diff --git a/plwn/bases.py b/plwn/bases.py
index 43f9712ef12cdbf7d08832f084686d047e078924..0e66bc704b21000b86afe8068fdb41a16d3d6926 100644
--- a/plwn/bases.py
+++ b/plwn/bases.py
@@ -939,6 +939,7 @@ class LexicalUnitBase(object):
"""
lu_dict = {
u'id': self.id,
+ u'legacy_id': self.legacy_id,
u'lemma': self.lemma,
u'pos': self.pos.value,
u'variant': self.variant,
diff --git a/plwn/enums.py b/plwn/enums.py
index 6dfcc70530dfb08d64189a99370872dacae0002e..de9f4d8086be89a0d12c2e42fd731e76190f0bdf 100644
--- a/plwn/enums.py
+++ b/plwn/enums.py
@@ -296,7 +296,7 @@ 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'
diff --git a/plwn/readers/nodes.py b/plwn/readers/nodes.py
index 9f58999afc307145030394a1894feafd2ef1a914..45fb60123bc992ecb43f0f3b50be2adc98ec0c0d 100644
--- a/plwn/readers/nodes.py
+++ b/plwn/readers/nodes.py
@@ -122,7 +122,7 @@ def make_relation_type_node(**props):
"""
rel = RelationTypeNode(
name=props.pop('name'),
- legacy_id=props.pop('legacy_id'),
+ 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/storages/sqlite.py b/plwn/storages/sqlite.py
index 904e5400ce632a08d27da49e89f9ebb99713ed27..940d4b11764bedbf70ef409771d257cd2f997de3 100644
--- a/plwn/storages/sqlite.py
+++ b/plwn/storages/sqlite.py
@@ -64,76 +64,72 @@ PRAGMA foreign_keys = ON;
-- Metadata table. Used for version number, currently
CREATE TABLE tbl_plwn_meta (
- name VARCHAR(255) NOT NULL ,
- value BLOB NULL
+ name TEXT UNIQUE NOT NULL ,
+ value BLOB
);
-- Tables for constant values
CREATE TABLE tbl_pos (
- id INTEGER NOT NULL ,
- value VARCHAR(255) NOT NULL ,
- PRIMARY KEY (id)
+ id INTEGER PRIMARY KEY,
+ value TEXT NOT NULL
);
CREATE TABLE tbl_verbaspect (
- id INTEGER NOT NULL ,
- value VARCHAR(255) NOT NULL ,
- PRIMARY KEY (id)
+ id INTEGER PRIMARY KEY,
+ value TEXT UNIQUE NOT NULL
);
CREATE TABLE tbl_emotionmark (
- id INTEGER NOT NULL ,
- value VARCHAR(255) NOT NULL ,
- PRIMARY KEY (id)
+ id INTEGER PRIMARY KEY,
+ value TEXT UNIQUE NOT NULL
);
CREATE TABLE tbl_emotionname (
- id INTEGER NOT NULL ,
- value VARCHAR(255) NOT NULL ,
- PRIMARY KEY (id)
+ id INTEGER PRIMARY KEY,
+ value TEXT UNIQUE NOT NULL COLLATE locale
);
CREATE TABLE tbl_emotionvaluation (
- id INTEGER NOT NULL ,
- value VARCHAR(255) NOT NULL ,
- PRIMARY KEY (id)
+ id INTEGER PRIMARY KEY,
+ value TEXT NOT NULL COLLATE locale
);
CREATE TABLE tbl_domain (
- id INTEGER NOT NULL ,
- value VARCHAR(255) NOT NULL ,
- PRIMARY KEY (id)
+ id INTEGER PRIMARY KEY,
+ value TEXT NOT NULL COLLATE locale
);
CREATE UNIQUE INDEX value ON tbl_domain (value);
-- Synset only gets one simple table
CREATE TABLE tbl_synset (
- id BLOB ,
+ id BLOB PRIMARY KEY,
legacy_id INTEGER NULL ,
- definition TEXT NULL ,
- isartificial INTEGER NOT NULL DEFAULT '0' ,
- PRIMARY KEY (id)
+ 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 tbl_lexicalunit (
- id BLOB ,
+ id BLOB PRIMARY KEY,
legacy_id INTEGER NULL ,
- lemma VARCHAR(255) NOT NULL ,
- pos INTEGER NOT NULL ,
+ lemma TEXT NOT NULL COLLATE locale,
+ pos INTEGER NOT NULL
+ REFERENCES tbl_pos (id),
variant INTEGER NOT NULL ,
- synset BLOB NOT NULL ,
+ synset BLOB NOT NULL
+ REFERENCES tbl_synset (id),
unitindex INTEGER NOT NULL ,
- definition TEXT NULL ,
- domain INTEGER NOT NULL ,
- verbaspect INTEGER NULL ,
- isemotional INTEGER NULL ,
- emotionmark INTEGER NULL ,
- emotionexample1 TEXT NULL ,
- emotionexample2 TEXT NULL ,
- PRIMARY KEY (id)
+ definition TEXT COLLATE locale,
+ domain INTEGER NOT NULL
+ REFERENCES tbl_domain (id),
+ verbaspect INTEGER
+ REFERENCES tbl_verbaspect (id),
+ isemotional INTEGER,
+ emotionmark INTEGER,
+ emotionexample1 TEXT COLLATE locale,
+ emotionexample2 TEXT COLLATE locale
);
CREATE UNIQUE INDEX lemma ON tbl_lexicalunit (lemma, pos, variant);
@@ -143,33 +139,40 @@ CREATE UNIQUE INDEX synset ON tbl_lexicalunit (synset, unitindex);
-- Tables dependant on lexicalunit
CREATE TABLE tbl_senseexample (
- unitid BLOB NOT NULL ,
- example TEXT NOT NULL ,
- source TEXT NOT NULL
+ unitid BLOB NOT NULL
+ REFERENCES tbl_lexicalunit (id),
+ example TEXT NOT NULL COLLATE locale,
+ source TEXT NOT NULL COLLATE locale
);
CREATE INDEX sen_i ON tbl_senseexample (unitid);
CREATE TABLE tbl_externallink (
- unitid BLOB NOT NULL ,
- link TEXT NOT NULL
+ unitid BLOB NOT NULL
+ REFERENCES tbl_lexicalunit (id),
+ link TEXT NOT NULL COLLATE locale
);
CREATE INDEX link_i ON tbl_externallink (unitid);
CREATE TABLE tbl_usagenote (
- unitid BLOB NOT NULL ,
- note TEXT NOT NULL
+ unitid BLOB NOT NULL
+ REFERENCES tbl_lexicalunit (id),
+ note TEXT NOT NULL COLLATE locale
);
CREATE INDEX note_i ON tbl_usagenote (unitid);
CREATE TABLE tbl_unitemotionname (
- unitid BLOB NOT NULL ,
- nameid INTEGER NOT NULL ,
+ unitid BLOB NOT NULL
+ REFERENCES tbl_lexicalunit (id),
+ nameid INTEGER NOT NULL
+ REFERENCES tbl_emotionname (id),
PRIMARY KEY (unitid, nameid)
);
CREATE TABLE tbl_unitemotionvaluation (
- unitid BLOB NOT NULL ,
- valuationid INTEGER NOT NULL ,
+ unitid BLOB NOT NULL
+ REFERENCES tbl_lexicalunit (id),
+ valuationid INTEGER NOT NULL
+ REFERENCES tbl_emotionvaluation (id),
PRIMARY KEY (unitid, valuationid)
);
@@ -178,24 +181,20 @@ CREATE TABLE tbl_unitemotionvaluation (
-- The for below are used to gather combinations of parent / child relation
-- names.
CREATE TABLE tbl_synsetrelationparentpart (
- id INTEGER NOT NULL ,
- name VARCHAR(255) NOT NULL ,
- PRIMARY KEY (id)
+ id INTEGER PRIMARY KEY,
+ name TEXT UNIQUE NOT NULL COLLATE locale
);
CREATE TABLE tbl_synsetrelationchildpart (
- id INTEGER NOT NULL ,
- name VARCHAR(255) NOT NULL ,
- PRIMARY KEY (id)
+ id INTEGER PRIMARY KEY,
+ name TEXT UNIQUE NOT NULL COLLATE locale
);
CREATE TABLE tbl_lexicalrelationparentpart (
- id INTEGER NOT NULL ,
- name VARCHAR(255) NOT NULL ,
- PRIMARY KEY (id)
+ id INTEGER PRIMARY KEY,
+ name TEXT UNIQUE NOT NULL COLLATE locale
);
CREATE TABLE tbl_lexicalrelationchildpart (
- id INTEGER NOT NULL ,
- name VARCHAR(255) NOT NULL ,
- PRIMARY KEY (id)
+ id INTEGER PRIMARY KEY,
+ name TEXT UNIQUE NOT NULL COLLATE locale
);
CREATE UNIQUE INDEX name ON tbl_lexicalrelationchildpart (name);
@@ -204,18 +203,22 @@ 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 ,
+ id INTEGER PRIMARY KEY,
legacy_id INTEGER NULL ,
- parentpart INTEGER NOT NULL ,
- childpart INTEGER NOT NULL ,
- PRIMARY KEY (id)
+ parentpart INTEGER NOT NULL
+ REFERENCES tbl_synsetrelationparentpart (id),
+ childpart INTEGER NOT NULL
+ REFERENCES tbl_synsetrelationchildpart (id),
+
+ UNIQUE (parentpart, childpart)
);
CREATE TABLE tbl_lexicalrelationtype (
- id BLOB ,
+ id INTEGER PRIMARY KEY ,
legacy_id INTEGER NULL ,
- parentpart INTEGER NOT NULL ,
- childpart INTEGER NOT NULL ,
- PRIMARY KEY (id)
+ parentpart INTEGER NOT NULL
+ REFERENCES tbl_lexicalrelationparentpart (id),
+ childpart INTEGER NOT NULL
+ REFERENCES tbl_lexicalrelationchildpart (id)
);
CREATE UNIQUE INDEX parentpart ON tbl_lexicalrelationtype (
parentpart,
@@ -225,17 +228,17 @@ CREATE UNIQUE INDEX parentpart ON tbl_lexicalrelationtype (
-- The below tables are simply maps of relation aliases to their main IDs.
-- Reverse indexes are needed, too.
CREATE TABLE tbl_synsetrelationalias (
- name VARCHAR(255) NOT NULL ,
- relationid BLOB NOT NULL ,
- PRIMARY KEY (name)
+ name TEXT PRIMARY KEY NOT NULL COLLATE locale,
+ relationid BLOB NOT NULL
+ REFERENCES tbl_synsetrelationtype (id)
);
CREATE INDEX synsetrelationalias_irev ON tbl_synsetrelationalias (
relationid
);
CREATE TABLE tbl_lexicalrelationalias (
- name VARCHAR(255) NOT NULL ,
- relationid BLOB NOT NULL ,
- PRIMARY KEY (name)
+ name TEXT PRIMARY KEY NOT NULL COLLATE locale,
+ relationid BLOB NOT NULL
+ REFERENCES tbl_lexicalrelationtype (id)
);
CREATE INDEX lexicalrelationalias_irev ON tbl_lexicalrelationalias (
relationid
@@ -243,15 +246,23 @@ CREATE INDEX lexicalrelationalias_irev ON tbl_lexicalrelationalias (
-- Next are finally the relation instances
CREATE TABLE tbl_synsetrelation (
- source BLOB NOT NULL ,
- relationtype BLOB NOT NULL ,
- target BLOB NOT NULL ,
+ source BLOB NOT NULL
+ REFERENCES tbl_synset (id),
+ relationtype BLOB NOT NULL
+ REFERENCES tbl_synsetrelationtype (id),
+ target BLOB NOT NULL
+ REFERENCES tbl_synset (id),
+
PRIMARY KEY (source, relationtype, target)
);
CREATE TABLE tbl_lexicalrelation (
- source BLOB NOT NULL ,
- relationtype BLOB NOT NULL ,
- target BLOB NOT NULL ,
+ source BLOB NOT NULL
+ REFERENCES tbl_lexicalunit (id),
+ relationtype BLOB NOT NULL
+ REFERENCES tbl_lexicalrelationtype (id),
+ target BLOB NOT NULL
+ REFERENCES tbl_lexicalunit (id),
+
PRIMARY KEY (source, relationtype, target)
);
@@ -404,7 +415,7 @@ class PLWordNet(bs.PLWordNetBase):
)
lu_q = u"""
- SELECT lemma, pos.value, variant, synset
+ SELECT lemma, tbl_pos.value, variant, synset
FROM tbl_lexicalunit
JOIN tbl_pos ON tbl_pos.id = tbl_lexicalunit.pos
WHERE tbl_lexicalunit.id = ?
@@ -1397,6 +1408,78 @@ class _DBBuilder(object):
en.RelationKind.lexical,
)
+ def _show(self):
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"""
+ SELECT * FROM tbl_synset
+ """
+ )
+ print(cur.fetchall())
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"""
+ SELECT * FROM tbl_synsetrelation
+ """
+ )
+ print(cur.fetchall())
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"""
+ SELECT * FROM tbl_synsetrelationtype
+ """
+ )
+ print(cur.fetchall())
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"""
+ SELECT * FROM tbl_synsetrelationparentpart
+ """
+ )
+ print(cur.fetchall())
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"""
+ SELECT * FROM tbl_synsetrelationchildpart
+ """
+ )
+ print(cur.fetchall())
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"""
+ SELECT * FROM tbl_lexicalunit
+ """
+ )
+ print(cur.fetchall())
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"""
+ SELECT * FROM tbl_lexicalrelation
+ """
+ )
+ print(cur.fetchall())
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"""
+ SELECT * FROM tbl_lexicalrelationtype
+ """
+ )
+ print(cur.fetchall())
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"""
+ SELECT * FROM tbl_lexicalrelationparentpart
+ """
+ )
+ print(cur.fetchall())
+ with closing(self._db.cursor()) as cur:
+ cur.execute(
+ u"""
+ SELECT * FROM tbl_lexicalrelationchildpart
+ """
+ )
+ print(cur.fetchall())
+
def _insert_synset(self, syn_node):
self._db.execute(
u"""
diff --git a/tests/abstract_cases/test_unit_and_synset.py b/tests/abstract_cases/test_unit_and_synset.py
index f3885dea2aae8871ef7ba7b48fe087ee43b98a05..4ac8513a1c366532e4b362d375dbb97783733728 100644
--- a/tests/abstract_cases/test_unit_and_synset.py
+++ b/tests/abstract_cases/test_unit_and_synset.py
@@ -94,6 +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,
@@ -103,6 +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,
@@ -112,6 +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,
@@ -258,6 +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,
@@ -267,6 +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,
@@ -276,6 +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,
@@ -285,6 +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,
@@ -294,6 +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,
@@ -303,6 +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,
@@ -312,6 +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,
@@ -321,6 +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,
@@ -446,6 +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,
@@ -455,6 +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,
@@ -464,6 +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,
@@ -473,6 +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,
@@ -529,6 +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,
@@ -645,6 +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,
@@ -658,6 +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,
@@ -668,6 +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,
@@ -767,6 +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,
@@ -776,6 +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,
@@ -785,6 +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,
@@ -794,6 +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,
@@ -844,6 +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,
@@ -867,6 +891,7 @@ class ToDictTest(ut.TestCase):
),
nd.make_lexical_unit_node(
id=21,
+ legacy_id=21,
lemma=u'bbb',
pos=en.PoS.n,
variant=1,
@@ -876,6 +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,
@@ -886,6 +912,7 @@ class ToDictTest(ut.TestCase):
))
self.__lex11_dict = {
u'id': 11,
+ u'legacy_id': 11,
u'lemma': u'aaa',
u'pos': u'czasownik',
u'variant': 1,