diff options
author | cedricbonhomme <devnull@localhost> | 2011-10-24 17:42:12 +0200 |
---|---|---|
committer | cedricbonhomme <devnull@localhost> | 2011-10-24 17:42:12 +0200 |
commit | 6f57793c3e9c6cfacf0bfc539da48ae44ae947ce (patch) | |
tree | ee964f5b586b54e1fccccab596e810784034873e /export.py | |
parent | Updated CherryPy configuration. (diff) | |
download | newspipe-6f57793c3e9c6cfacf0bfc539da48ae44ae947ce.tar.gz newspipe-6f57793c3e9c6cfacf0bfc539da48ae44ae947ce.tar.bz2 newspipe-6f57793c3e9c6cfacf0bfc539da48ae44ae947ce.zip |
Added export.py. This file will gather all the export functions.
Diffstat (limited to 'export.py')
-rw-r--r-- | export.py | 121 |
1 files changed, 121 insertions, 0 deletions
diff --git a/export.py b/export.py new file mode 100644 index 00000000..ee2a6938 --- /dev/null +++ b/export.py @@ -0,0 +1,121 @@ +#! /usr/bin/env python +#-*- coding: utf-8 -*- + +import os +import hashlib + +import utils + + +htmlheader = '<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">\n' + \ + '<head>' + \ + '\n\t<title>pyAggr3g470r - News aggregator</title>\n' + \ + '\t<link rel="stylesheet" type="text/css" href="/css/style.css" />' + \ + '\n\t<meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>\n' + \ + '\n\t<script type="text/javascript" src="https://apis.google.com/js/plusone.js"></script>\n' + \ + '</head>\n' + +htmlfooter = '<p>This software is under GPLv3 license. You are welcome to copy, modify or' + \ + ' redistribute the source code according to the' + \ + ' <a href="http://www.gnu.org/licenses/gpl-3.0.txt">GPLv3</a> license.</p></div>\n' + \ + '</body>\n</html>' + + + +def export_webzine(feeds): + """ + """ + index = htmlheader + index = "<br />\n<ul>" + for feed in feeds.values(): + # creates a folder for each stream + feed_folder = utils.path + "/var/export/webzine/" + \ + utils.normalize_filename(feed.feed_id) + try: + os.makedirs(feed_folder) + except OSError: + # directories already exists (not a problem) + pass + + + index += """<li><a href="%s">%s</a></a></li>\n""" % \ + (feed.feed_id, feed.feed_title) + + posts = htmlheader + for article in feed.articles.values(): + + post_file_name = os.path.normpath(feed_folder + "/" + article.article_id + ".html") + feed_index = os.path.normpath(feed_folder + "/index.html") + + posts += article.article_date + " - " + \ + """<a href="./%s.html">%s</a>""" % \ + (article.article_id, article.article_title[:150]) + "<br />\n" + + + a_post = htmlheader + a_post += '\n<div style="width: 50%; overflow:hidden; text-align: justify; margin:0 auto">\n' + a_post += """<h1><a href="%s">%s</a></h1><br />""" % \ + (article.article_link, article.article_title) + a_post += article.article_description + a_post += "</div>\n<hr />\n" + a_post += """<br />\n<a href="%s">Complete story</a>\n<br />\n""" % (article.article_link,) + a_post += "<hr />\n" + htmlfooter + + + with open(post_file_name, "w") as f: + f.write(a_post) + + posts += htmlfooter + with open(feed_index, "w") as f: + f.write(posts) + + index += "\n</ul>\n<br />" + index += htmlfooter + with open(utils.path + "/var/export/webzine/" + "index.html", "w") as f: + f.write(index) + + +def exports(feeds, export_method): + for feed in self.feeds.values(): + # creates folder for each stream + folder = utils.path + "/var/export/" + \ + utils.normalize_filename(feed.feed_title.strip().replace(':', '').lower()) + try: + os.makedirs(folder) + except OSError: + # directories already exists (not a problem) + pass + + for article in feed.articles.values(): + name = article.article_date.strip().replace(' ', '_') + + # Export all articles in HTML format + if export_method == "export_HTML": + name = os.path.normpath(folder + "/" + name + ".html") + content = htmlheader() + content += '\n<div style="width: 50%; overflow:hidden; text-align: justify; margin:0 auto">\n' + content += """<h1><a href="%s">%s</a></h1><br />""" % \ + (article.article_link, article.article_title) + content += article.article_description + content += "</div>\n<hr />\n" + content += htmlfooter + + # Export for dokuwiki + # example: http://wiki.cedricbonhomme.org/doku.php/news-archives + elif export_method == "export_dokuwiki": + name = os.path.normpath(folder + "/" + name.replace(':', '-') + ".txt") + content = "<html>" + content += '\n<div style="width: 50%; overflow:hidden; text-align: justify; margin:0 auto">\n' + content += """<h1><a href="%s">%s</a></h1><br />""" % \ + (article.article_link, article.article_title) + content += article.article_description + content += '</div>\n<hr />Generated with <a href="http://bitbucket.org/cedricbonhomme/pyaggr3g470r/">pyAggr3g470r</a>\n</html>' + + # Export all articles in raw text + elif export_method == "export_TXT": + content = "Title: " + article.article_title + "\n\n\n" + content += utils.clear_string(article.article_description) + name = os.path.normpath(folder + "/" + name + ".txt") + + with open(name, "w") as f: + f.write(content)
\ No newline at end of file |