Theodor-Springmann-Stiftung/lenz-briefe
1
0
Fork 0
mirror of https://github.com/Theodor-Springmann-Stiftung/lenz-briefe.git synced 2025-10-29 17:15:31 +00:00
Code Issues Packages Projects Releases Wiki Activity

Verweise linter

Browse Source
This commit is contained in:
Simon Martens
2025-02-05 20:48:54 +01:00
parent ff78eb7e6d
commit 9dfc3b86f9
2 changed files with 650 additions and 512 deletions
Download Patch File Download Diff File Expand all files Collapse all files

143
scripts/lint_verweise.py
View File

@@ -1,11 +1,17 @@
import os
import json
import sys
from lxml import etree
REPO_ROOT = os.path.abspath(os.path.join(os.path.dirname(__file__), '..'))
XML_DIR = os.path.join(REPO_ROOT, 'data', 'xml')
# Namespace map for your "lenz" default namespace
NAMESPACE_MAP = {"lenz": "https://lenz-archiv.de"}
def parse_xml_file(filepath):
"""
Parse an XML file using lxml and return the root element, or None on parse error.
"""
try:
parser = etree.XMLParser(remove_blank_text=True)
tree = etree.parse(filepath, parser)
@@ -14,74 +20,118 @@ def parse_xml_file(filepath):
print(f"Error parsing {filepath}: {e}")
return None
def get_all_ids(root, tag):
return set(elem.get('letter') for elem in root.xpath(f'.//{tag}'))
def get_letter_desc_ids(meta_root):
"""
Retrieve all 'letter' attribute values from <letterDesc letter="...">
in the meta.xml file, which uses the default namespace https://lenz-archiv.de.
Returns a set of string IDs.
"""
letter_ids = set()
# Find all <letterDesc> in the lenz namespace
letter_descs = meta_root.xpath("//lenz:letterDesc", namespaces=NAMESPACE_MAP)
for desc in letter_descs:
val = desc.get("letter")
if val:
letter_ids.add(val)
return letter_ids
def get_all_refs(root, tag, attr):
return set(elem.get(attr) for elem in root.xpath(f'.//{tag}'))
def get_person_def_ids(ref_root):
"""
Retrieve all 'index' attributes from <personDef index="..."> in references.xml.
If references.xml is un-namespaced, we can use a non-namespace XPath: //personDef
Adjust if references.xml also has a namespace.
"""
return set(elem.get("index") for elem in ref_root.xpath("//personDef"))
def get_location_def_ids(ref_root):
"""
Retrieve all 'index' attributes from <locationDef index="..."> in references.xml.
"""
return set(elem.get("index") for elem in ref_root.xpath("//locationDef"))
def get_app_def_ids(ref_root):
"""
Retrieve all 'index' attributes from <appDef index="..."> in references.xml.
"""
return set(elem.get("index") for elem in ref_root.xpath("//appDef"))
def check_references(root, reference_data, filepath):
"""
Check various references across briefe.xml, meta.xml, and traditions.xml
against known IDs from meta.xml (letterDesc) and references.xml (personDef, locationDef, appDef).
"""
errors = []
relative_path = os.path.relpath(filepath, REPO_ROOT)
def add_error(element, ref_type, ref_id):
line_number = element.sourceline
error_message = f"Invalid reference ({ref_type}:{ref_id})"
msg = f"Invalid reference ({ref_type}:{ref_id})"
errors.append({
"file": relative_path,
"line": line_number,
"message": error_message
"message": msg
})
# Check letterText in briefe to letterDesc in meta
for letter_text in root.xpath('//letterText'):
letter_id = letter_text.get('letter')
if letter_id not in reference_data['letterDesc']:
add_error(letter_text, 'letterText', letter_id)
# 1) Check <letterText letter="..."> references to meta.xml's letterDesc
# (Assumes <letterText> is un-namespaced, in briefe.xml or wherever.)
for letter_text in root.xpath("//letterText"):
letter_id = letter_text.get("letter")
if letter_id not in reference_data["letterDesc"]:
add_error(letter_text, "letterText", letter_id)
# Check letterTradition in traditions to letterDesc in meta
for letter_tradition in root.xpath('//letterTradition'):
letter_id = letter_tradition.get('letter')
if letter_id not in reference_data['letterDesc']:
add_error(letter_tradition, 'letterTradition', letter_id)
# 2) Check <letterTradition letter="..."> references to meta.xml's letterDesc
for letter_tradition in root.xpath("//letterTradition"):
letter_id = letter_tradition.get("letter")
if letter_id not in reference_data["letterDesc"]:
add_error(letter_tradition, "letterTradition", letter_id)
# Check hand in briefe with personDef in references
for hand in root.xpath('//hand'):
ref = hand.get('ref')
if ref not in reference_data['personDef']:
add_error(hand, 'hand', ref)
# 3) Check <location ref="..."> in the lenz namespace, referencing locationDef
# e.g. <location ref="3" />
for location_elem in root.xpath("//lenz:location", namespaces=NAMESPACE_MAP):
ref = location_elem.get("ref")
if ref not in reference_data["locationDef"]:
add_error(location_elem, "location", ref)
# Check sender and receiver in meta with personDef in references
for person in root.xpath('//sender | //receiver'):
ref = person.get('ref')
if ref not in reference_data['personDef']:
add_error(person, 'sender/receiver', ref)
# 4) Check <person ref="..."> in the lenz namespace, referencing personDef
# e.g. <person ref="1" />
for person_elem in root.xpath("//lenz:person", namespaces=NAMESPACE_MAP):
ref = person_elem.get("ref")
if ref not in reference_data["personDef"]:
add_error(person_elem, "person", ref)
# Check location in meta with locationDef in references
for location in root.xpath('//location'):
ref = location.get('ref')
if ref not in reference_data['locationDef']:
add_error(location, 'location', ref)
# Check app in traditions with appDef in references
for app in root.xpath('//app'):
ref = app.get('ref')
if ref not in reference_data['appDef']:
add_error(app, 'app', ref)
# 5) Check <app ref="..."> (un-namespaced?), referencing appDef
for app_elem in root.xpath("//app"):
ref = app_elem.get("ref")
if ref not in reference_data["appDef"]:
add_error(app_elem, "app", ref)
return errors
def main():
# Parse meta.xml (has letterDesc in default "lenz" namespace)
meta_root = parse_xml_file(os.path.join(XML_DIR, "meta.xml"))
if meta_root is None:
print("Could not parse meta.xml; aborting.")
sys.exit(1)
# Parse references.xml for personDef, locationDef, appDef
ref_root = parse_xml_file(os.path.join(XML_DIR, "references.xml"))
if ref_root is None:
print("Could not parse references.xml; aborting.")
sys.exit(1)
# Build our cross-file reference data
reference_data = {
'letterDesc': get_all_ids(parse_xml_file(os.path.join(XML_DIR, 'meta.xml')), 'letterDesc'),
'personDef': get_all_refs(parse_xml_file(os.path.join(XML_DIR, 'references.xml')), 'personDef', 'index'),
'locationDef': get_all_refs(parse_xml_file(os.path.join(XML_DIR, 'references.xml')), 'locationDef', 'index'),
'appDef': get_all_refs(parse_xml_file(os.path.join(XML_DIR, 'references.xml')), 'appDef', 'index'),
"letterDesc": get_letter_desc_ids(meta_root),
"personDef": get_person_def_ids(ref_root),
"locationDef": get_location_def_ids(ref_root),
"appDef": get_app_def_ids(ref_root),
}
all_errors = []
files_to_check = ['briefe.xml', 'meta.xml', 'traditions.xml']
# Check references in the following files
files_to_check = ["briefe.xml", "meta.xml", "traditions.xml"]
for filename in files_to_check:
filepath = os.path.join(XML_DIR, filename)
root = parse_xml_file(filepath)
@@ -89,16 +139,17 @@ def main():
errors = check_references(root, reference_data, filepath)
all_errors.extend(errors)
# Report any errors
if all_errors:
print("The linter found the following errors:")
for error in all_errors:
print(f"{error['file']}, Line {error['line']}: {error['message']}")
# GitHub Actions output
# Print GitHub Actions compatible error lines
for error in all_errors:
print(f"::error file={error['file']},line={error['line']}::{error['message']}")
exit(1) # Exit with error code if errors were found
sys.exit(1)
else:
print("No errors found.")