:imperf:neg | ||
672 | +407 ppas:sg:inst.loc:m1.m2.m3.n1.n2:imperf.perf:aff | ||
673 | +408 ppas:sg:inst.loc:m1.m2.m3.n1.n2:imperf.perf:neg | ||
674 | +409 ppas:sg:inst.loc:m1.m2.m3.n1.n2:perf:aff | ||
675 | +410 ppas:sg:inst.loc:m1.m2.m3.n1.n2:perf:neg | ||
676 | +411 ppas:sg:nom.acc.voc:n1.n2:imperf:aff | ||
677 | +412 ppas:sg:nom.acc.voc:n1.n2:imperf:neg | ||
678 | +413 ppas:sg:nom.acc.voc:n1.n2:imperf.perf:aff | ||
679 | +414 ppas:sg:nom.acc.voc:n1.n2:imperf.perf:neg | ||
680 | +415 ppas:sg:nom.acc.voc:n1.n2:perf:aff | ||
681 | +416 ppas:sg:nom.acc.voc:n1.n2:perf:neg | ||
682 | +417 ppas:sg:nom.voc:f:imperf:aff | ||
683 | +418 ppas:sg:nom.voc:f:imperf:neg | ||
684 | +419 ppas:sg:nom.voc:f:imperf.perf:aff | ||
685 | +420 ppas:sg:nom.voc:f:imperf.perf:neg | ||
686 | +421 ppas:sg:nom.voc:f:perf:aff | ||
687 | +422 ppas:sg:nom.voc:f:perf:neg | ||
688 | +423 ppas:sg:nom.voc:m1.m2.m3:imperf:aff | ||
689 | +424 ppas:sg:nom.voc:m1.m2.m3:imperf:neg | ||
690 | +425 ppas:sg:nom.voc:m1.m2.m3:imperf.perf:aff | ||
691 | +426 ppas:sg:nom.voc:m1.m2.m3:imperf.perf:neg | ||
692 | +427 ppas:sg:nom.voc:m1.m2.m3:perf:aff | ||
693 | +428 ppas:sg:nom.voc:m1.m2.m3:perf:neg | ||
694 | +# NUMERALS | ||
695 | +239 num:pl:acc:m1:rec | ||
696 | +240 num:pl:dat.loc:n1.p1.p2:congr.rec | ||
697 | +241 num:pl:dat:m1.m2.m3.n2.f:congr | ||
698 | +242 num:pl:gen.dat.inst.loc:m1.m2.m3.f.n1.n2.p1.p2:congr | ||
699 | +243 num:pl:gen.dat.inst.loc:m1.m2.m3.f.n2:congr | ||
700 | +244 num:pl:gen.dat.loc:m1.m2.m3.n2.f:congr | ||
701 | +245 num:pl:gen.loc:m1.m2.m3.f.n1.n2.p1.p2:congr | ||
702 | +246 num:pl:gen.loc:m1.m2.m3.n2.f:congr | ||
703 | +247 num:pl:gen:n1.p1.p2:rec | ||
704 | +248 num:pl:inst:f:congr | ||
705 | +249 num:pl:inst:m1.m2.m3.f.n1.n2.p1.p2:congr | ||
706 | +250 num:pl:inst:m1.m2.m3.f.n2:congr | ||
707 | +251 num:pl:inst:m1.m2.m3.n2:congr | ||
708 | +252 num:pl:inst:m1.m2.m3.n2.f:congr | ||
709 | +253 num:pl:inst:n1.p1.p2:rec | ||
710 | +254 num:pl:nom.acc:m1.m2.m3.f.n1.n2.p1.p2:rec | ||
711 | +255 num:pl:nom.acc.voc:f:congr | ||
712 | +256 num:pl:nom.acc.voc:m1:rec | ||
713 | +257 num:pl:nom.acc.voc:m2.m3.f.n1.n2.p1.p2:rec | ||
714 | +258 num:pl:nom.acc.voc:m2.m3.f.n2:rec | ||
715 | +259 num:pl:nom.acc.voc:m2.m3.n2:congr | ||
716 | +260 num:pl:nom.acc.voc:m2.m3.n2.f:congr | ||
717 | +261 num:pl:nom.acc.voc:n1.p1.p2:rec | ||
718 | +262 num:pl:nom.gen.dat.inst.acc.loc.voc:m1.m2.m3.f.n1.n2.p1.p2:rec | ||
719 | +263 num:pl:nom.voc:m1:congr | ||
720 | +264 num:pl:nom.voc:m1:rec | ||
721 | +265 num:sg:nom.gen.dat.inst.acc.loc.voc:f:rec | ||
722 | +266 num:sg:nom.gen.dat.inst.acc.loc.voc:m1.m2.m3.n1.n2:rec | ||
723 | +# numeral compounds forming form: | ||
724 | +238 num:comp | ||
725 | +# PREPOSITIONS | ||
726 | +578 prep:acc | ||
727 | +579 prep:acc:nwok | ||
728 | +580 prep:acc:wok | ||
729 | +581 prep:dat | ||
730 | +582 prep:gen | ||
731 | +583 prep:gen:nwok | ||
732 | +584 prep:gen:wok | ||
733 | +585 prep:inst | ||
734 | +586 prep:inst:nwok | ||
735 | +587 prep:inst:wok | ||
736 | +588 prep:loc | ||
737 | +589 prep:loc:nwok | ||
738 | +590 prep:loc:wok | ||
739 | +591 prep:nom | ||
740 | +# ADVERBS | ||
741 | +79 adv | ||
742 | +80 adv:com | ||
743 | +81 adv:pos | ||
744 | +82 adv:sup | ||
745 | +# OTHER | ||
746 | +# kubliki (particles): | ||
747 | +592 qub | ||
748 | +# conjunctions: | ||
749 | +148 conj | ||
750 | +# complementizers: | ||
751 | +99 comp | ||
752 | +# interjections: | ||
753 | +234 interj | ||
754 | +# burkinostki (bound words): | ||
755 | +98 burk | ||
756 | +# abbreviations: | ||
757 | +97 brev:pun | ||
758 | +97 brev:npun | ||
759 | +# punctuation: | ||
760 | +235 interp | ||
761 | +# digits: | ||
762 | +151 dig | ||
763 | +# Roman digits: | ||
764 | +593 romandig | ||
765 | +# emoticons: | ||
766 | +152 emoticon | ||
767 | +# prefixes: | ||
768 | +574 prefa | ||
769 | +575 prefppas | ||
770 | +576 prefs | ||
771 | +577 prefv | ||
772 | +# (special) | ||
773 | +236 naj | ||
774 | +237 nie |