plugins/af_sort_bayes/init.php

   1 <?php
   2
   3 class Af_Sort_Bayes extends Plugin {
   4
   5         private $host;
   6         private $filters = array();
   7         private $dbh;
   8         private $score_modifier = 50;
   9         private $sql_prefix = "ttrss_plugin_af_sort_bayes";
  10         private $auto_categorize_threshold = 10000;
  11
  12         function about() {
  13                 return array(1.0,
  14                         "Bayesian classifier for tt-rss (WIP)",
  15                         "fox");
  16         }
  17
  18         function init($host) {
  19                 require_once __DIR__ . "/lib/class.naivebayesian.php";
  20                 //require_once __DIR__ . "/lib/class.naivebayesian_ngram.php";
  21                 require_once __DIR__ . "/lib/class.naivebayesianstorage.php";
  22
  23                 $this->host = $host;
  24                 $this->dbh = Db::get();
  25
  26                 $this->init_database();
  27
  28                 $host->add_hook($host::HOOK_ARTICLE_FILTER, $this);
  29                 $host->add_hook($host::HOOK_PREFS_TAB, $this);
  30                 $host->add_hook($host::HOOK_ARTICLE_BUTTON, $this);
  31
  32         }
  33
  34         function trainArticle() {
  35                 $article_id = (int) $_REQUEST["article_id"];
  36                 $train_up = sql_bool_to_bool($_REQUEST["train_up"]);
  37
  38                 //$category = $train_up ? "GOOD" : "UGLY";
  39                 $dst_category = "UGLY";
  40
  41                 $nbs = new NaiveBayesianStorage($_SESSION["uid"]);
  42                 $nb = new NaiveBayesian($nbs);
  43
  44                 $result = $this->dbh->query("SELECT score, guid, title, content FROM ttrss_entries, ttrss_user_entries WHERE ref_id = id AND id = " .
  45                         $article_id . " AND owner_uid = " . $_SESSION["uid"]);
  46
  47                 if ($this->dbh->num_rows($result) != 0) {
  48                         $guid = $this->dbh->fetch_result($result, 0, "guid");
  49                         $title = $this->dbh->fetch_result($result, 0, "title");
  50                         $content = mb_strtolower($title . " " . strip_tags($this->dbh->fetch_result($result, 0, "content")));
  51                         $score = $this->dbh->fetch_result($result, 0, "score");
  52
  53                         $this->dbh->query("BEGIN");
  54
  55                         $ref = $nbs->getReference($guid, false);
  56
  57                         if (isset($ref['category_id'])) {
  58                                 $current_category = $nbs->getCategoryById($ref['category_id']);
  59                         } else {
  60                                 $current_category = "UGLY";
  61                         }
  62
  63                         // set score to fixed value for now
  64
  65                         if ($train_up) {
  66                                 switch ($current_category) {
  67                                         case "UGLY":
  68                                                 $dst_category = "GOOD";
  69                                                 $score = $this->score_modifier;
  70                                                 break;
  71                                         case "BAD":
  72                                                 $dst_category = "UGLY";
  73                                                 $score = 0;
  74                                                 break;
  75                                         case "GOOD":
  76                                                 $dst_category = "GOOD";
  77                                                 break;
  78                                 }
  79                         } else {
  80                                 switch ($current_category) {
  81                                         case "UGLY":
  82                                                 $dst_category = "BAD";
  83                                                 $score = -$this->score_modifier;
  84                                                 break;
  85                                         case "BAD":
  86                                                 $dst_category = "BAD";
  87                                                 break;
  88                                         case "GOOD":
  89                                                 $dst_category = "UGLY";
  90                                                 $score = 0;
  91                                                 break;
  92                                 }
  93                         }
  94
  95                         $nb->untrain($guid, $content);
  96                         $nb->train($guid, $nbs->getCategoryByName($dst_category), $content);
  97
  98                         $this->dbh->query("UPDATE ttrss_user_entries SET score = '$score' WHERE ref_id = $article_id AND owner_uid = " . $_SESSION["uid"]);
  99
 100                         $nb->updateProbabilities();
 101
 102                         $this->dbh->query("COMMIT");
 103
 104                 }
 105
 106                 print "$article_id :: $dst_category :: $score";
 107         }
 108
 109         function get_js() {
 110                 return file_get_contents(__DIR__ . "/init.js");
 111         }
 112
 113         function get_prefs_js() {
 114                 return file_get_contents(__DIR__ . "/init.js");
 115         }
 116
 117         function hook_article_button($line) {
 118                 return "<img src=\"plugins/af_sort_bayes/thumb_up.png\"
 119                         style=\"cursor : pointer\" style=\"cursor : pointer\"
 120                         onclick=\"bayesTrain(".$line["id"].", true)\"
 121                         class='tagsPic' title='".__('+1')."'>" .
 122                 "<img src=\"plugins/af_sort_bayes/thumb_down.png\"
 123                         style=\"cursor : pointer\" style=\"cursor : pointer\"
 124                         onclick=\"bayesTrain(".$line["id"].", false)\"
 125                         class='tagsPic' title='".__('-1')."'>";
 126
 127         }
 128
 129         function init_database() {
 130                 $prefix = $this->sql_prefix;
 131
 132                 // TODO there probably should be a way for plugins to determine their schema version to upgrade tables
 133
 134                 /*$this->dbh->query("DROP TABLE IF EXISTS ${prefix}_wordfreqs", false);
 135                 $this->dbh->query("DROP TABLE IF EXISTS ${prefix}_references", false);
 136                 $this->dbh->query("DROP TABLE IF EXISTS ${prefix}_categories", false);*/
 137
 138                 $this->dbh->query("BEGIN");
 139
 140                 // PG only for the time being
 141
 142                 if (DB_TYPE == "mysql") {
 143
 144                         $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_categories (
 145                                 id INTEGER NOT NULL PRIMARY KEY auto_increment,
 146                                 category varchar(100) NOT NULL DEFAULT '',
 147                                 probability DOUBLE NOT NULL DEFAULT '0',
 148                                 owner_uid INTEGER NOT NULL,
 149                                 FOREIGN KEY (owner_uid) REFERENCES ttrss_users(id) ON DELETE CASCADE,
 150                                 word_count BIGINT NOT NULL DEFAULT '0') ENGINE=InnoDB");
 151
 152                         $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_references (
 153                                 id INTEGER NOT NULL PRIMARY KEY auto_increment,
 154                                 document_id VARCHAR(255) NOT NULL,
 155                                 category_id INTEGER NOT NULL,
 156                                 FOREIGN KEY (category_id) REFERENCES ${prefix}_categories(id) ON DELETE CASCADE,
 157                                 owner_uid INTEGER NOT NULL,
 158                                 FOREIGN KEY (owner_uid) REFERENCES ttrss_users(id) ON DELETE CASCADE) ENGINE=InnoDB");
 159
 160                         $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_wordfreqs (
 161                                 word varchar(100) NOT NULL DEFAULT '',
 162                                 category_id INTEGER NOT NULL,
 163                                 FOREIGN KEY (category_id) REFERENCES ${prefix}_categories(id) ON DELETE CASCADE,
 164                                 owner_uid INTEGER NOT NULL,
 165                                 FOREIGN KEY (owner_uid) REFERENCES ttrss_users(id) ON DELETE CASCADE,
 166                                 count BIGINT NOT NULL DEFAULT '0') ENGINE=InnoDB");
 167
 168
 169                 } else {
 170                         $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_categories (
 171                                 id SERIAL NOT NULL PRIMARY KEY,
 172                                 category varchar(100) NOT NULL DEFAULT '',
 173                                 probability DOUBLE PRECISION NOT NULL DEFAULT '0',
 174                                 owner_uid INTEGER NOT NULL REFERENCES ttrss_users(id) ON DELETE CASCADE,
 175                                 word_count BIGINT NOT NULL DEFAULT '0')");
 176
 177                         $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_references (
 178                                 id SERIAL NOT NULL PRIMARY KEY,
 179                                 document_id VARCHAR(255) NOT NULL,
 180                                 category_id INTEGER NOT NULL REFERENCES ${prefix}_categories(id) ON DELETE CASCADE,
 181                                 owner_uid INTEGER NOT NULL REFERENCES ttrss_users(id) ON DELETE CASCADE)");
 182
 183                         $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_wordfreqs (
 184                                 word varchar(100) NOT NULL DEFAULT '',
 185                                 category_id INTEGER NOT NULL REFERENCES ${prefix}_categories(id) ON DELETE CASCADE,
 186                                 owner_uid INTEGER NOT NULL REFERENCES ttrss_users(id) ON DELETE CASCADE,
 187                                 count BIGINT NOT NULL DEFAULT '0')");
 188                 }
 189
 190                 $owner_uid = @$_SESSION["uid"];
 191
 192                 if ($owner_uid) {
 193                         $result = $this->dbh->query("SELECT id FROM ${prefix}_categories WHERE owner_uid = $owner_uid LIMIT 1");
 194
 195                         if ($this->dbh->num_rows($result) == 0) {
 196                                 $this->dbh->query("INSERT INTO ${prefix}_categories (category, owner_uid) VALUES ('GOOD', $owner_uid)");
 197                                 $this->dbh->query("INSERT INTO ${prefix}_categories (category, owner_uid) VALUES ('BAD', $owner_uid)");
 198                                 $this->dbh->query("INSERT INTO ${prefix}_categories (category, owner_uid) VALUES ('UGLY', $owner_uid)");
 199                         }
 200                 }
 201
 202                 $this->dbh->query("COMMIT");
 203         }
 204
 205         function renderPrefsUI() {
 206                 $result = $this->dbh->query("SELECT category, probability, word_count,
 207                         (SELECT COUNT(id) FROM {$this->sql_prefix}_references WHERE
 208                                 category_id = {$this->sql_prefix}_categories.id) as doc_count
 209                         FROM {$this->sql_prefix}_categories WHERE owner_uid = " . $_SESSION["uid"]);
 210
 211                 print "<h3>" . __("Statistics") . "</h3>";
 212
 213                 print "<p>".T_sprintf("Required UGLY word count for automatic matching: %d", $this->auto_categorize_threshold)."</p>";
 214
 215                 print "<table>";
 216                 print "<tr><th>Category</th><th>Probability</th><th>Words</th><th>Articles</th></tr>";
 217
 218                 while ($line = $this->dbh->fetch_assoc($result)) {
 219                         print "<tr>";
 220                         foreach ($line as $k => $v) {
 221                                 if ($k == "probability") $v = sprintf("%.3f", $v);
 222
 223                                 print "<td>$v</td>";
 224                         }
 225                         print "</tr>";
 226                 }
 227
 228                 print "</table>";
 229
 230                 print "<h3>" . __("Last matched articles") . "</h3>";
 231
 232                 $result = $this->dbh->query("SELECT te.title, category, tf.title AS feed_title
 233                         FROM ttrss_entries AS te, ttrss_user_entries AS tu, ttrss_feeds AS tf, {$this->sql_prefix}_references AS tr, {$this->sql_prefix}_categories AS tc
 234                         WHERE tf.id = tu.feed_id AND tu.ref_id = te.id AND tc.id = tr.category_id AND tr.document_id = te.guid ORDER BY te.id DESC LIMIT 20");
 235
 236                 print "<ul class=\"browseFeedList\" style=\"border-width : 1px\">";
 237
 238                 while ($line = $this->dbh->fetch_assoc($result)) {
 239                         print "<li>" . $line["category"] . ": " . $line["title"] . " (" . $line["feed_title"] . ")</li>";
 240                 }
 241
 242                 print "</ul>";
 243
 244                 print "<button dojoType=\"dijit.form.Button\" onclick=\"return bayesUpdateUI()\">".
 245                         __('Refresh')."</button> ";
 246
 247                 print "<button dojoType=\"dijit.form.Button\" onclick=\"return bayesClearDatabase()\">".
 248                         __('Clear database')."</button> ";
 249
 250                 //
 251         }
 252
 253         function hook_prefs_tab($args) {
 254                 if ($args != "prefPrefs") return;
 255
 256                 print "<div id=\"af_sort_bayes_prefs\" dojoType=\"dijit.layout.AccordionPane\" title=\"".__('Bayesian classifier (af_sort_bayes)')."\">";
 257
 258                 $this->renderPrefsUI();
 259
 260                 print "</div>";
 261         }
 262
 263         function hook_article_filter($article) {
 264                 $owner_uid = $article["owner_uid"];
 265
 266                 // guid already includes owner_uid so we don't need to include it
 267                 $result = $this->dbh->query("SELECT id FROM {$this->sql_prefix}_references WHERE
 268                         document_id = '" . $this->dbh->escape_string($article['guid_hashed']) . "'");
 269
 270                 if (db_num_rows($result) != 0) {
 271                         _debug("bayes: article already categorized");
 272                         return $article;
 273                 }
 274
 275                 $nbs = new NaiveBayesianStorage($owner_uid);
 276                 $nb = new NaiveBayesian($nbs);
 277
 278                 $categories = $nbs->getCategories();
 279
 280                 if (count($categories) > 0) {
 281
 282                         $count_neutral = 0;
 283
 284                         $id_good = 0;
 285                         $id_ugly = 0;
 286                         $id_bad = 0;
 287
 288                         foreach ($categories as $id => $cat) {
 289                                 if ($cat["category"] == "GOOD") {
 290                                         $id_good = $id;
 291                                 } else if ($cat["category"] == "UGLY") {
 292                                         $id_ugly = $id;
 293                                         $count_neutral += $cat["word_count"];
 294                                 } else if ($cat["category"] == "BAD") {
 295                                         $id_bad = $id;
 296                                 }
 297                         }
 298
 299                         $dst_category = $id_ugly;
 300
 301                         $bayes_content = mb_strtolower($article["title"] . " " . strip_tags($article["content"]));
 302
 303                         if ($count_neutral >= $this->auto_categorize_threshold) {
 304                                 // enable automatic categorization
 305
 306                                 $result = $nb->categorize($bayes_content);
 307
 308                                 //print_r($result);
 309
 310                                 if (count($result) == 3) {
 311                                         $prob_good = $result[$id_good];
 312                                         $prob_bad = $result[$id_bad];
 313
 314                                         if ($prob_good > 0.90) {
 315                                                 $dst_category = $id_good;
 316                                                 $article["score_modifier"] += $this->score_modifier;
 317                                         } else if ($prob_bad > 0.90) {
 318                                                 $dst_category = $id_bad;
 319                                                 $article["score_modifier"] -= $this->score_modifier;
 320                                         }
 321                                 }
 322
 323                                 _debug("bayes, dst category: $dst_category");
 324                         }
 325
 326                         $nb->train($article["guid_hashed"], $dst_category, $bayes_content);
 327
 328                         $nb->updateProbabilities();
 329                 }
 330
 331                 return $article;
 332
 333         }
 334
 335         function clearDatabase() {
 336                 $prefix = $this->sql_prefix;
 337
 338                 $this->dbh->query("BEGIN");
 339                 $this->dbh->query("DELETE FROM ${prefix}_references WHERE owner_uid = " . $_SESSION["uid"]);
 340                 $this->dbh->query("DELETE FROM ${prefix}_wordfreqs WHERE owner_uid = " . $_SESSION["uid"]);
 341                 $this->dbh->query("COMMIT");
 342
 343                 $nbs = new NaiveBayesianStorage($_SESSION["uid"]);
 344                 $nb = new NaiveBayesian($nbs);
 345                 $nb->updateProbabilities();
 346         }
 347
 348         function api_version() {
 349                 return 2;
 350         }
 351
 352 }
 353 ?>