view mod_ogp/test.lua @ 5296:0f5657db1cfc

mod_isolate_host: handle server-generated stanzas The hook for setting the no_host_isolation is only called for c2s sessions. This does not work for stanzas generated by the server, such as PEP notifications or presence probe answers. To handle that, we do per-stanza checks for the case that the origin is local.
author Jonas Schäfer <jonas@wielicki.name>
date Sat, 01 Apr 2023 12:03:08 +0200
parents 38da10e4b593
children
line wrap: on
line source

local html = [[
<meta property="og:title" content="Example 1 A">
<meta property=og:title content="Example 2 B">
<meta property="og:title" content="Example 3 C" >
<meta property="og:title" content="Example 4 D" />
<meta property="og:title" content="Example 5 E"/>
<meta property=og:title content=Example 6 F/>
<meta property="og:title" content= "Example 7 G" />
<meta property="og:title" itemprop="image primaryImageOfPage" content="Example 8 H" />
<meta property='og:title' content='Example 9 I' />
<meta content="Example 10 J" property="og:title" >
<meta content="Example 11 K" property="og:title">
<meta content="Example 12 L" property="og:title"/>
<meta content="Example 13 M" property="og:title" />
<meta content="Example 14 N" property=og:title >
<meta content=Example 15 O property=og:title >
<meta content= "Example 16 P" property="og:title" />
<meta content="Example 17 Q" itemprop="image primaryImageOfPage"  property="og:title" />
<meta content= 'Example 18 R' property='og:title' />
]]



local meta_pattern = [[<meta (.-)/?>]]
for match in html:gmatch(meta_pattern) do
    local property = match:match([[property=%s*["']?(og:.-)["']?%s]])
    if not property then
        property = match:match([[property=["']?(og:.-)["']$]])
    end

    local content = match:match([[content=%s*["'](.-)["']%s]])
    if not content then
        content = match:match([[content=["']?(.-)["']$]])
    end
    if not content then
        content = match:match([[content=(.-) property]])
    end
    if not content then
        content = match:match([[content=(.-)$]])
    end

    print(property, '\t', content, '\t', match .. "|")
end