ScraperWiki oEmbed provider API v1.2.

scraperwiki_oembed_v1.2.provider.py

import scraperwiki

"""
 ScraperWiki oEmbed provider API, v1.2.

 Copyright 2013-01-16 Nick Freear. All rights reserved.
 License:  GNU General Public License <http://gnu.org/licenses/gpl-3.0.html>

 Usage:
 https://views.scraperwiki.com/run/scraperwiki_oembed_v1/?callback=_jsonp12&format=json&url=https%3A//views.scraperwiki.com/run/cloudworks_mindmap/%3FcloudscapeID=2451
 Source:  <https://gist.github.com/4551718>
"""

class Scraperwiki_Oembed:
    regex = '^https://views.scraperwiki.com/run/([\w_]+)/?'
    about = ''
    displayname = 'ScraperWiki - write code, get data'
    type = 'rich'
    about_url = 'https://scraperwiki.com/'
    width = '100%'
    height = 500


#----

import sys, os, cgi, re, json
from urlparse import urlparse


def getServiceName():
    # --scraper=NAME
    return sys.argv[2][sys.argv[2].find('=') + 1 :]


def oembedGetUrl(regex):
    p_url = re.compile(regex)

    try:
        qsenv = dict(cgi.parse_qsl(os.getenv("QUERY_STRING")))
        if 'url' in qsenv:
            r_url = qsenv['url']
            m_url = p_url.match(r_url)
            if not m_url:
                print "Error, invalid 'url' parameter: ", url
                exit(-400)

            o = urlparse(r_url)

            if o.scheme != 'https':
                print "Error, invalid 'url' parameter (2): ", url
                exit(-3)
        else:
            print "Error, 'url' parameter is missing."
            exit(-2)
    except:
        exit(-1)

    # print "OK, URL found, ", r_url
    return r_url, m_url.group(1)


def oembedRender(oembed_rsp): 
    p_cb = re.compile('^[a-zA-Z_][\w_]+$')
    qsenv = dict(cgi.parse_qsl(os.getenv("QUERY_STRING")))

    if 'callback' in qsenv:
        callback = qsenv['callback']
        m_cb = p_cb.match(callback)
        if not m_cb:
            print "Error, invalid 'callback' parameter: ", callback
            exit(-4)

    scraperwiki.utils.httpresponseheader("Content-Type", "application/json")

    if 'callback' in qsenv:
        print callback, '(', json.dumps(oembed_rsp), ')'
    else:
        print json.dumps(oembed_rsp)

    return


#----


provider = Scraperwiki_Oembed

url, short_name = oembedGetUrl(provider.regex)

html_template = ("<div class='scraperwiki embed-rsp'>"
                 "<iframe width='{0}' height='{1}' frameborder='0' src='{2}'></iframe></div>".format(provider.width, provider.height, url))

oembed_rsp = {'version' : '1.0',
              'type' : provider.type,
              'provider_name': provider.displayname,
              'provider_url' : provider.about_url,
              'original_url' : url,
              'title' : short_name,
              'service_name' : getServiceName(),
              'html' : html_template,
              'width': provider.width,
              'height':provider.height}

oembedRender(oembed_rsp)