diff --git a/scripts/maintenance/unidata.py b/scripts/maintenance/unidata.py
new file mode 100644
index 0000000000..cf20dca746
--- /dev/null
+++ b/scripts/maintenance/unidata.py
@@ -0,0 +1,151 @@
+#!/usr/bin/env python3
+"""Script to update :mod:`pywikibot.tools._unidata`.
+
+This script is for updating ``_first_upper_exception_dict``. Run this
+module multiple times using different python versions.
+
+.. note:: I seems that running under the latest version of Python gives
+   a superse of the older version and should be enough. But this is not
+   tested completely.
+"""
+#
+# (C) Pywikibot team, 2018-2023
+#
+# Distributed under the terms of the MIT license.
+#
+from sys import maxunicode
+from re import findall
+from json import dump, load
+from queue import Queue
+from threading import Thread
+
+from scripts.maintenance.wikimedia_sites import families_list
+from pywikibot.family import Family
+from pywikibot import Site
+from pywikibot.comms.http import session
+
+
+NUMBER_OF_THREADS = 26
+FILEPATH = '/data/firstup_excepts.json'
+
+
+def chars_uppers_wikilinks():
+    """Retrieve upper chars from MediaWiki using page titles."""
+    n = 0
+    chars = []
+    uppers = []
+    wikilinks = ''
+    for i in range(0, maxunicode + 1):
+        c = chr(i)
+        uc = c.upper()
+        if uc != c:
+            n += 1
+            chars.append(c)
+            uppers.append(uc)
+            # MediaWiki is first-letter case
+            wikilinks += '[[MediaWiki:' + c + ']]\n'
+    return chars, uppers, wikilinks
+
+
+def process_site(fam_name, site_code):
+    """Process title for a single site."""
+    j = session.post(
+        f'https://{site_code}.{fam_name}.org/w/api.php?'
+        f'action=parse&contentmodel=wikitext&prop=text'
+        f'&format=json&utf8',
+        data={'text': wikilinks},
+        timeout=10,
+    ).json()
+    pased_text = j['parse']['text']['*']
+    titles = findall(r'title="[^:]*:(.)', pased_text)
+    site_excepts = {}
+    for i, original_char in enumerate(chars):
+        title_char = titles[i]
+        if uppers[i] != title_char:
+            site_excepts[original_char] = title_char
+    return site_excepts
+
+
+def threads_target(q):
+    """Thread processing a single site."""
+    global families_excepts
+    while True:
+        try:
+            fam, code = q.get()
+        except TypeError:  # non-iterable NoneType object
+            break
+        site_excepts = process_site(fam, code)
+        families_excepts[fam].setdefault(code, {}).update(site_excepts)
+        q.task_done()
+
+
+def spawn_threads(q):
+    """Prepare several threads."""
+    # TODO: use ThreadList instead
+    threads = []
+    for i in range(NUMBER_OF_THREADS):
+        t = Thread(target=threads_target, args=(q,))
+        t.start()
+        threads.append(t)
+    return threads
+
+
+def stop_threads(q, threads):
+    """Stop threads."""
+    for i in range(NUMBER_OF_THREADS):
+        q.put(None)
+    for t in threads:
+        t.join()
+
+
+def main():
+    """Main loop processing sites."""
+    global families_excepts
+    q = Queue()
+    threads = spawn_threads(q)
+    for fam_name in families_list:
+        family = Family.load(fam_name)
+        families_excepts.setdefault(fam_name, {})
+        for site_code in family.languages_by_size:
+            site = Site(site_code, family)
+            if site.namespaces[8].case != 'first-letter':
+                raise ValueError('MW namespace case is not first-letter')
+            fam_code = (fam_name, site_code)
+            if fam_code in {
+                ('wikisource', 'www'),
+                ('wikisource', 'mul'),
+                ('wikiversity', 'test'),
+            }:
+                continue  # the API of these codes does not respond as expected
+            q.put(fam_code)
+    # block until all tasks are done
+    q.join()
+    stop_threads(q, threads)
+
+
+def save_json(obj, path):
+    """Save data to file."""
+    with open(path, 'w', encoding='utf8') as f:
+        dump(obj, f)
+
+
+def load_json(path):
+    """Load data from file."""
+    try:
+        with open(path, 'r', encoding='utf8') as f:
+            return load(f)
+    except OSError:
+        print('File not found:', path)  # noqa: T001, T201
+        return {}
+
+
+if __name__ == '__main__':
+    chars, uppers, wikilinks = chars_uppers_wikilinks()
+    # save_json({'chars': chars, 'uppers': uppers, 'wikilinks': wikilinks},
+    #           'user-temp-save.json')
+    # j = load_json('user-temp-save.json')
+    # chars, uppers, wikilinks = j['chars'], j['uppers'], j['wikilinks']
+    # families_excepts = load_json(FILEPATH)
+    # main()
+    # save_json(families_excepts, FILEPATH)
+    print(process_site('wiktionary', 'fr'))  # noqa: T001, T201