aboutsummaryrefslogtreecommitdiffhomepage
path: root/application
diff options
context:
space:
mode:
authorArthur <arthur@hoa.ro>2016-01-06 19:57:42 +0100
committerArthur <arthur@hoa.ro>2016-01-06 19:57:42 +0100
commit88c15abb2a2b24f05ca926c0ddbdec18a407e47d (patch)
tree90f783c69dc51866dc01dc34e740023bd1e0c4ec /application
parenta327d891b3a2762bb6aabba3a6572b077f6003c0 (diff)
parenteefb636cea7acef9ddfd02a90749820f5fafc9f6 (diff)
downloadShaarli-88c15abb2a2b24f05ca926c0ddbdec18a407e47d.tar.gz
Shaarli-88c15abb2a2b24f05ca926c0ddbdec18a407e47d.tar.zst
Shaarli-88c15abb2a2b24f05ca926c0ddbdec18a407e47d.zip
Merge pull request #424 from ArthurHoaro/search
Link filter refactoring
Diffstat (limited to 'application')
-rw-r--r--application/LinkDB.php120
-rw-r--r--application/LinkFilter.php259
-rw-r--r--application/Utils.php12
3 files changed, 281 insertions, 110 deletions
diff --git a/application/LinkDB.php b/application/LinkDB.php
index 51fa926d..16848519 100644
--- a/application/LinkDB.php
+++ b/application/LinkDB.php
@@ -63,6 +63,11 @@ class LinkDB implements Iterator, Countable, ArrayAccess
63 private $_redirector; 63 private $_redirector;
64 64
65 /** 65 /**
66 * @var LinkFilter instance.
67 */
68 private $linkFilter;
69
70 /**
66 * Creates a new LinkDB 71 * Creates a new LinkDB
67 * 72 *
68 * Checks if the datastore exists; else, attempts to create a dummy one. 73 * Checks if the datastore exists; else, attempts to create a dummy one.
@@ -80,6 +85,7 @@ class LinkDB implements Iterator, Countable, ArrayAccess
80 $this->_redirector = $redirector; 85 $this->_redirector = $redirector;
81 $this->_checkDB(); 86 $this->_checkDB();
82 $this->_readDB(); 87 $this->_readDB();
88 $this->linkFilter = new LinkFilter($this->_links);
83 } 89 }
84 90
85 /** 91 /**
@@ -334,114 +340,18 @@ You use the community supported version of the original Shaarli project, by Seba
334 } 340 }
335 341
336 /** 342 /**
337 * Returns the list of links corresponding to a full-text search 343 * Filter links.
338 * 344 *
339 * Searches: 345 * @param string $type Type of filter.
340 * - in the URLs, title and description; 346 * @param mixed $request Search request, string or array.
341 * - are case-insensitive. 347 * @param bool $casesensitive Optional: Perform case sensitive filter
342 * 348 * @param bool $privateonly Optional: Returns private links only if true.
343 * Example:
344 * print_r($mydb->filterFulltext('hollandais'));
345 *
346 * mb_convert_case($val, MB_CASE_LOWER, 'UTF-8')
347 * - allows to perform searches on Unicode text
348 * - see https://github.com/shaarli/Shaarli/issues/75 for examples
349 */
350 public function filterFulltext($searchterms)
351 {
352 // FIXME: explode(' ',$searchterms) and perform a AND search.
353 // FIXME: accept double-quotes to search for a string "as is"?
354 $filtered = array();
355 $search = mb_convert_case($searchterms, MB_CASE_LOWER, 'UTF-8');
356 $keys = array('title', 'description', 'url', 'tags');
357
358 foreach ($this->_links as $link) {
359 $found = false;
360
361 foreach ($keys as $key) {
362 if (strpos(mb_convert_case($link[$key], MB_CASE_LOWER, 'UTF-8'),
363 $search) !== false) {
364 $found = true;
365 }
366 }
367
368 if ($found) {
369 $filtered[$link['linkdate']] = $link;
370 }
371 }
372 krsort($filtered);
373 return $filtered;
374 }
375
376 /**
377 * Returns the list of links associated with a given list of tags
378 * 349 *
379 * You can specify one or more tags, separated by space or a comma, e.g. 350 * @return array filtered links
380 * print_r($mydb->filterTags('linux programming'));
381 */ 351 */
382 public function filterTags($tags, $casesensitive=false) 352 public function filter($type, $request, $casesensitive = false, $privateonly = false) {
383 { 353 $requestFilter = is_array($request) ? implode(' ', $request) : $request;
384 // Same as above, we use UTF-8 conversion to handle various graphemes (i.e. cyrillic, or greek) 354 return $this->linkFilter->filter($type, trim($requestFilter), $casesensitive, $privateonly);
385 // FIXME: is $casesensitive ever true?
386 $t = str_replace(
387 ',', ' ',
388 ($casesensitive ? $tags : mb_convert_case($tags, MB_CASE_LOWER, 'UTF-8'))
389 );
390
391 $searchtags = explode(' ', $t);
392 $filtered = array();
393
394 foreach ($this->_links as $l) {
395 $linktags = explode(
396 ' ',
397 ($casesensitive ? $l['tags']:mb_convert_case($l['tags'], MB_CASE_LOWER, 'UTF-8'))
398 );
399
400 if (count(array_intersect($linktags, $searchtags)) == count($searchtags)) {
401 $filtered[$l['linkdate']] = $l;
402 }
403 }
404 krsort($filtered);
405 return $filtered;
406 }
407
408
409 /**
410 * Returns the list of articles for a given day, chronologically sorted
411 *
412 * Day must be in the form 'YYYYMMDD' (e.g. '20120125'), e.g.
413 * print_r($mydb->filterDay('20120125'));
414 */
415 public function filterDay($day)
416 {
417 if (! checkDateFormat('Ymd', $day)) {
418 throw new Exception('Invalid date format');
419 }
420
421 $filtered = array();
422 foreach ($this->_links as $l) {
423 if (startsWith($l['linkdate'], $day)) {
424 $filtered[$l['linkdate']] = $l;
425 }
426 }
427 ksort($filtered);
428 return $filtered;
429 }
430
431 /**
432 * Returns the article corresponding to a smallHash
433 */
434 public function filterSmallHash($smallHash)
435 {
436 $filtered = array();
437 foreach ($this->_links as $l) {
438 if ($smallHash == smallHash($l['linkdate'])) {
439 // Yes, this is ugly and slow
440 $filtered[$l['linkdate']] = $l;
441 return $filtered;
442 }
443 }
444 return $filtered;
445 } 355 }
446 356
447 /** 357 /**
diff --git a/application/LinkFilter.php b/application/LinkFilter.php
new file mode 100644
index 00000000..cf647371
--- /dev/null
+++ b/application/LinkFilter.php
@@ -0,0 +1,259 @@
1<?php
2
3/**
4 * Class LinkFilter.
5 *
6 * Perform search and filter operation on link data list.
7 */
8class LinkFilter
9{
10 /**
11 * @var string permalinks.
12 */
13 public static $FILTER_HASH = 'permalink';
14
15 /**
16 * @var string text search.
17 */
18 public static $FILTER_TEXT = 'fulltext';
19
20 /**
21 * @var string tag filter.
22 */
23 public static $FILTER_TAG = 'tags';
24
25 /**
26 * @var string filter by day.
27 */
28 public static $FILTER_DAY = 'FILTER_DAY';
29
30 /**
31 * @var array all available links.
32 */
33 private $links;
34
35 /**
36 * @param array $links initialization.
37 */
38 public function __construct($links)
39 {
40 $this->links = $links;
41 }
42
43 /**
44 * Filter links according to parameters.
45 *
46 * @param string $type Type of filter (eg. tags, permalink, etc.).
47 * @param string $request Filter content.
48 * @param bool $casesensitive Optional: Perform case sensitive filter if true.
49 * @param bool $privateonly Optional: Only returns private links if true.
50 *
51 * @return array filtered link list.
52 */
53 public function filter($type, $request, $casesensitive = false, $privateonly = false)
54 {
55 switch($type) {
56 case self::$FILTER_HASH:
57 return $this->filterSmallHash($request);
58 break;
59 case self::$FILTER_TEXT:
60 return $this->filterFulltext($request, $privateonly);
61 break;
62 case self::$FILTER_TAG:
63 return $this->filterTags($request, $casesensitive, $privateonly);
64 break;
65 case self::$FILTER_DAY:
66 return $this->filterDay($request);
67 break;
68 default:
69 return $this->noFilter($privateonly);
70 }
71 }
72
73 /**
74 * Unknown filter, but handle private only.
75 *
76 * @param bool $privateonly returns private link only if true.
77 *
78 * @return array filtered links.
79 */
80 private function noFilter($privateonly = false)
81 {
82 if (! $privateonly) {
83 krsort($this->links);
84 return $this->links;
85 }
86
87 $out = array();
88 foreach ($this->links as $value) {
89 if ($value['private']) {
90 $out[$value['linkdate']] = $value;
91 }
92 }
93
94 krsort($out);
95 return $out;
96 }
97
98 /**
99 * Returns the shaare corresponding to a smallHash.
100 *
101 * @param string $smallHash permalink hash.
102 *
103 * @return array $filtered array containing permalink data.
104 */
105 private function filterSmallHash($smallHash)
106 {
107 $filtered = array();
108 foreach ($this->links as $l) {
109 if ($smallHash == smallHash($l['linkdate'])) {
110 // Yes, this is ugly and slow
111 $filtered[$l['linkdate']] = $l;
112 return $filtered;
113 }
114 }
115 return $filtered;
116 }
117
118 /**
119 * Returns the list of links corresponding to a full-text search
120 *
121 * Searches:
122 * - in the URLs, title and description;
123 * - are case-insensitive.
124 *
125 * Example:
126 * print_r($mydb->filterFulltext('hollandais'));
127 *
128 * mb_convert_case($val, MB_CASE_LOWER, 'UTF-8')
129 * - allows to perform searches on Unicode text
130 * - see https://github.com/shaarli/Shaarli/issues/75 for examples
131 *
132 * @param string $searchterms search query.
133 * @param bool $privateonly return only private links if true.
134 *
135 * @return array search results.
136 */
137 private function filterFulltext($searchterms, $privateonly = false)
138 {
139 // FIXME: explode(' ',$searchterms) and perform a AND search.
140 // FIXME: accept double-quotes to search for a string "as is"?
141 $filtered = array();
142 $search = mb_convert_case($searchterms, MB_CASE_LOWER, 'UTF-8');
143 $explodedSearch = explode(' ', trim($search));
144 $keys = array('title', 'description', 'url', 'tags');
145
146 // Iterate over every stored link.
147 foreach ($this->links as $link) {
148 $found = false;
149
150 // ignore non private links when 'privatonly' is on.
151 if (! $link['private'] && $privateonly === true) {
152 continue;
153 }
154
155 // Iterate over searchable link fields.
156 foreach ($keys as $key) {
157 // Search full expression.
158 if (strpos(
159 mb_convert_case($link[$key], MB_CASE_LOWER, 'UTF-8'),
160 $search
161 ) !== false) {
162 $found = true;
163 }
164
165 if ($found) {
166 break;
167 }
168 }
169
170 if ($found) {
171 $filtered[$link['linkdate']] = $link;
172 }
173 }
174
175 krsort($filtered);
176 return $filtered;
177 }
178
179 /**
180 * Returns the list of links associated with a given list of tags
181 *
182 * You can specify one or more tags, separated by space or a comma, e.g.
183 * print_r($mydb->filterTags('linux programming'));
184 *
185 * @param string $tags list of tags separated by commas or blank spaces.
186 * @param bool $casesensitive ignore case if false.
187 * @param bool $privateonly returns private links only.
188 *
189 * @return array filtered links.
190 */
191 public function filterTags($tags, $casesensitive = false, $privateonly = false)
192 {
193 $searchtags = $this->tagsStrToArray($tags, $casesensitive);
194 $filtered = array();
195
196 foreach ($this->links as $l) {
197 // ignore non private links when 'privatonly' is on.
198 if (! $l['private'] && $privateonly === true) {
199 continue;
200 }
201
202 $linktags = $this->tagsStrToArray($l['tags'], $casesensitive);
203
204 if (count(array_intersect($linktags, $searchtags)) == count($searchtags)) {
205 $filtered[$l['linkdate']] = $l;
206 }
207 }
208 krsort($filtered);
209 return $filtered;
210 }
211
212 /**
213 * Returns the list of articles for a given day, chronologically sorted
214 *
215 * Day must be in the form 'YYYYMMDD' (e.g. '20120125'), e.g.
216 * print_r($mydb->filterDay('20120125'));
217 *
218 * @param string $day day to filter.
219 *
220 * @return array all link matching given day.
221 *
222 * @throws Exception if date format is invalid.
223 */
224 public function filterDay($day)
225 {
226 if (! checkDateFormat('Ymd', $day)) {
227 throw new Exception('Invalid date format');
228 }
229
230 $filtered = array();
231 foreach ($this->links as $l) {
232 if (startsWith($l['linkdate'], $day)) {
233 $filtered[$l['linkdate']] = $l;
234 }
235 }
236 ksort($filtered);
237 return $filtered;
238 }
239
240 /**
241 * Convert a list of tags (str) to an array. Also
242 * - handle case sensitivity.
243 * - accepts spaces commas as separator.
244 * - remove private tags for loggedout users.
245 *
246 * @param string $tags string containing a list of tags.
247 * @param bool $casesensitive will convert everything to lowercase if false.
248 *
249 * @return array filtered tags string.
250 */
251 public function tagsStrToArray($tags, $casesensitive)
252 {
253 // We use UTF-8 conversion to handle various graphemes (i.e. cyrillic, or greek)
254 $tagsOut = $casesensitive ? $tags : mb_convert_case($tags, MB_CASE_LOWER, 'UTF-8');
255 $tagsOut = str_replace(',', ' ', $tagsOut);
256
257 return explode(' ', trim($tagsOut));
258 }
259}
diff --git a/application/Utils.php b/application/Utils.php
index f84f70e4..aeaef9ff 100644
--- a/application/Utils.php
+++ b/application/Utils.php
@@ -72,12 +72,14 @@ function sanitizeLink(&$link)
72 72
73/** 73/**
74 * Checks if a string represents a valid date 74 * Checks if a string represents a valid date
75
76 * @param string $format The expected DateTime format of the string
77 * @param string $string A string-formatted date
78 *
79 * @return bool whether the string is a valid date
75 * 80 *
76 * @param string a string-formatted date 81 * @see http://php.net/manual/en/class.datetime.php
77 * @param format the expected DateTime format of the string 82 * @see http://php.net/manual/en/datetime.createfromformat.php
78 * @return whether the string is a valid date
79 * @see http://php.net/manual/en/class.datetime.php
80 * @see http://php.net/manual/en/datetime.createfromformat.php
81 */ 83 */
82function checkDateFormat($format, $string) 84function checkDateFormat($format, $string)
83{ 85{