Download Twitter favorite images

getimg.py

import json
import sys
import os
import re
import urllib.request
import time
import datetime

# Install twint: https://github.com/twintproject/twint
# Grab favorites: twint --username <myuser> --images --favorites --json -o tweets.json
# Fetch images: getimg.py tweets.json

UNSAFE_RE=re.compile(r'[^-_a-zA-Z0-9]')
IMGDIR='images'

infile=sys.argv[1]

os.makedirs(IMGDIR, exist_ok=True)

images=[]
for line in open(infile):
    entry = json.loads(line)
    username = UNSAFE_RE.sub('', entry['username'])
    entry_id = entry['id']

    tstamp_str = f"{entry['date']}T{entry['time']}"
    tstamp_dt = datetime.datetime.strptime(tstamp_str, '%Y-%m-%dT%H:%M:%S')
    tstamp = tstamp_dt.timestamp()
    
    for idx, p_url in enumerate(entry['photos']):
        if p_url.endswith('png'):
            suffix = 'png'
        else:
            suffix = 'jpg'
        fname = f'{username}_{entry_id}_{idx+1}.{suffix}'
        images.append((fname, p_url, tstamp))

print(f'found {len(images)} images')

def geturl(dpath, url):
    try:
        resp = urllib.request.urlopen(url)
        data = resp.read()
        with open(dpath, 'wb') as fp:
            fp.write(data)

        return len(data)
    except Exception as exc:
        print(f'failed to download {url}: {exc}') 
        return False

    return 0

atime = time.time()
def settime(dpath, mtime):
    os.utime(dpath, (atime, mtime))

for fname, src, tstamp in images:
    dst = os.path.join(IMGDIR, fname)
    if os.path.exists(dst):
        print(f'image {fname} exists, skipping')
        settime(dst, tstamp)
        continue

    dlen = geturl(dst, src)
    if dlen:
        print(f'image {fname} downloaded ({dlen} bytes)')
        settime(dst, tstamp)