Import to Pinboard from Pocket while preserving read status and multi-word tags.

pinboard_from_pocket.py

import sys
import time
import os
import re

import requests


def main():
    to_read = 'yes'
    LINE_RE = re.compile(r'href="(.*?)" time_added="(.*?)" tags="(.*?)">(.*?)</a>')
    session = requests.Session()
    api_token = os.environ['PINBOARD_TOKEN']
    api_endpoint = 'https://api.pinboard.in/v1/posts/add'
    for line in sys.stdin:
        if '<h1>Read Archive</h1>' in line:
            to_read = 'no'
        line = line.strip()
        match = LINE_RE.search(line)
        if match:
            link, time_added, raw_tags, title = match.groups()
            tags = parse_tags(raw_tags)
            params = {
                'auth_token': api_token,
                'url': link,
                'description': title,
                'tags': ','.join(tags),
                'dt': time_added,
                'replace': 'yes',
                'shared': 'no',
                'toread': to_read,
            }
            response = session.get(api_endpoint, params=params)
            timeout = 3.5
            while response.status_code != 200:
                if response.status_code == 429:
                    timeout *= 2
                time.sleep(timeout)
                response = session.get(api_endpoint, params=params)
            print('Submitted %s' % title)
            time.sleep(timeout)




def parse_tags(tag_string):
    tags = []
    raw_tags = tag_string.split(',')
    for tag in raw_tags:
        if not tag:
            continue
        tags.append(tag.lower().replace(' ', '-'))
    return tags



if __name__ == '__main__':
    main()