Annotate

mod_pubsub_feeds/feeds.lib.lua @ 4282:281a864e7472

mod_pubsub_feeds: Don't skip publishing items after an existing one I encountered a feed which was backwards, such that older entries were considered first and then it would skip newer entries. This may however run into trouble if the feed contains more items than what's persisted in pubsub.
author Kim Alvefur <zash@zash.se>
date Mon, 30 Nov 2020 15:17:29 +0100
parent 2378:649f733aa3dc
child 4427:c402b273f2e3
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2132
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
1 local st = require "util.stanza";
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
2 -- RSS->Atom translator
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
3 -- http://code.matthewwild.co.uk/lua-feeds/
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
4
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
5 -- Helpers to translate item child elements
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
6 local rss2atom = {};
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
7 function rss2atom.title(atom_entry, tag)
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
8 atom_entry:tag("title"):text(tag:get_text()):up();
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
9 end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
10
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
11 function rss2atom.link(atom_entry, tag)
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
12 atom_entry:tag("link", { href = tag:get_text() }):up();
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
13 end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
14
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
15 function rss2atom.author(atom_entry, tag)
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
16 atom_entry:tag("author")
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
17 :tag("email"):text(tag:get_text()):up()
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
18 :up();
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
19 end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
20
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
21 function rss2atom.guid(atom_entry, tag)
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
22 atom_entry:tag("id"):text(tag:get_text()):up();
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
23 end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
24
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
25 function rss2atom.category(atom_entry, tag)
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
26 atom_entry:tag("category", { term = tag:get_text(), scheme = tag.attr.domain }):up();
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
27 end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
28
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
29 function rss2atom.description(atom_entry, tag)
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
30 atom_entry:tag("summary"):text(tag:get_text()):up();
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
31 end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
32
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
33 local months = {
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
34 jan = "01", feb = "02", mar = "03", apr = "04", may = "05", jun = "06";
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
35 jul = "07", aug = "08", sep = "09", oct = "10", nov = "11", dec = "12";
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
36 };
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
37
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
38 function rss2atom.pubDate(atom_entry, tag)
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
39 local pubdate = tag:get_text():gsub("^%a+,", ""):gsub("^%s*", "");
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
40 local date, month, year, hour, minute, second, zone =
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
41 pubdate:match("^(%d%d?) (%a+) (%d+) (%d+):(%d+):?(%d*) ?(.*)$");
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
42 if not date then return; end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
43 if #date == 1 then
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
44 date = "0"..date;
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
45 end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
46 month = months[month:sub(1,3):lower()];
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
47 if #year == 2 then -- GAH!
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
48 if tonumber(year) > 80 then
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
49 year = "19"..year;
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
50 else
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
51 year = "20"..year;
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
52 end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
53 end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
54 if zone == "UT" or zone == "GMT" then zone = "Z"; end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
55 if #second == 0 then
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
56 second = "00";
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
57 end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
58 local date_string = string.format("%s-%s-%sT%s:%s:%s%s", year, month, date, hour, minute, second, zone);
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
59 atom_entry:tag("published"):text(date_string):up();
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
60 end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
61
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
62 -- Translate a single item to atom
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
63 local function translate_rss(rss_feed)
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
64 local feed = st.stanza("feed", { xmlns = "http://www.w3.org/2005/Atom" });
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
65 local channel = rss_feed:get_child("channel");
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
66 -- TODO channel properties
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
67 for item in channel:childtags("item") do
2378
649f733aa3dc mod_pubsub_feeds/feeds.lib: Fix converting RSS items to more than one Atom entry (thanks walduhu)
Kim Alvefur <zash@zash.se>
parents: 2133
diff changeset
68 feed:tag("entry");
2133
85762420a2c0 mod_pubsub_feeds: Use correct loop variable
Kim Alvefur <zash@zash.se>
parents: 2132
diff changeset
69 for tag in item:childtags() do
2132
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
70 local translator = rss2atom[tag.name];
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
71 if translator then
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
72 translator(feed, tag);
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
73 end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
74 end
2378
649f733aa3dc mod_pubsub_feeds/feeds.lib: Fix converting RSS items to more than one Atom entry (thanks walduhu)
Kim Alvefur <zash@zash.se>
parents: 2133
diff changeset
75 feed:reset();
2132
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
76 end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
77 return feed;
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
78 end
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
79
b149ea428b81 mod_pubsub_feeds: Switch to use util.xml for parsing feeds and include RSS to Atom translation code from lua-feeds
Kim Alvefur <zash@zash.se>
parents:
diff changeset
80 return { translate_rss = translate_rss }