Merge pull request #485 from VNW22/Update-Check-Query-Forms

Update check query forms

src/scribe_data/check/check_query_forms.py

@@ -162,6 +162,34 @@ def extract_form_qids(form_text: str):
         return [q.split("wd:")[1].split(" .")[0] for q in match[0].split(", ")]
 
 
+# MARK: Punctuation
+
+
+def check_query_formatting(form_text: str):
+    """
+    Checks the formatting of the given SPARQL query text for common formatting issues.
+
+    Parameters
+    ----------
+        query_text : str
+            The SPARQL query text to check.
+
+    Returns
+    -------
+        bool
+            Whether there are formatting errors with the query.
+    """
+    # Check for spaces before commas that should not exist.
+    if re.search(r"\s,", form_text):
+        return False
+
+    # Check for non space characters before periods and semicolons that should not exist.
+    if re.search(r"\S[.;]", form_text):
+        return False
+
+    return True
+
+
 # MARK: Correct Label
 
 
@@ -450,25 +478,33 @@ def check_query_forms() -> None:
                     "ontolex:lexicalForm" in form_text
                     and "ontolex:representation" in form_text
                 ):
+                    correct_formatting = check_query_formatting(form_text=form_text)
                     form_rep_label = extract_form_rep_label(form_text=form_text)
                     check = check_form_label(form_text=form_text)
                     qids = extract_form_qids(form_text=form_text)
                     correct_form_rep_label = return_correct_form_label(qids=qids)
 
                     query_form_check_dict[form_rep_label] = {
                         "form_rep_match": check,
+                        "correct_formatting": correct_formatting,
                         "qids": qids,
                         "correct_form_rep_label": correct_form_rep_label,
                     }
 
             if query_form_check_dict:
                 incorrect_query_labels = []
-                for k in query_form_check_dict:
-                    if k != query_form_check_dict[k]["correct_form_rep_label"]:
+                for k, v in query_form_check_dict.items():
+                    if k != v["correct_formatting"] is False:
+                        incorrect_query_labels.append(
+                            (
+                                k,
+                                "Invalid query formatting found - please put spaces before all periods and semicolons and also remove spaces before commas.",
+                            )
+                        )
+                    elif k != query_form_check_dict[k]["correct_form_rep_label"]:
                         incorrect_query_labels.append(
                             (k, query_form_check_dict[k]["correct_form_rep_label"])
                         )
-
                     elif query_form_check_dict[k]["form_rep_match"] is False:
                         incorrect_query_labels.append(
                             (k, "Form and representation labels don't match")

src/scribe_data/wikidata/language_data_extraction/persian/verbs/query_verbs_1.sparql

@@ -11,24 +11,24 @@ SELECT
   ?pastWordStem
 
 WHERE {
-  ?lexeme dct:language wd:Q9168;
-    wikibase:lexicalCategory wd:Q24905;
-    wikibase:lemma ?infinitive.
+  ?lexeme dct:language wd:Q9168 ;
+    wikibase:lexicalCategory wd:Q24905 ;
+    wikibase:lemma ?infinitive .
 
   #MARK: Past Participle
 
   OPTIONAL {
     ?lexeme ontolex:lexicalForm ?presentParticipleForm .
     ?presentParticipleForm ontolex:representation ?presentParticiple ;
       wikibase:grammaticalFeature wd:Q192613, wd:Q814722 .
-    FILTER(lang(?presentParticiple) = "fa").
+    FILTER(lang(?presentParticiple) = "fa") .
   }
 
   OPTIONAL {
     ?lexeme ontolex:lexicalForm ?pastParticipleForm .
     ?pastParticipleForm ontolex:representation ?pastParticiple ;
       wikibase:grammaticalFeature wd:Q814722, wd:Q1994301 .
-    FILTER(lang(?pastParticiple) = "fa").
+    FILTER(lang(?pastParticiple) = "fa") .
   }
 
   #MARK: Word Stem
@@ -37,13 +37,13 @@ WHERE {
     ?lexeme ontolex:lexicalForm ?presentWordStemForm .
     ?presentWordStemForm ontolex:representation ?presentWordStem ;
       wikibase:grammaticalFeature wd:Q192613, wd:Q210523 .
-    FILTER(lang(?presentWordStem) = "fa").
+    FILTER(lang(?presentWordStem) = "fa") .
   }
 
   OPTIONAL {
     ?lexeme ontolex:lexicalForm ?pastWordStemForm .
     ?pastWordStemForm ontolex:representation ?pastWordStem ;
       wikibase:grammaticalFeature wd:Q1994301, wd:Q210523 .
-    FILTER(lang(?pastWordStem) = "fa").
+    FILTER(lang(?pastWordStem) = "fa") .
   }
 }

src/scribe_data/wikidata/language_data_extraction/persian/verbs/query_verbs_2.sparql

@@ -13,51 +13,51 @@ SELECT
   ?indicativeThirdPersonAoristPlural
 
 WHERE {
-  ?lexeme dct:language wd:Q9168;
-    wikibase:lexicalCategory wd:Q24905;
-    wikibase:lemma ?infinitive.
+  ?lexeme dct:language wd:Q9168 ;
+    wikibase:lexicalCategory wd:Q24905 ;
+    wikibase:lemma ?infinitive .
 
   #MARK: Indicative Aorist
 
   OPTIONAL {
     ?lexeme ontolex:lexicalForm ?indicativeFirstPersonAoristSingularForm .
     ?indicativeFirstPersonAoristSingularForm ontolex:representation ?indicativeFirstPersonAoristSingular ;
       wikibase:grammaticalFeature wd:Q21714344, wd:Q110786, wd:Q682111, wd:Q216497 .
-    FILTER(lang(?indicativeFirstPersonAoristSingular) = "fa").
+    FILTER(lang(?indicativeFirstPersonAoristSingular) = "fa") .
   }
 
   OPTIONAL {
     ?lexeme ontolex:lexicalForm ?indicativeSecondPersonAoristSingularForm .
     ?indicativeSecondPersonAoristSingularForm ontolex:representation ?indicativeSecondPersonAoristSingular ;
       wikibase:grammaticalFeature wd:Q51929049, wd:Q110786, wd:Q682111, wd:Q216497 .
-    FILTER(lang(?indicativeSecondPersonAoristSingular) = "fa").
+    FILTER(lang(?indicativeSecondPersonAoristSingular) = "fa") .
   }
 
   OPTIONAL {
     ?lexeme ontolex:lexicalForm ?indicativeThirdPersonAoristSingularForm .
     ?indicativeThirdPersonAoristSingularForm ontolex:representation ?indicativeThirdPersonAoristSingular ;
       wikibase:grammaticalFeature wd:Q51929074, wd:Q110786, wd:Q682111, wd:Q216497 .
-    FILTER(lang(?indicativeThirdPersonAoristSingular) = "fa").
+    FILTER(lang(?indicativeThirdPersonAoristSingular) = "fa") .
   }
 
   OPTIONAL {
     ?lexeme ontolex:lexicalForm ?indicativeFirstPersonAoristPluralForm .
     ?indicativeFirstPersonAoristPluralForm ontolex:representation ?indicativeFirstPersonAoristPlural ;
       wikibase:grammaticalFeature wd:Q21714344, wd:Q146786, wd:Q682111, wd:Q216497 .
-    FILTER(lang(?indicativeFirstPersonAoristPlural) = "fa").
+    FILTER(lang(?indicativeFirstPersonAoristPlural) = "fa") .
   }
 
   OPTIONAL {
     ?lexeme ontolex:lexicalForm ?indicativeSecondPersonAoristPluralForm .
     ?indicativeSecondPersonAoristPluralForm ontolex:representation ?indicativeSecondPersonAoristPlural ;
       wikibase:grammaticalFeature wd:Q51929049, wd:Q146786, wd:Q682111, wd:Q216497 .
-    FILTER(lang(?indicativeSecondPersonAoristPlural) = "fa").
+    FILTER(lang(?indicativeSecondPersonAoristPlural) = "fa") .
   }
 
   OPTIONAL {
     ?lexeme ontolex:lexicalForm ?indicativeThirdPersonAoristPluralForm .
     ?indicativeThirdPersonAoristPluralForm ontolex:representation ?indicativeThirdPersonAoristPlural ;
       wikibase:grammaticalFeature wd:Q51929074, wd:Q146786, wd:Q682111, wd:Q216497 .
-    FILTER(lang(?indicativeThirdPersonAoristPlural) = "fa").
+    FILTER(lang(?indicativeThirdPersonAoristPlural) = "fa") .
   }
 }