diff --git a/plwn/readers/nodes.py b/plwn/readers/nodes.py
index b1bf8d3cafafd96ae67ee3c0012489bd71d6fb7e..9f58999afc307145030394a1894feafd2ef1a914 100644
--- a/plwn/readers/nodes.py
+++ b/plwn/readers/nodes.py
@@ -31,19 +31,19 @@ __all__ = (
SynsetNode = namedtuple(
"SynsetNode",
- ["id", "definition", "related", "is_artificial"],
+ ["id", "legacy_id", "definition", "related", "is_artificial"],
)
LexicalUnitNode = namedtuple(
"LexicalUnitNode",
- ["id", "lemma", "pos", "variant", "synset", "unit_index", "definition",
- "usage_notes", "external_links", "examples", "examples_sources",
- "domain", "related", "verb_aspect", "is_emotional", "emotion_markedness",
- "emotion_names", "emotion_valuations", "emotion_example_1",
- "emotion_example_2"]
+ ["id", "legacy_id", "lemma", "pos", "variant", "synset", "unit_index",
+ "definition", "usage_notes", "external_links", "examples",
+ "examples_sources", "domain", "related", "verb_aspect", "is_emotional",
+ "emotion_markedness", "emotion_names", "emotion_valuations",
+ "emotion_example_1", "emotion_example_2"]
)
RelationTypeNode = namedtuple(
"RelationTypeNode",
- ["kind", "name", "parent", "aliases"],
+ ["kind", "legacy_id", "name", "parent", "aliases"],
)
@@ -58,6 +58,7 @@ def make_synset_node(**props):
"""
syn = SynsetNode(
id=props.pop('id'),
+ legacy_id=props.pop('legacy_id', None),
definition=props.pop('definition', None),
related=props.pop('related', ()),
is_artificial=props.pop('is_artificial', False),
@@ -80,6 +81,7 @@ def make_lexical_unit_node(**props):
"""
lex = LexicalUnitNode(
id=props.pop('id'),
+ legacy_id=props.pop('legacy_id', None),
lemma=props.pop('lemma'),
pos=props.pop('pos'),
variant=props.pop('variant'),
@@ -120,6 +122,7 @@ def make_relation_type_node(**props):
"""
rel = RelationTypeNode(
name=props.pop('name'),
+ legacy_id=props.pop('legacy_id'),
kind=props.pop('kind'),
parent=props.pop('parent', None),
aliases=props.pop('aliases', frozenset()),
diff --git a/plwn/readers/ubylmf.py b/plwn/readers/ubylmf.py
index a3859ef5bc91737206c05d4c442103d09ad10a00..f91d2d3210c0d9a81565d8c2cd86432b2d434bdd 100644
--- a/plwn/readers/ubylmf.py
+++ b/plwn/readers/ubylmf.py
@@ -136,6 +136,7 @@ def _make_lexicalunit(xml_lexicalentry, xml_sense):
)
return make_lexical_unit_node(
id=lu_id,
+ legacy_id=None,
lemma=lu_lemma,
pos=PoS(lu_pos),
synset=lu_synset,
diff --git a/plwn/storages/sqlite.py b/plwn/storages/sqlite.py
index 6f48144e9079de4eef93216b220df6b4bcd31854..904e5400ce632a08d27da49e89f9ebb99713ed27 100644
--- a/plwn/storages/sqlite.py
+++ b/plwn/storages/sqlite.py
@@ -108,7 +108,7 @@ CREATE UNIQUE INDEX value ON tbl_domain (value);
-- Synset only gets one simple table
CREATE TABLE tbl_synset (
- id BLOB NOT NULL ,
+ id BLOB ,
legacy_id INTEGER NULL ,
definition TEXT NULL ,
isartificial INTEGER NOT NULL DEFAULT '0' ,
@@ -119,7 +119,7 @@ CREATE TABLE tbl_synset (
-- properties. They also need indexes for lookup.
CREATE TABLE tbl_lexicalunit (
- id BLOB NOT NULL ,
+ id BLOB ,
legacy_id INTEGER NULL ,
lemma VARCHAR(255) NOT NULL ,
pos INTEGER NOT NULL ,
@@ -204,14 +204,14 @@ CREATE UNIQUE INDEX name ON tbl_lexicalrelationchildpart (name);
-- Parent can't be NULL - the no-parent case will be handled by a special empty
-- string parent. This is so that UNIQUE works correctly.
CREATE TABLE tbl_synsetrelationtype (
- id BLOB NOT NULL ,
+ id BLOB ,
legacy_id INTEGER NULL ,
parentpart INTEGER NOT NULL ,
childpart INTEGER NOT NULL ,
PRIMARY KEY (id)
);
CREATE TABLE tbl_lexicalrelationtype (
- id BLOB NOT NULL ,
+ id BLOB ,
legacy_id INTEGER NULL ,
parentpart INTEGER NOT NULL ,
childpart INTEGER NOT NULL ,
@@ -1265,6 +1265,19 @@ class RelationInfo(bs.RelationInfoBase):
self._par = _UNFETCHED
self._name = _UNFETCHED
self._aliases = _UNFETCHED
+ self._leg_id = _UNFETCHED
+
+ @property
+ def legacy_id(self):
+ if self._leg_id is _UNFETCHED:
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"SELECT legacy_id FROM {} WHERE id = ?"
+ .format(_RELTYPE_TABLES[self._kind]),
+ (self._id,),
+ )
+ self._leg_id = cur.fetchone()[0]
+ return self._leg_id
@property
def kind(self):
@@ -1387,10 +1400,11 @@ class _DBBuilder(object):
def _insert_synset(self, syn_node):
self._db.execute(
u"""
- INSERT INTO tbl_synset (id, definition, isartificial)
- VALUES (?, ?, ?)
+ INSERT INTO tbl_synset (id, legacy_id, definition, isartificial)
+ VALUES (?, ?, ?, ?)
""",
- (syn_node.id, syn_node.definition, syn_node.is_artificial),
+ (syn_node.id, syn_node.legacy_id, syn_node.definition,
+ syn_node.is_artificial),
).close()
# Related go into temp storage
self._adhoc_synrels[syn_node.id] = syn_node.related
@@ -1416,9 +1430,11 @@ class _DBBuilder(object):
childname_id = self._ensure_rel_part_name(child_tbl, rel_node.name)
# And now the relation itself
cur.execute(
- u"INSERT INTO {} (parentpart, childpart) VALUES (?, ?)"
- .format(type_tbl),
- (parname_id, childname_id),
+ u"""
+ INSERT INTO {} (legacy_id, parentpart, childpart)
+ VALUES (?, ?, ?)
+ """.format(type_tbl),
+ (rel_node.legacy_id, parname_id, childname_id),
)
# Do aliases if present
if rel_node.aliases:
@@ -1439,14 +1455,14 @@ class _DBBuilder(object):
cur.execute(
u"""
INSERT INTO tbl_lexicalunit (
- id, lemma, pos, variant,
+ id, legacy_id, lemma, pos, variant,
synset, unitindex,
definition, domain, verbaspect,
isemotional, emotionmark,
emotionexample1, emotionexample2
)
VALUES (
- :id, :lemma, :pos, :var,
+ :id, :legacy_id, :lemma, :pos, :var,
:syn, :uidx,
:def, :dom, :va,
:emo_is, :emo_m,
@@ -1455,6 +1471,7 @@ class _DBBuilder(object):
""",
{
u'id': lu_node.id,
+ u'legacy_id': lu_node.legacy_id,
u'lemma': lu_node.lemma,
u'pos': self._posids[lu_node.pos.value],
u'var': lu_node.variant,
@@ -1607,7 +1624,10 @@ class _DBBuilder(object):
return row[0]
cur.execute(
- u"INSERT INTO {} (parentpart, childpart) VALUES (?, ?)"
+ u"""
+ INSERT INTO {} (parentpart, childpart)
+ VALUES (?, ?)
+ """
.format(type_tbl),
(empty_parent_id, child_id),
)
diff --git a/tests/abstract_cases/test_unit_and_synset.py b/tests/abstract_cases/test_unit_and_synset.py
index 626e50f2161d79f5026a2152a3becc2392617fd9..f3885dea2aae8871ef7ba7b48fe087ee43b98a05 100644
--- a/tests/abstract_cases/test_unit_and_synset.py
+++ b/tests/abstract_cases/test_unit_and_synset.py
@@ -29,9 +29,10 @@ class SynsetPropertiesTest(ut.TestCase):
def setUp(self):
self.__plwn = self._PLWNClass.from_reader((
- nd.make_synset_node(id=1, definition=u'foobar'),
+ nd.make_synset_node(id=1, legacy_id=1, definition=u'foobar'),
nd.make_lexical_unit_node(
id=11,
+ legacy_id=11,
lemma=u'aaa',
pos=en.PoS.n,
variant=1,
@@ -858,10 +859,10 @@ class ToDictTest(ut.TestCase):
verb_aspect=en.VerbAspect.pred,
emotion_markedness=en.EmotionMarkedness.strong_negative,
emotion_names=(en.EmotionName.surprise,),
- emotion_valuations=(
- en.EmotionValuation.ugliness,
+ emotion_valuations=[
en.EmotionValuation.error,
- ),
+ en.EmotionValuation.ugliness,
+ ],
emotion_example_1=u'Bad thing.',
),
nd.make_lexical_unit_node(
@@ -886,7 +887,7 @@ class ToDictTest(ut.TestCase):
self.__lex11_dict = {
u'id': 11,
u'lemma': u'aaa',
- u'pos': u'verb',
+ u'pos': u'czasownik',
u'variant': 1,
u'synset': 1,
u'definition': u'bar',
diff --git a/tests/cases/test_ubylmf_reader.py b/tests/cases/test_ubylmf_reader.py
index a8d7aba9944d7b84dab9e87c67291e7d6211ca1b..735c91c832890214572ba8aa2cd8ceadbdd2e7ef 100644
--- a/tests/cases/test_ubylmf_reader.py
+++ b/tests/cases/test_ubylmf_reader.py
@@ -28,7 +28,7 @@ test_xml = u"""<?xml version="1.0" encoding="UTF-8" ?>
<LexicalResource dtdVersion="ubyDTD_1_0.dtd" name="plWordnet">
<Lexicon languageIdentifier="pl" id="1" name="Słowosieć 2.2">
-<LexicalEntry id="15" partOfSpeech="noun">
+<LexicalEntry id="15" partOfSpeech="rzeczownik">
<Lemma>
<FormRepresentation writtenForm="'patafizyka"/>
</Lemma>
@@ -173,7 +173,7 @@ class UBYLMFReaderTest(unittest.TestCase):
# Missing <Lemma>
xml_lu = et.fromstring(
u"""
- <LexicalEntry id="15" partOfSpeech="noun">
+ <LexicalEntry id="15" partOfSpeech="rzeczownik">
</LexicalEntry>
""".encode(ENCODING)
)
@@ -192,7 +192,7 @@ class UBYLMFReaderTest(unittest.TestCase):
# Empty <Lemma>
xml_lu = et.fromstring(
u"""
- <LexicalEntry id="15" partOfSpeech="noun">
+ <LexicalEntry id="15" partOfSpeech="rzeczownik">
<Lemma>
<FormRepresentation writtenForm=""/>
</Lemma>
@@ -227,7 +227,7 @@ class UBYLMFReaderTest(unittest.TestCase):
# Incorrect unit index
xml_lu = et.fromstring(
u"""
- <LexicalEntry id="15" partOfSpeech="noun">
+ <LexicalEntry id="15" partOfSpeech="rzeczownik">
<Lemma>
<FormRepresentation writtenForm="'patafizyka"/>
</Lemma>