Mercurial > prosody-modules
annotate mod_pastebin/mod_pastebin.lua @ 501:e851f386c904
mod_pastebin: Threshold is now UTF-8 characters (thanks Grom_PE for initial patch)
| author | Matthew Wild <mwild1@gmail.com> |
|---|---|
| date | Thu, 08 Dec 2011 22:01:54 +0000 |
| parents | 3f101f7a26d0 |
| children | c819365ac6ab |
| 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 |
|
444
82ccfba5ac2f
mod_pastebin: Support for changing the summary length (pastebin_summary_length), and including it even in the plaintext version of the generated message
Matthew Wild <mwild1@gmail.com>
parents:
438
diff
changeset
|
9 local utf8_pattern = "[\194-\244][\128-\191]*$"; |
|
190
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
10 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
|
11 local start = seq:byte(); |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
12 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
|
13 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
|
14 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
|
15 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
|
16 or (start > 244) then |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
17 return ""; |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
18 end |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
19 return seq; |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
20 end |
|
7a695ee3884b
mod_pastebin: Better fix for stripping truncated UFT-8 sequences
Matthew Wild <mwild1@gmail.com>
parents:
189
diff
changeset
|
21 |
|
501
e851f386c904
mod_pastebin: Threshold is now UTF-8 characters (thanks Grom_PE for initial patch)
Matthew Wild <mwild1@gmail.com>
parents:
454
diff
changeset
|
22 local function utf8_length(str) |
|
e851f386c904
mod_pastebin: Threshold is now UTF-8 characters (thanks Grom_PE for initial patch)
Matthew Wild <mwild1@gmail.com>
parents:
454
diff
changeset
|
23 local _, count = string.gsub(str, "[^\128-\193]", ""); |
|
e851f386c904
mod_pastebin: Threshold is now UTF-8 characters (thanks Grom_PE for initial patch)
Matthew Wild <mwild1@gmail.com>
parents:
454
diff
changeset
|
24 return count; |
|
e851f386c904
mod_pastebin: Threshold is now UTF-8 characters (thanks Grom_PE for initial patch)
Matthew Wild <mwild1@gmail.com>
parents:
454
diff
changeset
|
25 end |
|
e851f386c904
mod_pastebin: Threshold is now UTF-8 characters (thanks Grom_PE for initial patch)
Matthew Wild <mwild1@gmail.com>
parents:
454
diff
changeset
|
26 |
|
444
82ccfba5ac2f
mod_pastebin: Support for changing the summary length (pastebin_summary_length), and including it even in the plaintext version of the generated message
Matthew Wild <mwild1@gmail.com>
parents:
438
diff
changeset
|
27 local pastebin_private_messages = module:get_option_boolean("pastebin_private_messages", hosts[module.host].type ~= "component"); |
|
445
66ca5e4d9f7b
mod_pastebin: Update to use module:get_option()
Matthew Wild <mwild1@gmail.com>
parents:
444
diff
changeset
|
28 local length_threshold = module:get_option_number("pastebin_threshold", 500); |
|
66ca5e4d9f7b
mod_pastebin: Update to use module:get_option()
Matthew Wild <mwild1@gmail.com>
parents:
444
diff
changeset
|
29 local line_threshold = module:get_option_number("pastebin_line_threshold", 4); |
|
446
56f1c29ee7f1
mod_pastebin: Some experimental summary changes
Matthew Wild <mwild1@gmail.com>
parents:
445
diff
changeset
|
30 local max_summary_length = module:get_option_number("pastebin_summary_length", 150); |
|
56f1c29ee7f1
mod_pastebin: Some experimental summary changes
Matthew Wild <mwild1@gmail.com>
parents:
445
diff
changeset
|
31 local html_preview = module:get_option_boolean("pastebin_html_preview", true); |
| 5 | 32 |
|
445
66ca5e4d9f7b
mod_pastebin: Update to use module:get_option()
Matthew Wild <mwild1@gmail.com>
parents:
444
diff
changeset
|
33 local base_url = module:get_option_string("pastebin_url"); |
| 5 | 34 |
|
24
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
35 -- Seconds a paste should live for in seconds (config is in hours), default 24 hours |
|
445
66ca5e4d9f7b
mod_pastebin: Update to use module:get_option()
Matthew Wild <mwild1@gmail.com>
parents:
444
diff
changeset
|
36 local expire_after = math.floor(module:get_option_number("pastebin_expire_after", 24) * 3600); |
|
24
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
37 |
|
445
66ca5e4d9f7b
mod_pastebin: Update to use module:get_option()
Matthew Wild <mwild1@gmail.com>
parents:
444
diff
changeset
|
38 local trigger_string = module:get_option_string("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
|
39 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
|
40 |
| 5 | 41 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
|
42 local default_headers = { ["Content-Type"] = "text/plain; charset=utf-8" }; |
| 5 | 43 |
| 44 local xmlns_xhtmlim = "http://jabber.org/protocol/xhtml-im"; | |
| 45 local xmlns_xhtml = "http://www.w3.org/1999/xhtml"; | |
| 46 | |
|
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
|
47 function pastebin_text(text) |
| 5 | 48 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
|
49 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
|
50 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
|
51 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
|
52 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
|
53 end |
| 5 | 54 return base_url..uuid; |
| 55 end | |
| 56 | |
| 57 function handle_request(method, body, request) | |
| 58 local pasteid = request.url.path:match("[^/]+$"); | |
| 59 if not pasteid or not pastes[pasteid] then | |
| 60 return "Invalid paste id, perhaps it expired?"; | |
| 61 end | |
| 62 | |
| 63 --module:log("debug", "Received request, replying: %s", pastes[pasteid].text); | |
| 64 | |
|
76
1fc4e8dc66a6
mod_pastebin: Send Content-Type header to specify plain UTF-8 text
Matthew Wild <mwild1@gmail.com>
parents:
75
diff
changeset
|
65 return pastes[pasteid]; |
| 5 | 66 end |
| 67 | |
| 68 function check_message(data) | |
| 69 local origin, stanza = data.origin, data.stanza; | |
| 70 | |
| 71 local body, bodyindex, htmlindex; | |
| 72 for k,v in ipairs(stanza) do | |
| 73 if v.name == "body" then | |
| 74 body, bodyindex = v, k; | |
|
71
3c18c2d03bc2
mod_pastebin: Fix finding of XHTML content.
Paul Aurich <paul@darkrain42.org>
parents:
25
diff
changeset
|
75 elseif v.name == "html" and v.attr.xmlns == xmlns_xhtmlim then |
| 5 | 76 htmlindex = k; |
| 77 end | |
| 78 end | |
| 79 | |
| 80 if not body then return; end | |
| 81 body = body:get_text(); | |
| 82 | |
|
25
ea59a8d98b03
mod_pastebin: Comment some debug logging on every message
Matthew Wild <mwild1@gmail.com>
parents:
24
diff
changeset
|
83 --module:log("debug", "Body(%s) length: %d", type(body), #(body or "")); |
| 5 | 84 |
|
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
|
85 if body and ( |
|
501
e851f386c904
mod_pastebin: Threshold is now UTF-8 characters (thanks Grom_PE for initial patch)
Matthew Wild <mwild1@gmail.com>
parents:
454
diff
changeset
|
86 ((#body > length_threshold) |
|
e851f386c904
mod_pastebin: Threshold is now UTF-8 characters (thanks Grom_PE for initial patch)
Matthew Wild <mwild1@gmail.com>
parents:
454
diff
changeset
|
87 and (length_utf8(body) > length_threshold)) or |
|
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
|
88 (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
|
89 (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
|
90 ) 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
|
91 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
|
92 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
|
93 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
|
94 local url = pastebin_text(body); |
| 5 | 95 module:log("debug", "Pasted message as %s", url); |
| 96 --module:log("debug", " stanza[bodyindex] = %q", tostring( stanza[bodyindex])); | |
|
446
56f1c29ee7f1
mod_pastebin: Some experimental summary changes
Matthew Wild <mwild1@gmail.com>
parents:
445
diff
changeset
|
97 local summary = (body:sub(1, max_summary_length):gsub(utf8_pattern, drop_invalid_utf8) or ""):match("[^\n]+") or ""; |
|
56f1c29ee7f1
mod_pastebin: Some experimental summary changes
Matthew Wild <mwild1@gmail.com>
parents:
445
diff
changeset
|
98 summary = summary:match("^%s*(.-)%s*$"); |
|
56f1c29ee7f1
mod_pastebin: Some experimental summary changes
Matthew Wild <mwild1@gmail.com>
parents:
445
diff
changeset
|
99 local summary_prefixed = summary:match("[,:]$"); |
|
56f1c29ee7f1
mod_pastebin: Some experimental summary changes
Matthew Wild <mwild1@gmail.com>
parents:
445
diff
changeset
|
100 stanza[bodyindex][1] = (summary_prefixed and (summary.." ") or "")..url; |
|
56f1c29ee7f1
mod_pastebin: Some experimental summary changes
Matthew Wild <mwild1@gmail.com>
parents:
445
diff
changeset
|
101 |
|
56f1c29ee7f1
mod_pastebin: Some experimental summary changes
Matthew Wild <mwild1@gmail.com>
parents:
445
diff
changeset
|
102 if html_preview then |
|
56f1c29ee7f1
mod_pastebin: Some experimental summary changes
Matthew Wild <mwild1@gmail.com>
parents:
445
diff
changeset
|
103 local line_count = select(2, body:gsub("\n", "%0")); |
|
56f1c29ee7f1
mod_pastebin: Some experimental summary changes
Matthew Wild <mwild1@gmail.com>
parents:
445
diff
changeset
|
104 local link_text = ("[view %spaste (%d line%s)]"):format(summary_prefixed and "" or "rest of ", line_count, line_count == 1 and "" or "s"); |
|
56f1c29ee7f1
mod_pastebin: Some experimental summary changes
Matthew Wild <mwild1@gmail.com>
parents:
445
diff
changeset
|
105 local html = st.stanza("html", { xmlns = xmlns_xhtmlim }):tag("body", { xmlns = xmlns_xhtml }); |
|
56f1c29ee7f1
mod_pastebin: Some experimental summary changes
Matthew Wild <mwild1@gmail.com>
parents:
445
diff
changeset
|
106 html:tag("p"):text(summary.." "):up(); |
|
56f1c29ee7f1
mod_pastebin: Some experimental summary changes
Matthew Wild <mwild1@gmail.com>
parents:
445
diff
changeset
|
107 html:tag("a", { href = url }):text(link_text):up(); |
|
56f1c29ee7f1
mod_pastebin: Some experimental summary changes
Matthew Wild <mwild1@gmail.com>
parents:
445
diff
changeset
|
108 stanza[htmlindex or #stanza+1] = html; |
|
56f1c29ee7f1
mod_pastebin: Some experimental summary changes
Matthew Wild <mwild1@gmail.com>
parents:
445
diff
changeset
|
109 end |
| 5 | 110 end |
| 111 end | |
| 112 | |
| 113 module:hook("message/bare", check_message); | |
|
438
7f0cdde1e42a
mod_pastebin: Add option 'pastebin_private_messages', defaults to false for components and true for other hosts (thanks Kelden/Maranda)
Matthew Wild <mwild1@gmail.com>
parents:
190
diff
changeset
|
114 if pastebin_private_messages then |
|
7f0cdde1e42a
mod_pastebin: Add option 'pastebin_private_messages', defaults to false for components and true for other hosts (thanks Kelden/Maranda)
Matthew Wild <mwild1@gmail.com>
parents:
190
diff
changeset
|
115 module:hook("message/full", check_message); |
|
7f0cdde1e42a
mod_pastebin: Add option 'pastebin_private_messages', defaults to false for components and true for other hosts (thanks Kelden/Maranda)
Matthew Wild <mwild1@gmail.com>
parents:
190
diff
changeset
|
116 end |
| 5 | 117 |
|
24
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
118 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
|
119 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
|
120 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
|
121 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
|
122 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
|
123 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
|
124 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
|
125 end |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
126 end |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
127 end |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
128 |
|
72bcc0475e2f
mod_pastebin: Expire pastes after 'pastebin_expire_after' hours, 24 by default
Matthew Wild <mwild1@gmail.com>
parents:
23
diff
changeset
|
129 |
|
445
66ca5e4d9f7b
mod_pastebin: Update to use module:get_option()
Matthew Wild <mwild1@gmail.com>
parents:
444
diff
changeset
|
130 local ports = module:get_option("pastebin_ports", { 5280 }); |
| 5 | 131 for _, options in ipairs(ports) do |
| 132 local port, base, ssl, interface = 5280, "pastebin", false, nil; | |
| 133 if type(options) == "number" then | |
| 134 port = options; | |
| 135 elseif type(options) == "table" then | |
| 136 port, base, ssl, interface = options.port or 5280, options.path or "pastebin", options.ssl or false, options.interface; | |
| 137 elseif type(options) == "string" then | |
| 138 base = options; | |
| 139 end | |
| 140 | |
| 141 base_url = base_url or ("http://"..module:get_host()..(port ~= 80 and (":"..port) or "").."/"..base.."/"); | |
| 142 | |
| 143 httpserver.new{ port = port, base = base, handler = handle_request, ssl = ssl } | |
| 144 end | |
|
454
3f101f7a26d0
mod_pastebin: Preserve pastes through a reload
Matthew Wild <mwild1@gmail.com>
parents:
446
diff
changeset
|
145 |
|
3f101f7a26d0
mod_pastebin: Preserve pastes through a reload
Matthew Wild <mwild1@gmail.com>
parents:
446
diff
changeset
|
146 function module.save() |
|
3f101f7a26d0
mod_pastebin: Preserve pastes through a reload
Matthew Wild <mwild1@gmail.com>
parents:
446
diff
changeset
|
147 return { pastes = pastes }; |
|
3f101f7a26d0
mod_pastebin: Preserve pastes through a reload
Matthew Wild <mwild1@gmail.com>
parents:
446
diff
changeset
|
148 end |
|
3f101f7a26d0
mod_pastebin: Preserve pastes through a reload
Matthew Wild <mwild1@gmail.com>
parents:
446
diff
changeset
|
149 |
|
3f101f7a26d0
mod_pastebin: Preserve pastes through a reload
Matthew Wild <mwild1@gmail.com>
parents:
446
diff
changeset
|
150 function module.restore(data) |
|
3f101f7a26d0
mod_pastebin: Preserve pastes through a reload
Matthew Wild <mwild1@gmail.com>
parents:
446
diff
changeset
|
151 pastes = data.pastes or pastes; |
|
3f101f7a26d0
mod_pastebin: Preserve pastes through a reload
Matthew Wild <mwild1@gmail.com>
parents:
446
diff
changeset
|
152 end |