liquid_feedback_frontend
view env/util/html_to_text.lua @ 1428:469195242190
Fixed wrong variable name
| author | bsw |
|---|---|
| date | Fri Oct 05 21:56:46 2018 +0200 (2018-10-05) |
| parents | 32cc544d5a5b |
| children |
line source
1 function util.html_to_text(str)
2 str = string.gsub(str, "[\0-\32]", " ")
3 str = string.gsub(str, "<[Bb][Rr] */?>", "\n")
4 str = string.gsub(str, "</?[Pp] *>", "\n\n")
5 str = string.gsub(str, "</?[Bb] *>", "**")
6 str = string.gsub(str, "</?[Ii] *>", "//")
7 str = string.gsub(str, "</?[Ss][Uu][Bb] *>", "__")
8 str = string.gsub(str, "</?[Ss][Uu][Pp] *>", "^^")
9 str = string.gsub(str, '<[Aa] *[Hh][Rr][Ee][Ff] *= *"([^"]*)" *>', "[[%1 ")
10 str = string.gsub(str, "<[Aa] *[Hh][Rr][Ee][Ff] *= *'([^']*)' *>", "[[%1 ")
11 str = string.gsub(str, "<[Aa] *[Hh][Rr][Ee][Ff] *= *([^ <>\"']*) *>", "[[%1 ")
12 str = string.gsub(str, "</[Aa] *>", "]]")
13 str = string.gsub(str, "<[Hh]1 *>", "\n\n###### ")
14 str = string.gsub(str, "<[Hh]2 *>", "\n\n##### ")
15 str = string.gsub(str, "<[Hh]3 *>", "\n\n#### ")
16 str = string.gsub(str, "<[Hh]4 *>", "\n\n### ")
17 str = string.gsub(str, "<[Hh]5 *>", "\n\n## ")
18 str = string.gsub(str, "<[Hh]6 *>", "\n\n# ")
19 str = string.gsub(str, "</[Hh]1 *>", " ######\n\n")
20 str = string.gsub(str, "</[Hh]2 *>", " #####\n\n")
21 str = string.gsub(str, "</[Hh]3 *>", " ####\n\n")
22 str = string.gsub(str, "</[Hh]4 *>", " ###\n\n")
23 str = string.gsub(str, "</[Hh]5 *>", " ##\n\n")
24 str = string.gsub(str, "</[Hh]6 *>", " #\n\n")
25 local li_info = {}
26 local pos = 1
27 local counters = {}
28 while true do
29 local list_start, list_stop, list_tagname = string.find(str, "<(/?[OoUu]l) *>", pos)
30 if list_tagname then
31 list_tagname = string.lower(list_tagname)
32 end
33 local elem_start, elem_stop = string.find(str, "<[Ll][Ii] *>", pos)
34 if list_start and not elem_start then
35 pos = list_stop
36 elseif elem_start and not list_start then
37 pos = elem_stop
38 elseif list_start and elem_start then
39 if list_start < elem_start then
40 pos = list_stop
41 else
42 pos = elem_stop
43 list_tagname = nil
44 end
45 else
46 break
47 end
48 if list_tagname == "ol" then
49 counters[#counters+1] = 0
50 elseif list_tagname == "ul" then
51 counters[#counters+1] = false
52 elseif list_tagname then
53 counters[#counters] = nil
54 else
55 if counters[#counters] then
56 counters[#counters] = counters[#counters] + 1
57 end
58 local string_parts = {}
59 for idx, counter in ipairs(counters) do
60 if counter then
61 string_parts[idx] = tostring(counter) .. ". "
62 else
63 string_parts[idx] = "* "
64 end
65 end
66 li_info[#li_info+1] = table.concat(string_parts)
67 end
68 end
69 str = string.gsub(str, "</?[OoUu]l *>", "\n\n")
70 local li_index = 0
71 str = string.gsub(str, "<[Ll][Ii] *>", function()
72 li_index = li_index + 1
73 return li_info[li_index]
74 end)
75 str = string.gsub(str, "</[Ll][Ii] *>", "\n")
76 str = string.gsub(str, "<[^<>]*>", "")
77 str = string.gsub(str, "<", "<")
78 str = string.gsub(str, ">", ">")
79 str = string.gsub(str, " +", " ")
80 str = string.gsub(str, "%f[^\0\n] ", "")
81 str = string.gsub(str, " %f[\0\n]", "")
82 str = string.gsub(str, "\n\n\n+", "\n\n")
83 str = string.gsub(str, "^\n+", "")
84 str = string.gsub(str, "\n*$", "\n")
85 return str
86 end