zulip/static/js/unread.js

329 lines
9.7 KiB
JavaScript

// See http://zulip.readthedocs.io/en/latest/pointer.html for notes on
// how this system is designed.
var unread = (function () {
var exports = {};
var unread_mentioned = new Dict();
exports.suppress_unread_counts = true;
exports.messages_read_in_narrow = false;
exports.unread_pm_counter = (function () {
var self = {};
var unread_privates = new Dict(); // indexed by user_ids_string like 5,7,9
self.clear = function () {
unread_privates = new Dict();
};
self.add = function (message) {
var user_ids_string = people.pm_reply_user_string(message);
if (user_ids_string) {
unread_privates.setdefault(user_ids_string, new Dict());
unread_privates.get(user_ids_string).set(message.id, true);
}
};
self.del = function (message) {
var user_ids_string = people.pm_reply_user_string(message);
if (user_ids_string) {
var dict = unread_privates.get(user_ids_string);
if (dict) {
dict.del(message.id);
}
}
};
self.get_counts = function () {
var pm_dict = new Dict(); // Hash by user_ids_string -> count
var total_count = 0;
unread_privates.each(function (obj, user_ids_string) {
var count = obj.num_items();
pm_dict.set(user_ids_string, count);
total_count += count;
});
return {
total_count: total_count,
pm_dict: pm_dict,
};
};
self.num_unread = function (user_ids_string) {
if (!user_ids_string) {
return 0;
}
if (!unread_privates.has(user_ids_string)) {
return 0;
}
return unread_privates.get(user_ids_string).num_items();
};
return self;
}());
exports.unread_topic_counter = (function () {
var self = {};
function str_dict() {
// Use this when keys are topics
return new Dict({fold_case: true});
}
function num_dict() {
// Use this for message ids.
return new Dict();
}
var unread_topics = num_dict(); // dict of stream -> topic -> msg id
self.clear = function () {
unread_topics = num_dict();
};
self.update = function (stream_id, topic, new_topic, msg_id) {
if (unread_topics.has(stream_id) &&
unread_topics.get(stream_id).has(topic) &&
unread_topics.get(stream_id).get(topic).get(msg_id)) {
// Move the unread topic count to the new topic
unread_topics.get(stream_id).get(topic).del(msg_id);
if (unread_topics.get(stream_id).get(topic).num_items() === 0) {
unread_topics.get(stream_id).del(topic);
}
unread_topics.get(stream_id).setdefault(new_topic, num_dict());
unread_topics.get(stream_id).get(new_topic).set(msg_id, true);
}
};
self.add = function (stream_id, topic, msg_id) {
unread_topics.setdefault(stream_id, str_dict());
unread_topics.get(stream_id).setdefault(topic, num_dict());
unread_topics.get(stream_id).get(topic).set(msg_id, true);
};
self.del = function (stream_id, topic, msg_id) {
var stream_dict = unread_topics.get(stream_id);
if (stream_dict) {
var topic_dict = stream_dict.get(topic);
if (topic_dict) {
topic_dict.del(msg_id);
}
}
};
self.get_counts = function () {
var res = {};
res.stream_unread_messages = 0;
res.stream_count = num_dict(); // hash by stream_id -> count
res.topic_count = num_dict(); // hash of hashes (stream_id, then topic -> count)
unread_topics.each(function (_, stream_id) {
// We track unread counts for streams that may be currently
// unsubscribed. Since users may re-subscribe, we don't
// completely throw away the data. But we do ignore it here,
// so that callers have a view of the **current** world.
var sub = stream_data.get_sub_by_id(stream_id);
if (!sub || !stream_data.is_subscribed(sub.name)) {
return true;
}
if (unread_topics.has(stream_id)) {
res.topic_count.set(stream_id, str_dict());
var stream_count = 0;
unread_topics.get(stream_id).each(function (msgs, topic) {
var topic_count = msgs.num_items();
res.topic_count.get(stream_id).set(topic, topic_count);
if (!muting.is_topic_muted(sub.name, topic)) {
stream_count += topic_count;
}
});
res.stream_count.set(stream_id, stream_count);
if (stream_data.in_home_view(stream_id)) {
res.stream_unread_messages += stream_count;
}
}
});
return res;
};
self.get_stream_count = function (stream_id) {
var stream_count = 0;
if (!unread_topics.has(stream_id)) {
return 0;
}
unread_topics.get(stream_id).each(function (msgs, topic) {
var sub = stream_data.get_sub_by_id(stream_id);
if (sub && !muting.is_topic_muted(sub.name, topic)) {
stream_count += msgs.num_items();
}
});
return stream_count;
};
self.get = function (stream_id, topic) {
var num_unread = 0;
if (unread_topics.has(stream_id) &&
unread_topics.get(stream_id).has(topic)) {
num_unread = unread_topics.get(stream_id).get(topic).num_items();
}
return num_unread;
};
self.topic_has_any_unread = function (stream_id, topic) {
var stream_dct = unread_topics.get(stream_id);
if (!stream_dct) {
return false;
}
var topic_dct = stream_dct.get(topic);
if (!topic_dct) {
return false;
}
return !topic_dct.is_empty();
};
return self;
}());
exports.message_unread = function (message) {
if (message === undefined) {
return false;
}
return message.flags === undefined ||
message.flags.indexOf('read') === -1;
};
exports.update_unread_topics = function (msg, event) {
if (event.subject !== undefined) {
exports.unread_topic_counter.update(
msg.stream_id,
msg.subject,
event.subject,
msg.id
);
}
};
exports.process_loaded_messages = function (messages) {
_.each(messages, function (message) {
var unread = exports.message_unread(message);
if (!unread) {
return;
}
if (message.type === 'private') {
exports.unread_pm_counter.add(message);
}
if (message.type === 'stream') {
exports.unread_topic_counter.add(
message.stream_id,
message.subject,
message.id
);
}
if (message.mentioned) {
unread_mentioned.set(message.id, true);
}
});
};
exports.process_read_message = function (message) {
if (message.type === 'private') {
exports.unread_pm_counter.del(message);
}
if (message.type === 'stream') {
if (message.stream_id === undefined) {
blueslip.error('No stream_id found for message ' + message.id);
return;
}
exports.unread_topic_counter.del(
message.stream_id,
message.subject,
message.id
);
}
unread_mentioned.del(message.id);
};
exports.declare_bankruptcy = function () {
exports.unread_pm_counter.clear();
exports.unread_topic_counter.clear();
unread_mentioned = new Dict();
};
exports.num_unread_current_messages = function () {
var num_unread = 0;
_.each(current_msg_list.all_messages(), function (msg) {
if ((msg.id > current_msg_list.selected_id()) && exports.message_unread(msg)) {
num_unread += 1;
}
});
return num_unread;
};
exports.get_counts = function () {
var res = {};
// Return a data structure with various counts. This function should be
// pretty cheap, even if you don't care about all the counts, and you
// should strive to keep it free of side effects on globals or DOM.
res.private_message_count = 0;
res.mentioned_message_count = unread_mentioned.num_items();
// This sets stream_count, topic_count, and home_unread_messages
var topic_res = exports.unread_topic_counter.get_counts();
res.home_unread_messages = topic_res.stream_unread_messages;
res.stream_count = topic_res.stream_count;
res.topic_count = topic_res.topic_count;
var pm_res = exports.unread_pm_counter.get_counts();
res.pm_count = pm_res.pm_dict;
res.private_message_count = pm_res.total_count;
res.home_unread_messages += pm_res.total_count;
if (narrow_state.active()) {
res.unread_in_current_view = exports.num_unread_current_messages();
} else {
res.unread_in_current_view = res.home_unread_messages;
}
return res;
};
exports.num_unread_for_stream = function (stream_id) {
return exports.unread_topic_counter.get_stream_count(stream_id);
};
exports.num_unread_for_topic = function (stream_id, subject) {
return exports.unread_topic_counter.get(stream_id, subject);
};
exports.topic_has_any_unread = function (stream_id, topic) {
return exports.unread_topic_counter.topic_has_any_unread(stream_id, topic);
};
exports.num_unread_for_person = function (user_ids_string) {
return exports.unread_pm_counter.num_unread(user_ids_string);
};
return exports;
}());
if (typeof module !== 'undefined') {
module.exports = unread;
}