Merge pull request #679 from biolink/go-site-2210-gorule-0000027-full…

…-regex-match For #2210
biolink · Jul 19, 2024 · 5f865b1 · 5f865b1
2 parents 726162e + 19b72a0
commit 5f865b1
Show file tree

Hide file tree

Showing 3 changed files with 66 additions and 2 deletions.
diff --git a/ontobio/io/assocparser.py b/ontobio/io/assocparser.py
@@ -826,7 +826,7 @@ def _validate_id(self, id, line: SplitLine, allowed_ids=None, context=None):
                 type_name_regex_patterns = self.config.db_type_name_regex_id_syntax[id_prefix]
                 identity_matches_pattern = False
                 for regex in type_name_regex_patterns.values():
-                    if regex.match(right):
+                    if regex.fullmatch(right):
                         identity_matches_pattern = True
                         break
                 if identity_matches_pattern == False:

diff --git a/tests/test_gafparser.py b/tests/test_gafparser.py
@@ -638,6 +638,24 @@ def test_id_syntax():
     messages = p.report.to_report_json()["messages"]
     assert len(messages["gorule-0000027"]) == 1
     assert messages["gorule-0000027"][0]["obj"] == "PMID:PMID:18422602" 
+
+    p = GafParser(config=assocparser.AssocParserConfig(
+        ontology=OntologyFactory().create(ONT), db_type_name_regex_id_syntax=database_id_syntax_lookups))
+    assoc_result = p.parse_line("PomBase\tSPBC1289.03c\tspi1\t\tGO:0005515\tPMID:0.\tIPI\tPomBase:SPAC25A8.01c\tF\tRan GTPase Spi1\t\tprotein\ttaxon:4896\t20080718\tPomBase\t")
+    assert len(assoc_result.associations) == 1
+    assert assoc_result.skipped == False
+    messages = p.report.to_report_json()["messages"]
+    assert len(messages["gorule-0000027"]) == 1
+    assert messages["gorule-0000027"][0]["obj"] == "PMID:0."
+
+    p = GafParser(config=assocparser.AssocParserConfig(
+        ontology=OntologyFactory().create(ONT), db_type_name_regex_id_syntax=database_id_syntax_lookups))
+    assoc_result = p.parse_line("PomBase\tSPBC1289.03c\tspi1\t\tGO:0005515\tPMID:x18422602\tIPI\tPomBase:SPAC25A8.01c\tF\tRan GTPase Spi1\t\tprotein\ttaxon:4896\t20080718\tPomBase\t")
+    assert len(assoc_result.associations) == 1
+    assert assoc_result.skipped == False
+    messages = p.report.to_report_json()["messages"]
+    assert len(messages["gorule-0000027"]) == 1
+    assert messages["gorule-0000027"][0]["obj"] == "PMID:x18422602"
 
     p = GafParser(config=assocparser.AssocParserConfig(
         ontology=OntologyFactory().create(ONT), db_type_name_regex_id_syntax=database_id_syntax_lookups))

diff --git a/tests/test_gpad_parser.py b/tests/test_gpad_parser.py
@@ -382,7 +382,53 @@ def test_id_syntax():
     assert result.skipped == False
     messages = p.report.to_report_json()["messages"]
     assert len(messages["gorule-0000027"]) == 1
-    assert messages["gorule-0000027"][0]["obj"] == "PMID:PMID:15494018" 
+    assert messages["gorule-0000027"][0]["obj"] == "PMID:PMID:15494018"
+
+    vals = ["PomBase",
+            "SPAC25A8.01c",
+            "acts_upstream_of_or_within",
+            "GO:0007155",
+            "PMID:9.",
+            "ECO:0000305",
+            "GO:0005913",
+            "",
+            "20041026",
+            "ZFIN",
+            "",
+            "PomBase"
+            ]
+
+    p = GpadParser(config=config)
+    result = p.parse_line("\t".join(vals))
+    assert len(result.associations) == 1
+    assert result.skipped == False
+    messages = p.report.to_report_json()["messages"]
+    assert len(messages["gorule-0000027"]) == 1
+    assert messages["gorule-0000027"][0]["obj"] == "PMID:9." 
+
+
+    vals = ["PomBase",
+            "SPAC25A8.01c",
+            "acts_upstream_of_or_within",
+            "GO:0007155",
+            "PMID:a1549418",
+            "ECO:0000305",
+            "GO:0005913",
+            "",
+            "20041026",
+            "ZFIN",
+            "",
+            "PomBase"
+            ]
+
+    p = GpadParser(config=config)
+    result = p.parse_line("\t".join(vals))
+    assert len(result.associations) == 1
+    assert result.skipped == False
+    messages = p.report.to_report_json()["messages"]
+    assert len(messages["gorule-0000027"]) == 1
+    assert messages["gorule-0000027"][0]["obj"] == "PMID:a1549418" 
+
 
     vals = ["PomBase",
             "SPAC25A8.01c",