From e5dce42df2c545f0040298c55cdfb1368c5eb76c Mon Sep 17 00:00:00 2001
From: Benedikt Seeger <benedikt.seeger@ptb.de>
Date: Mon, 17 Jun 2024 11:41:06 +0200
Subject: [PATCH] added logging for dsiUnits and regEx missmatch
---
XMLUnitExtractor.py | 31 +++++++++++++++++++++++++++++++
1 file changed, 31 insertions(+)
diff --git a/XMLUnitExtractor.py b/XMLUnitExtractor.py
index 7e493d3..0570067 100644
--- a/XMLUnitExtractor.py
+++ b/XMLUnitExtractor.py
@@ -1,6 +1,15 @@
import re
from dsiUnits import dsiUnit
+import logging
+import json
+from regexGenerator import generateRegex
+dsiregExStr = generateRegex()
+# Compile the regex pattern
+dsiregExPattern = re.compile(dsiregExStr)
+
+# Configure logging to log to a file
+logging.basicConfig(filename='unitValidationLog.log', level=logging.DEBUG, format='%(asctime)s %(levelname)s %(message)s')
def parse_plain_utf8_xml(xml_string):
result = {}
@@ -28,26 +37,48 @@ def parse_plain_utf8_xml(xml_string):
def process_units(unit_dict):
+ # Static regex parser function
+ def validate_dsi_unit(dsi_unit_str):
+ return dsiregExPattern.fullmatch(dsi_unit_str) is not None
+
valid_units = {}
invalid_units = {}
for key, value in unit_dict.items():
try:
unit = dsiUnit(value)
+ regExresult = validate_dsi_unit(value)
if unit.valid:
valid_units[key] = value # Assuming you want to return the string value
+ if not regExresult:
+ discrepancy = {
+ "type": "Regex Error",
+ "message": "Unit parsed as valid by dsiUnit constructor but invalid by regex",
+ "key": key,
+ "value": value
+ }
+ logging.debug(json.dumps(discrepancy))
else:
invalid_units[key] = {
"unit": value,
"warnings": unit.warnings
}
print(f"Warning: Invalid unit at {key} with value: {value}")
+ if regExresult:
+ discrepancy = {
+ "type": "Regex Error",
+ "message": "Unit parsed as invalid by dsiUnit constructor but valid by regex",
+ "key": key,
+ "value": value
+ }
+ logging.debug(json.dumps(discrepancy))
except Exception as e:
print(f"Error processing unit at {key} with value: {value}. Error: {e}")
invalid_units[key] = {
"unit": value,
"error": str(e)
}
+
return valid_units, invalid_units
def parse_and_process(xml_string):
--
GitLab