Merge remote-tracking branch 'upstream/master'

[nominatim.git] / lib-php / Phrase.php

diff --git a/lib-php/Phrase.php b/lib-php/Phrase.php
index e2643e878ed40a3bb22533600603647418e98011..4ed4d402f68c7c4e1c8dff5abd808087ef7ccfaa 100644 (file)
--- a/lib-php/Phrase.php
+++ b/lib-php/Phrase.php
@@ -1,4 +1,12 @@
 <?php
 <?php

+/**
+ * SPDX-License-Identifier: GPL-2.0-only
+ *
+ * This file is part of Nominatim. (https://nominatim.org)
+ *
+ * Copyright (C) 2022 by the Nominatim developer community.
+ * For a full list of authors see the git log.
+ */

 
 namespace Nominatim;
 
 
 namespace Nominatim;
 


@@ -9,36 +17,26 @@ namespace Nominatim;
  */
 class Phrase
 {
  */
 class Phrase
 {

-    const MAX_WORDSET_LEN = 20;
-    const MAX_WORDSETS = 100;
-
-    // Complete phrase as a string.
+    // Complete phrase as a string (guaranteed to have no leading or trailing
+    // spaces).

     private $sPhrase;
     // Element type for structured searches.
     private $sPhraseType;
     private $sPhrase;
     // Element type for structured searches.
     private $sPhraseType;

-    // Space-separated words of the phrase.
-    private $aWords;

     // Possible segmentations of the phrase.
     private $aWordSets;
 
     // Possible segmentations of the phrase.
     private $aWordSets;
 

-    public static function cmpByArraylen($aA, $aB)
-    {
-        $iALen = count($aA);
-        $iBLen = count($aB);
-
-        if ($iALen == $iBLen) {
-            return 0;
-        }
-
-        return ($iALen < $iBLen) ? -1 : 1;
-    }
-
-

     public function __construct($sPhrase, $sPhraseType)
     {
         $this->sPhrase = trim($sPhrase);
         $this->sPhraseType = $sPhraseType;
     public function __construct($sPhrase, $sPhraseType)
     {
         $this->sPhrase = trim($sPhrase);
         $this->sPhraseType = $sPhraseType;

-        $this->aWords = explode(' ', $this->sPhrase);
+    }
+
+    /**
+     * Get the original phrase of the string.
+     */
+    public function getPhrase()
+    {
+        return $this->sPhrase;

     }
 
     /**
     }
 
     /**


@@ -52,6 +50,11 @@ class Phrase
         return $this->sPhraseType;
     }
 
         return $this->sPhraseType;
     }
 

+    public function setWordSets($aWordSets)
+    {
+        $this->aWordSets = $aWordSets;
+    }
+

     /**
      * Return the array of possible segmentations of the phrase.
      *
     /**
      * Return the array of possible segmentations of the phrase.
      *


@@ -63,30 +66,6 @@ class Phrase
         return $this->aWordSets;
     }
 
         return $this->aWordSets;
     }
 

-    /**
-     * Add the tokens from this phrase to the given list of tokens.
-     *
-     * @param string[] $aTokens List of tokens to append.
-     *
-     * @return void
-     */
-    public function addTokens(&$aTokens)
-    {
-        $iNumWords = count($this->aWords);
-
-        for ($i = 0; $i < $iNumWords; $i++) {
-            $sPhrase = $this->aWords[$i];
-            $aTokens[' '.$sPhrase] = ' '.$sPhrase;
-            $aTokens[$sPhrase] = $sPhrase;
-
-            for ($j = $i + 1; $j < $iNumWords; $j++) {
-                $sPhrase .= ' '.$this->aWords[$j];
-                $aTokens[' '.$sPhrase] = ' '.$sPhrase;
-                $aTokens[$sPhrase] = $sPhrase;
-            }
-        }
-    }
-

     /**
      * Invert the set of possible segmentations.
      *
     /**
      * Invert the set of possible segmentations.
      *


@@ -99,61 +78,11 @@ class Phrase
         }
     }
 
         }
     }
 

-    public function computeWordSets($oTokens)
-    {
-        $iNumWords = count($this->aWords);
-        // Caches the word set for the partial phrase up to word i.
-        $aSetCache = array_fill(0, $iNumWords, array());
-
-        // Initialise first element of cache. There can only be the word.
-        if ($oTokens->containsAny($this->aWords[0])) {
-            $aSetCache[0][] = array($this->aWords[0]);
-        }
-
-        // Now do the next elements using what we already have.
-        for ($i = 1; $i < $iNumWords; $i++) {
-            for ($j = $i; $j > 0; $j--) {
-                $sPartial = $j == $i ? $this->aWords[$j] : $this->aWords[$j].' '.$sPartial;
-                if (!empty($aSetCache[$j - 1]) && $oTokens->containsAny($sPartial)) {
-                    $aPartial = array($sPartial);
-                    foreach ($aSetCache[$j - 1] as $aSet) {
-                        if (count($aSet) < Phrase::MAX_WORDSET_LEN) {
-                            $aSetCache[$i][] = array_merge($aSet, $aPartial);
-                        }
-                    }
-                    if (count($aSetCache[$i]) > 2 * Phrase::MAX_WORDSETS) {
-                        usort(
-                            $aSetCache[$i],
-                            array('\Nominatim\Phrase', 'cmpByArraylen')
-                        );
-                        $aSetCache[$i] = array_slice(
-                            $aSetCache[$i],
-                            0,
-                            Phrase::MAX_WORDSETS
-                        );
-                    }
-                }
-            }
-
-            // finally the current full phrase
-            $sPartial = $this->aWords[0].' '.$sPartial;
-            if ($oTokens->containsAny($sPartial)) {
-                $aSetCache[$i][] = array($sPartial);
-            }
-        }
-
-        $this->aWordSets = $aSetCache[$iNumWords - 1];
-        usort($this->aWordSets, array('\Nominatim\Phrase', 'cmpByArraylen'));
-        $this->aWordSets = array_slice($this->aWordSets, 0, Phrase::MAX_WORDSETS);
-    }
-
-

     public function debugInfo()
     {
         return array(
                 'Type' => $this->sPhraseType,
                 'Phrase' => $this->sPhrase,
     public function debugInfo()
     {
         return array(
                 'Type' => $this->sPhraseType,
                 'Phrase' => $this->sPhrase,

-                'Words' => $this->aWords,

                 'WordSets' => $this->aWordSets
                );
     }
                 'WordSets' => $this->aWordSets
                );
     }