Mercurial > prosody-modules

comparison mod_ogp/test.lua @ 4255:38da10e4b593

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
mod_ogp: Update parsing logic to catch more cases
author Seve Ferrer <seve@delape.net>
date Wed, 18 Nov 2020 13:48:07 +0100
parents a4e182d7ff0a
children
comparison
equal deleted inserted replaced
4254:a4e182d7ff0a 4255:38da10e4b593
1 local html = [[ 1 local html = [[
2 <meta property="og:title" content="Example 1"> 2 <meta property="og:title" content="Example 1 A">
3 <meta property=og:title content="Example 2"> 3 <meta property=og:title content="Example 2 B">
4 <meta property="og:title" content="Example 3" > 4 <meta property="og:title" content="Example 3 C" >
5 <meta property="og:title" content="Example 4" /> 5 <meta property="og:title" content="Example 4 D" />
6 <meta property="og:title" content="Example 5"/> 6 <meta property="og:title" content="Example 5 E"/>
7 <meta property=og:title content=Example 6/> 7 <meta property=og:title content=Example 6 F/>
8 <meta property="og:title" content= "Example 7" /> 8 <meta property="og:title" content= "Example 7 G" />
9 <meta property="og:title" itemprop="image primaryImageOfPage" content="Example 8" /> 9 <meta property="og:title" itemprop="image primaryImageOfPage" content="Example 8 H" />
10 <meta content="Example 9" property="og:title" > 10 <meta property='og:title' content='Example 9 I' />
11 <meta content="Example 10" property="og:title"> 11 <meta content="Example 10 J" property="og:title" >
12 <meta content="Example 11" property="og:title"/> 12 <meta content="Example 11 K" property="og:title">
13 <meta content="Example 12" property="og:title" /> 13 <meta content="Example 12 L" property="og:title"/>
14 <meta content="Example 13" property=og:title > 14 <meta content="Example 13 M" property="og:title" />
15 <meta content=Example 14 property=og:title > 15 <meta content="Example 14 N" property=og:title >
16 <meta content= "Example 15" property="og:title" /> 16 <meta content=Example 15 O property=og:title >
17 <meta content="Example 16" itemprop="image primaryImageOfPage" property="og:title" /> 17 <meta content= "Example 16 P" property="og:title" />
18 <meta content="Example 17 Q" itemprop="image primaryImageOfPage" property="og:title" />
19 <meta content= 'Example 18 R' property='og:title' />
18 ]] 20 ]]
19 21
20 22
21 local ogp_pattern = [[<meta property=["']?(og:.-)["']? content=%s*["']?(.-)["']?%s-/?>]]
22 local ogp_pattern2 = [[<meta content=%s*["']?(.-)["']? property=["']?(og:.-)["']?%s-/?>]]
23 23
24 for property, content in html:gmatch(ogp_pattern) do 24 local meta_pattern = [[<meta (.-)/?>]]
25 print("Pattern 1|", property, content, "|Pattern 1") 25 for match in html:gmatch(meta_pattern) do
26 local property = match:match([[property=%s*["']?(og:.-)["']?%s]])
27 if not property then
28 property = match:match([[property=["']?(og:.-)["']$]])
29 end
30
31 local content = match:match([[content=%s*["'](.-)["']%s]])
32 if not content then
33 content = match:match([[content=["']?(.-)["']$]])
34 end
35 if not content then
36 content = match:match([[content=(.-) property]])
37 end
38 if not content then
39 content = match:match([[content=(.-)$]])
40 end
41
42 print(property, '\t', content, '\t', match .. "|")
26 end 43 end
27 print('-------------------------------------------------------------')
28 for content, property in html:gmatch(ogp_pattern2) do
29 print("Pattern 2|", property, content, "|Pattern 2")
30 end