Software / code / prosody-modules
File
mod_remote_roster/mod_remote_roster.lua @ 6301:fa45ae704b79
mod_cloud_notify: Update Readme
diff --git a/mod_cloud_notify/README.md b/mod_cloud_notify/README.md
--- a/mod_cloud_notify/README.md
+++ b/mod_cloud_notify/README.md
@@ -1,109 +1,106 @@
----
-labels:
-- 'Stage-Beta'
-summary: 'XEP-0357: Cloud push notifications'
----
+# Introduction
-Introduction
-============
+This module enables support for sending "push notifications" to clients
+that need it, typically those running on certain mobile devices.
-This module enables support for sending "push notifications" to clients that
-need it, typically those running on certain mobile devices.
+As well as this module, your client must support push notifications (the
+apps that need it generally do, of course) and the app developer's push
+gateway must be reachable from your Prosody server (this happens over a
+normal XMPP server-to-server 's2s' connection).
-As well as this module, your client must support push notifications (the apps
-that need it generally do, of course) and the app developer's push gateway
-must be reachable from your Prosody server (this happens over a normal XMPP
-server-to-server 's2s' connection).
-
-Details
-=======
+# Details
Some platforms, notably Apple's iOS and many versions of Android, impose
-limits that prevent applications from running or accessing the network in the
-background. This makes it difficult or impossible for an XMPP application to
-remain reliably connected to a server to receive messages.
-
-In order for messaging and other apps to receive notifications, the OS vendors
-run proprietary servers that their OS maintains a permanent connection to in
-the background. Then they provide APIs to application developers that allow
-sending notifications to specific devices via those servers.
+limits that prevent applications from running or accessing the network
+in the background. This makes it difficult or impossible for an XMPP
+application to remain reliably connected to a server to receive
+messages.
-When you connect to your server with an app that requires push notifications,
-it will use this module to set up a "push registration". When you receive
-a message but your device is not connected to the server, this module will
-generate a notification and send it to the push gateway operated by your
-application's developers). Their gateway will then connect to your device's
-OS vendor and ask them to forward the notification to your device. When your
-device receives the notification, it will display it or wake up the app so it
-can connect to XMPP and receive any pending messages.
+In order for messaging and other apps to receive notifications, the OS
+vendors run proprietary servers that their OS maintains a permanent
+connection to in the background. Then they provide APIs to application
+developers that allow sending notifications to specific devices via
+those servers.
-This protocol is described for developers in [XEP-0357: Push Notifications].
+When you connect to your server with an app that requires push
+notifications, it will use this module to set up a "push registration".
+When you receive a message but your device is not connected to the
+server, this module will generate a notification and send it to the push
+gateway operated by your application's developers). Their gateway will
+then connect to your device's OS vendor and ask them to forward the
+notification to your device. When your device receives the notification,
+it will display it or wake up the app so it can connect to XMPP and
+receive any pending messages.
-For this module to work reliably, you must have [mod_smacks], [mod_mam] and
-[mod_carbons] also enabled on your server.
+This protocol is described for developers in \[XEP-0357: Push
+Notifications\].
+
+For this module to work reliably, you must have \[mod_smacks\],
+\[mod_mam\] and \[mod_carbons\] also enabled on your server.
-Some clients, notably Siskin and Snikket iOS need some additional extensions
-that are not currently defined in a standard XEP. To support these clients,
-see [mod_cloud_notify_extensions].
+Some clients, notably Siskin and Snikket iOS need some additional
+extensions that are not currently defined in a standard XEP. To support
+these clients, see \[mod_cloud_notify_extensions\].
-Configuration
-=============
+# Configuration
- Option Default Description
- ------------------------------------ ----------------- -------------------------------------------------------------------------------------------------------------------
- `push_notification_important_body` `New Message!` The body text to use when the stanza is important (see above), no message body is sent if this is empty
- `push_max_errors` `16` How much persistent push errors are tolerated before notifications for the identifier in question are disabled
- `push_max_devices` `5` The number of allowed devices per user (the oldest devices are automatically removed if this threshold is reached)
- `push_max_hibernation_timeout` `259200` (72h) Number of seconds to extend the smacks timeout if no push was triggered yet (default: 72 hours)
- `push_notification_with_body` (\*) `false` Whether or not to send the real message body to remote pubsub node. Without end-to-end encryption, enabling this may expose your message contents to your client developers and OS vendor. Not recommended.
- `push_notification_with_sender` (\*) `false` Whether or not to send the real message sender to remote pubsub node. Enabling this may expose your contacts to your client developers and OS vendor. Not recommended.
+ Option Default Description
+ -------------------------------------- ---------------- -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
+ `push_notification_important_body` `New Message!` The body text to use when the stanza is important (see above), no message body is sent if this is empty
+ `push_max_errors` `16` How much persistent push errors are tolerated before notifications for the identifier in question are disabled
+ `push_max_devices` `5` The number of allowed devices per user (the oldest devices are automatically removed if this threshold is reached)
+ `push_max_hibernation_timeout` `259200` (72h) Number of seconds to extend the smacks timeout if no push was triggered yet (default: 72 hours)
+ `push_notification_with_body` (\*) `false` Whether or not to send the real message body to remote pubsub node. Without end-to-end encryption, enabling this may expose your message contents to your client developers and OS vendor. Not recommended.
+ `push_notification_with_sender` (\*) `false` Whether or not to send the real message sender to remote pubsub node. Enabling this may expose your contacts to your client developers and OS vendor. Not recommended.
-(\*) There are privacy implications for enabling these options.
+(\*) There are privacy implications for enabling these options.[^1]
-Internal design notes
-=====================
+# Internal design notes
-App servers are notified about offline messages, messages stored by [mod_mam]
-or messages waiting in the smacks queue.
-The business rules outlined [here](//mail.jabber.org/pipermail/standards/2016-February/030925.html) are all honored[^2].
+App servers are notified about offline messages, messages stored by
+\[mod_mam\] or messages waiting in the smacks queue. The business rules
+outlined
+[here](//mail.jabber.org/pipermail/standards/2016-February/030925.html)
+are all honored[^2].
-To cooperate with [mod_smacks] this module consumes some events:
-`smacks-ack-delayed`, `smacks-hibernation-start` and `smacks-hibernation-end`.
-These events allow this module to send out notifications for messages received
-while the session is hibernated by [mod_smacks] or even when smacks
-acknowledgements for messages are delayed by a certain amount of seconds
-configurable with the [mod_smacks] setting `smacks_max_ack_delay`.
+To cooperate with \[mod_smacks\] this module consumes some events:
+`smacks-ack-delayed`, `smacks-hibernation-start` and
+`smacks-hibernation-end`. These events allow this module to send out
+notifications for messages received while the session is hibernated by
+\[mod_smacks\] or even when smacks acknowledgements for messages are
+delayed by a certain amount of seconds configurable with the
+\[mod_smacks\] setting `smacks_max_ack_delay`.
-The `smacks_max_ack_delay` setting allows to send out notifications to clients
-which aren't already in smacks hibernation state (because the read timeout or
-connection close didn't already happen) but also aren't responding to acknowledgement
-request in a timely manner. This setting thus allows conversations to be smoother
-under such circumstances.
+The `smacks_max_ack_delay` setting allows to send out notifications to
+clients which aren't already in smacks hibernation state (because the
+read timeout or connection close didn't already happen) but also aren't
+responding to acknowledgement request in a timely manner. This setting
+thus allows conversations to be smoother under such circumstances.
-The new event `cloud-notify-ping` can be used by any module to send out a cloud
-notification to either all registered endpoints for the given user or only the endpoints
-given in the event data.
+The new event `cloud-notify-ping` can be used by any module to send out
+a cloud notification to either all registered endpoints for the given
+user or only the endpoints given in the event data.
-The config setting `push_notification_important_body` can be used to specify an alternative
-body text to send to the remote pubsub node if the stanza is encrypted or has a body.
-This way the real contents of the message aren't revealed to the push appserver but it
-can still see that the push is important.
-This is used by Chatsecure on iOS to send out high priority pushes in those cases for example.
+The config setting `push_notification_important_body` can be used to
+specify an alternative body text to send to the remote pubsub node if
+the stanza is encrypted or has a body. This way the real contents of the
+message aren't revealed to the push appserver but it can still see that
+the push is important. This is used by Chatsecure on iOS to send out
+high priority pushes in those cases for example.
-Compatibility
-=============
-
-**Note:** This module should be used with Lua 5.2 and higher. Using it with
-Lua 5.1 may cause push notifications to not be sent to some clients.
+# Compatibility
------- -----------------------------------------------------------------------------
- trunk Works
- 0.12 Works
- 0.11 Works
- 0.10 Works
- 0.9 Support dropped, use last supported version [675726ab06d3](//hg.prosody.im/prosody-modules/raw-file/675726ab06d3/mod_cloud_notify/mod_cloud_notify.lua)
------- -----------------------------------------------------------------------------
+**Note:** This module should be used with Lua 5.2 and higher. Using it
+with Lua 5.1 may cause push notifications to not be sent to some
+clients.
+ ------- -----------------------------------------------------------------
+ trunk Works as of 25-06-13
+ 13 Works
+ 0.12 Works
+ ------- -----------------------------------------------------------------
-[^1]: The service which is expected to forward notifications to something like Google Cloud Messaging or Apple Notification Service
-[^2]: [business_rules.markdown](//hg.prosody.im/prosody-modules/file/tip/mod_cloud_notify/business_rules.markdown)
+[^1]: The service which is expected to forward notifications to
+ something like Google Cloud Messaging or Apple Notification Service
+
+[^2]: [business_rules.md](//hg.prosody.im/prosody-modules/file/tip/mod_cloud_notify/business_rules.md)
| author | Menel <menel@snikket.de> |
|---|---|
| date | Fri, 13 Jun 2025 10:36:52 +0200 |
| parent | 5376:ad6e9b0fd15b |
line wrap: on
line source
-- -- mod_remote_roster -- -- This is an experimental implementation of http://jkaluza.fedorapeople.org/remote-roster.html -- local st = require "util.stanza"; local jid_split = require "util.jid".split; local jid_prep = require "util.jid".prep; local t_concat = table.concat; local tonumber = tonumber; local pairs, ipairs = pairs, ipairs; local hosts = hosts; local load_roster = require "core.rostermanager".load_roster; local save_roster = require "core.rostermanager".save_roster; local rm_remove_from_roster = require "core.rostermanager".remove_from_roster; local rm_add_to_roster = require "core.rostermanager".add_to_roster; local rm_roster_push = require "core.rostermanager".roster_push; local user_exists = require "core.usermanager".user_exists; local add_task = require "util.timer".add_task; local new_id = require "util.id".short; module:hook("iq-get/bare/jabber:iq:roster:query", function(event) local origin, stanza = event.origin, event.stanza; if origin.type == "component" and stanza.attr.from == origin.host then local node, host = jid_split(stanza.attr.to); local roster = load_roster(node, host); local reply = st.reply(stanza):query("jabber:iq:roster"); for jid, item in pairs(roster) do if jid ~= "pending" and jid then local node, host = jid_split(jid); if host == origin.host then -- only include contacts which are on this component reply:tag("item", { jid = jid, subscription = item.subscription, ask = item.ask, name = item.name, }); for group in pairs(item.groups) do reply:tag("group"):text(group):up(); end reply:up(); -- move out from item end end end origin.send(reply); --origin.interested = true; -- resource is interested in roster updates return true; end end); module:hook("iq-set/bare/jabber:iq:roster:query", function(event) local session, stanza = event.origin, event.stanza; if not(session.type == "component" and stanza.attr.from == session.host) then return; end local from_node, from_host = jid_split(stanza.attr.to); if not(user_exists(from_node, from_host)) then return; end local roster = load_roster(from_node, from_host); if not(roster) then return; end local query = stanza.tags[1]; if #query.tags == 1 and query.tags[1].name == "item" and query.tags[1].attr.xmlns == "jabber:iq:roster" and query.tags[1].attr.jid -- Protection against overwriting roster.pending, until we move it and query.tags[1].attr.jid ~= "pending" then local item = query.tags[1]; local jid = jid_prep(item.attr.jid); local node, host, resource = jid_split(jid); if not resource and host == session.host then if jid ~= stanza.attr.to then -- not self-jid if item.attr.subscription == "remove" then local r_item = roster[jid]; if r_item then local to_bare = node and (node.."@"..host) or host; -- bare JID --if r_item.subscription == "both" or r_item.subscription == "from" or (roster.pending and roster.pending[jid]) then -- module:send(st.presence({type="unsubscribed", from=stanza.attr.to, to=to_bare})); --end --if r_item.subscription == "both" or r_item.subscription == "to" or r_item.ask then -- module:send(st.presence({type="unsubscribe", from=stanza.attr.to, to=to_bare})); --end roster[jid] = nil; if save_roster(from_node, from_host, roster) then session.send(st.reply(stanza)); rm_roster_push(from_node, from_host, jid); else roster[jid] = item; session.send(st.error_reply(stanza, "wait", "internal-server-error", "Unable to save roster")); end else session.send(st.error_reply(stanza, "modify", "item-not-found")); end else local subscription = item.attr.subscription; if subscription ~= "both" and subscription ~= "to" and subscription ~= "from" and subscription ~= "none" then -- TODO error on invalid subscription = roster[jid] and roster[jid].subscription or "none"; end local r_item = {name = item.attr.name, groups = {}}; if r_item.name == "" then r_item.name = nil; end r_item.subscription = subscription; if subscription ~= "both" and subscription ~= "to" then r_item.ask = roster[jid] and roster[jid].ask; end for _, child in ipairs(item) do if child.name == "group" then local text = t_concat(child); if text and text ~= "" then r_item.groups[text] = true; end end end local olditem = roster[jid]; roster[jid] = r_item; if save_roster(from_node, from_host, roster) then -- Ok, send success session.send(st.reply(stanza)); -- and push change to all resources rm_roster_push(from_node, from_host, jid); else -- Adding to roster failed roster[jid] = olditem; session.send(st.error_reply(stanza, "wait", "internal-server-error", "Unable to save roster")); end end else -- Trying to add self to roster session.send(st.error_reply(stanza, "cancel", "not-allowed")); end else -- Invalid JID added to roster session.send(st.error_reply(stanza, "modify", "bad-request")); -- FIXME what's the correct error? end else -- Roster set didn't include a single item, or its name wasn't 'item' session.send(st.error_reply(stanza, "modify", "bad-request")); end return true; end); function component_roster_push(node, host, jid) local roster = load_roster(node, host); if roster then local item = roster[jid]; local contact_node, contact_host = jid_split(jid); local stanza = st.iq({ type="set", from=node.."@"..host, to=contact_host, id = new_id() }):query("jabber:iq:roster"); if item then stanza:tag("item", { jid = jid, subscription = item.subscription, name = item.name, ask = item.ask }); for group in pairs(item.groups) do stanza:tag("group"):text(group):up(); end else stanza:tag("item", {jid = jid, subscription = "remove"}); end stanza:up(); -- move out from item stanza:up(); -- move out from stanza module:send(stanza); end end module:hook("iq-set/bare/jabber:iq:roster:query", function(event) local origin, stanza = event.origin, event.stanza; local query = stanza.tags[1]; local item = query.tags[1]; local contact_jid = item and item.name == "item" and item.attr.jid ~= "pending" and item.attr.jid; if contact_jid then local contact_node, contact_host = jid_split(contact_jid); if hosts[contact_host] and hosts[contact_host].type == "component" then local node, host = jid_split(stanza.attr.to or origin.full_jid); add_task(0, function() component_roster_push(node, host, contact_jid); end); end end end, 100);