Transform nicknames to avoid issues with @username markdown syntax

cli/_lib/users.js

@@ -3,17 +3,22 @@
 
 'use strict';
 
-const Promise       = require('bluebird');
-const mongoose      = require('mongoose');
-const html_unescape = require('nodeca.vbconvert/lib/html_unescape_entities');
-const progress      = require('./utils').progress;
+const Promise        = require('bluebird');
+const mongoose       = require('mongoose');
+const html_unescape  = require('nodeca.vbconvert/lib/html_unescape_entities');
+const nick_transform = require('nodeca.vbconvert/lib/nick_transform');
+const progress       = require('./utils').progress;
 
 const UNCONFIRMED = 3;
 const MEMBERS     = 11;
 const VIOLATORS   = 12;
 
 
 module.exports = async function (N) {
+  let bot = await N.models.users.User.findOne()
+                      .where('hid').equals(N.config.bots.default_bot_hid)
+                      .lean(true);
+
   let usergroups = await N.models.vbconvert.UserGroupMapping.find().lean(true);
 
   let mongoid = {};
@@ -24,6 +29,24 @@ module.exports = async function (N) {
 
   let conn = await N.vbconvert.getConnection();
 
+  let rows = (await conn.query('SELECT userid, username FROM user ORDER BY userid ASC'))[0];
+  let username_mapping = {};
+  let nicks_seen = new Set(rows.map(row => nick_transform.normalize(row.username)));
+
+  for (let { userid, username } of rows) {
+    let nick = html_unescape(username);
+
+    if (!nick_transform.valid(nick)) {
+      username_mapping[userid] = nick_transform(nick, nicks_seen, nick_transform.rules);
+    }
+  }
+
+  N.logger.info('Username mapping created ' +
+    `(changing ${Object.keys(username_mapping).length} out of ${rows.length} nicknames)`);
+
+  // remove old records of name change in case import is restarted
+  await N.models.users.UserNickChange.deleteMany({});
+
   let gi_rows = (await conn.query('SELECT value FROM setting WHERE varname = "globalignore" LIMIT 1'))[0];
 
   let hellbanned_ids = [];
@@ -32,7 +55,7 @@ module.exports = async function (N) {
     hellbanned_ids = gi_rows[0].value.split(' ').map(Number);
   }
 
-  let rows = (await conn.query(`
+  rows = (await conn.query(`
     SELECT userid,usergroupid,membergroupids,username,email,password,salt,
            passworddate,ipaddress,joindate,lastactivity,icq,skype,
            CAST(birthday_search as char) as birthday,
@@ -58,7 +81,7 @@ module.exports = async function (N) {
     }
 
     user.hid            = row.userid;
-    user.nick           = html_unescape(row.username);
+    user.nick           = username_mapping[row.userid] || html_unescape(row.username);
     user.email          = row.email;
     user.joined_ts      = new Date(row.joindate * 1000);
     user.joined_ip      = row.ipaddress;
@@ -124,6 +147,15 @@ module.exports = async function (N) {
 
     await user.save();
 
+    if (username_mapping[row.userid]) {
+      await new N.models.users.UserNickChange({
+        from:     bot._id,
+        user:     user._id,
+        old_nick: html_unescape(row.username),
+        new_nick: username_mapping[row.userid]
+      }).save();
+    }
+
     let authProvider = new N.models.users.AuthProvider();
 
     authProvider.user    = user._id;

lib/nick_transform.js

@@ -0,0 +1,91 @@
+'use strict';
+
+const regexp = /(?=.{2,})^=?[\p{L}\d]+(?:[-_=][\p{L}\d]+)*[#=]*$/u;
+
+let rules = [];
+
+rules.push([ n => n.replace(/\$+/, m => 'S'.repeat(m.length)), 10 ]);
+rules.push([ n => n.replace(/^Ⓔⓤⓖⓔⓔ$/, 'Eugee'), 10 ]);
+rules.push([ n => n.replace(/\][I|]\[|\}[I|]\{|\)[I|]\(/ug, 'Ж'), 10 ]);
+rules.push([ n => n.replace(/\]\[|\}\{/ug, 'X'), 10 ]);
+rules.push([ n => n.replace(/\(\)/ug, 'O'), 10 ]);
+rules.push([ n => n.replace(/\|\{/ug, 'K'), 11 ]);
+rules.push([ n => n.replace(/@/g, () => (n.indexOf('.') === -1 ? 'A' : '@')), 10 ]);
+rules.push([ n => n.replace(/@[a-z0-9]+\.[a-z]*$/i, ''), 10 ]);
+rules.push([ n => n.replace(/^&&&$/, 'AAA'), 10 ]);
+rules.push([ n => n.replace(/^\^\^\^$/, 'aaa'), 10 ]);
+
+rules.push([ n => n.replace(/^[^-_=\p{L}\d]+/ug, ''), 100 ]);
+rules.push([ n => n.replace(/[^-_=\p{L}\d]+$/ug, ''), 1000 ]);
+
+rules.push([ n => n.replace(/[^\$@\-_=\p{L}\d]/ug, '_'), 10000 ]);
+rules.push([ n => n.replace(/[^\$@\-_=\p{L}\d]/ug, '-'), 11000 ]);
+rules.push([ n => n.replace(/(\p{L}\d)[^\$@\-_=\p{L}\d](\p{L}\d)/ug, '$1=$2'), 12000 ]);
+
+rules.push([ n => n.replace(/^[-_]+|[-_]+$/ug, ''), 100 ]);
+rules.push([ n => n.replace(/[-_=]{2,}/ug, '_'), 1000 ]);
+rules.push([ n => n.replace(/[-_=]{2,}/ug, '-'), 1100 ]);
+rules.push([ n => n.replace(/[-_=]{2,}/ug, '='), 20000 ]);
+
+rules.push([ n => n.replace(/$/, '='), 100000 ]);
+rules.push([ n => n.replace(/$/, '#'), 110000 ]);
+rules.push([ n => n.replace(/^(.*)$/, '=$1#'), 150000 ]);
+rules.push([ n => n.replace(/^(.*)$/, '=$1='), 160000 ]);
+
+rules = rules.sort((a, b) => a[1] - b[1]);
+
+function valid(nick)     { return nick.match(regexp); }
+function normalize(nick) { return nick.toUpperCase().toLowerCase(); }
+
+function transform_nick(nick, nicks_seen, rules/*, debug = null*/) {
+  let visited = new Set();
+
+  let weights = new Map();
+  weights.set(nick, 0);
+
+  /*let paths;
+
+  if (debug) {
+    paths = new Map();
+    paths.set(nick, []);
+  }*/
+
+  for (;;) {
+    let min_entry = [ '', Infinity ];
+
+    for (let entry of weights.entries()) {
+      if (entry[1] >= min_entry[1]) continue;
+      if (visited.has(entry[0])) continue;
+      min_entry = entry;
+    }
+
+    let [ current_nick, current_weight ] = min_entry;
+    let nick_norm = normalize(current_nick);
+
+    if (valid(current_nick) && !nicks_seen.has(nick_norm)) {
+      nicks_seen.add(nick_norm);
+      //if (debug) return [ current_nick, current_weight, paths.get(current_nick) ];
+      return current_nick;
+    }
+
+    for (let i = 0; i < rules.length; i++) {
+      let [ fn, weight ] = rules[i];
+      let new_nick = fn(current_nick);
+      let new_weight = current_weight + weight;
+      let existing_weight = weights.get(new_nick);
+
+      if (typeof existing_weight === 'undefined' || new_weight < existing_weight) {
+        weights.set(new_nick, new_weight);
+        //if (debug) paths.set(new_nick, paths.get(current_nick).concat([ i ]));
+      }
+    }
+
+    visited.add(current_nick);
+  }
+}
+
+module.exports = transform_nick;
+
+module.exports.valid     = valid;
+module.exports.normalize = normalize;
+module.exports.rules     = rules;