Expand Twitter URLs from a twitter archive manifest

expander.py

# This script replaces all strings matching https://t.co/ links in a the manifest with their expanded versions
import json
import re
import urlexpander

# Open the file containing twitter bookmarks in json
# and convert to a dictionary
with open('bookmark-manifest.json', 'r') as fh:
    contents = fh.read()
    bookmarks = json.loads(contents)

# Go through each bookmark and collect all the short url links
for mark in bookmarks:
    text = mark['text']
    links = re.findall(r'https://t.co/\w+', text)
    if links:
        print(links)
    # Use urlexpander.expand() to expand any found links
    # then replace them in `text`   
    for link in links:
        expanded = urlexpander.expand(link)
        text = text.replace(link, expanded)
    # Save the expanded text back to the bookmark
    mark['text'] = text

# Write the updated bookmarks back to the file
with open('bookmark-manifest-expanded.json', 'w') as fh:
    json.dump(bookmarks, fh, indent=4)

Authored mostly by Codex all bugs should be reported to OpenAI