Software / code / prosody-modules
Annotate
mod_pastebin/mod_pastebin.lua @ 190:7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
| author | Matthew Wild <mwild1@gmail.com> |
|---|---|
| date | Mon, 28 Jun 2010 15:34:28 +0100 |
| parent | 189:16d367e3c04e |
| child | 438:7f0cdde1e42a |
| rev | line source |
|---|---|
| 5 | 1 |
| 2 local st = require "util.stanza"; | |
| 3 local httpserver = require "net.httpserver"; | |
| 4 local uuid_new = require "util.uuid".generate; | |
|
12
316e8437f233
mod_pastebin: Allow configurable message length threshold
Matthew Wild <mwild1@gmail.com>
parents:
5
diff
changeset
|
5 local os_time = os.time; |
|
24
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
6 local t_insert, t_remove = table.insert, table.remove; |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
7 local add_task = require "util.timer".add_task; |
| 5 | 8 |
|
190
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
9 local function drop_invalid_utf8(seq) |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
10 local start = seq:byte(); |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
11 module:log("utf8: %d, %d", start, #seq); |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
12 if (start <= 223 and #seq < 2) |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
13 or (start >= 224 and start <= 239 and #seq < 3) |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
14 or (start >= 240 and start <= 244 and #seq < 4) |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
15 or (start > 244) then |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
16 return ""; |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
17 end |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
18 return seq; |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
19 end |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
20 |
|
23
92b1e6592d36
mod_pastebin: Allow per-host pastebin_threshold
Matthew Wild <mwild1@gmail.com>
parents:
21
diff
changeset
|
21 local length_threshold = config.get(module.host, "core", "pastebin_threshold") or 500; |
|
167
0d37d18ea073
mod_pastebin: Fix trigger_string matching when no trigger is set, and add support for counting lines (pastebin_line_threshold, default: 4)
Matthew Wild <mwild1@gmail.com>
parents:
156
diff
changeset
|
22 local line_threshold = config.get(module.host, "core", "pastebin_line_threshold") or 4; |
| 5 | 23 |
|
21
4f18696f043a
mod_pastebin: Small fix to read the pastebin URL from the config
Matthew Wild <mwild1@gmail.com>
parents:
13
diff
changeset
|
24 local base_url = config.get(module.host, "core", "pastebin_url"); |
| 5 | 25 |
|
24
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
26 -- Seconds a paste should live for in seconds (config is in hours), default 24 hours |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
27 local expire_after = math.floor((config.get(module.host, "core", "pastebin_expire_after") or 24) * 3600); |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
28 |
|
156
b51741b7e86d
mod_pastebin: Optionally bin if message starts with a configurable trigger string
Florian Zeitz <florob@babelmonkeys.de>
parents:
76
diff
changeset
|
29 local trigger_string = config.get(module.host, "core", "pastebin_trigger"); |
|
167
0d37d18ea073
mod_pastebin: Fix trigger_string matching when no trigger is set, and add support for counting lines (pastebin_line_threshold, default: 4)
Matthew Wild <mwild1@gmail.com>
parents:
156
diff
changeset
|
30 trigger_string = (trigger_string and trigger_string .. " "); |
|
156
b51741b7e86d
mod_pastebin: Optionally bin if message starts with a configurable trigger string
Florian Zeitz <florob@babelmonkeys.de>
parents:
76
diff
changeset
|
31 |
| 5 | 32 local pastes = {}; |
|
76
1fc4e8dc66a6
mod_pastebin: Send Content-Type header to specify plain UTF-8 text
Matthew Wild <mwild1@gmail.com>
parents:
75
diff
changeset
|
33 local default_headers = { ["Content-Type"] = "text/plain; charset=utf-8" }; |
| 5 | 34 |
| 35 local xmlns_xhtmlim = "http://jabber.org/protocol/xhtml-im"; | |
| 36 local xmlns_xhtml = "http://www.w3.org/1999/xhtml"; | |
| 37 | |
|
75
3c7189e26848
mod_pastebin: Rename pastebin_message() to pastebin_text() and make it global so it can be called by other plugins
Matthew Wild <mwild1@gmail.com>
parents:
71
diff
changeset
|
38 function pastebin_text(text) |
| 5 | 39 local uuid = uuid_new(); |
|
76
1fc4e8dc66a6
mod_pastebin: Send Content-Type header to specify plain UTF-8 text
Matthew Wild <mwild1@gmail.com>
parents:
75
diff
changeset
|
40 pastes[uuid] = { body = text, time = os_time(), headers = default_headers }; |
|
24
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
41 pastes[#pastes+1] = uuid; |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
42 if not pastes[2] then -- No other pastes, give the timer a kick |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
43 add_task(expire_after, expire_pastes); |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
44 end |
| 5 | 45 return base_url..uuid; |
| 46 end | |
| 47 | |
| 48 function handle_request(method, body, request) | |
| 49 local pasteid = request.url.path:match("[^/]+$"); | |
| 50 if not pasteid or not pastes[pasteid] then | |
| 51 return "Invalid paste id, perhaps it expired?"; | |
| 52 end | |
| 53 | |
| 54 --module:log("debug", "Received request, replying: %s", pastes[pasteid].text); | |
| 55 | |
|
76
1fc4e8dc66a6
mod_pastebin: Send Content-Type header to specify plain UTF-8 text
Matthew Wild <mwild1@gmail.com>
parents:
75
diff
changeset
|
56 return pastes[pasteid]; |
| 5 | 57 end |
| 58 | |
| 59 function check_message(data) | |
| 60 local origin, stanza = data.origin, data.stanza; | |
| 61 | |
| 62 local body, bodyindex, htmlindex; | |
| 63 for k,v in ipairs(stanza) do | |
| 64 if v.name == "body" then | |
| 65 body, bodyindex = v, k; | |
|
71
3c18c2d03bc2
mod_pastebin: Fix finding of XHTML content.
Paul Aurich <paul@darkrain42.org>
parents:
25
diff
changeset
|
66 elseif v.name == "html" and v.attr.xmlns == xmlns_xhtmlim then |
| 5 | 67 htmlindex = k; |
| 68 end | |
| 69 end | |
| 70 | |
| 71 if not body then return; end | |
| 72 body = body:get_text(); | |
| 73 | |
|
25
ea59a8d98b03
mod_pastebin: Comment some debug logging on every message
Matthew Wild <mwild1@gmail.com>
parents:
24
diff
changeset
|
74 --module:log("debug", "Body(%s) length: %d", type(body), #(body or "")); |
| 5 | 75 |
|
167
0d37d18ea073
mod_pastebin: Fix trigger_string matching when no trigger is set, and add support for counting lines (pastebin_line_threshold, default: 4)
Matthew Wild <mwild1@gmail.com>
parents:
156
diff
changeset
|
76 if body and ( |
|
0d37d18ea073
mod_pastebin: Fix trigger_string matching when no trigger is set, and add support for counting lines (pastebin_line_threshold, default: 4)
Matthew Wild <mwild1@gmail.com>
parents:
156
diff
changeset
|
77 (#body > length_threshold) or |
|
0d37d18ea073
mod_pastebin: Fix trigger_string matching when no trigger is set, and add support for counting lines (pastebin_line_threshold, default: 4)
Matthew Wild <mwild1@gmail.com>
parents:
156
diff
changeset
|
78 (trigger_string and body:find(trigger_string, 1, true) == 1) or |
|
0d37d18ea073
mod_pastebin: Fix trigger_string matching when no trigger is set, and add support for counting lines (pastebin_line_threshold, default: 4)
Matthew Wild <mwild1@gmail.com>
parents:
156
diff
changeset
|
79 (select(2, body:gsub("\n", "%0")) >= line_threshold) |
|
0d37d18ea073
mod_pastebin: Fix trigger_string matching when no trigger is set, and add support for counting lines (pastebin_line_threshold, default: 4)
Matthew Wild <mwild1@gmail.com>
parents:
156
diff
changeset
|
80 ) then |
|
0d37d18ea073
mod_pastebin: Fix trigger_string matching when no trigger is set, and add support for counting lines (pastebin_line_threshold, default: 4)
Matthew Wild <mwild1@gmail.com>
parents:
156
diff
changeset
|
81 if trigger_string then |
|
0d37d18ea073
mod_pastebin: Fix trigger_string matching when no trigger is set, and add support for counting lines (pastebin_line_threshold, default: 4)
Matthew Wild <mwild1@gmail.com>
parents:
156
diff
changeset
|
82 body = body:gsub("^" .. trigger_string, "", 1); |
|
0d37d18ea073
mod_pastebin: Fix trigger_string matching when no trigger is set, and add support for counting lines (pastebin_line_threshold, default: 4)
Matthew Wild <mwild1@gmail.com>
parents:
156
diff
changeset
|
83 end |
|
75
3c7189e26848
mod_pastebin: Rename pastebin_message() to pastebin_text() and make it global so it can be called by other plugins
Matthew Wild <mwild1@gmail.com>
parents:
71
diff
changeset
|
84 local url = pastebin_text(body); |
| 5 | 85 module:log("debug", "Pasted message as %s", url); |
| 86 --module:log("debug", " stanza[bodyindex] = %q", tostring( stanza[bodyindex])); | |
| 87 stanza[bodyindex][1] = url; | |
| 88 local html = st.stanza("html", { xmlns = xmlns_xhtmlim }):tag("body", { xmlns = xmlns_xhtml }); | |
|
190
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
89 html:tag("p"):text(body:sub(1,150):gsub("[\194-\244][\128-\191]*$", drop_invalid_utf8)):up(); |
| 5 | 90 html:tag("a", { href = url }):text("[...]"):up(); |
| 91 stanza[htmlindex or #stanza+1] = html; | |
| 92 end | |
| 93 end | |
| 94 | |
| 95 module:hook("message/bare", check_message); | |
| 96 | |
|
24
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
97 function expire_pastes(time) |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
98 time = time or os_time(); -- COMPAT with 0.5 |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
99 if pastes[1] then |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
100 pastes[pastes[1]] = nil; |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
101 t_remove(pastes, 1); |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
102 if pastes[1] then |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
103 return (expire_after - (time - pastes[pastes[1]].time)) + 1; |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
104 end |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
105 end |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
106 end |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
107 |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
108 |
| 5 | 109 local ports = config.get(module.host, "core", "pastebin_ports") or { 5280 }; |
| 110 for _, options in ipairs(ports) do | |
| 111 local port, base, ssl, interface = 5280, "pastebin", false, nil; | |
| 112 if type(options) == "number" then | |
| 113 port = options; | |
| 114 elseif type(options) == "table" then | |
| 115 port, base, ssl, interface = options.port or 5280, options.path or "pastebin", options.ssl or false, options.interface; | |
| 116 elseif type(options) == "string" then | |
| 117 base = options; | |
| 118 end | |
| 119 | |
| 120 base_url = base_url or ("http://"..module:get_host()..(port ~= 80 and (":"..port) or "").."/"..base.."/"); | |
| 121 | |
| 122 httpserver.new{ port = port, base = base, handler = handle_request, ssl = ssl } | |
| 123 end |
