Upload preprocess_wiki_xml.py

preprocess_wiki_xml.py

@@ -0,0 +1,68 @@
+import json
+
+import mwxml
+from mwxml.errors import MalformedXML
+from tqdm import tqdm
+
+"""
+fix the error in the lines below, before running the script:
+xml.etree.ElementTree.ParseError: not well-formed (invalid token): line 23871045, column 4
+"""
+
+# specify the data path here
+file_path = "path/to/your-history.xml"
+
+raw_data = []
+
+import xml.etree.ElementTree as ET
+
+
+def strip_tag_name(t):
+    t = elem.tag
+    idx = k = t.rfind("}")
+    if idx != -1:
+        t = t[idx + 1:]
+    return t
+
+
+if __name__ == '__main__':
+    count = 0
+    prev_title = ""
+    current_data = None
+    ns = ""
+
+    # get an iterable
+    context = ET.iterparse(file_path, events=("start", "end"))
+
+    # turn it into an iterator
+    context = iter(context)
+
+    # get the root element
+    event, root = next(context)
+
+    with open("raw_data.jsonl", "w", encoding="utf-8") as f:
+        for event, elem in tqdm(context):
+            tname = strip_tag_name(elem.tag)
+
+            if event == 'end':
+                if tname == 'title':
+                    title = elem.text
+                if tname == "ns":
+                    ns = elem.text
+                elif tname == 'text':
+                    content = elem.text
+                    if ns == "0" and content:  # only namespace = "0" is what we need
+                        # only account for the latest
+                        if title != prev_title and current_data:
+                            print(json.dumps(current_data, ensure_ascii=False), file=f)
+                            count += 1
+                            if count % 1000 == 0:
+                                print(count)
+
+                        prev_title = title
+                        current_data = {
+                            "title": title,
+                            "content": content,
+                        }
+
+                root.clear()