Get content length and type via HTTP
This commit is contained in:
parent
1c982636c5
commit
13b42783ef
1 changed files with 19 additions and 7 deletions
|
@ -4,6 +4,7 @@ require 'mediawiki/dotfile'
|
|||
require 'cgi'
|
||||
require 'date'
|
||||
require 'erb'
|
||||
require 'net/http'
|
||||
|
||||
#----------------------------------------------------------------------------
|
||||
# Constants
|
||||
|
@ -65,7 +66,7 @@ TPL_ITEM = '
|
|||
<link><%= episode["url"] %></link>
|
||||
<description><%= CGI::escapeHTML(episode["title"]) %></description>
|
||||
<category>Talk Radio</category>
|
||||
<enclosure url="<%= episode["enclosure_url"] %>" length="<%= episode["length"] %>" type="audio/mpeg" />
|
||||
<enclosure url="<%= episode["enclosure_url"] %>" length="<%= episode["length"] %>" type="<%= episode["type"] %>" />
|
||||
|
||||
<guid isPermaLink="true"><%= episode["url"] %></guid>
|
||||
<pubDate><%= episode["pubdate"] %></pubDate>
|
||||
|
@ -115,6 +116,14 @@ def datum2isodate(datum)
|
|||
return datum[6..9] + "-" + datum[3..4] + "-" + datum[0..1]
|
||||
end
|
||||
|
||||
def head(url)
|
||||
uri = URI.parse(url)
|
||||
|
||||
Net::HTTP.start(uri.host, uri.port) do |http|
|
||||
http.head(uri.path)
|
||||
end
|
||||
end
|
||||
|
||||
#----------------------------------------------------------------------------
|
||||
# Get episodes from wiki
|
||||
|
||||
|
@ -127,8 +136,8 @@ episodes = []
|
|||
if erste_sendung = parse_template(page, template)["erste_sendung"]
|
||||
episode = {
|
||||
"title" => page,
|
||||
"url" => "http://entropia.de/wiki/#{CGI::escape(page)}",
|
||||
"discussion" => "http://entropia.de/wiki/Diskussion:#{CGI::escape(page)}",
|
||||
"url" => @wiki.article_url(page),
|
||||
"discussion" => @wiki.article_url("Diskussion:#{page}"),
|
||||
"enclosure_url" => parse_template(page, template)["download"],
|
||||
"date" => Date.today(), # fallback
|
||||
}
|
||||
|
@ -144,14 +153,17 @@ episodes = []
|
|||
episode["pubdate"] = episode["date"].strftime(RFC822)
|
||||
|
||||
|
||||
# FIXME
|
||||
episode["length"] = 57671680
|
||||
episode["duration"] = "00:59:59"
|
||||
# Get content type and length
|
||||
head = head(episode["enclosure_url"])
|
||||
episode["length"] = head["content-length"]
|
||||
episode["type"] = head["content-type"]
|
||||
|
||||
#FIXME
|
||||
episode["duration"] = "00:59:59"
|
||||
|
||||
episodes.push(episode)
|
||||
end
|
||||
# break # DEBUG
|
||||
$stderr.puts "DEBUG: Only crawling one episode"; break
|
||||
end
|
||||
|
||||
episodes = episodes.sort do |x,y|
|
||||
|
|
Reference in a new issue