]>
Commit | Line | Data |
---|---|---|
853cc128 AD |
1 | <?php |
2 | ||
3 | class Af_Sort_Bayes extends Plugin { | |
4 | ||
5 | private $host; | |
6 | private $filters = array(); | |
7 | private $dbh; | |
59e83455 | 8 | private $score_modifier = 50; |
3c43def9 | 9 | private $sql_prefix = "ttrss_plugin_af_sort_bayes"; |
fa05356a | 10 | private $auto_categorize_threshold = 10000; |
853cc128 AD |
11 | |
12 | function about() { | |
13 | return array(1.0, | |
14 | "Bayesian classifier for tt-rss (WIP)", | |
15 | "fox"); | |
16 | } | |
17 | ||
18 | function init($host) { | |
19 | require_once __DIR__ . "/lib/class.naivebayesian.php"; | |
4dbd303b | 20 | //require_once __DIR__ . "/lib/class.naivebayesian_ngram.php"; |
853cc128 AD |
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 | ||
08cfcba4 AD |
38 | //$category = $train_up ? "GOOD" : "UGLY"; |
39 | $dst_category = "UGLY"; | |
853cc128 | 40 | |
59e83455 | 41 | $nbs = new NaiveBayesianStorage($_SESSION["uid"]); |
4dbd303b | 42 | $nb = new NaiveBayesian($nbs); |
59e83455 AD |
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 | ||
08cfcba4 AD |
55 | $ref = $nbs->getReference($guid, false); |
56 | ||
57 | if (isset($ref['category_id'])) { | |
58 | $current_category = $nbs->getCategoryById($ref['category_id']); | |
59 | } else { | |
60 | $current_category = "UGLY"; | |
59e83455 AD |
61 | } |
62 | ||
08cfcba4 AD |
63 | // set score to fixed value for now |
64 | ||
65 | if ($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"; | |
21111f62 | 83 | $score = -$this->score_modifier; |
08cfcba4 AD |
84 | break; |
85 | case "BAD": | |
86 | $dst_category = "BAD"; | |
87 | break; | |
88 | case "GOOD": | |
89 | $dst_category = "UGLY"; | |
a72cd54c | 90 | $score = 0; |
08cfcba4 AD |
91 | break; |
92 | } | |
93 | } | |
59e83455 | 94 | |
08cfcba4 AD |
95 | $nb->untrain($guid, $content); |
96 | $nb->train($guid, $nbs->getCategoryByName($dst_category), $content); | |
59e83455 AD |
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 | $this->dbh->query("COMMIT"); | |
103 | ||
104 | } | |
105 | ||
08cfcba4 | 106 | print "$article_id :: $dst_category :: $score"; |
853cc128 AD |
107 | } |
108 | ||
109 | function get_js() { | |
110 | return file_get_contents(__DIR__ . "/init.js"); | |
111 | } | |
112 | ||
3c43def9 AD |
113 | function get_prefs_js() { |
114 | return file_get_contents(__DIR__ . "/init.js"); | |
115 | } | |
116 | ||
853cc128 AD |
117 | function hook_article_button($line) { |
118 | return "<img src=\"plugins/af_sort_bayes/thumb_up.png\" | |
119 | style=\"cursor : pointer\" style=\"cursor : pointer\" | |
5b0cb9aa | 120 | onclick=\"bayesTrain(".$line["id"].", true, event)\" |
853cc128 AD |
121 | class='tagsPic' title='".__('+1')."'>" . |
122 | "<img src=\"plugins/af_sort_bayes/thumb_down.png\" | |
123 | style=\"cursor : pointer\" style=\"cursor : pointer\" | |
5b0cb9aa | 124 | onclick=\"bayesTrain(".$line["id"].", false, event)\" |
168e32b9 AD |
125 | class='tagsPic' title='".__('-1')."'>" . |
126 | "<img src=\"plugins/af_sort_bayes/chart_bar.png\" | |
127 | style=\"cursor : pointer\" style=\"cursor : pointer\" | |
128 | onclick=\"bayesShow(".$line["id"].")\" | |
129 | class='tagsPic' title='".__('Show classifier info')."'>"; | |
853cc128 AD |
130 | |
131 | } | |
132 | ||
133 | function init_database() { | |
3c43def9 | 134 | $prefix = $this->sql_prefix; |
853cc128 | 135 | |
59e83455 AD |
136 | // TODO there probably should be a way for plugins to determine their schema version to upgrade tables |
137 | ||
138 | /*$this->dbh->query("DROP TABLE IF EXISTS ${prefix}_wordfreqs", false); | |
139 | $this->dbh->query("DROP TABLE IF EXISTS ${prefix}_references", false); | |
140 | $this->dbh->query("DROP TABLE IF EXISTS ${prefix}_categories", false);*/ | |
853cc128 AD |
141 | |
142 | $this->dbh->query("BEGIN"); | |
143 | ||
144 | // PG only for the time being | |
145 | ||
b02e8bc8 AD |
146 | if (DB_TYPE == "mysql") { |
147 | ||
148 | $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_categories ( | |
149 | id INTEGER NOT NULL PRIMARY KEY auto_increment, | |
150 | category varchar(100) NOT NULL DEFAULT '', | |
151 | probability DOUBLE NOT NULL DEFAULT '0', | |
152 | owner_uid INTEGER NOT NULL, | |
153 | FOREIGN KEY (owner_uid) REFERENCES ttrss_users(id) ON DELETE CASCADE, | |
154 | word_count BIGINT NOT NULL DEFAULT '0') ENGINE=InnoDB"); | |
155 | ||
156 | $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_references ( | |
157 | id INTEGER NOT NULL PRIMARY KEY auto_increment, | |
158 | document_id VARCHAR(255) NOT NULL, | |
159 | category_id INTEGER NOT NULL, | |
160 | FOREIGN KEY (category_id) REFERENCES ${prefix}_categories(id) ON DELETE CASCADE, | |
161 | owner_uid INTEGER NOT NULL, | |
61c9ce6a | 162 | FOREIGN KEY (owner_uid) REFERENCES ttrss_users(id) ON DELETE CASCADE) ENGINE=InnoDB"); |
b02e8bc8 AD |
163 | |
164 | $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_wordfreqs ( | |
165 | word varchar(100) NOT NULL DEFAULT '', | |
166 | category_id INTEGER NOT NULL, | |
167 | FOREIGN KEY (category_id) REFERENCES ${prefix}_categories(id) ON DELETE CASCADE, | |
168 | owner_uid INTEGER NOT NULL, | |
169 | FOREIGN KEY (owner_uid) REFERENCES ttrss_users(id) ON DELETE CASCADE, | |
170 | count BIGINT NOT NULL DEFAULT '0') ENGINE=InnoDB"); | |
171 | ||
172 | ||
173 | } else { | |
174 | $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_categories ( | |
175 | id SERIAL NOT NULL PRIMARY KEY, | |
176 | category varchar(100) NOT NULL DEFAULT '', | |
d62a5e0c | 177 | probability DOUBLE PRECISION NOT NULL DEFAULT '0', |
b02e8bc8 AD |
178 | owner_uid INTEGER NOT NULL REFERENCES ttrss_users(id) ON DELETE CASCADE, |
179 | word_count BIGINT NOT NULL DEFAULT '0')"); | |
180 | ||
181 | $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_references ( | |
182 | id SERIAL NOT NULL PRIMARY KEY, | |
183 | document_id VARCHAR(255) NOT NULL, | |
184 | category_id INTEGER NOT NULL REFERENCES ${prefix}_categories(id) ON DELETE CASCADE, | |
61c9ce6a | 185 | owner_uid INTEGER NOT NULL REFERENCES ttrss_users(id) ON DELETE CASCADE)"); |
b02e8bc8 AD |
186 | |
187 | $this->dbh->query("CREATE TABLE IF NOT EXISTS ${prefix}_wordfreqs ( | |
188 | word varchar(100) NOT NULL DEFAULT '', | |
189 | category_id INTEGER NOT NULL REFERENCES ${prefix}_categories(id) ON DELETE CASCADE, | |
190 | owner_uid INTEGER NOT NULL REFERENCES ttrss_users(id) ON DELETE CASCADE, | |
191 | count BIGINT NOT NULL DEFAULT '0')"); | |
192 | } | |
853cc128 | 193 | |
59e83455 AD |
194 | $owner_uid = @$_SESSION["uid"]; |
195 | ||
196 | if ($owner_uid) { | |
197 | $result = $this->dbh->query("SELECT id FROM ${prefix}_categories WHERE owner_uid = $owner_uid LIMIT 1"); | |
198 | ||
199 | if ($this->dbh->num_rows($result) == 0) { | |
200 | $this->dbh->query("INSERT INTO ${prefix}_categories (category, owner_uid) VALUES ('GOOD', $owner_uid)"); | |
08cfcba4 AD |
201 | $this->dbh->query("INSERT INTO ${prefix}_categories (category, owner_uid) VALUES ('BAD', $owner_uid)"); |
202 | $this->dbh->query("INSERT INTO ${prefix}_categories (category, owner_uid) VALUES ('UGLY', $owner_uid)"); | |
59e83455 AD |
203 | } |
204 | } | |
205 | ||
853cc128 AD |
206 | $this->dbh->query("COMMIT"); |
207 | } | |
208 | ||
308c55c0 | 209 | function renderPrefsUI() { |
4947c02e AD |
210 | $result = $this->dbh->query("SELECT category, probability, word_count, |
211 | (SELECT COUNT(id) FROM {$this->sql_prefix}_references WHERE | |
212 | category_id = {$this->sql_prefix}_categories.id) as doc_count | |
213 | FROM {$this->sql_prefix}_categories WHERE owner_uid = " . $_SESSION["uid"]); | |
3c43def9 | 214 | |
308c55c0 AD |
215 | print "<h3>" . __("Statistics") . "</h3>"; |
216 | ||
fa05356a AD |
217 | print "<p>".T_sprintf("Required UGLY word count for automatic matching: %d", $this->auto_categorize_threshold)."</p>"; |
218 | ||
3c43def9 | 219 | print "<table>"; |
308c55c0 | 220 | print "<tr><th>Category</th><th>Probability</th><th>Words</th><th>Articles</th></tr>"; |
3c43def9 AD |
221 | |
222 | while ($line = $this->dbh->fetch_assoc($result)) { | |
223 | print "<tr>"; | |
224 | foreach ($line as $k => $v) { | |
225 | if ($k == "probability") $v = sprintf("%.3f", $v); | |
226 | ||
227 | print "<td>$v</td>"; | |
228 | } | |
229 | print "</tr>"; | |
230 | } | |
231 | ||
232 | print "</table>"; | |
233 | ||
308c55c0 AD |
234 | print "<h3>" . __("Last matched articles") . "</h3>"; |
235 | ||
236 | $result = $this->dbh->query("SELECT te.title, category, tf.title AS feed_title | |
237 | FROM ttrss_entries AS te, ttrss_user_entries AS tu, ttrss_feeds AS tf, {$this->sql_prefix}_references AS tr, {$this->sql_prefix}_categories AS tc | |
238 | WHERE tf.id = tu.feed_id AND tu.ref_id = te.id AND tc.id = tr.category_id AND tr.document_id = te.guid ORDER BY te.id DESC LIMIT 20"); | |
239 | ||
240 | print "<ul class=\"browseFeedList\" style=\"border-width : 1px\">"; | |
241 | ||
242 | while ($line = $this->dbh->fetch_assoc($result)) { | |
243 | print "<li>" . $line["category"] . ": " . $line["title"] . " (" . $line["feed_title"] . ")</li>"; | |
244 | } | |
245 | ||
246 | print "</ul>"; | |
247 | ||
248 | print "<button dojoType=\"dijit.form.Button\" onclick=\"return bayesUpdateUI()\">". | |
249 | __('Refresh')."</button> "; | |
250 | ||
3c43def9 AD |
251 | print "<button dojoType=\"dijit.form.Button\" onclick=\"return bayesClearDatabase()\">". |
252 | __('Clear database')."</button> "; | |
853cc128 AD |
253 | |
254 | // | |
308c55c0 AD |
255 | } |
256 | ||
257 | function hook_prefs_tab($args) { | |
258 | if ($args != "prefPrefs") return; | |
259 | ||
260 | print "<div id=\"af_sort_bayes_prefs\" dojoType=\"dijit.layout.AccordionPane\" title=\"".__('Bayesian classifier (af_sort_bayes)')."\">"; | |
261 | ||
262 | $this->renderPrefsUI(); | |
853cc128 AD |
263 | |
264 | print "</div>"; | |
265 | } | |
266 | ||
267 | function hook_article_filter($article) { | |
268 | $owner_uid = $article["owner_uid"]; | |
269 | ||
5df13f31 AD |
270 | // guid already includes owner_uid so we don't need to include it |
271 | $result = $this->dbh->query("SELECT id FROM {$this->sql_prefix}_references WHERE | |
272 | document_id = '" . $this->dbh->escape_string($article['guid_hashed']) . "'"); | |
59e83455 | 273 | |
13167d2a AD |
274 | if (db_num_rows($result) != 0) { |
275 | _debug("bayes: article already categorized"); | |
276 | return $article; | |
277 | } | |
278 | ||
5df13f31 | 279 | $nbs = new NaiveBayesianStorage($owner_uid); |
4dbd303b | 280 | $nb = new NaiveBayesian($nbs); |
fe4535e6 | 281 | |
59e83455 AD |
282 | $categories = $nbs->getCategories(); |
283 | ||
284 | if (count($categories) > 0) { | |
285 | ||
286 | $count_neutral = 0; | |
08cfcba4 | 287 | |
59e83455 | 288 | $id_good = 0; |
08cfcba4 AD |
289 | $id_ugly = 0; |
290 | $id_bad = 0; | |
59e83455 AD |
291 | |
292 | foreach ($categories as $id => $cat) { | |
293 | if ($cat["category"] == "GOOD") { | |
294 | $id_good = $id; | |
08cfcba4 AD |
295 | } else if ($cat["category"] == "UGLY") { |
296 | $id_ugly = $id; | |
59e83455 | 297 | $count_neutral += $cat["word_count"]; |
08cfcba4 AD |
298 | } else if ($cat["category"] == "BAD") { |
299 | $id_bad = $id; | |
59e83455 AD |
300 | } |
301 | } | |
302 | ||
08cfcba4 | 303 | $dst_category = $id_ugly; |
59e83455 AD |
304 | |
305 | $bayes_content = mb_strtolower($article["title"] . " " . strip_tags($article["content"])); | |
306 | ||
fa05356a | 307 | if ($count_neutral >= $this->auto_categorize_threshold) { |
59e83455 AD |
308 | // enable automatic categorization |
309 | ||
310 | $result = $nb->categorize($bayes_content); | |
311 | ||
47cd0bd5 | 312 | //print_r($result); |
e0ae194a | 313 | |
08cfcba4 | 314 | if (count($result) == 3) { |
59e83455 | 315 | $prob_good = $result[$id_good]; |
08cfcba4 | 316 | $prob_bad = $result[$id_bad]; |
59e83455 | 317 | |
bc0e565a | 318 | if (!is_nan($prob_good) && $prob_good > 0.90) { |
e0ae194a | 319 | $dst_category = $id_good; |
59e83455 | 320 | $article["score_modifier"] += $this->score_modifier; |
bc0e565a | 321 | } else if (!is_nan($prob_bad) && $prob_bad > 0.90) { |
e0ae194a | 322 | $dst_category = $id_bad; |
08cfcba4 | 323 | $article["score_modifier"] -= $this->score_modifier; |
59e83455 AD |
324 | } |
325 | } | |
e0ae194a AD |
326 | |
327 | _debug("bayes, dst category: $dst_category"); | |
59e83455 AD |
328 | } |
329 | ||
330 | $nb->train($article["guid_hashed"], $dst_category, $bayes_content); | |
331 | ||
332 | $nb->updateProbabilities(); | |
333 | } | |
853cc128 AD |
334 | |
335 | return $article; | |
336 | ||
337 | } | |
338 | ||
3c43def9 AD |
339 | function clearDatabase() { |
340 | $prefix = $this->sql_prefix; | |
341 | ||
342 | $this->dbh->query("BEGIN"); | |
343 | $this->dbh->query("DELETE FROM ${prefix}_references WHERE owner_uid = " . $_SESSION["uid"]); | |
344 | $this->dbh->query("DELETE FROM ${prefix}_wordfreqs WHERE owner_uid = " . $_SESSION["uid"]); | |
345 | $this->dbh->query("COMMIT"); | |
346 | ||
347 | $nbs = new NaiveBayesianStorage($_SESSION["uid"]); | |
4dbd303b | 348 | $nb = new NaiveBayesian($nbs); |
3c43def9 AD |
349 | $nb->updateProbabilities(); |
350 | } | |
351 | ||
168e32b9 AD |
352 | function showArticleStats() { |
353 | $article_id = (int) $_REQUEST["article_id"]; | |
354 | ||
355 | $result = $this->dbh->query("SELECT score, guid, title, content FROM ttrss_entries, ttrss_user_entries WHERE ref_id = id AND id = " . | |
356 | $article_id . " AND owner_uid = " . $_SESSION["uid"]); | |
357 | ||
358 | if ($this->dbh->num_rows($result) != 0) { | |
359 | $guid = $this->dbh->fetch_result($result, 0, "guid"); | |
360 | $title = $this->dbh->fetch_result($result, 0, "title"); | |
361 | $content = mb_strtolower($title . " " . strip_tags($this->dbh->fetch_result($result, 0, "content"))); | |
362 | ||
363 | print "<h2>" . $title . "</h2>"; | |
364 | ||
365 | $nbs = new NaiveBayesianStorage($_SESSION["uid"]); | |
366 | $nb = new NaiveBayesian($nbs); | |
367 | ||
368 | $categories = $nbs->getCategories(); | |
369 | ||
370 | $ref = $nbs->getReference($guid, false); | |
371 | ||
372 | $current_cat = isset($ref["category_id"]) ? $categories[$ref["category_id"]]["category"] : "N/A"; | |
373 | ||
374 | print "<p>" . T_sprintf("Currently stored as: %s", $current_cat) . "</p>"; | |
375 | ||
376 | $result = $nb->categorize($content); | |
377 | ||
378 | print "<h3>" . __("Classifier result") . "</h3>"; | |
379 | ||
380 | print "<table>"; | |
381 | print "<tr><th>Category</th><th>Probability</th></tr>"; | |
382 | ||
383 | foreach ($result as $k => $v) { | |
384 | print "<tr>"; | |
385 | print "<td>" . $categories[$k]["category"] . "</td>"; | |
386 | print "<td>" . $v . "</td>"; | |
387 | ||
388 | print "</tr>"; | |
389 | } | |
390 | ||
391 | print "</table>"; | |
392 | ||
393 | } else { | |
394 | print_error("Article not found"); | |
395 | } | |
396 | ||
397 | print "<div align='center'>"; | |
398 | ||
399 | print "<button dojoType=\"dijit.form.Button\" onclick=\"return dijit.byId('bayesShowDlg').hide()\">". | |
400 | __('Close this window')."</button>"; | |
401 | ||
402 | print "</div>"; | |
403 | ||
404 | } | |
405 | ||
853cc128 AD |
406 | function api_version() { |
407 | return 2; | |
408 | } | |
409 | ||
410 | } | |
411 | ?> |