Software /
code /
prosody-modules
File
mod_pubsub_summary/mod_pubsub_summary.lua @ 4651:8231774f5bfd
mod_cloud_notify_encrypted: Ensure body substring remains valid UTF-8
The `body:sub()` call risks splitting the string in the middle of a
multi-byte UTF-8 sequence. This should have been caught by util.stanza
validation, but that would have caused some havoc, at the very least causing
the notification to not be sent.
There have been no reports of this happening. Likely because this module
isn't widely deployed among users with languages that use many longer UTF-8
sequences.
The util.encodings.utf8.valid() function is O(n) where only the last
sequence really needs to be checked, but it's in C and expected to be fast.
author | Kim Alvefur <zash@zash.se> |
---|---|
date | Sun, 22 Aug 2021 13:22:59 +0200 |
parent | 4609:fcfe691d6322 |
child | 5129:cde38b7de04a |
line wrap: on
line source
-- No, not trying to parse HTML here. It's an illusion. Just trying to read RSS feeds. -- -- Compose a textual representation of Atom payloads module:hook("pubsub-summary/http://www.w3.org/2005/Atom", function (event) local payload = event.payload; local title = payload:get_child_text("title"); -- Note: This prefers content over summary, it was made for a news feed where -- the interesting stuff was in the content and the summary was .. meh. local content_tag = payload:get_child("content") or payload:get_child("summary"); local content = content_tag and content_tag:get_text(); if content and content_tag.attr.type == "html" then content = content:gsub("\n*<p[^>]*>\n*(.-)\n*</p>\n*", "%1\n\n"); content = content:gsub("<li>(.-)</li>\n", "* %1\n"); content = content:gsub("<a[^>]*href=[\"'](.-)[\"'][^>]*>(.-)</a>", "\1%1\2%2\3"); content = content:gsub("<b>(.-)</b>", "*%1*"); content = content:gsub("<strong>(.-)</strong>", "*%1*"); content = content:gsub("<em>(.-)</em>", "_%1_"); content = content:gsub("<i>(.-)</i>", "_%1_"); content = content:gsub("<img[^>]*src=[\"'](.-)[\"'][^>]*>", " %1 "); -- TODO alt= would have been nice to grab content = content:gsub("<br[^>]*>", "\n"); content = content:gsub("<[^>]+>", ""); content = content:gsub("\1(.-)\2(.-)\3", "%2 <%1>"); content = content:gsub("^%s*", ""):gsub("%s*$", ""); content = content:gsub("\n\n\n+", "\n\n"); content = content:gsub("&(%w+);", { apos = "'"; quot = '"'; lt = "<"; gt = ">"; amp = "&"; nbsp = "\194\160"; -- U+00A0 }); end local summary; if title and content and content:sub(1, #title) ~= title then summary = "*" .. title .. "*\n\n" .. content; elseif title or content then summary = content or title; end for link in payload:childtags("link") do if link and link.attr.href and link.attr.href ~= content then summary = (summary and summary .. "\n" or "") .. link.attr.href; if link.attr.rel then summary = summary .. " [" .. link.attr.rel .. "]" end end end return summary; end, 1);