Expand lexeme metadata and fix forms

scribe-org · Oct 22, 2024 · ca15554 · ca15554
1 parent ef5f6a4
commit ca15554
Show file tree

Hide file tree

Showing 7 changed files with 162 additions and 164 deletions.
diff --git a/src/scribe_data/check/check_query_forms.py b/src/scribe_data/check/check_query_forms.py
@@ -96,6 +96,9 @@ def check_form_label(form_text: str):
             form_label = label_match[1].strip()
             current_form_rep_label = form_label.split("Form")[0]
 
+    if not line_match:
+        return False
+
     onto_rep_pattern = r"{form_label} ontolex:representation .* ;".format(
         form_label=form_label
     )

diff --git a/src/scribe_data/language_data_extraction/Persian/verbs/query_verbs_1.sparql b/src/scribe_data/language_data_extraction/Persian/verbs/query_verbs_1.sparql
@@ -1,54 +1,49 @@
- # tool: scribe-data
+# tool: scribe-data
 # All Persian (Q9168) verbs (Q24905) and the given forms.
 # Enter this query at https://query.wikidata.org/.
 
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?infinitive
-  ?presentTenseParticiple
-  ?pastTenseParticiple
-  ?presentTenseWordStem
-  ?pastTenseWordStem
+  ?presentParticiple
+  ?pastParticiple
+  ?presentWordStem
+  ?pastWordStem
 
 WHERE {
   ?lexeme dct:language wd:Q9168;
-          wikibase:lexicalCategory wd:Q24905;
-          wikibase:lemma ?infinitiveForm.
+    wikibase:lexicalCategory wd:Q24905;
+    wikibase:lemma ?infinitive.
 
-  #MARK: infinitive
-  OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?infinitiveLexForm.
-    ?infinitiveLexForm ontolex:representation ?infinitive;
-      wikibase:grammaticalFeature wd:Q179230.
-  }
+  #MARK: Past Participle
 
-  #MARK: present tense and past participle
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?presentParticipleForm.
-    ?presentParticipleForm ontolex:representation ?presentTenseParticiple;
-      wikibase:grammaticalFeature wd:Q192613 , wd:Q814722.
-    FILTER(lang(?presentTenseParticiple) = "fa").
+    ?lexeme ontolex:lexicalForm ?presentParticipleForm .
+    ?presentParticipleForm ontolex:representation ?presentParticiple ;
+      wikibase:grammaticalFeature wd:Q192613, wd:Q814722 .
+    FILTER(lang(?presentParticiple) = "fa").
   }
 
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?pastParticipleForm.
-    ?pastParticipleForm ontolex:representation ?pastTenseParticiple;
-      wikibase:grammaticalFeature wd:Q814722 , wd:Q1994301.
-    FILTER(lang(?pastTenseParticiple) = "fa").
+    ?lexeme ontolex:lexicalForm ?pastParticipleForm .
+    ?pastParticipleForm ontolex:representation ?pastParticiple ;
+      wikibase:grammaticalFeature wd:Q814722, wd:Q1994301 .
+    FILTER(lang(?pastParticiple) = "fa").
   }
 
-  #MARK: present tense and past word stem
+  #MARK: Word Stem
+
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?presentWordStemForm.
-    ?presentWordStemForm ontolex:representation ?presentTenseWordStem;
-      wikibase:grammaticalFeature wd:Q192613 , wd:Q210523.
-    FILTER(lang(?presentTenseWordStem) = "fa").
+    ?lexeme ontolex:lexicalForm ?presentWordStemForm .
+    ?presentWordStemForm ontolex:representation ?presentWordStem ;
+      wikibase:grammaticalFeature wd:Q192613, wd:Q210523 .
+    FILTER(lang(?presentWordStem) = "fa").
   }
 
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?pastWordStemForm.
-    ?pastWordStemForm ontolex:representation ?pastTenseWordStem;
-      wikibase:grammaticalFeature wd:Q1994301, wd:Q210523.
-    FILTER(lang(?pastTenseWordStem) = "fa").
+    ?lexeme ontolex:lexicalForm ?pastWordStemForm .
+    ?pastWordStemForm ontolex:representation ?pastWordStem ;
+      wikibase:grammaticalFeature wd:Q1994301, wd:Q210523 .
+    FILTER(lang(?pastWordStem) = "fa").
   }
 }
diff --git a/src/scribe_data/language_data_extraction/Persian/verbs/query_verbs_2.sparql b/src/scribe_data/language_data_extraction/Persian/verbs/query_verbs_2.sparql
@@ -5,65 +5,59 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?infinitive
-  ?firstPersonSingularIndicativeAorist
-  ?secondPersonSingularIndicativeAorist
-  ?thirdPersonSingularIndicativeAorist
-  ?firstPersonPluralIndicativeAorist
-  ?secondPersonPluralIndicativeAorist
-  ?thirdPersonPluralIndicativeAorist
+  ?indicativeFirstPersonAoristSingular
+  ?indicativeSecondPersonAoristSingular
+  ?indicativeThirdPersonAoristSingular
+  ?indicativeFirstPersonAoristPlural
+  ?indicativeSecondPersonAoristPlural
+  ?indicativeThirdPersonAoristPlural
 
 WHERE {
   ?lexeme dct:language wd:Q9168;
-          wikibase:lexicalCategory wd:Q24905;
-          wikibase:lemma ?infinitiveForm.
+    wikibase:lexicalCategory wd:Q24905;
+    wikibase:lemma ?infinitive.
 
-  #MARK: infinitive
-  OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?infinitiveLexForm.
-    ?infinitiveLexForm ontolex:representation ?infinitive;
-      wikibase:grammaticalFeature wd:Q179230.
-  }
+  #MARK: Indicative Aorist
 
-  #MARK: indicative aorist
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?firstSingularAoristForm.
-    ?firstSingularAoristForm ontolex:representation ?firstPersonSingularIndicativeAorist;
-      wikibase:grammaticalFeature wd:Q21714344 , wd:Q110786 , wd:Q682111 , wd:Q216497.
-    FILTER(lang(?firstPersonSingularIndicativeAorist) = "fa").
+    ?lexeme ontolex:lexicalForm ?indicativeFirstPersonAoristSingularForm .
+    ?indicativeFirstPersonAoristSingularForm ontolex:representation ?indicativeFirstPersonAoristSingular ;
+      wikibase:grammaticalFeature wd:Q21714344, wd:Q110786, wd:Q682111, wd:Q216497 .
+    FILTER(lang(?indicativeFirstPersonAoristSingular) = "fa").
   }
 
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?secondSingularAoristForm.
-    ?secondSingularAoristForm ontolex:representation ?secondPersonSingularIndicativeAorist;
-      wikibase:grammaticalFeature wd:Q51929049 , wd:Q110786, wd:Q682111 , wd:Q216497.
-    FILTER(lang(?secondPersonSingularIndicativeAorist) = "fa").
+    ?lexeme ontolex:lexicalForm ?indicativeSecondPersonAoristSingularForm .
+    ?indicativeSecondPersonAoristSingularForm ontolex:representation ?indicativeSecondPersonAoristSingular ;
+      wikibase:grammaticalFeature wd:Q51929049, wd:Q110786, wd:Q682111, wd:Q216497 .
+    FILTER(lang(?indicativeSecondPersonAoristSingular) = "fa").
   }
 
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?thirdSingularAoristForm.
-    ?thirdSingularAoristForm ontolex:representation ?thirdPersonSingularIndicativeAorist;
-      wikibase:grammaticalFeature wd:Q51929074 , wd:Q110786 , wd:Q682111 , wd:Q216497.
-    FILTER(lang(?thirdPersonSingularIndicativeAorist) = "fa").
+    ?lexeme ontolex:lexicalForm ?indicativeThirdPersonAoristSingularForm .
+    ?indicativeThirdPersonAoristSingularForm ontolex:representation ?indicativeThirdPersonAoristSingular ;
+      wikibase:grammaticalFeature wd:Q51929074, wd:Q110786, wd:Q682111, wd:Q216497 .
+    FILTER(lang(?indicativeThirdPersonAoristSingular) = "fa").
   }
 
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?firstPluralAoristForm.
-    ?firstPluralAoristForm ontolex:representation ?firstPersonPluralIndicativeAorist;
-      wikibase:grammaticalFeature wd:Q21714344 ,wd:Q146786 , wd:Q682111 , wd:Q216497.
-    FILTER(lang(?firstPersonPluralIndicativeAorist) = "fa").
+    ?lexeme ontolex:lexicalForm ?indicativeFirstPersonAoristPluralForm .
+    ?indicativeFirstPersonAoristPluralForm ontolex:representation ?indicativeFirstPersonAoristPlural ;
+      wikibase:grammaticalFeature wd:Q21714344, wd:Q146786, wd:Q682111, wd:Q216497 .
+    FILTER(lang(?indicativeFirstPersonAoristPlural) = "fa").
   }
 
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?secondPluralAoristForm.
-    ?secondPluralAoristForm ontolex:representation ?secondPersonPluralIndicativeAorist;
-      wikibase:grammaticalFeature wd:Q51929049 ,wd:Q146786 , wd:Q682111 ,wd:Q216497.
-    FILTER(lang(?secondPersonPluralIndicativeAorist) = "fa").
+    ?lexeme ontolex:lexicalForm ?indicativeSecondPersonAoristPluralForm .
+    ?indicativeSecondPersonAoristPluralForm ontolex:representation ?indicativeSecondPersonAoristPlural ;
+      wikibase:grammaticalFeature wd:Q51929049, wd:Q146786, wd:Q682111, wd:Q216497 .
+    FILTER(lang(?indicativeSecondPersonAoristPlural) = "fa").
   }
 
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?thirdPluralAoristForm.
-    ?thirdPluralAoristForm ontolex:representation ?thirdPersonPluralIndicativeAorist;
-      wikibase:grammaticalFeature wd:Q51929074 , wd:Q146786 ,wd:Q682111 ,wd:Q216497.
-    FILTER(lang(?thirdPersonPluralIndicativeAorist) = "fa").
+    ?lexeme ontolex:lexicalForm ?indicativeThirdPersonAoristPluralForm .
+    ?indicativeThirdPersonAoristPluralForm ontolex:representation ?indicativeThirdPersonAoristPlural ;
+      wikibase:grammaticalFeature wd:Q51929074, wd:Q146786, wd:Q682111, wd:Q216497 .
+    FILTER(lang(?indicativeThirdPersonAoristPlural) = "fa").
   }
 }
diff --git a/src/scribe_data/language_data_extraction/Persian/verbs/query_verbs_3.sparql b/src/scribe_data/language_data_extraction/Persian/verbs/query_verbs_3.sparql
@@ -5,56 +5,53 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?infinitive
-  ?firstPersonSingularPastTenseIndicative
-  ?secondPersonSingularPastTenseIndicative
-  ?thirdPersonSingularPastTenseIndicative
-  ?firstPersonPluralPastTenseIndicative
-  ?secondPersonPluralPastTenseIndicative
-  ?thirdPersonPluralPastTenseIndicative
+  ?indicativePastFirstPersonSingular
+  ?indicativePastSecondPersonSingular
+  ?indicativePastThirdPersonSingular
+  ?indicativePastFirstPersonPlural
+  ?indicativePastSecondPersonPlural
+  ?indicativePastThirdPersonPlural
 
 WHERE {
-  ?lexeme dct:language wd:Q9168 ;
-          wikibase:lexicalCategory wd:Q24905 .
-
-  #MARK: infinitive
-  ?lexeme ontolex:lexicalForm ?infinitiveForm .
-  ?infinitiveForm ontolex:representation ?infinitive ;
-    wikibase:grammaticalFeature wd:Q179230 .
+  ?lexeme dct:language wd:Q9168;
+    wikibase:lexicalCategory wd:Q24905;
+    wikibase:lemma ?infinitive.
 
   # MARK: Past and Present Indicative
+
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?firstPersonSingularPastTenseForm .
-    ?firstPersonSingularPastTenseForm ontolex:representation ?firstPersonSingularPastTenseIndicative ;
-      wikibase:grammaticalFeature wd:Q21714344, wd:Q110786, wd:Q1994301 , wd:Q682111;
+    ?lexeme ontolex:lexicalForm ?indicativePastFirstPersonSingularForm .
+    ?indicativePastFirstPersonSingularForm ontolex:representation ?indicativePastFirstPersonSingular ;
+      wikibase:grammaticalFeature wd:Q21714344, wd:Q110786, wd:Q1994301, wd:Q682111 .
   }
 
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?secondPersonSingularPastTenseForm .
-    ?secondPersonSingularPastTenseForm ontolex:representation ?secondPersonSingularPastTenseIndicative ;
-      wikibase:grammaticalFeature wd:Q51929049, wd:Q110786, wd:Q1994301 , wd:Q682111;
+    ?lexeme ontolex:lexicalForm ?indicativePastSecondPersonSingularForm .
+    ?indicativePastSecondPersonSingularForm ontolex:representation ?indicativePastSecondPersonSingular ;
+      wikibase:grammaticalFeature wd:Q51929049, wd:Q110786, wd:Q1994301, wd:Q682111 .
   }
 
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?thirdPersonSingularPastTenseForm .
-    ?thirdPersonSingularPastTenseForm ontolex:representation ?thirdPersonSingularPastTenseIndicative ;
-      wikibase:grammaticalFeature wd:Q51929074, wd:Q110786, wd:Q1994301 , wd:Q682111;
+    ?lexeme ontolex:lexicalForm ?indicativePastThirdPersonSingularForm .
+    ?indicativePastThirdPersonSingularForm ontolex:representation ?indicativePastThirdPersonSingular ;
+      wikibase:grammaticalFeature wd:Q51929074, wd:Q110786, wd:Q1994301, wd:Q682111 .
   }
 
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?firstPersonPluralPastTenseForm .
-    ?firstPersonPluralPastTenseForm ontolex:representation ?firstPersonPluralPastTenseIndicative ;
-      wikibase:grammaticalFeature wd:Q21714344, wd:Q146786, wd:Q1994301 , wd:Q682111;
+    ?lexeme ontolex:lexicalForm ?indicativePastFirstPersonPluralForm .
+    ?indicativePastFirstPersonPluralForm ontolex:representation ?indicativePastFirstPersonPlural ;
+      wikibase:grammaticalFeature wd:Q21714344, wd:Q146786, wd:Q1994301, wd:Q682111 .
   }
 
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?secondPersonPluralPastTenseForm .
-    ?secondPersonPluralPastTenseForm ontolex:representation ?secondPersonPluralPastTenseIndicative ;
-      wikibase:grammaticalFeature wd:Q51929049, wd:Q146786, wd:Q1994301 , wd:Q682111;
+    ?lexeme ontolex:lexicalForm ?indicativePastSecondPersonPluralForm .
+    ?indicativePastSecondPersonPluralForm ontolex:representation ?indicativePastSecondPersonPlural ;
+      wikibase:grammaticalFeature wd:Q51929049, wd:Q146786, wd:Q1994301, wd:Q682111 .
   }
 
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?thirdPersonPluralPastTenseForm .
-    ?thirdPersonPluralPastTenseForm ontolex:representation ?thirdPersonPluralPastTenseIndicative ;
-      wikibase:grammaticalFeature wd:Q51929074, wd:Q146786, wd:Q1994301 , wd:Q682111;
+    ?lexeme ontolex:lexicalForm ?indicativePastThirdPersonPluralForm .
+    ?indicativePastThirdPersonPluralForm ontolex:representation ?indicativePastThirdPersonPlural ;
+      wikibase:grammaticalFeature wd:Q51929074, wd:Q146786, wd:Q1994301, wd:Q682111 .
   }
 }
diff --git a/src/scribe_data/language_data_extraction/Persian/verbs/query_verbs_4.sparql b/src/scribe_data/language_data_extraction/Persian/verbs/query_verbs_4.sparql
@@ -5,51 +5,53 @@
 SELECT
   (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?infinitive
-  ?firstPersonSingPresentPerfect
-  ?secondPersonSingPresentPerfect
-  ?thirdPersonSingPresentPerfect
-  ?firstPersonPluralPresentPerfect
-  ?secondPersonPluralPresentPerfect
-  ?thirdPersonPluralPresentPerfect
+  ?presentPerfectFirstPersonSingular
+  ?presentPerfectSecondPersonSingular
+  ?presentPerfectThirdPersonSingular
+  ?presentPerfectFirstPersonPlural
+  ?presentPerfectSecondPersonPlural
+  ?presentPerfectThirdPersonPlural
 
 WHERE {
   ?lexeme dct:language wd:Q9168;
-          wikibase:lexicalCategory wd:Q24905.
+    wikibase:lexicalCategory wd:Q24905;
+    wikibase:lemma ?infinitive.
 
-  # MARK: Infinitive
-  ?lexeme ontolex:lexicalForm ?infinitiveForm .
-  ?infinitiveForm ontolex:representation ?infinitive ;
-    wikibase:grammaticalFeature wd:Q179230 .
+  # MARK: Present Perfect
 
-  # MARK: Present Perfect Forms
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?firstPersonSingPresentPerfectForm.
-    ?firstPersonSingPresentPerfectForm ontolex:representation ?firstPersonSingPresentPerfect;
-      wikibase:grammaticalFeature wd:Q625420 , wd:Q21714344 , wd:Q192613 , wd:Q110786.
+    ?lexeme ontolex:lexicalForm ?presentPerfectFirstPersonSingularForm .
+    ?presentPerfectFirstPersonSingularForm ontolex:representation ?presentPerfectFirstPersonSingular ;
+      wikibase:grammaticalFeature wd:Q625420, wd:Q21714344, wd:Q192613, wd:Q110786 .
   }
+
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?secondPersonSingPresentPerfectForm.
-    ?secondPersonSingPresentPerfectForm ontolex:representation ?secondPersonSingPresentPerfect;
-      wikibase:grammaticalFeature wd:Q625420 , wd:Q51929049 ,wd:Q192613 , wd:Q110786.
+    ?lexeme ontolex:lexicalForm ?presentPerfectSecondPersonSingularForm .
+    ?presentPerfectSecondPersonSingularForm ontolex:representation ?presentPerfectSecondPersonSingular ;
+      wikibase:grammaticalFeature wd:Q625420, wd:Q51929049, wd:Q192613, wd:Q110786 .
   }
+
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?thirdPersonSingPresentPerfectForm.
-    ?thirdPersonSingPresentPerfectForm ontolex:representation ?thirdPersonSingPresentPerfect;
-      wikibase:grammaticalFeature wd:Q625420 , wd:Q51929074 ,wd:Q192613 , wd:Q110786.
+    ?lexeme ontolex:lexicalForm ?presentPerfectThirdPersonSingularForm .
+    ?presentPerfectThirdPersonSingularForm ontolex:representation ?presentPerfectThirdPersonSingular ;
+      wikibase:grammaticalFeature wd:Q625420, wd:Q51929074, wd:Q192613, wd:Q110786 .
   }
+
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?firstPersonPluralPresentPerfectForm.
-    ?firstPersonPluralPresentPerfectForm ontolex:representation ?firstPersonPluralPresentPerfect;
-      wikibase:grammaticalFeature wd:Q625420 , wd:Q21714344 ,wd:Q192613 , wd:Q146786.
+    ?lexeme ontolex:lexicalForm ?presentPerfectFirstPersonPluralForm .
+    ?presentPerfectFirstPersonPluralForm ontolex:representation ?presentPerfectFirstPersonPlural ;
+      wikibase:grammaticalFeature wd:Q625420, wd:Q21714344, wd:Q192613, wd:Q146786 .
   }
+
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?secondPersonPluralPresentPerfectForm.
-    ?secondPersonPluralPresentPerfectForm ontolex:representation ?secondPersonPluralPresentPerfect;
-      wikibase:grammaticalFeature wd:Q625420 , wd:Q51929049 ,wd:Q192613 , wd:Q146786.
+    ?lexeme ontolex:lexicalForm ?presentPerfectSecondPersonPluralForm .
+    ?presentPerfectSecondPersonPluralForm ontolex:representation ?presentPerfectSecondPersonPlural ;
+      wikibase:grammaticalFeature wd:Q625420, wd:Q51929049, wd:Q192613, wd:Q146786 .
   }
+
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?thirdPersonPluralPresentPerfectForm.
-    ?thirdPersonPluralPresentPerfectForm ontolex:representation ?thirdPersonPluralPresentPerfect;
-      wikibase:grammaticalFeature wd:Q625420 , wd:Q51929074 ,wd:Q192613 , wd:Q146786.
+    ?lexeme ontolex:lexicalForm ?presentPerfectThirdPersonPluralForm .
+    ?presentPerfectThirdPersonPluralForm ontolex:representation ?presentPerfectThirdPersonPlural ;
+      wikibase:grammaticalFeature wd:Q625420, wd:Q51929074, wd:Q192613, wd:Q146786 .
   }
 }