author | Matthew Wild <mwild1@gmail.com> |
Tue, 16 Apr 2024 12:41:52 +0100 | |
changeset 5888 | 49b0873932ca |
parent 4825 | 17fbe82d4bfe |
permissions | -rw-r--r-- |
4819
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
1 |
|
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
2 |
local t_insert = table.insert; |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
3 |
local t_sort = table.sort; |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
4 |
|
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
5 |
local sm = require "core.storagemanager"; |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
6 |
local um = require "core.usermanager"; |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
7 |
|
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
8 |
local argparse = require "util.argparse"; |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
9 |
local dt = require "util.datetime"; |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
10 |
local jid = require "util.jid"; |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
11 |
local st = require "util.stanza"; |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
12 |
|
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
13 |
local function skeleton(s) |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
14 |
local o = st.stanza(s.name, { xmlns = s.attr.xmlns }); |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
15 |
|
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
16 |
local children = {}; |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
17 |
for _, child in ipairs(s.tags) do t_insert(children, skeleton(child)) end |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
18 |
t_sort(children, function(a, b) |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
19 |
if a.attr.xmlns == b.attr.xmlns then return a.name < b.name; end |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
20 |
return (a.attr.xmlns or "") < (b.attr.xmlns or ""); |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
21 |
end); |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
22 |
for _, child in ipairs(children) do o:add_direct_child(child); end |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
23 |
return o; |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
24 |
end |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
25 |
|
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
26 |
local function classify_jid(s) |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
27 |
if not s then return "" end |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
28 |
local u, h, r = jid.split(s); |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
29 |
if r then |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
30 |
return "full" |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
31 |
elseif u then |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
32 |
return "bare" |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
33 |
elseif h then |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
34 |
return "host" |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
35 |
else |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
36 |
return "invalid" |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
37 |
end |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
38 |
end |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
39 |
|
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
40 |
function module.command(arg) |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
41 |
local opts = argparse.parse(arg, { value_params = { store = true; with = true; start = true; ["end"] = true } }); |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
42 |
local store = opts.store or "archive"; -- so you can pass 'archive2' |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
43 |
opts.store = nil; |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
44 |
local query = { with = jid.prep(opts.with); start = dt.parse(opts.start); ["end"] = dt.parse(opts["end"]) }; |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
45 |
local host_initialized = {}; |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
46 |
for _, export_jid in ipairs(arg) do |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
47 |
|
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
48 |
local username, host = jid.split(export_jid); |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
49 |
if not host_initialized[host] then |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
50 |
sm.initialize_host(host); |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
51 |
um.initialize_host(host); |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
52 |
host_initialized[host] = true; |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
53 |
end |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
54 |
|
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
55 |
local archive = module:context(host):open_store(store, "archive"); |
4825
17fbe82d4bfe
mod_export_skeletons: Remove wildcard query
Kim Alvefur <zash@zash.se>
parents:
4824
diff
changeset
|
56 |
local iter, total = assert(archive:find(username, query)) |
4819
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
57 |
if total then io.stderr:write(string.format("Processing %d entries\n", total)); end |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
58 |
for _, item in iter do |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
59 |
local clean = skeleton(item); |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
60 |
|
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
61 |
-- Normalize top level attributes |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
62 |
clean.attr.type = item.attr.type; |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
63 |
if clean.attr.type == nil and clean.name == "message" then clean.attr.type = "normal"; end |
4822
d66162e850cd
mod_export_skeletons: Generate ids based on log2 of the original length
Kim Alvefur <zash@zash.se>
parents:
4819
diff
changeset
|
64 |
clean.attr.id = string.rep("x", math.floor(math.log(1+#(item.attr.id or ""), 2))); |
4819
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
65 |
clean.attr.from = classify_jid(item.attr.from); |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
66 |
clean.attr.to = classify_jid(item.attr.to); |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
67 |
print(clean); |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
68 |
end |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
69 |
|
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
70 |
end |
9c2af2146ee2
mod_export_skeletons: Command to aid in analysis of archive contents
Kim Alvefur <zash@zash.se>
parents:
diff
changeset
|
71 |
end |