fixup(data-pipeline): PR review, use saxutils built in quoteattr func…

…tion
broadinstitute · Nov 18, 2024 · 01d67b0 · 01d67b0
1 parent bacbc86
commit 01d67b0
Showing 1 changed file with 4 additions and 11 deletions.
diff --git a/data-pipeline/src/data_pipeline/datasets/clinvar.py b/data-pipeline/src/data_pipeline/datasets/clinvar.py
@@ -5,7 +5,7 @@
 import subprocess
 import sys
 from xml.etree import ElementTree
-import xml.sax.saxutils as saxutils
+from xml.sax.saxutils import quoteattr
 
 import hail as hl
 from tqdm import tqdm
@@ -36,20 +36,13 @@
 }
 
 
-def escape_xpath_value(value):
-    if value is None:
-        return ""
-    value = value.replace('"', "&quot;").replace("'", "&apos;")
-    return saxutils.escape(value)
-
-
 def find_mapping_elements_by_xref(trait_element, submission_element, trait_mapping_list_element):
     if trait_mapping_list_element is None:
         return None
 
     xref_elements = trait_element.findall("XRef")
     for xref_element in xref_elements:
-        selector = f"./TraitMapping[@ClinicalAssertionID='{submission_element.attrib['ID']}'][@TraitType='{trait_element.attrib['Type']}'][@MappingType='XRef'][@MappingValue='{escape_xpath_value(xref_element.attrib['ID'])}']"  # noqa
+        selector = f"./TraitMapping[@ClinicalAssertionID='{submission_element.attrib['ID']}'][@TraitType='{trait_element.attrib['Type']}'][@MappingType='XRef'][@MappingValue={quoteattr(xref_element.attrib['ID'])}]"  # noqa
         mapping_element = trait_mapping_list_element.find(selector)
         if mapping_element is not None:
             return mapping_element
@@ -60,7 +53,7 @@ def find_mapping_elements_by_preferred_name(trait_element, submission_element, t
     preferred_name_element = trait_element.find("./Name/ElementValue[@Type='Preferred']")
 
     if preferred_name_element is not None and trait_mapping_list_element is not None:
-        selector = f"./TraitMapping[@ClinicalAssertionID='{submission_element.attrib['ID']}'][@TraitType='{trait_element.attrib['Type']}'][@MappingType='Name'][@MappingValue='{escape_xpath_value(preferred_name_element.text)}']"  # noqa
+        selector = f"./TraitMapping[@ClinicalAssertionID='{submission_element.attrib['ID']}'][@TraitType='{trait_element.attrib['Type']}'][@MappingType='Name'][@MappingValue={quoteattr(preferred_name_element.text)}]"  # noqa
         mapping_element = trait_mapping_list_element.find(selector)
         return mapping_element, preferred_name_element
 
@@ -76,7 +69,7 @@ def find_mapping_elements_by_name(trait_element, submission_element, trait_mappi
             preferred_name_element = name_element
 
         if trait_mapping_list_element is not None:
-            selector = f"./TraitMapping[@ClinicalAssertionID='{submission_element.attrib['ID']}'][@TraitType='{trait_element.attrib['Type']}'][@MappingType='Name'][@MappingValue='{escape_xpath_value(name_element.text)}']"  # noqa
+            selector = f"./TraitMapping[@ClinicalAssertionID='{submission_element.attrib['ID']}'][@TraitType='{trait_element.attrib['Type']}'][@MappingType='Name'][@MappingValue={quoteattr(name_element.text)}]"  # noqa
             mapping_element = trait_mapping_list_element.find(selector)
             if mapping_element:
                 return mapping_element, preferred_name_element