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 $sql_prefix = "ttrss_plugin_af_sort_bayes";
  10
  11         function about() {
  12                 return array(1.0,
  13                         "Bayesian classifier for tt-rss (WIP)",
  14                         "fox");
  15         }
  16
  17         function init($host) {
  18                 require_once __DIR__ . "/lib/class.naivebayesian.php";
  19                 require_once __DIR__ . "/lib/class.naivebayesianstorage.php";
  20
  21                 $this->host = $host;
  22                 $this->dbh = Db::get();
  23
  24                 $this->init_database();
  25
  26                 $host->add_hook($host::HOOK_ARTICLE_FILTER, $this);
  27                 $host->add_hook($host::HOOK_PREFS_TAB, $this);
  28                 $host->add_hook($host::HOOK_ARTICLE_BUTTON, $this);
  29
  30         }
  31
  32         function trainArticle() {
  33                 $article_id = (int) $_REQUEST["article_id"];
  34                 $train_up = sql_bool_to_bool($_REQUEST["train_up"]);
  35
  36                 $category = $train_up ? "GOOD" : "NEUTRAL";
  37
  38                 $nbs = new NaiveBayesianStorage($_SESSION["uid"]);
  39                 $nb = new NaiveBayesian($nbs);
  40
  41                 $result = $this->dbh->query("SELECT score, guid, title, content FROM ttrss_entries, ttrss_user_entries WHERE ref_id = id AND id = " .
  42                         $article_id . " AND owner_uid = " . $_SESSION["uid"]);
  43
  44                 if ($this->dbh->num_rows($result) != 0) {
  45                         $guid = $this->dbh->fetch_result($result, 0, "guid");
  46                         $title = $this->dbh->fetch_result($result, 0, "title");
  47                         $content = mb_strtolower($title . " " . strip_tags($this->dbh->fetch_result($result, 0, "content")));
  48                         $score = $this->dbh->fetch_result($result, 0, "score");
  49
  50                         $this->dbh->query("BEGIN");
  51
  52                         if ($nb->untrain($guid, $content)) {
  53                                 if ($score >= $this->score_modifier) $score -= $this->score_modifier;
  54                         }
  55
  56                         $nb->train($guid, $nbs->getCategoryByName($category), $content);
  57
  58                         if ($category == "GOOD") $score += $this->score_modifier;
  59
  60                         $this->dbh->query("UPDATE ttrss_user_entries SET score = '$score' WHERE ref_id = $article_id AND owner_uid = " . $_SESSION["uid"]);
  61
  62                         $nb->updateProbabilities();
  63
  64                         $this->dbh->query("COMMIT");
  65
  66                 }
  67
  68                 print "$article_id :: $category";
  69         }
  70
  71         function get_js() {
  72                 return file_get_contents(__DIR__ . "/init.js");
  73         }
  74
  75         function get_prefs_js() {
  76                 return file_get_contents(__DIR__ . "/init.js");
  77         }
  78
  79         function hook_article_button($line) {
  80                 return "<img src=\"plugins/af_sort_bayes/thumb_up.png\"
  81                         style=\"cursor : pointer\" style=\"cursor : pointer\"
  82                         onclick=\"bayesTrain(".$line["id"].", true)\"
  83                         class='tagsPic' title='".__('+1')."'>" .
  84                 "<img src=\"plugins/af_sort_bayes/thumb_down.png\"
  85                         style=\"cursor : pointer\" style=\"cursor : pointer\"
  86                         onclick=\"bayesTrain(".$line["id"].", false)\"
  87                         class='tagsPic' title='".__('-1')."'>";
  88
  89         }
  90
  91         function init_database() {
  92                 $prefix = $this->sql_prefix;
  93
  94                 // TODO there probably should be a way for plugins to determine their schema version to upgrade tables
  95
  96                 /*$this->dbh->query("DROP TABLE IF EXISTS ${prefix}_wordfreqs", false);
  97                 $this->dbh->query("DROP TABLE IF EXISTS ${prefix}_references", false);
  98                 $this->dbh->query("DROP TABLE IF EXISTS ${prefix}_categories", false);*/
  99
 100                 $this->dbh->query("BEGIN");
 101
 102                 // PG only for the time being
 103
 104                 if (DB_TYPE == "mysql") {
 105
 106                         $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_categories (
 107                                 id INTEGER NOT NULL PRIMARY KEY auto_increment,
 108                                 category varchar(100) NOT NULL DEFAULT '',
 109                                 probability DOUBLE NOT NULL DEFAULT '0',
 110                                 owner_uid INTEGER NOT NULL,
 111                                 FOREIGN KEY (owner_uid) REFERENCES ttrss_users(id) ON DELETE CASCADE,
 112                                 word_count BIGINT NOT NULL DEFAULT '0') ENGINE=InnoDB");
 113
 114                         $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_references (
 115                                 id INTEGER NOT NULL PRIMARY KEY auto_increment,
 116                                 document_id VARCHAR(255) NOT NULL,
 117                                 category_id INTEGER NOT NULL,
 118                                 FOREIGN KEY (category_id) REFERENCES ${prefix}_categories(id) ON DELETE CASCADE,
 119                                 owner_uid INTEGER NOT NULL,
 120                                 FOREIGN KEY (owner_uid) REFERENCES ttrss_users(id) ON DELETE CASCADE,
 121                                 content text NOT NULL) ENGINE=InnoDB");
 122
 123                         $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_wordfreqs (
 124                                 word varchar(100) NOT NULL DEFAULT '',
 125                                 category_id INTEGER NOT NULL,
 126                                 FOREIGN KEY (category_id) REFERENCES ${prefix}_categories(id) ON DELETE CASCADE,
 127                                 owner_uid INTEGER NOT NULL,
 128                                 FOREIGN KEY (owner_uid) REFERENCES ttrss_users(id) ON DELETE CASCADE,
 129                                 count BIGINT NOT NULL DEFAULT '0') ENGINE=InnoDB");
 130
 131
 132                 } else {
 133                         $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_categories (
 134                                 id SERIAL NOT NULL PRIMARY KEY,
 135                                 category varchar(100) NOT NULL DEFAULT '',
 136                                 probability DOUBLE NOT NULL DEFAULT '0',
 137                                 owner_uid INTEGER NOT NULL REFERENCES ttrss_users(id) ON DELETE CASCADE,
 138                                 word_count BIGINT NOT NULL DEFAULT '0')");
 139
 140                         $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_references (
 141                                 id SERIAL NOT NULL PRIMARY KEY,
 142                                 document_id VARCHAR(255) NOT NULL,
 143                                 category_id INTEGER NOT NULL REFERENCES ${prefix}_categories(id) ON DELETE CASCADE,
 144                                 owner_uid INTEGER NOT NULL REFERENCES ttrss_users(id) ON DELETE CASCADE,
 145                                 content text NOT NULL)");
 146
 147                         $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_wordfreqs (
 148                                 word varchar(100) NOT NULL DEFAULT '',
 149                                 category_id INTEGER NOT NULL REFERENCES ${prefix}_categories(id) ON DELETE CASCADE,
 150                                 owner_uid INTEGER NOT NULL REFERENCES ttrss_users(id) ON DELETE CASCADE,
 151                                 count BIGINT NOT NULL DEFAULT '0')");
 152                 }
 153
 154                 $owner_uid = @$_SESSION["uid"];
 155
 156                 if ($owner_uid) {
 157                         $result = $this->dbh->query("SELECT id FROM ${prefix}_categories WHERE owner_uid = $owner_uid LIMIT 1");
 158
 159                         if ($this->dbh->num_rows($result) == 0) {
 160                                 $this->dbh->query("INSERT INTO ${prefix}_categories (category, owner_uid) VALUES ('GOOD', $owner_uid)");
 161                                 $this->dbh->query("INSERT INTO ${prefix}_categories (category, owner_uid) VALUES ('NEUTRAL', $owner_uid)");
 162                         }
 163                 }
 164
 165                 $this->dbh->query("COMMIT");
 166         }
 167
 168         function hook_prefs_tab($args) {
 169                 if ($args != "prefPrefs") return;
 170
 171                 print "<div dojoType=\"dijit.layout.AccordionPane\" title=\"".__('Bayesian classifier (af_sort_bayes)')."\">";
 172
 173                 $result = $this->dbh->query("SELECT category, probability, word_count FROM {$this->sql_prefix}_categories WHERE owner_uid = " . $_SESSION["uid"]);
 174
 175                 print "<table>";
 176                 print "<tr><th>Category</th><th>Probability</th><th>Word count</th></tr>";
 177
 178                 while ($line = $this->dbh->fetch_assoc($result)) {
 179                         print "<tr>";
 180                         foreach ($line as $k => $v) {
 181                                 if ($k == "probability") $v = sprintf("%.3f", $v);
 182
 183                                 print "<td>$v</td>";
 184                         }
 185                         print "</tr>";
 186                 }
 187
 188                 print "</table>";
 189
 190                 print "<button dojoType=\"dijit.form.Button\" onclick=\"return bayesClearDatabase()\">".
 191                         __('Clear database')."</button> ";
 192
 193                 //
 194
 195                 print "</div>";
 196         }
 197
 198         function hook_article_filter($article) {
 199                 $owner_uid = $article["owner_uid"];
 200
 201                 $nbs = new NaiveBayesianStorage($owner_uid);
 202                 $nb = new NaiveBayesian($nbs);
 203
 204                 $categories = $nbs->getCategories();
 205
 206                 if (count($categories) > 0) {
 207
 208                         $count_neutral = 0;
 209                         $count_good = 0;
 210                         $id_good = 0;
 211                         $id_neutral = 0;
 212
 213                         foreach ($categories as $id => $cat) {
 214                                 if ($cat["category"] == "GOOD") {
 215                                         $id_good = $id;
 216                                         $count_good += $cat["word_count"];
 217                                 } else if ($cat["category"] == "NEUTRAL") {
 218                                         $id_neutral = $id;
 219                                         $count_neutral += $cat["word_count"];
 220                                 }
 221                         }
 222
 223                         $dst_category = $id_neutral;
 224
 225                         $bayes_content = mb_strtolower($article["title"] . " " . strip_tags($article["content"]));
 226
 227                         if ($count_neutral >= 3000 && $count_good >= 1000) {
 228                                 // enable automatic categorization
 229
 230                                 $result = $nb->categorize($bayes_content);
 231
 232                                 if (count($result) == 2) {
 233                                         $prob_good = $result[$id_good];
 234                                         $prob_neutral = $result[$id_neutral];
 235
 236                                         if ($prob_good > 0.90 && $prob_good > $prob_neutral) {
 237                                                 $dst_category = $id_good; // should we autofile as good or not? idk
 238                                                 $article["score_modifier"] += $this->score_modifier;
 239                                         }
 240                                 }
 241                         }
 242
 243                         $nb->train($article["guid_hashed"], $dst_category, $bayes_content);
 244
 245                         $nb->updateProbabilities();
 246                 }
 247
 248                 return $article;
 249
 250         }
 251
 252         function clearDatabase() {
 253                 $prefix = $this->sql_prefix;
 254
 255                 $this->dbh->query("BEGIN");
 256                 $this->dbh->query("DELETE FROM ${prefix}_references WHERE owner_uid = " . $_SESSION["uid"]);
 257                 $this->dbh->query("DELETE FROM ${prefix}_wordfreqs WHERE owner_uid = " . $_SESSION["uid"]);
 258                 $this->dbh->query("COMMIT");
 259
 260                 $nbs = new NaiveBayesianStorage($_SESSION["uid"]);
 261                 $nb = new NaiveBayesian($nbs);
 262                 $nb->updateProbabilities();
 263         }
 264
 265         function api_version() {
 266                 return 2;
 267         }
 268
 269 }
 270 ?>