Rework advanced search options, and default search behavior (#833)

* Stop splitting words in a sentence by default when searching
* Add "Each Word" option to search all word in a sentence (old default behavior)
* Rename "Perfect march" to "Entire string", "Whole Words" to "Entire Words"
* Reword and reverse API promotion links
* Ignore words shorter than 2 characters when searching for each word
* Added tooltips to search options

Author: flodolo

app/classes/Transvision/API.php

@@ -15,7 +15,7 @@
  * Calls are like this:
  * api/<version>/<service>/<repository>/<search type>/<source locale>/<target locale>/<url escaped search>/?optional_parameter1=foo&optional_parameter2=bar
  * Example for an entity search containing bookmark:
- * https://transvision.mozfr.org/api/v1/tm/release/entity/en-US/fr/bookmark/?case_sensitive=1
+ * https://transvision.mozfr.org/api/v1/tm/release/entity/en-US/fr/bookmark/?case_sensitive=case_sensitive
  * (tm = translation memory service)
  *
  * Example for the list of locales supported for a repo:
@@ -245,7 +245,7 @@ private function isValidServiceCall($service)
 
                 break;
             case 'search':
-            // ex: /api/v1/search/string/central/en-US/fr/Bookmark/?case_sensitive=1
+            // ex: /api/v1/search/string/central/en-US/fr/Bookmark/?case_sensitive=case_sensitive
                 if (! $this->verifyEnoughParameters(7)) {
                     return false;
                 }

app/classes/Transvision/Search.php

@@ -10,9 +10,10 @@
  * e.g.:
  * $search = (new Search)
  *     ->setSearchTerms('Bookmark this page')
- *     ->setRegexWholeWords(true)
  *     ->setRegexCaseInsensitive(true)
- *     ->setRegexPerfectMatch(false)
+ *     ->setRegexEntireString(false)
+ *     ->setEachWord(false)
+ *     ->setEntireWords(false)
  *     ->setRepository('release')
  *     ->setSearchType('strings')
  *     ->setLocales(['en-US', 'fr']);
@@ -38,16 +39,22 @@ class Search
     protected $regex_case;
 
     /**
-     * Consider the space separated string as a single word for search
-     * @var string
+     * Only return strings that entirely match the search (case excluded)
+     * @var boolean
+     */
+    protected $regex_entire_string;
+
+    /**
+     * Only return strings where entire words match the search (case excluded)
+     * @var boolean
      */
-    protected $regex_whole_words;
+    protected $regex_entire_words;
 
     /**
-     * Only return strings that match the search perfectly (case excluded)
+     * Set to search for each word in the query instead of using it as a whole.
      * @var boolean
      */
-    protected $regex_perfect_match;
+    protected $each_word;
 
     /**
      * The search terms for the regex, these differ from $search_terms as
@@ -79,7 +86,8 @@ class Search
      * @var array
      */
     protected $form_search_options = [
-        'case_sensitive', 'perfect_match', 'repo', 'search_type', 't2t', 'whole_word',
+        'case_sensitive', 'entire_string', 'repo',
+        'search_type', 't2t', 'each_word', 'entire_words',
     ];
 
     /**
@@ -102,8 +110,9 @@ public function __construct()
         $this->search_terms = '';
         $this->regex = '';
         $this->regex_case = 'i';
-        $this->regex_whole_words = '';
-        $this->regex_perfect_match = false;
+        $this->regex_entire_string = false;
+        $this->regex_entire_words = false;
+        $this->each_word = false;
         $this->regex_search_terms = '';
         $this->repository = 'aurora'; // Most locales work on Aurora
         $this->search_type = 'strings';
@@ -161,35 +170,51 @@ public function setRegexCaseInsensitive($flag)
     }
 
     /**
-     * Set the regex to only return perfect matches for the searched string.
+     * Set the regex to only return strings that entirely match the
+     * searched string.
      * We cast the value to a boolean because we usually get it from a GET.
      *
-     * @param  boolean $flag Set to True for a perfect match
+     * @param  boolean $flag Set to True for an entire string match
      * @return $this
      */
-    public function setRegexPerfectMatch($flag)
+    public function setRegexEntireString($flag)
     {
-        $this->regex_perfect_match = (boolean) $flag;
+        $this->regex_entire_string = (boolean) $flag;
         $this->updateRegex();
 
         return $this;
     }
 
     /**
-     * Set the regex so as that a multi-word search is taken as a single word.
+     * Set the regex to only return strings where entire words match
+     * the searched string.
      * We cast the value to a boolean because we usually get it from a GET.
      *
-     * @param  boolean $flag A string evaluated to True will add \b to the regex
+     * @param  boolean $flag Set to True for an entire words match
      * @return $this
      */
-    public function setRegexWholeWords($flag)
+    public function setRegexEntireWords($flag)
     {
-        $this->regex_whole_words = (boolean) $flag ? '\b' : '';
+        $this->regex_entire_words = (boolean) $flag ? '\b' : '';
         $this->updateRegex();
 
         return $this;
     }
 
+    /**
+     * Set to search for each word in the query instead of using it as a whole.
+     * We cast the value to a boolean because we usually get it from a GET.
+     *
+     * @param  boolean $flag Set to True to search for each word.
+     * @return $this
+     */
+    public function setEachWord($flag)
+    {
+        $this->each_word = (boolean) $flag;
+
+        return $this;
+    }
+
     /**
      * Update the $regex_search_terms value every time a setter to the regex
      * is called.
@@ -199,15 +224,15 @@ public function setRegexWholeWords($flag)
     private function updateRegex()
     {
         $search = preg_quote($this->regex_search_terms);
-        if ($this->regex_perfect_match) {
+        if ($this->regex_entire_string) {
             $search = "^{$search}$";
         }
 
         $this->regex =
             '~'
-            . $this->regex_whole_words
+            . $this->regex_entire_words
             . $search
-            . $this->regex_whole_words
+            . $this->regex_entire_words
             . '~'
             . $this->regex_case
             . 'u';
@@ -226,53 +251,63 @@ public function getRegex()
     }
 
     /**
-     * Get the state of regex_perfect_match
+     * Get the state of regex_entire_string
      *
-     * @return boolean True if the regex searches for a perfect string match
+     * @return boolean True if the regex searches for an entire string match
      */
-    public function isPerfectMatch()
+    public function isEntireString()
     {
-        return $this->regex_perfect_match;
+        return $this->regex_entire_string;
     }
 
     /**
-     * Get search terms
+     * Get the state of each_word
      *
-     * @return string Searched terms
+     * @return boolean True if the search should be for each word.
      */
-    public function getSearchTerms()
+    public function isEachWord()
     {
-        return $this->search_terms;
+        return $this->each_word;
     }
 
     /**
-     * Get search terms in regex
+     * Get the state of entire_words
      *
-     * @return string Searched terms in regex
+     * @return boolean True if the search should be only for entire word.
      */
-    public function getRegexSearchTerms()
+    public function isEntireWords()
     {
-        return $this->regex_search_terms;
+        return $this->regex_entire_words == '\b' ? true : false;
     }
 
     /**
-     * Get the regex case
+     * Get the state of case_sensitive
      *
-     * @return string Return 'i' for case insensitive search, '' for sensitive
+     * @return boolean False if the search should be case sensitive
      */
-    public function getRegexCase()
+    public function isCaseSensitive()
     {
-        return $this->regex_case;
+        return $this->regex_case == 'i' ? false : true;
     }
 
     /**
-     * Get the regex whole words
+     * Get search terms
      *
-     * @return boolean True if we have the 'whole words' option for the regex
+     * @return string Searched terms
      */
-    public function isWholeWords()
+    public function getSearchTerms()
     {
-        return $this->regex_whole_words;
+        return $this->search_terms;
+    }
+
+    /**
+     * Get search terms in regex
+     *
+     * @return string Searched terms in regex
+     */
+    public function getRegexSearchTerms()
+    {
+        return $this->regex_search_terms;
     }
 
     /**

app/classes/Transvision/ShowResults.php

@@ -270,10 +270,10 @@ public static function resultsTable($search_object, $search_results, $page)
                      </thead>
                      <tbody>\n";
 
-        if (! $search_object->isWholeWords() && ! $search_object->isPerfectMatch()) {
-            $search = Utils::uniqueWords($search_object->getSearchTerms());
+        if ($search_object->isEachWord()) {
+            $search_terms = Utils::uniqueWords($search_object->getSearchTerms());
         } else {
-            $search = [$search_object->getSearchTerms()];
+            $search_terms = [$search_object->getSearchTerms()];
         }
 
         $current_repo = $search_object->getRepository();
@@ -284,7 +284,7 @@ public static function resultsTable($search_object, $search_results, $page)
             if ($search_object->getSearchType() == 'strings') {
                 $result_entity = self::formatEntity($key);
             } else {
-                $result_entity = self::formatEntity($key, $search[0]);
+                $result_entity = self::formatEntity($key, $search_terms[0]);
             }
 
             $component = explode('/', $key)[0];
@@ -296,7 +296,7 @@ public static function resultsTable($search_object, $search_results, $page)
             . "&locale={$locale2}"
             . "&repo={$current_repo}"
             . "&search_type=entities&recherche={$key}"
-            . "&perfect_match=perfect_match";
+            . "&entire_string=entire_string";
 
             $bz_link = [Bugzilla::reportErrorLink(
                 $locale2, $key, $source_string, $target_string, $current_repo, $entity_link
@@ -308,7 +308,7 @@ public static function resultsTable($search_object, $search_results, $page)
                                 . "&locale={$search_object->getLocale('extra')}"
                                 . "&repo={$current_repo}"
                                 . "&search_type=entities&recherche={$key}"
-                                . "&perfect_match=perfect_match";
+                                . "&entire_string=entire_string";
                 $bz_link[] = Bugzilla::reportErrorLink(
                     $search_object->getLocale('extra'), $key, $source_string, $target_string2, $current_repo, $entity_link
                 );
@@ -338,14 +338,14 @@ public static function resultsTable($search_object, $search_results, $page)
                 $transliterate_string_id = 'transliterate_' . $string_id;
             }
 
-            foreach ($search as $val) {
-                $source_string = Strings::markString($val, $source_string);
-                $target_string = Strings::markString($val, $target_string);
+            foreach ($search_terms as $search_term) {
+                $source_string = Strings::markString($search_term, $source_string);
+                $target_string = Strings::markString($search_term, $target_string);
                 if ($extra_locale) {
-                    $target_string2 = Strings::markString($val, $target_string2);
+                    $target_string2 = Strings::markString($search_term, $target_string2);
                 }
                 if ($transliterate) {
-                    $transliterated_string = Strings::markString($val, $transliterated_string);
+                    $transliterated_string = Strings::markString($search_term, $transliterated_string);
                 }
             }
 
@@ -531,7 +531,7 @@ public static function searchEntities($source_strings, $regex)
 
         /*
             If there are no results, search also through the entity names.
-            This is needed for "perfect match" when only the entity name is
+            This is needed for "entire string" when only the entity name is
             provided.
         */
         if (empty($entities)) {

app/classes/Transvision/Utils.php

@@ -124,7 +124,8 @@ public static function printSimpleTable(
     }
 
     /**
-     * Split a sentence in words from longest to shortest
+     * Split a sentence in words from longest to shortest, ignoring
+     * words shorter than 2 characters.
      *
      * @param  string $sentence
      * @return array  all the words in the sentence sorted by length
@@ -133,7 +134,14 @@ public static function uniqueWords($sentence)
     {
         $words = explode(' ', $sentence);
         $words = array_filter($words); // Filter out extra spaces
-        $words = array_unique($words); // Remove duplicate words
+        // Filter out 1-character words
+        $words = array_filter($words, function ($a) {
+            return (mb_strlen($a) >= 2);
+        });
+
+        // Remove duplicate words
+        $words = array_unique($words);
+
         // Sort words from longest to shortest
         usort(
             $words,
@@ -444,8 +452,8 @@ public static function APIPromotion($source_locale, $target_locale)
             array_map($sanitize, array_values($args))
         );
 
-        $args['locale'] = $source_locale;
-        $args['sourcelocale'] = $target_locale;
+        $args['locale'] = $target_locale;
+        $args['sourcelocale'] = $source_locale;
         $args['json'] = 'true';
 
         // We don't want to encode slashes in searches for entity names

app/controllers/mainsearch.php

@@ -31,9 +31,10 @@
         : '';
 
     $regex = [];
-    $regex['whole'] = isset($_GET['whole_word']) ? 'whole_word=1' : '';
-    $regex['case'] = isset($_GET['case_sensitive']) ? 'case_sensitive=1' : '';
-    $regex['perfect'] = isset($_GET['perfect_match']) ? 'perfect_match=1' : '';
+    $regex['each_word'] = isset($_GET['each_word']) ? 'each_word=each_word' : '';
+    $regex['case_sensitive'] = isset($_GET['case_sensitive']) ? 'case_sensitive=case_sensitive' : '';
+    $regex['entire_string'] = isset($_GET['entire_string']) ? 'entire_string=entire_string' : '';
+    $regex['entire_words'] = isset($_GET['entire_words']) ? 'entire_words=entire_words' : '';
     $regex = array_filter($regex);
     $regex = count($regex) > 0 ? '?' . implode('&', $regex) : '';
 

app/inc/search_options.php

@@ -28,9 +28,10 @@
 // Define our regex and search parameters
 $search
     ->setSearchTerms($my_search)
-    ->setRegexWholeWords($check['whole_word'])
+    ->setEachWord($check['each_word'])
     ->setRegexCaseInsensitive($check['case_sensitive'])
-    ->setRegexPerfectMatch($check['perfect_match'])
+    ->setRegexEntireString($check['entire_string'])
+    ->setRegexEntireWords($check['entire_words'])
     ->setRepository($repo)
     ->setSearchType($search_type)
     ->setLocales([$source_locale, $locale, $locale2]);

app/models/3locales_search.php

@@ -3,11 +3,16 @@
 
 $tmx_target2 = Utils::getRepoStrings($locale2, $search->getRepository());
 
-if ($search->isPerfectMatch()) {
+if ($search->isEntireString()) {
     $locale3_strings = $search->grep($tmx_target2);
 } else {
     $locale3_strings = $tmx_target2;
-    foreach (Utils::uniqueWords($search->getSearchTerms()) as $word) {
+
+    $search_terms = $search->isEachWord()
+        ? Utils::uniqueWords($search->getSearchTerms())
+        : [$search->getSearchTerms()];
+
+    foreach ($search_terms as $word) {
         $search->setRegexSearchTerms($word);
         $locale3_strings = $search->grep($locale3_strings);
     }