Allow word IDs on synset members list (#256)

The OEWN uses word IDs while OMW uses sense IDs for the synset members
attribute. This meant that OEWN synset members were not being returned
in the right order.

Fixes #255

Author: goodmami

CHANGELOG.md

@@ -2,6 +2,11 @@
 
 ## [Unreleased][unreleased]
 
+## Changed
+
+* `wn.add()` allows synset members to be lexical entry IDs for rank
+  calculations ([#255])
+
 
 ## [v0.12.0]
 
@@ -771,3 +776,4 @@ abandoned, but this is an entirely new codebase.
 [#241]: https://github.com/goodmami/wn/issues/241
 [#246]: https://github.com/goodmami/wn/issues/246
 [#250]: https://github.com/goodmami/wn/issues/250
+[#255]: https://github.com/goodmami/wn/issues/255

tests/data/sense-member-order.xml

@@ -0,0 +1,33 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE LexicalResource SYSTEM "http://globalwordnet.github.io/schemas/WN-LMF-1.1.dtd">
+<LexicalResource xmlns:dc="http://purl.org/dc/elements/1.1/">
+
+<!-- duplicate ID in synsets -->
+
+  <Lexicon id="test"
+           label="Testing Sense Member Orders"
+           language="en"
+           email="[email protected]"
+           license="https://creativecommons.org/licenses/by/4.0/"
+           version="1">
+
+    <LexicalEntry id="test-foo-n">
+      <Lemma partOfSpeech="n" writtenForm="foo" />
+      <Sense id="test-01-foo-n" synset="test-01-n" />
+      <Sense id="test-02-foo-n" synset="test-02-n" />
+    </LexicalEntry>
+
+    <LexicalEntry id="test-bar-n">
+      <Lemma partOfSpeech="n" writtenForm="bar" />
+      <Sense id="test-02-bar-n" synset="test-02-n" />
+      <Sense id="test-01-bar-n" synset="test-01-n" />
+    </LexicalEntry>
+
+    <!-- sense IDs as members -->
+    <Synset id="test-01-n" ili="i12345" partOfSpeech="n" members="test-01-bar-n test-01-foo-n"/>
+    <!-- word IDs as members -->
+    <Synset id="test-02-n" ili="i12346" partOfSpeech="n" members="test-bar-n test-foo-n" />
+
+  </Lexicon>
+
+</LexicalResource>

tests/secondary_query_test.py

@@ -146,3 +146,25 @@ def test_synset_lexicalized():
 def test_synset_translate():
     assert len(wn.synset('test-en-0001-n').translate(lang='es')) == 1
     assert len(wn.synset('test-es-0001-n').translate(lang='en')) == 1
+
+
+@pytest.mark.usefixtures('uninitialized_datadir')
+def test_word_sense_order(datadir):
+    wn.add(datadir / 'sense-member-order.xml')
+    assert [s.id for s in wn.word('test-foo-n').senses()] == [
+        "test-01-foo-n", "test-02-foo-n",
+    ]
+    assert [s.id for s in wn.word('test-bar-n').senses()] == [
+        "test-02-bar-n", "test-01-bar-n",
+    ]
+
+
+@pytest.mark.usefixtures('uninitialized_datadir')
+def test_synset_member_order(datadir):
+    wn.add(datadir / 'sense-member-order.xml')
+    assert [s.id for s in wn.synset('test-01-n').senses()] == [
+        "test-01-bar-n", "test-01-foo-n",
+    ]
+    assert [s.id for s in wn.synset('test-02-n').senses()] == [
+        "test-02-bar-n", "test-02-foo-n",
+    ]

wn/_add.py

@@ -665,9 +665,11 @@ def _insert_senses(
     progress: ProgressHandler
 ) -> None:
     progress.set(status='Senses')
-    ssrank = {s: i
-              for ss in _local_synsets(synsets)
-              for i, s in enumerate(ss.get('members', []))}
+    ssrank = {
+        (ss['id'], _id): i
+        for ss in _local_synsets(synsets)
+        for i, _id in enumerate(ss.get('members', []))
+    }
     query = f'''
         INSERT INTO senses
         VALUES (null,
@@ -682,14 +684,23 @@ def _insert_senses(
     '''
     for batch in _batch(entries):
         data = [
-            (sense['id'],
-             lexid,
-             entry['id'], lexidmap.get(entry['id'], lexid),
-             i,
-             sense['synset'], lexidmap.get(sense['synset'], lexid),
-             ssrank.get(sense['id'], DEFAULT_MEMBER_RANK),
-             sense.get('lexicalized', True),
-             sense['meta'])
+            (
+                sense['id'],
+                lexid,
+                entry['id'], lexidmap.get(entry['id'], lexid),
+                i,
+                sense['synset'], lexidmap.get(sense['synset'], lexid),
+                # members can be sense or entry IDs
+                ssrank.get(
+                    (sense['synset'], sense['id']),
+                    ssrank.get(
+                        (sense['synset'], entry['id']),
+                        DEFAULT_MEMBER_RANK
+                    )
+                ),
+                sense.get('lexicalized', True),
+                sense['meta']
+            )
             for entry in batch
             for i, sense in enumerate(_local_senses(_senses(entry)))
         ]