gniquyij.github.io
Reap the Wire
(a crawler setup record)
thewiremagazine said they would open the Wire’s complete archive of back issues to everyone until 28 Mar 2020, for free. I saw this on the night of 27 Mar and as a music enthusiast, I was like, I should do something.
What am I going to do? A crawler? It could not be better than getting all the issues in my local.
Planning
Here’re several screenshots of the site, with API information:
and what a specific issue looked like in HTML:
and a page:
Page urls followed a pattern. I realized it maybe possible for me to directly get to any page I want with the following keys provided in the API:
- i - issue id
- p - publish time of the issue
- policy - guess for AAA?
then based on these information, I could make the Wire saved in my local in this way:
$ tree
.
|---issue 0
| |---page 0
| |---page 1
| |---...
| |---page n
|---issue 1
|---...
|---issue n
What I did
- create a dir called
wreaper- the Wire Reaper - save the issue summary as a json file called
theWire.json - create a py file called
wreaper.py - trigger the py file
wreaper.py details:
import json
import os
with open('theWire.json') as f:
db = json.load(f)
policy = db['policy']
issues = db['issues']
wreaper_path = '/Users/yuqing.ji/wreaper'
for issue in issues:
issue_i = issue['i']
issue_p = issue['p']
os.mkdir(os.path.join(wreaper_path, '%s' % issue_i))
os.chdir(os.path.join(wreaper_path, '%s' % issue_i))
n = 1
while n < 120: # each issue with around 100 pages
n_str = '%s'.rjust(7, '0') % n
url = 'curl "https://d1zfca9r0ctlm4.cloudfront.net/435/493/%s/%s/images/3/page-%s.jpg%s" -o page-%s.jpg; sleep 5;' % (issue_i, issue_p, n_str, policy, n_str)
os.system(url)
n += 1
The end
Do not be fooled by my screenshot above ;)
Those are the only pages I got in the end - I started the script, saw ‘everything works well’, then went to sleep. When I checked it out the next day, I found e.g. issue id=87020 with the pages only for preview! The most likely problem I guess was the IP ban since I needed proxy to open any issue page in my web browser, which means I should make my script over the Wall. Will try if there’s a next time.
