getwp

#! /usr/bin/env python
import argparse
from BeautifulSoup import BeautifulSoup, BeautifulStoneSoup
import feedparser
import json
import sys
from pprint import pprint


def main(args):
    parser = argparse.ArgumentParser(
        description="Retrieve worpress posts.",
    )

    parser.add_argument(
        'target',
        help='Target user to pull public data from.',
    )

    options = parser.parse_args(args)

    print(json.dumps(get_posts(options.target)))


def get_posts(url):
    full_url = '%s/feed/' % url
    wp = feedparser.parse(full_url)

    entries = []
    for entry in wp.entries:
        if 'content' in entry.keys():
            content = entry['content']
            if not isinstance(content, list):
                content = [entry['content']]
            else:
                content = entry['content']
            for post in content:
                post = post['value']
                soup = BeautifulSoup(
                    post,
                    convertEntities=BeautifulSoup.HTML_ENTITIES,
                )
                post = ''.join(soup.findAll(text=True)).replace('\n', '')
                entries.append(post)

    return entries

if __name__ == '__main__':
    main(sys.argv[1:])