diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index c85983e56ca1a3aafdea2352683c531cfe42c1f6..fe40822f862d95b0dffdfa4329f7aca7540b3efb 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -10,6 +10,7 @@ cache:
stages:
- check_style
- tests
+ - test_load
- push_wheel
pep8:
@@ -45,3 +46,10 @@ push_wheel:
--repository-url https://pypi.clarin-pl.eu/
-u $PIPY_USER -p $PIPY_PASS dist/plwn_api*.whl
+test_load:
+ stage: test_load
+ before_script:
+ - pip install --upgrade pip
+ - pip install .
+ script:
+ - python -c 'import plwn; print(plwn.load_default().lexical_unit("pies", plwn.PoS.noun_pl, 2).definition)'
diff --git a/plwn/bases.py b/plwn/bases.py
index 43f9712ef12cdbf7d08832f084686d047e078924..09e84e3da0d283c463ebae82412e0a812305f245 100644
--- a/plwn/bases.py
+++ b/plwn/bases.py
@@ -588,7 +588,7 @@ class SynsetBase(object):
:meth:`LexicalUnitBase.to_dict`.
"""
syn_dict = {
- u'id': self.id,
+ u'id': self.uuid,
u'definition': self.definition,
u'is_artificial': self.is_artificial,
u'units': tuple(
@@ -602,7 +602,7 @@ class SynsetBase(object):
if include_related:
syn_dict[u'related'] = {
six.text_type(rel): tuple(
- (target.id, target.short_str())
+ (target.uuid, target.short_str())
for target in self.related(rel)
)
for rel in self.relations
@@ -631,7 +631,7 @@ class SynsetBase(object):
def __repr__(self):
head = self.lexical_units[0]
rstr = '<Synset id={!r} lemma={!r} pos={!r} variant={!r}'.format(
- self.id,
+ str(self.uuid),
head.lemma,
head.pos,
head.variant,
@@ -938,7 +938,8 @@ class LexicalUnitBase(object):
}
"""
lu_dict = {
- u'id': self.id,
+ u'id': self.uuid,
+ u'legacy_id': self.legacy_id,
u'lemma': self.lemma,
u'pos': self.pos.value,
u'variant': self.variant,
@@ -948,7 +949,7 @@ class LexicalUnitBase(object):
u'external_links': tuple(self.external_links),
u'usage_notes': tuple(self.usage_notes),
u'domain': self.domain.value,
- u'synset': self.synset.id,
+ u'synset': self.synset.uuid,
u'verb_aspect': None
if self.verb_aspect is None
else self.verb_aspect.value,
@@ -965,7 +966,7 @@ class LexicalUnitBase(object):
if include_related:
lu_dict[u'related'] = {
six.text_type(rel): tuple(
- (target.id, six.text_type(target))
+ (target.uuid, six.text_type(target))
for target in self.related(rel)
)
for rel in self.relations
@@ -1001,7 +1002,7 @@ class LexicalUnitBase(object):
def __repr__(self):
return '<LexicalUnit id={!r} lemma={!r} pos={!r} variant={!r}>'.format(
- self.id,
+ str(self.uuid),
self.lemma,
self.pos,
self.variant,
diff --git a/plwn/config.ini b/plwn/config.ini
index ec3c6ffbc546b7cc4832deaca53288cb9a1a5482..23a2ffb4d237a8c0ff9c2b93b7d2eead0c844ebf 100644
--- a/plwn/config.ini
+++ b/plwn/config.ini
@@ -1,2 +1,2 @@
[DOWNLOAD]
-default_model = https://minio.clarin-pl.eu/public/models/plwn_api_dumps/plwn_dump_27-03-2018.sqlite
\ No newline at end of file
+default_model = https://minio.clarin-pl.eu/public/models/plwn_api_dumps/plwn_dump_new_07-12-2022.sqlite
diff --git a/plwn/download.py b/plwn/download.py
index f5647645180dea0dcf739020ef29b272d113a24a..9fac08a192f613c75d7bb14063e00df9e9bdd149 100644
--- a/plwn/download.py
+++ b/plwn/download.py
@@ -42,7 +42,7 @@ def download(name="default_model"):
return
if name in models:
url = config["DOWNLOAD"]["default_model"]
- url = url.replace("plwn_dump_27-03-2018.sqlite", name)
+ url = url.replace("plwn_dump_new_07-12-2022.sqlite", name)
r = requests.get(url)
with open(name, "wb") as f:
f.write(r.content)
diff --git a/plwn/enums.py b/plwn/enums.py
index 246fc3c8290958889f6d55fa5bf0d1098045bc2f..f99afb6198bc3b667683f6886c61aea9354b310d 100644
--- a/plwn/enums.py
+++ b/plwn/enums.py
@@ -104,10 +104,10 @@ class PoS(Enum):
__order__ = 'verb noun adverb adjective ' \
'verb_en noun_en adverb_en adjective_en'
- verb = u'verb'
- noun = u'noun'
- adverb = u'adverb'
- adjective = u'adjective'
+ verb = u'czasownik'
+ noun = u'rzeczownik'
+ adverb = u'przysłówek'
+ adjective = u'przymiotnik'
# English (PWN) PoSes
verb_en = u'verb_en'
@@ -170,8 +170,9 @@ class VerbAspect(Enum):
"""Defines verb aspect values used in plWordNet."""
if six.PY2:
- __order__ = 'perfective imperfective predicative two_aspect'
+ __order__ = 'perfective imperfective predicative two_aspect no'
+ no = u'no'
perfective = u'perf'
imperfective = u'imperf'
predicative = u'pred'
@@ -296,7 +297,7 @@ class Domain(Enum):
bhp = u'najwyższe w hierarchii'
czy = u'czynności (nazwy)'
- wytw = u'wytwory ludzkie (nazwy)'
+ wytw = u'wytwory ludzkie(nazwy)'
cech = u'cechy ludzi i zwierzÄ…t'
czc = u'części ciała'
umy = u'związane z myśleniem'
@@ -346,8 +347,8 @@ class Domain(Enum):
sys = u'systematyka, klasyfikacja'
- adj = u'PWN: all adjective clusters'
- adv = u'PWN: all adverbs'
+ adj = u'all adjective clusters'
+ adv = u'all adverbs'
mat = u'przymiotniki materiałowe'
diff --git a/plwn/readers/nodes.py b/plwn/readers/nodes.py
index b1bf8d3cafafd96ae67ee3c0012489bd71d6fb7e..45fb60123bc992ecb43f0f3b50be2adc98ec0c0d 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', None),
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 3983068dc6058ad0343ff03e6660e8a4bcb078a6..7d411384f158c0a55dffe09b2cbbdacee2091ffa 100644
--- a/plwn/storages/sqlite.py
+++ b/plwn/storages/sqlite.py
@@ -63,121 +63,116 @@ _DB_SCHEMA_SCRIPT = u"""
PRAGMA foreign_keys = ON;
-- Metadata table. Used for version number, currently
-CREATE TABLE plwn_meta (
- name TEXT UNIQUE NOT NULL,
+CREATE TABLE tbl_plwn_meta (
+ name TEXT UNIQUE NOT NULL ,
value BLOB
);
-- Tables for constant values
-CREATE TABLE pos (
+CREATE TABLE tbl_pos (
id INTEGER PRIMARY KEY,
- value TEXT UNIQUE NOT NULL
+ value TEXT NOT NULL
);
-CREATE TABLE verbaspect (
+CREATE TABLE tbl_verbaspect (
id INTEGER PRIMARY KEY,
value TEXT UNIQUE NOT NULL
);
-CREATE TABLE emotionmark (
+CREATE TABLE tbl_emotionmark (
id INTEGER PRIMARY KEY,
value TEXT UNIQUE NOT NULL
);
-CREATE TABLE emotionname (
+CREATE TABLE tbl_emotionname (
id INTEGER PRIMARY KEY,
value TEXT UNIQUE NOT NULL COLLATE locale
);
-CREATE TABLE emotionvaluation (
+CREATE TABLE tbl_emotionvaluation (
id INTEGER PRIMARY KEY,
- value TEXT UNIQUE NOT NULL COLLATE locale
+ value TEXT NOT NULL COLLATE locale
);
-CREATE TABLE domain (
+CREATE TABLE tbl_domain (
id INTEGER PRIMARY KEY,
- value TEXT UNIQUE NOT NULL COLLATE locale
+ value TEXT NOT NULL COLLATE locale
);
+CREATE UNIQUE INDEX value ON tbl_domain (value);
-- Synset only gets one simple table
-CREATE TABLE synset (
+CREATE TABLE tbl_synset (
id INTEGER PRIMARY KEY,
+ legacy_id INTEGER NULL ,
definition TEXT COLLATE locale,
isartificial INTEGER NOT NULL DEFAULT 0
);
-- Lexical units have several tables, since they have several list-like
-- properties. They also need indexes for lookup.
-CREATE TABLE lexicalunit (
+
+CREATE TABLE tbl_lexicalunit (
id INTEGER PRIMARY KEY,
+ legacy_id INTEGER NULL ,
lemma TEXT NOT NULL COLLATE locale,
pos INTEGER NOT NULL
- REFERENCES pos (id),
- variant INTEGER NOT NULL,
+ REFERENCES tbl_pos (id),
+ variant INTEGER NOT NULL ,
synset INTEGER NOT NULL
- REFERENCES synset (id),
- unitindex INTEGER NOT NULL,
+ REFERENCES tbl_synset (id),
+ unitindex INTEGER NOT NULL ,
definition TEXT COLLATE locale,
domain INTEGER NOT NULL
- REFERENCES domain (id),
+ REFERENCES tbl_domain (id),
verbaspect INTEGER
- REFERENCES verbaspect (id),
+ REFERENCES tbl_verbaspect (id),
isemotional INTEGER,
- emotionmark INTEGER
- REFERENCES emotionmark (id),
+ emotionmark INTEGER,
emotionexample1 TEXT COLLATE locale,
- emotionexample2 TEXT COLLATE locale,
-
- UNIQUE (lemma, pos, variant),
- -- Also, each unit needs its of place in synset
- UNIQUE (synset, unitindex)
+ emotionexample2 TEXT COLLATE locale
);
--- lem-pos-var and synset-unitindex indexes (and partial ones) are
--- automatically made because of UNIQUE constraint, but additional indexes
--- need to be created.
-CREATE INDEX lex_i_lem_var ON lexicalunit (lemma, variant);
-CREATE INDEX lex_i_pos ON lexicalunit (pos);
--- No index for variant itself - it's not an useful use case
+CREATE UNIQUE INDEX lemma ON tbl_lexicalunit (lemma, pos, variant);
+CREATE INDEX lex_i_lem_var ON tbl_lexicalunit (lemma, variant);
+CREATE INDEX lex_i_pos ON tbl_lexicalunit (pos);
+CREATE UNIQUE INDEX synset ON tbl_lexicalunit (synset, unitindex);
-- Tables dependant on lexicalunit
-CREATE TABLE senseexample (
+CREATE TABLE tbl_senseexample (
unitid INTEGER NOT NULL
- REFERENCES lexicalunit (id),
+ REFERENCES tbl_lexicalunit (id),
example TEXT NOT NULL COLLATE locale,
source TEXT NOT NULL COLLATE locale
);
-CREATE INDEX sen_i ON senseexample (unitid);
+CREATE INDEX sen_i ON tbl_senseexample (unitid);
-CREATE TABLE externallink (
+CREATE TABLE tbl_externallink (
unitid INTEGER NOT NULL
- REFERENCES lexicalunit (id),
+ REFERENCES tbl_lexicalunit (id),
link TEXT NOT NULL COLLATE locale
);
-CREATE INDEX link_i ON externallink (unitid);
+CREATE INDEX link_i ON tbl_externallink (unitid);
-CREATE TABLE usagenote (
+CREATE TABLE tbl_usagenote (
unitid INTEGER NOT NULL
- REFERENCES lexicalunit (id),
+ REFERENCES tbl_lexicalunit (id),
note TEXT NOT NULL COLLATE locale
);
-CREATE INDEX note_i ON usagenote (unitid);
+CREATE INDEX note_i ON tbl_usagenote (unitid);
-CREATE TABLE unitemotionname (
+CREATE TABLE tbl_unitemotionname (
unitid INTEGER NOT NULL
- REFERENCES lexicalunit (id),
+ REFERENCES tbl_lexicalunit (id),
nameid INTEGER NOT NULL
- REFERENCES emotionname (id),
-
+ REFERENCES tbl_emotionname (id),
PRIMARY KEY (unitid, nameid)
);
-CREATE TABLE unitemotionvaluation (
+CREATE TABLE tbl_unitemotionvaluation (
unitid INTEGER NOT NULL
- REFERENCES lexicalunit (id),
+ REFERENCES tbl_lexicalunit (id),
valuationid INTEGER NOT NULL
- REFERENCES emotionvaluation (id),
-
+ REFERENCES tbl_emotionvaluation (id),
PRIMARY KEY (unitid, valuationid)
);
@@ -185,113 +180,123 @@ CREATE TABLE unitemotionvaluation (
-- The for below are used to gather combinations of parent / child relation
-- names.
-CREATE TABLE synsetrelationparentpart (
+CREATE TABLE tbl_synsetrelationparentpart (
id INTEGER PRIMARY KEY,
name TEXT UNIQUE NOT NULL COLLATE locale
);
-CREATE TABLE synsetrelationchildpart (
+CREATE TABLE tbl_synsetrelationchildpart (
id INTEGER PRIMARY KEY,
name TEXT UNIQUE NOT NULL COLLATE locale
);
-CREATE TABLE lexicalrelationparentpart (
+CREATE TABLE tbl_lexicalrelationparentpart (
id INTEGER PRIMARY KEY,
name TEXT UNIQUE NOT NULL COLLATE locale
);
-CREATE TABLE lexicalrelationchildpart (
+CREATE TABLE tbl_lexicalrelationchildpart (
id INTEGER PRIMARY KEY,
name TEXT UNIQUE NOT NULL COLLATE locale
);
+CREATE UNIQUE INDEX name ON tbl_lexicalrelationchildpart (name);
+
-- Next, gather these parts into relation types themselves.
-- 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 synsetrelationtype (
+CREATE TABLE tbl_synsetrelationtype (
id INTEGER PRIMARY KEY,
+ legacy_id INTEGER NULL ,
parentpart INTEGER NOT NULL
- REFERENCES synsetrelationparentpart (id),
+ REFERENCES tbl_synsetrelationparentpart (id),
childpart INTEGER NOT NULL
- REFERENCES synsetrelationchildpart (id),
+ REFERENCES tbl_synsetrelationchildpart (id),
UNIQUE (parentpart, childpart)
);
-CREATE TABLE lexicalrelationtype (
- id INTEGER PRIMARY KEY,
+CREATE TABLE tbl_lexicalrelationtype (
+ id INTEGER PRIMARY KEY ,
+ legacy_id INTEGER NULL ,
parentpart INTEGER NOT NULL
- REFERENCES lexicalrelationparentpart (id),
+ REFERENCES tbl_lexicalrelationparentpart (id),
childpart INTEGER NOT NULL
- REFERENCES lexicalrelationchildpart (id),
-
- UNIQUE (parentpart, childpart)
+ REFERENCES tbl_lexicalrelationchildpart (id)
+);
+CREATE UNIQUE INDEX parentpart ON tbl_lexicalrelationtype (
+ parentpart,
+ childpart
);
-- The below tables are simply maps of relation aliases to their main IDs.
-- Reverse indexes are needed, too.
-CREATE TABLE synsetrelationalias (
+CREATE TABLE tbl_synsetrelationalias (
name TEXT PRIMARY KEY NOT NULL COLLATE locale,
relationid INTEGER NOT NULL
- REFERENCES synsetrelationtype (id)
+ REFERENCES tbl_synsetrelationtype (id)
+);
+CREATE INDEX synsetrelationalias_irev ON tbl_synsetrelationalias (
+ relationid
);
-CREATE INDEX synsetrelationalias_irev ON synsetrelationalias (relationid);
-CREATE TABLE lexicalrelationalias (
+CREATE TABLE tbl_lexicalrelationalias (
name TEXT PRIMARY KEY NOT NULL COLLATE locale,
relationid INTEGER NOT NULL
- REFERENCES lexicalrelationtype (id)
+ REFERENCES tbl_lexicalrelationtype (id)
+);
+CREATE INDEX lexicalrelationalias_irev ON tbl_lexicalrelationalias (
+ relationid
);
-CREATE INDEX lexicalrelationalias_irev ON lexicalrelationalias (relationid);
-- Next are finally the relation instances
-CREATE TABLE synsetrelation (
+CREATE TABLE tbl_synsetrelation (
source INTEGER NOT NULL
- REFERENCES synset (id),
+ REFERENCES tbl_synset (id),
relationtype INTEGER NOT NULL
- REFERENCES synsetrelationtype (id),
+ REFERENCES tbl_synsetrelationtype (id),
target INTEGER NOT NULL
- REFERENCES synset (id),
+ REFERENCES tbl_synset (id),
PRIMARY KEY (source, relationtype, target)
);
-CREATE TABLE lexicalrelation (
+CREATE TABLE tbl_lexicalrelation (
source INTEGER NOT NULL
- REFERENCES lexicalunit (id),
+ REFERENCES tbl_lexicalunit (id),
relationtype INTEGER NOT NULL
- REFERENCES lexicalrelationtype (id),
+ REFERENCES tbl_lexicalrelationtype (id),
target INTEGER NOT NULL
- REFERENCES lexicalunit (id),
+ REFERENCES tbl_lexicalunit (id),
PRIMARY KEY (source, relationtype, target)
);
-- Insert the special empty values for the parent part tables
-INSERT INTO synsetrelationparentpart (name) VALUES ('');
-INSERT INTO lexicalrelationparentpart (name) VALUES ('');
+INSERT INTO tbl_synsetrelationparentpart (name) VALUES ('');
+INSERT INTO tbl_lexicalrelationparentpart (name) VALUES ('');
""" # }}}
_RELTYPE_TABLES = {
- en.RelationKind.synset: u'synsetrelationtype',
- en.RelationKind.lexical: u'lexicalrelationtype',
+ en.RelationKind.synset: u'tbl_synsetrelationtype',
+ en.RelationKind.lexical: u'tbl_lexicalrelationtype',
}
_RELALIAS_TABLES = {
- en.RelationKind.synset: u'synsetrelationalias',
- en.RelationKind.lexical: u'lexicalrelationalias',
+ en.RelationKind.synset: u'tbl_synsetrelationalias',
+ en.RelationKind.lexical: u'tbl_lexicalrelationalias',
}
_RELPARENTPART_TABLES = {
- en.RelationKind.synset: u'synsetrelationparentpart',
- en.RelationKind.lexical: u'lexicalrelationparentpart',
+ en.RelationKind.synset: u'tbl_synsetrelationparentpart',
+ en.RelationKind.lexical: u'tbl_lexicalrelationparentpart',
}
_RELCHILDPART_TABLES = {
- en.RelationKind.synset: u'synsetrelationchildpart',
- en.RelationKind.lexical: u'lexicalrelationchildpart',
+ en.RelationKind.synset: u'tbl_synsetrelationchildpart',
+ en.RelationKind.lexical: u'tbl_lexicalrelationchildpart',
}
_RELINST_TABLES = {
- en.RelationKind.synset: u'synsetrelation',
- en.RelationKind.lexical: u'lexicalrelation',
+ en.RelationKind.synset: u'tbl_synsetrelation',
+ en.RelationKind.lexical: u'tbl_lexicalrelation',
}
class PLWordNet(bs.PLWordNetBase):
_STORAGE_NAME = 'sqlite3'
- _SCHEMA_VERSION = 4
+ _SCHEMA_VERSION = '4'
@classmethod
def from_reader(cls, reader, dump_to=None):
@@ -370,12 +375,31 @@ class PLWordNet(bs.PLWordNetBase):
)
def lexical_unit_by_id(self, id_):
+
+ if isinstance(id_, str):
+ uuid = "X'" + id_.replace('-', '').upper() + "'"
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"""
+ SELECT id
+ FROM tbl_lexicalunit
+ WHERE quote(tbl_lexicalunit.id) = ?
+ """,
+ (uuid,)
+ )
+ temp_id = cur.fetchone()
+ if temp_id is None:
+ raise exc.LexicalUnitNotFound('id=' + repr(id_))
+ else:
+ id_ = temp_id[0]
+
with closing(self._db.cursor()) as cur:
cur.execute(
u"""
- SELECT lemma, pos.value, variant, synset
- FROM lexicalunit JOIN pos ON lexicalunit.pos = pos.id
- WHERE lexicalunit.id = ?
+ SELECT lemma, tbl_pos.value, variant, synset
+ FROM tbl_lexicalunit
+ JOIN tbl_pos ON tbl_lexicalunit.pos = tbl_pos.id
+ WHERE tbl_lexicalunit.id = ?
""",
(id_,)
)
@@ -403,16 +427,16 @@ class PLWordNet(bs.PLWordNetBase):
cur.execute(
u"""
SELECT source, target, relationtype
- FROM lexicalrelation
+ FROM tbl_lexicalrelation
""" + where_clause,
param_tuple,
)
lu_q = u"""
- SELECT lemma, pos.value, variant, synset
- FROM lexicalunit
- JOIN pos ON pos.id = lexicalunit.pos
- WHERE lexicalunit.id = ?
+ SELECT lemma, tbl_pos.value, variant, synset
+ FROM tbl_lexicalunit
+ JOIN tbl_pos ON tbl_pos.id = tbl_lexicalunit.pos
+ WHERE tbl_lexicalunit.id = ?
"""
edges = []
@@ -467,9 +491,26 @@ class PLWordNet(bs.PLWordNetBase):
)
def synset_by_id(self, id_):
+ if isinstance(id_, str):
+ uuid = "X'" + id_.replace('-', '').upper() + "'"
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"""
+ SELECT id
+ FROM tbl_synset
+ WHERE quote(tbl_synset.id) = ?
+ """,
+ (uuid,)
+ )
+ temp_id = cur.fetchone()
+ if temp_id is None:
+ raise exc.LexicalUnitNotFound('id=' + repr(id_))
+ else:
+ id_ = temp_id[0]
+
with closing(self._db.cursor()) as cur:
cur.execute(
- u"SELECT EXISTS (SELECT 1 FROM synset WHERE id = ?)",
+ u"SELECT EXISTS (SELECT 1 FROM tbl_synset WHERE id = ?)",
(id_,),
)
if not cur.fetchone()[0]:
@@ -495,14 +536,14 @@ class PLWordNet(bs.PLWordNetBase):
)
select_clause = u"SELECT source, target, relationtype"
- from_clause = u"FROM synsetrelation"
+ from_clause = u"FROM tbl_synsetrelation"
# Pre-fetch artificial status if skipping is necessary
if skip_artificial:
select_clause += u", parentsyn.isartificial, childsyn.isartificial"
from_clause += (
- u" JOIN synset AS parentsyn ON parentsyn.id = source"
- u" JOIN synset AS childsyn ON childsyn.id = target"
+ u" JOIN tbl_synset AS parentsyn ON parentsyn.id = source"
+ u" JOIN tbl_synset AS childsyn ON childsyn.id = target"
)
yield_edges = self.__syn_edges_withskip
else:
@@ -552,11 +593,12 @@ class PLWordNet(bs.PLWordNetBase):
with closing(self._db.cursor()) as cur:
cur.execute(
u"""
- SELECT lexicalunit.id, lemma, pos.value, variant, synset
- FROM lexicalunit
- JOIN pos ON lexicalunit.pos = pos.id
+ SELECT tbl_lexicalunit.id, lemma, tbl_pos.value,
+ variant, synset
+ FROM tbl_lexicalunit
+ JOIN tbl_pos ON tbl_lexicalunit.pos = tbl_pos.id
WHERE COALESCE(lemma = :lem, :defval)
- AND COALESCE(pos.value = :pos, :defval)
+ AND COALESCE(tbl_pos.value = :pos, :defval)
AND COALESCE(variant = :var, :defval)
""",
{
@@ -589,34 +631,37 @@ class PLWordNet(bs.PLWordNetBase):
with self._db:
self._db.executemany(
- u"INSERT OR IGNORE INTO pos (value) VALUES (?)",
+ u"INSERT OR IGNORE INTO tbl_pos (value) VALUES (?)",
((p.value,) for p in en.PoS),
).close()
self._db.executemany(
- u"INSERT OR IGNORE INTO verbaspect (value) VALUES (?)",
+ u"INSERT OR IGNORE INTO tbl_verbaspect (value) VALUES (?)",
((va.value,) for va in en.VerbAspect),
).close()
self._db.executemany(
- u"INSERT OR IGNORE INTO emotionmark (value) VALUES (?)",
+ u"INSERT OR IGNORE INTO tbl_emotionmark (value) VALUES (?)",
((em.value,) for em in en.EmotionMarkedness),
).close()
self._db.executemany(
- u"INSERT OR IGNORE INTO emotionname (value) VALUES (?)",
+ u"INSERT OR IGNORE INTO tbl_emotionname (value) VALUES (?)",
((en.value,) for en in en.EmotionName),
).close()
self._db.executemany(
- u"INSERT OR IGNORE INTO emotionvaluation (value) VALUES (?)",
+ u"""
+ INSERT OR IGNORE INTO tbl_emotionvaluation (value)
+ VALUES (?)
+ """,
((ev.value,) for ev in en.EmotionValuation),
).close()
self._db.executemany(
- u"INSERT OR IGNORE INTO domain (value) VALUES (?)",
+ u"INSERT OR IGNORE INTO tbl_domain (value) VALUES (?)",
((dm.value,) for dm in en.Domain),
).close()
# Insert version if the database is new
self._db.execute(
u"""
- INSERT OR IGNORE INTO plwn_meta (name, value)
+ INSERT OR IGNORE INTO tbl_plwn_meta (name, value)
VALUES ('version', ?)
""",
(self._SCHEMA_VERSION,),
@@ -626,7 +671,7 @@ class PLWordNet(bs.PLWordNetBase):
with closing(self._db.cursor()) as cur:
try:
cur.execute(
- u"SELECT value FROM plwn_meta WHERE name = 'version'",
+ u"SELECT value FROM tbl_plwn_meta WHERE name = 'version'",
)
except sqlite3.OperationalError:
raise exc.LoadException(
@@ -684,6 +729,8 @@ class LexicalUnit(bs.LexicalUnitBase):
self._var = variant
self._synid = synid
# Rest is unitialized
+ self._uuid = _UNFETCHED
+ self._leg_id = _UNFETCHED
self._syn = _UNFETCHED
self._def = _UNFETCHED
self._usn = _UNFETCHED
@@ -723,6 +770,35 @@ class LexicalUnit(bs.LexicalUnitBase):
def is_english(self):
return self._pos.is_english
+ @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 tbl_lexicalunit WHERE id = ?",
+ (self._id,),
+ )
+ self._leg_id = cur.fetchone()[0]
+ return self._leg_id
+
+ @property
+ def uuid(self):
+ if self._uuid is _UNFETCHED:
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"SELECT quote(id) from tbl_lexicalunit WHERE id = ?",
+ (self._id,),
+ )
+ temp = cur.fetchone()[0]
+ if isinstance(temp, str):
+ indices = [2, 10, 14, 18, 22]
+ self._uuid = ('-').join([temp[i:j] for i, j in
+ zip(indices, indices[1:] + [None])
+ ]).lower()[:-1]
+ else:
+ self._uuid = temp
+ return self._uuid
+
@property
def synset(self):
if self._syn is _UNFETCHED or self._syn() is None:
@@ -737,7 +813,7 @@ class LexicalUnit(bs.LexicalUnitBase):
if self._def is _UNFETCHED:
with closing(self._db.cursor()) as cur:
cur.execute(
- u"SELECT definition FROM lexicalunit WHERE id = ?",
+ u"SELECT definition FROM tbl_lexicalunit WHERE id = ?",
(self._id,),
)
self._def = cur.fetchone()[0]
@@ -748,7 +824,7 @@ class LexicalUnit(bs.LexicalUnitBase):
if self._exms is _UNFETCHED:
with closing(self._db.cursor()) as cur:
cur.execute(
- u"SELECT example FROM senseexample WHERE unitid = ?",
+ u"SELECT example FROM tbl_senseexample WHERE unitid = ?",
(self._id,),
)
self._exms = tuple(row[0] for row in cur)
@@ -759,7 +835,7 @@ class LexicalUnit(bs.LexicalUnitBase):
if self._exms_srcs is _UNFETCHED:
with closing(self._db.cursor()) as cur:
cur.execute(
- u"SELECT source FROM senseexample WHERE unitid = ?",
+ u"SELECT source FROM tbl_senseexample WHERE unitid = ?",
(self._id,),
)
self._exms_srcs = tuple(row[0] for row in cur)
@@ -770,7 +846,7 @@ class LexicalUnit(bs.LexicalUnitBase):
if self._extl is _UNFETCHED:
with closing(self._db.cursor()) as cur:
cur.execute(
- u"SELECT link FROM externallink WHERE unitid = ?",
+ u"SELECT link FROM tbl_externallink WHERE unitid = ?",
(self._id,),
)
self._extl = tuple(row[0] for row in cur)
@@ -781,7 +857,7 @@ class LexicalUnit(bs.LexicalUnitBase):
if self._usn is _UNFETCHED:
with closing(self._db.cursor()) as cur:
cur.execute(
- u"SELECT note FROM usagenote WHERE unitid = ?",
+ u"SELECT note FROM tbl_usagenote WHERE unitid = ?",
(self._id,),
)
self._usn = tuple(row[0] for row in cur)
@@ -793,10 +869,10 @@ class LexicalUnit(bs.LexicalUnitBase):
with closing(self._db.cursor()) as cur:
cur.execute(
u"""
- SELECT domain.value
- FROM lexicalunit JOIN domain
- ON lexicalunit.domain = domain.id
- WHERE lexicalunit.id = ?
+ SELECT tbl_domain.value
+ FROM tbl_lexicalunit JOIN tbl_domain
+ ON tbl_lexicalunit.domain = tbl_domain.id
+ WHERE tbl_lexicalunit.id = ?
""",
(self._id,),
)
@@ -809,10 +885,10 @@ class LexicalUnit(bs.LexicalUnitBase):
with closing(self._db.cursor()) as cur:
cur.execute(
u"""
- SELECT verbaspect.value
- FROM lexicalunit JOIN verbaspect
- ON lexicalunit.verbaspect = verbaspect.id
- WHERE lexicalunit.id = ?
+ SELECT tbl_verbaspect.value
+ FROM tbl_lexicalunit JOIN tbl_verbaspect
+ ON tbl_lexicalunit.verbaspect = tbl_verbaspect.id
+ WHERE tbl_lexicalunit.id = ?
""",
(self._id,),
)
@@ -825,7 +901,7 @@ class LexicalUnit(bs.LexicalUnitBase):
if self._is_emo is _UNFETCHED:
with closing(self._db.cursor()) as cur:
cur.execute(
- u"SELECT isemotional FROM lexicalunit WHERE id = ?",
+ u"SELECT isemotional FROM tbl_lexicalunit WHERE id = ?",
(self._id,),
)
rowval = cur.fetchone()[0]
@@ -838,10 +914,10 @@ class LexicalUnit(bs.LexicalUnitBase):
with closing(self._db.cursor()) as cur:
cur.execute(
u"""
- SELECT emotionmark.value
- FROM lexicalunit JOIN emotionmark
- ON lexicalunit.emotionmark = emotionmark.id
- WHERE lexicalunit.id = ?
+ SELECT tbl_emotionmark.value
+ FROM tbl_lexicalunit JOIN tbl_emotionmark
+ ON tbl_lexicalunit.emotionmark = tbl_emotionmark.id
+ WHERE tbl_lexicalunit.id = ?
""",
(self._id,),
)
@@ -859,11 +935,11 @@ class LexicalUnit(bs.LexicalUnitBase):
with closing(self._db.cursor()) as cur:
cur.execute(
u"""
- SELECT emotionname.value
- FROM emotionname JOIN unitemotionname
- ON emotionname.id = unitemotionname.nameid
- WHERE unitemotionname.unitid = ?
- ORDER BY emotionname.value
+ SELECT tbl_emotionname.value
+ FROM tbl_emotionname JOIN tbl_unitemotionname
+ ON tbl_emotionname.id = tbl_unitemotionname.nameid
+ WHERE tbl_unitemotionname.unitid = ?
+ ORDER BY tbl_emotionname.value
""",
(self._id,),
)
@@ -876,12 +952,12 @@ class LexicalUnit(bs.LexicalUnitBase):
with closing(self._db.cursor()) as cur:
cur.execute(
u"""
- SELECT emotionvaluation.value
- FROM emotionvaluation JOIN unitemotionvaluation
- ON emotionvaluation.id =
- unitemotionvaluation.valuationid
- WHERE unitemotionvaluation.unitid = ?
- ORDER BY emotionvaluation.value
+ SELECT tbl_emotionvaluation.value
+ FROM tbl_emotionvaluation JOIN tbl_unitemotionvaluation
+ ON tbl_emotionvaluation.id =
+ tbl_unitemotionvaluation.valuationid
+ WHERE tbl_unitemotionvaluation.unitid = ?
+ ORDER BY tbl_emotionvaluation.value
""",
(self._id,),
)
@@ -896,7 +972,7 @@ class LexicalUnit(bs.LexicalUnitBase):
if self._emo_ex1 is _UNFETCHED:
with closing(self._db.cursor()) as cur:
cur.execute(
- u"SELECT emotionexample1 FROM lexicalunit WHERE id = ?",
+ u"SELECT emotionexample1 FROM tbl_lexicalunit WHERE id = ?",
(self._id,),
)
self._emo_ex1 = cur.fetchone()[0]
@@ -907,7 +983,7 @@ class LexicalUnit(bs.LexicalUnitBase):
if self._emo_ex2 is _UNFETCHED:
with closing(self._db.cursor()) as cur:
cur.execute(
- u"SELECT emotionexample2 FROM lexicalunit WHERE id = ?",
+ u"SELECT emotionexample2 FROM tbl_lexicalunit WHERE id = ?",
(self._id,),
)
self._emo_ex2 = cur.fetchone()[0]
@@ -921,7 +997,7 @@ class LexicalUnit(bs.LexicalUnitBase):
cur.execute(
u"""
SELECT DISTINCT relationtype
- FROM lexicalrelation
+ FROM tbl_lexicalrelation
WHERE source = ?
""",
(self._id,),
@@ -940,10 +1016,11 @@ class LexicalUnit(bs.LexicalUnitBase):
with closing(self._db.cursor()) as cur:
cur.execute(
u"""
- SELECT lexicalunit.id, lemma, pos.value, variant, synset
- FROM lexicalrelation
- JOIN lexicalunit ON lexicalunit.id = target
- JOIN pos ON lexicalunit.pos = pos.id
+ SELECT tbl_lexicalunit.id,
+ lemma, tbl_pos.value, variant, synset
+ FROM tbl_lexicalrelation
+ JOIN tbl_lexicalunit ON tbl_lexicalunit.id = target
+ JOIN tbl_pos ON tbl_lexicalunit.pos = tbl_pos.id
WHERE source = ? {}
""".format(_make_relationtype_where(relinfos)),
tuple(itt.chain(
@@ -966,10 +1043,10 @@ class LexicalUnit(bs.LexicalUnitBase):
cur.execute(
u"""
SELECT relationtype,
- lexicalunit.id, lemma, pos.value, variant, synset
- FROM lexicalrelation
- JOIN lexicalunit ON lexicalunit.id = target
- JOIN pos ON lexicalunit.pos = pos.id
+ tbl_lexicalunit.id, lemma, tbl_pos.value, variant, synset
+ FROM tbl_lexicalrelation
+ JOIN tbl_lexicalunit ON tbl_lexicalunit.id = target
+ JOIN tbl_pos ON tbl_lexicalunit.pos = tbl_pos.id
WHERE source = ? {}
""".format(_make_relationtype_where(relinfos)),
tuple(itt.chain(
@@ -1001,8 +1078,10 @@ class Synset(bs.SynsetBase):
self._id = syn_id
self._isart = syn_art
+ self._uuid = _UNFETCHED
self._units = _UNFETCHED
self._def = _UNFETCHED
+ self._leg_id = _UNFETCHED
self._pos = _UNFETCHED
self._is_polish = _UNFETCHED
@@ -1038,8 +1117,9 @@ class Synset(bs.SynsetBase):
with closing(self._db.cursor()) as cur:
cur.execute(
u"""
- SELECT lexicalunit.id, lemma, pos.value, variant
- FROM lexicalunit JOIN pos ON lexicalunit.pos = pos.id
+ SELECT tbl_lexicalunit.id, lemma, tbl_pos.value, variant
+ FROM tbl_lexicalunit
+ JOIN tbl_pos ON tbl_lexicalunit.pos = tbl_pos.id
WHERE synset = ?
ORDER BY unitindex
""",
@@ -1060,12 +1140,41 @@ class Synset(bs.SynsetBase):
assert self._units
return self._units
+ @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 tbl_synset WHERE id = ?",
+ (self._id,),
+ )
+ self._leg_id = cur.fetchone()[0]
+ return self._leg_id
+
+ @property
+ def uuid(self):
+ if self._uuid is _UNFETCHED:
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"SELECT quote(id) from tbl_synset WHERE id = ?",
+ (self._id,),
+ )
+ temp = cur.fetchone()[0]
+ if isinstance(temp, str):
+ indices = [2, 10, 14, 18, 22]
+ self._uuid = ('-').join([temp[i:j] for i, j in
+ zip(indices, indices[1:] + [None])
+ ]).lower()[:-1]
+ else:
+ self._uuid = temp
+ return self._uuid
+
@property
def definition(self):
if self._def is _UNFETCHED:
with closing(self._db.cursor()) as cur:
cur.execute(
- u"SELECT definition FROM synset WHERE id = ?",
+ u"SELECT definition FROM tbl_synset WHERE id = ?",
(self._id,),
)
row = cur.fetchone()
@@ -1078,7 +1187,7 @@ class Synset(bs.SynsetBase):
if self._isart is _UNFETCHED:
with closing(self._db.cursor()) as cur:
cur.execute(
- u"SELECT isartificial FROM synset WHERE id = ?",
+ u"SELECT isartificial FROM tbl_synset WHERE id = ?",
(self._id,),
)
row = cur.fetchone()
@@ -1094,7 +1203,7 @@ class Synset(bs.SynsetBase):
cur.execute(
u"""
SELECT DISTINCT relationtype
- FROM synsetrelation
+ FROM tbl_synsetrelation
WHERE source = ?
""",
(self._id,),
@@ -1116,11 +1225,11 @@ class Synset(bs.SynsetBase):
en.RelationKind.synset,
)
select_clause = u"SELECT target"
- from_clause = u"FROM synsetrelation"
+ from_clause = u"FROM tbl_synsetrelation"
if skip_artificial:
- select_clause += u", synset.isartificial, relationtype"
- from_clause += u" JOIN synset ON target = synset.id"
+ select_clause += u", tbl_synset.isartificial, relationtype"
+ from_clause += u" JOIN tbl_synset ON target = tbl_synset.id"
yield_related = self.__related_withskip
else:
yield_related = self.__related_noskip
@@ -1153,11 +1262,11 @@ class Synset(bs.SynsetBase):
en.RelationKind.synset,
)
select_clause = u"SELECT relationtype, target"
- from_clause = u"FROM synsetrelation"
+ from_clause = u"FROM tbl_synsetrelation"
if skip_artificial:
- select_clause += u", synset.isartificial"
- from_clause += u" JOIN synset ON target = synset.id"
+ select_clause += u", tbl_synset.isartificial"
+ from_clause += u" JOIN tbl_synset ON target = tbl_synset.id"
yield_related = self.__related_withskip_pairs
else:
yield_related = self.__related_noskip_pairs
@@ -1240,6 +1349,19 @@ 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):
@@ -1319,18 +1441,23 @@ class _DBBuilder(object):
# Synset to lexical units relations also need to be deferred.
self._synid2lexids = coll.defaultdict(list)
# Cache IDs of constant values
- with closing(db.execute(u"SELECT value, id FROM pos")) as cur:
+ with closing(db.execute(u"SELECT value, id FROM tbl_pos")) as cur:
self._posids = dict(cur)
- with closing(db.execute(u"SELECT value, id FROM verbaspect")) as cur:
+ with closing(db.execute(u"SELECT value, id FROM tbl_verbaspect")) \
+ as cur:
self._vaids = dict(cur)
- with closing(db.execute(u"SELECT value, id FROM emotionmark")) as cur:
+ with closing(db.execute(u"SELECT value, id FROM tbl_emotionmark")) \
+ as cur:
self._emids = dict(cur)
- with closing(db.execute(u"SELECT value, id FROM emotionname")) as cur:
- self._enids = dict(cur)
- with closing(db.execute(u"SELECT value, id FROM emotionvaluation")) \
+ with closing(db.execute(u"SELECT value, id FROM tbl_emotionname")) \
as cur:
+ self._enids = dict(cur)
+ with closing(db.execute(u"""
+ SELECT value, id
+ FROM tbl_emotionvaluation
+ """)) as cur:
self._evids = dict(cur)
- with closing(db.execute(u"SELECT value, id FROM domain")) as cur:
+ with closing(db.execute(u"SELECT value, id FROM tbl_domain")) as cur:
self._dmids = dict(cur)
def __call__(self, reader):
@@ -1357,10 +1484,11 @@ class _DBBuilder(object):
def _insert_synset(self, syn_node):
self._db.execute(
u"""
- INSERT INTO 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
@@ -1386,9 +1514,11 @@ 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:
@@ -1408,15 +1538,15 @@ class _DBBuilder(object):
try:
cur.execute(
u"""
- INSERT INTO lexicalunit (
- id, lemma, pos, variant,
+ INSERT INTO tbl_lexicalunit (
+ 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,
@@ -1425,6 +1555,7 @@ 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,
@@ -1459,7 +1590,7 @@ class _DBBuilder(object):
cur.executemany(
u"""
- INSERT INTO senseexample (unitid, example, source)
+ INSERT INTO tbl_senseexample (unitid, example, source)
VALUES (?, ?, ?)
""",
(
@@ -1470,14 +1601,14 @@ class _DBBuilder(object):
)
cur.executemany(
u"""
- INSERT INTO usagenote (unitid, note)
+ INSERT INTO tbl_usagenote (unitid, note)
VALUES (?, ?)
""",
((lu_node.id, note) for note in lu_node.usage_notes),
)
cur.executemany(
u"""
- INSERT INTO externallink (unitid, link)
+ INSERT INTO tbl_externallink (unitid, link)
VALUES (?, ?)
""",
((lu_node.id, link)
@@ -1485,7 +1616,7 @@ class _DBBuilder(object):
)
cur.executemany(
u"""
- INSERT INTO unitemotionname (unitid, nameid)
+ INSERT INTO tbl_unitemotionname (unitid, nameid)
VALUES (?, ?)
""",
(
@@ -1495,7 +1626,8 @@ class _DBBuilder(object):
)
cur.executemany(
u"""
- INSERT INTO unitemotionvaluation (unitid, valuationid)
+ INSERT INTO tbl_unitemotionvaluation
+ (unitid, valuationid)
VALUES (?, ?)
""",
(
@@ -1576,7 +1708,10 @@ 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),
)
@@ -1632,10 +1767,11 @@ class _DBBuilder(object):
with closing(self._db.cursor()) as cur:
cur.execute(
u"""
- SELECT synset.id
- FROM synset
- LEFT JOIN lexicalunit ON synset.id = lexicalunit.synset
- WHERE lexicalunit.synset IS NULL
+ SELECT tbl_synset.id
+ FROM tbl_synset
+ LEFT JOIN tbl_lexicalunit
+ ON tbl_synset.id = tbl_lexicalunit.synset
+ WHERE tbl_lexicalunit.synset IS NULL
""",
)
empties = tuple(row[0] for row in cur)
diff --git a/tests/abstract_cases/test_unit_and_synset.py b/tests/abstract_cases/test_unit_and_synset.py
index 626e50f2161d79f5026a2152a3becc2392617fd9..4ac8513a1c366532e4b362d375dbb97783733728 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,
@@ -93,6 +94,7 @@ class SynsetRelationsTest(ut.TestCase):
nd.make_synset_node(id=3),
nd.make_lexical_unit_node(
id=11,
+ legacy_id=11,
lemma=u'aaa',
pos=en.PoS.n,
variant=1,
@@ -102,6 +104,7 @@ class SynsetRelationsTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=21,
+ legacy_id=21,
lemma=u'aaa',
pos=en.PoS.n,
variant=2,
@@ -111,6 +114,7 @@ class SynsetRelationsTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=31,
+ legacy_id=31,
lemma=u'aaa',
pos=en.PoS.n,
variant=3,
@@ -257,6 +261,7 @@ class SynsetRelationsWithArtificialTest(ut.TestCase):
nd.make_synset_node(id=8),
nd.make_lexical_unit_node(
id=11,
+ legacy_id=11,
lemma=u'aaa',
pos=en.PoS.n,
variant=1,
@@ -266,6 +271,7 @@ class SynsetRelationsWithArtificialTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=21,
+ legacy_id=21,
lemma=u'aaa',
pos=en.PoS.n,
variant=2,
@@ -275,6 +281,7 @@ class SynsetRelationsWithArtificialTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=31,
+ legacy_id=31,
lemma=u'aaa',
pos=en.PoS.n,
variant=3,
@@ -284,6 +291,7 @@ class SynsetRelationsWithArtificialTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=41,
+ legacy_id=41,
lemma=u'aaa',
pos=en.PoS.n,
variant=4,
@@ -293,6 +301,7 @@ class SynsetRelationsWithArtificialTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=51,
+ legacy_id=51,
lemma=u'aaa',
pos=en.PoS.n,
variant=5,
@@ -302,6 +311,7 @@ class SynsetRelationsWithArtificialTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=61,
+ legacy_id=61,
lemma=u'aaa',
pos=en.PoS.n,
variant=6,
@@ -311,6 +321,7 @@ class SynsetRelationsWithArtificialTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=71,
+ legacy_id=71,
lemma=u'aaa',
pos=en.PoS.n,
variant=7,
@@ -320,6 +331,7 @@ class SynsetRelationsWithArtificialTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=81,
+ legacy_id=81,
lemma=u'aaa',
pos=en.PoS.n,
variant=8,
@@ -445,6 +457,7 @@ class SynsetRelationsWithArtificialLoopTest(ut.TestCase):
nd.make_synset_node(id=4),
nd.make_lexical_unit_node(
id=11,
+ legacy_id=11,
lemma=u'aaa',
pos=en.PoS.n,
variant=1,
@@ -454,6 +467,7 @@ class SynsetRelationsWithArtificialLoopTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=21,
+ legacy_id=21,
lemma=u'aaa',
pos=en.PoS.n,
variant=2,
@@ -463,6 +477,7 @@ class SynsetRelationsWithArtificialLoopTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=31,
+ legacy_id=31,
lemma=u'aaa',
pos=en.PoS.n,
variant=3,
@@ -472,6 +487,7 @@ class SynsetRelationsWithArtificialLoopTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=41,
+ legacy_id=41,
lemma=u'aaa',
pos=en.PoS.n,
variant=4,
@@ -528,6 +544,7 @@ class LexicalUnitPropertiesTest(ut.TestCase):
nd.make_synset_node(id=1),
nd.make_lexical_unit_node(
id=11,
+ legacy_id=11,
lemma=u'aaa',
pos=en.PoS.n,
variant=1,
@@ -644,6 +661,7 @@ class LexicalUnitRelationsTest(ut.TestCase):
nd.make_synset_node(id=1),
nd.make_lexical_unit_node(
id=11,
+ legacy_id=11,
lemma=u'aaa',
pos=en.PoS.n,
variant=1,
@@ -657,6 +675,7 @@ class LexicalUnitRelationsTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=12,
+ legacy_id=12,
lemma=u'bbb',
pos=en.PoS.n,
variant=1,
@@ -667,6 +686,7 @@ class LexicalUnitRelationsTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=13,
+ legacy_id=13,
lemma=u'ccc',
pos=en.PoS.n,
variant=1,
@@ -766,6 +786,7 @@ class ItemOrderingTest(ut.TestCase):
nd.make_synset_node(id=3),
nd.make_lexical_unit_node(
id=11,
+ legacy_id=11,
lemma=u'aaa',
pos=en.PoS.n,
variant=1,
@@ -775,6 +796,7 @@ class ItemOrderingTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=21,
+ legacy_id=21,
lemma=u'bbb',
pos=en.PoS.n,
variant=1,
@@ -784,6 +806,7 @@ class ItemOrderingTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=22,
+ legacy_id=22,
lemma=u'Ä…Ä…Ä…',
pos=en.PoS.n,
variant=2,
@@ -793,6 +816,7 @@ class ItemOrderingTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=31,
+ legacy_id=31,
lemma=u'Ä…Ä…Ä…',
pos=en.PoS.n,
variant=1,
@@ -843,6 +867,7 @@ class ToDictTest(ut.TestCase):
nd.make_synset_node(id=3),
nd.make_lexical_unit_node(
id=11,
+ legacy_id=11,
lemma=u'aaa',
pos=en.PoS.v,
variant=1,
@@ -858,14 +883,15 @@ 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(
id=21,
+ legacy_id=21,
lemma=u'bbb',
pos=en.PoS.n,
variant=1,
@@ -875,6 +901,7 @@ class ToDictTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=31,
+ legacy_id=31,
lemma=u'ccc',
pos=en.PoS.n,
variant=1,
@@ -885,8 +912,9 @@ class ToDictTest(ut.TestCase):
))
self.__lex11_dict = {
u'id': 11,
+ u'legacy_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>