Merge pull request #566 from axif0/lastModified

Format Query Data adding Last Modified Date
scribe-org · Feb 2, 2025 · 7f5c39c · 7f5c39c
2 parents 03101e7 + 8f6baeb
commit 7f5c39c
Show file tree

Hide file tree

Showing 299 changed files with 1,073 additions and 474 deletions.
diff --git a/src/scribe_data/wikidata/format_data.py b/src/scribe_data/wikidata/format_data.py
@@ -51,16 +51,18 @@ def format_data(
 
     for data_vals in data_list:
         lexeme_id = data_vals["lexemeID"]
+        modified_date = data_vals["lastModified"]
 
         if lexeme_id not in data_formatted:
-            data_formatted[lexeme_id] = {}
+            data_formatted[lexeme_id] = {modified_date: {}}
 
         # Reverse to make sure that we're getting the same order as the query.
         query_identifiers = list(reversed(data_vals.keys()))
         query_identifiers.remove("lexemeID")
+        query_identifiers.remove("lastModified")
 
         for k in query_identifiers:
-            data_formatted[lexeme_id][k] = data_vals[k]
+            data_formatted[lexeme_id][modified_date][k] = data_vals[k]
 
     data_formatted = collections.OrderedDict(sorted(data_formatted.items()))
 

diff --git a/src/scribe_data/wikidata/language_data_extraction/arabic/adjectives/query_adjectives.sparql b/src/scribe_data/wikidata/language_data_extraction/arabic/adjectives/query_adjectives.sparql
@@ -33,11 +33,13 @@ SELECT
   ?pausalMasculineIndefiniteSingular
   ?pausalMasculineIndefinitePlural
   ?pausalMasculineIndefiniteDual
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q13955 ;
     wikibase:lexicalCategory wd:Q34698 ;
-    wikibase:lemma ?adjective .
+    wikibase:lemma ?adjective ;
+    schema:dateModified ?lastModified .
 
   # MARK: Nominative
 

diff --git a/src/scribe_data/wikidata/language_data_extraction/arabic/adverbs/query_adverbs.sparql b/src/scribe_data/wikidata/language_data_extraction/arabic/adverbs/query_adverbs.sparql
@@ -5,9 +5,11 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?adverb
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q13955 ;
     wikibase:lexicalCategory wd:Q380057 ;
-    wikibase:lemma ?adverb .
+    wikibase:lemma ?adverb ;
+    schema:dateModified ?lastModified .
 }
diff --git a/src/scribe_data/wikidata/language_data_extraction/arabic/nouns/query_nouns.sparql b/src/scribe_data/wikidata/language_data_extraction/arabic/nouns/query_nouns.sparql
@@ -5,6 +5,7 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?noun
+  ?lastModified
 
   ?nominativeFeminineIndefiniteSingular
   ?nominativeFeminineIndefinitePlural
@@ -37,7 +38,8 @@ SELECT
 WHERE {
   ?lexeme dct:language wd:Q13955 ;
     wikibase:lexicalCategory wd:Q1084 ;
-    wikibase:lemma ?noun .
+    wikibase:lemma ?noun ;
+    schema:dateModified ?lastModified .
 
   # MARK: Nominative
 

diff --git a/...ribe_data/wikidata/language_data_extraction/arabic/proper_nouns/query_proper_nouns.sparql b/...ribe_data/wikidata/language_data_extraction/arabic/proper_nouns/query_proper_nouns.sparql
@@ -5,9 +5,11 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?properNoun
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q13955 ;
     wikibase:lexicalCategory wd:Q147276 ;
-    wikibase:lemma ?properNoun .
+    wikibase:lemma ?properNoun ;
+    schema:dateModified ?lastModified .
 }
diff --git a/src/scribe_data/wikidata/language_data_extraction/arabic/verbs/query_verbs_1.sparql b/src/scribe_data/wikidata/language_data_extraction/arabic/verbs/query_verbs_1.sparql
@@ -5,6 +5,7 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?verb
+  ?lastModified
 
   ?indicativeFirstPersonSingularFiilMudari
   ?indicativeFirstPersonPluralFiilMudari
@@ -23,7 +24,8 @@ SELECT
 WHERE {
   ?lexeme dct:language wd:Q13955 ;
     wikibase:lexicalCategory wd:Q24905 ;
-    wikibase:lemma ?verb .
+    wikibase:lemma ?verb ;
+    schema:dateModified ?lastModified .
 
   # MARK: Indicative Present
   OPTIONAL {

diff --git a/src/scribe_data/wikidata/language_data_extraction/arabic/verbs/query_verbs_2.sparql b/src/scribe_data/wikidata/language_data_extraction/arabic/verbs/query_verbs_2.sparql
@@ -4,26 +4,28 @@
 
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
-?verb
+  ?verb
+  ?lastModified
 
-?activePerformativeFirstPersonSingular
-?activePerformativeFirstPersonPlural
-?activePerformativeSecondPersonDual
+  ?activePerformativeFirstPersonSingular
+  ?activePerformativeFirstPersonPlural
+  ?activePerformativeSecondPersonDual
 
-?feminineActivePerformativeSecondPersonSingular
-?feminineActivePerformativeSecondPersonPlural
-?feminineActivePerformativeThirdPersonSingular
-?feminineActivePerformativeThirdPersonDual
+  ?feminineActivePerformativeSecondPersonSingular
+  ?feminineActivePerformativeSecondPersonPlural
+  ?feminineActivePerformativeThirdPersonSingular
+  ?feminineActivePerformativeThirdPersonDual
 
-?masculineActivePerformativeSecondPersonSingular
-?masculineActivePerformativeSecondPersonPlural
-?masculineActivePerformativeThirdPersonSingular
-?masculineActivePerformativeThirdPersonDual
+  ?masculineActivePerformativeSecondPersonSingular
+  ?masculineActivePerformativeSecondPersonPlural
+  ?masculineActivePerformativeThirdPersonSingular
+  ?masculineActivePerformativeThirdPersonDual
 
 WHERE {
   ?lexeme dct:language wd:Q13955 ;
     wikibase:lexicalCategory wd:Q24905 ;
-    wikibase:lemma ?verb .
+    wikibase:lemma ?verb ;
+    schema:dateModified ?lastModified .
 
   # MARK: Performative Past
 

diff --git a/src/scribe_data/wikidata/language_data_extraction/basque/adjectives/query_adjectives.sparql b/src/scribe_data/wikidata/language_data_extraction/basque/adjectives/query_adjectives.sparql
@@ -5,9 +5,11 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?adjective
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q8752 ;
     wikibase:lexicalCategory wd:Q34698 ;
-    wikibase:lemma ?adjective .
+    wikibase:lemma ?adjective ;
+    schema:dateModified ?lastModified .
 }
diff --git a/src/scribe_data/wikidata/language_data_extraction/basque/adverbs/query_adverbs.sparql b/src/scribe_data/wikidata/language_data_extraction/basque/adverbs/query_adverbs.sparql
@@ -5,9 +5,11 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?adverb
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q8752;
     wikibase:lexicalCategory wd:Q380057 ;
-    wikibase:lemma ?adverb .
+    wikibase:lemma ?adverb ;
+    schema:dateModified ?lastModified .
 }
diff --git a/src/scribe_data/wikidata/language_data_extraction/basque/nouns/query_nouns.sparql b/src/scribe_data/wikidata/language_data_extraction/basque/nouns/query_nouns.sparql
@@ -7,11 +7,13 @@ SELECT
   ?absIndefinite
   ?absolutiveSingular
   ?absolutivePlural
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q8752 ;
     wikibase:lexicalCategory wd:Q1084 ;
-    wikibase:lemma ?absIndefinite .
+    wikibase:lemma ?absIndefinite ;
+    schema:dateModified ?lastModified .
 
   # MARK: Absolutive Singular
 

diff --git a/...cribe_data/wikidata/language_data_extraction/basque/prepositions/query_preposition.sparql b/...cribe_data/wikidata/language_data_extraction/basque/prepositions/query_preposition.sparql
@@ -5,9 +5,11 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?preposition
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q8752 ;
     wikibase:lexicalCategory wd:Q4833830 ;
-    wikibase:lemma ?preposition .
+    wikibase:lemma ?preposition ;
+    schema:dateModified ?lastModified .
 }
diff --git a/...ribe_data/wikidata/language_data_extraction/basque/proper_nouns/query_proper_nouns.sparql b/...ribe_data/wikidata/language_data_extraction/basque/proper_nouns/query_proper_nouns.sparql
@@ -5,9 +5,11 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?properNoun
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q8752 ;
     wikibase:lexicalCategory wd:Q147276 ;
-    wikibase:lemma ?properNoun .
+    wikibase:lemma ?properNoun ;
+    schema:dateModified ?lastModified .
 }
diff --git a/src/scribe_data/wikidata/language_data_extraction/basque/verbs/query_verbs.sparql b/src/scribe_data/wikidata/language_data_extraction/basque/verbs/query_verbs.sparql
@@ -10,13 +10,15 @@ SELECT
   ?imperfective
   ?nominalized
   ?participle
+  ?lastModified
 
 WHERE {
   # MARK: Infinitive
 
   ?lexeme dct:language wd:Q8752 ;
     wikibase:lexicalCategory wd:Q24905 ;
-    wikibase:lemma ?infinitive .
+    wikibase:lemma ?infinitive ;
+    schema:dateModified ?lastModified .
 
   # MARK: Future
 

diff --git a/src/scribe_data/wikidata/language_data_extraction/bengali/adjectives/query_adjectives.sparql b/src/scribe_data/wikidata/language_data_extraction/bengali/adjectives/query_adjectives.sparql
@@ -5,9 +5,11 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?adjective
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q9610 ;
     wikibase:lexicalCategory wd:Q34698 ;
-    wikibase:lemma ?adjective .
+    wikibase:lemma ?adjective ;
+    schema:dateModified ?lastModified .
 }
diff --git a/src/scribe_data/wikidata/language_data_extraction/bengali/adverbs/query_adverbs.sparql b/src/scribe_data/wikidata/language_data_extraction/bengali/adverbs/query_adverbs.sparql
@@ -5,9 +5,11 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?adverb
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q9610 ;
     wikibase:lexicalCategory wd:Q380057 ;
-    wikibase:lemma ?adverb .
+    wikibase:lemma ?adverb ;
+    schema:dateModified ?lastModified .
 }
diff --git a/src/scribe_data/wikidata/language_data_extraction/bengali/nouns/query_nouns.sparql b/src/scribe_data/wikidata/language_data_extraction/bengali/nouns/query_nouns.sparql
@@ -8,10 +8,12 @@ SELECT
   ?genitive
   ?accusative
   ?locative
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q9610 ;
     wikibase:lexicalCategory wd:Q1084 ;
+    schema:dateModified ?lastModified .
 
   # MARK: Nominative
 

diff --git a/...e_data/wikidata/language_data_extraction/bengali/postpositions/query_postpositions.sparql b/...e_data/wikidata/language_data_extraction/bengali/postpositions/query_postpositions.sparql
@@ -7,11 +7,13 @@ SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?preposition
   ?grammaticalCase
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q9610 ;
     wikibase:lexicalCategory wd:Q161873 ;
-    wikibase:lemma ?preposition .
+    wikibase:lemma ?preposition ;
+    schema:dateModified ?lastModified .
 
   # MARK: Case
 

diff --git a/...ibe_data/wikidata/language_data_extraction/bengali/prepositions/query_prepositions.sparql b/...ibe_data/wikidata/language_data_extraction/bengali/prepositions/query_prepositions.sparql
@@ -6,11 +6,13 @@ SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?preposition
   ?grammaticalCase
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q9610 ;
     wikibase:lexicalCategory wd:Q4833830 ;
-    wikibase:lemma ?preposition .
+    wikibase:lemma ?preposition ;
+    schema:dateModified ?lastModified .
 
   # MARK: Case
 

diff --git a/...ibe_data/wikidata/language_data_extraction/bengali/proper_nouns/query_proper_nouns.sparql b/...ibe_data/wikidata/language_data_extraction/bengali/proper_nouns/query_proper_nouns.sparql
@@ -8,10 +8,12 @@ SELECT
   ?genitive
   ?accusative
   ?locative
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q9610 ;
     wikibase:lexicalCategory wd:Q147276 ;
+    schema:dateModified ?lastModified .
 
   # MARK: Nominative
 

diff --git a/src/scribe_data/wikidata/language_data_extraction/bengali/verbs/query_verbs.sparql b/src/scribe_data/wikidata/language_data_extraction/bengali/verbs/query_verbs.sparql
@@ -5,9 +5,11 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?verb
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q9610 ;
     wikibase:lexicalCategory wd:Q24905 ;
-    wikibase:lemma ?verb .
+    wikibase:lemma ?verb ;
+    schema:dateModified ?lastModified .
 }
diff --git a/...ata/wikidata/language_data_extraction/chinese/mandarin/adjectives/query_adjectives.sparql b/...ata/wikidata/language_data_extraction/chinese/mandarin/adjectives/query_adjectives.sparql
@@ -5,9 +5,11 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?adjective
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q727694 ;
     wikibase:lexicalCategory wd:Q34698 ;
-    wikibase:lemma ?adjective .
+    wikibase:lemma ?adjective ;
+    schema:dateModified ?lastModified .
 }
diff --git a/...ribe_data/wikidata/language_data_extraction/chinese/mandarin/adverbs/query_adverbs.sparql b/...ribe_data/wikidata/language_data_extraction/chinese/mandarin/adverbs/query_adverbs.sparql
@@ -5,10 +5,12 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?adverb
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q727694 ;
     wikibase:lexicalCategory wd:Q380057 ;
-    wikibase:lemma ?adverb .
+    wikibase:lemma ?adverb ;
+    schema:dateModified ?lastModified .
     FILTER(LANG(?adverb) = "zh")
 }
diff --git a/src/scribe_data/wikidata/language_data_extraction/chinese/mandarin/nouns/query_nouns.sparql b/src/scribe_data/wikidata/language_data_extraction/chinese/mandarin/nouns/query_nouns.sparql
@@ -5,9 +5,11 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?noun
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q727694 ;
     wikibase:lexicalCategory wd:Q1084 ;
-    wikibase:lemma ?noun .
+    wikibase:lemma ?noun ;
+    schema:dateModified ?lastModified .
 }
diff --git a/...wikidata/language_data_extraction/chinese/mandarin/prepositions/query_prepositions.sparql b/...wikidata/language_data_extraction/chinese/mandarin/prepositions/query_prepositions.sparql
@@ -5,9 +5,11 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?preposition
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q727694 ;
     wikibase:lexicalCategory wd:Q4833830 ;
-    wikibase:lemma ?preposition .
+    wikibase:lemma ?preposition ;
+    schema:dateModified ?lastModified .
 }
diff --git a/...wikidata/language_data_extraction/chinese/mandarin/proper_nouns/query_proper_nouns.sparql b/...wikidata/language_data_extraction/chinese/mandarin/proper_nouns/query_proper_nouns.sparql
@@ -5,9 +5,11 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?properNoun
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q727694 ;
     wikibase:lexicalCategory wd:Q147276 ;
-    wikibase:lemma ?properNoun .
+    wikibase:lemma ?properNoun ;
+    schema:dateModified ?lastModified .
 }
diff --git a/src/scribe_data/wikidata/language_data_extraction/chinese/mandarin/verbs/query_verbs.sparql b/src/scribe_data/wikidata/language_data_extraction/chinese/mandarin/verbs/query_verbs.sparql
@@ -5,9 +5,11 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?verb
+  ?lastModified
 
 WHERE {
   ?lexeme dct:language wd:Q727694 ;
     wikibase:lexicalCategory wd:Q24905 ;
-    wikibase:lemma ?verb .
+    wikibase:lemma ?verb ;
+    schema:dateModified ?lastModified .
 }