Add ia-files-xml-to-jsonl

ia-files-xml-to-jsonl

@@ -0,0 +1,17 @@
+#!/usr/bin/env python3
+import json
+import sys
+import xml.etree.ElementTree
+
+
+root = xml.etree.ElementTree.fromstring(sys.stdin.read())
+assert root.tag == 'files'
+for file in root:
+	assert file.tag == 'file'
+	attributes = file.attrib
+	childrenTags = [child.tag for child in file]
+	assert sorted(childrenTags) == sorted(set(childrenTags)), 'duplicate children'
+	children = {child.tag: child.text for child in file}
+	assert not any(k in children for k in attributes), 'attribute found in children'
+	assert not any(k in attributes for k in children), 'child found in attributes'
+	print(json.dumps({**attributes, **children}))