Библиотека для cis, online, cms1

core/search/lexer.php

@@ -0,0 +1,93 @@
+<?php
+
+/**
+ * Ðàçáèðâåò ñòðîêó çàïðîñà íà òîêåíû
+ */
+class Lexer
+{
+    const TOKEN_NOT    = 1;
+    const TOKEN_OR     = 2;
+    const TOKEN_LPAREN = 3;
+    const TOKEN_RPAREN = 4;
+    const TOKEN_AND    = 5;
+    const TOKEN_WORD   = 6;
+    const TOKEN_EOL    = 7;
+
+    protected $src;
+    private   $offset = 0;
+    public    $token;
+
+    public function __construct ()
+    {
+    }
+
+    function setSource ($src)
+    {
+        $this->src = $src;
+        $this->offset;
+    }
+
+    private function skipSpace ()
+    {
+        while (!$this->isEOL() && $this->getChar() == " ") { 
+            $this->offset++; 
+        }
+    }
+
+    private function getChar ()
+    {
+        return $this->src [$this->offset]; 
+    }
+    
+    /**
+     * Ïðîâåðÿåò íà êîíåö ñòðîêè
+     */
+    private function isEOL () {
+        return $this->offset >= strlen($this->src);
+    }
+
+    /**
+     * Îäíîñèìâîëüíûé òîêåí
+     */
+    private function easyToken () {
+        $ch = $this->getChar ();
+        switch ($ch) {
+            case '~': $token = array(self::TOKEN_NOT, $ch); break;
+            case '|': $token = array(self::TOKEN_OR, $ch); break;
+            case '(': $token = array(self::TOKEN_LPAREN, $ch); break;
+            case ')': $token = array(self::TOKEN_RPAREN, $ch); break; 
+            case '&': $token = array(self::TOKEN_AND, $ch); break; 
+            default:
+                $this->offset++;
+                $token = $this->getToken();
+        }
+        $this->offset++;
+        return $token;
+    } 
+
+    /**
+     * Âîçâðàùàåò ñëåäóþùèé òîêåí
+     */
+    public function getToken ()
+    {
+        $this->skipSpace ();
+        if ($this->isEOL()) {
+            return array(self::TOKEN_EOL, "");
+        }
+        if (ctype_alpha($this->getChar())) {
+            $start = $this->offset;
+            while (!$this->isEOL() && ctype_alpha($this->getChar())) {
+                $this->offset ++;
+            }
+            return array(self::TOKEN_WORD, substr ($this->src, $start, $this->offset-$start));
+        }
+        return $this->easyToken();
+    }
+
+    public function nextToken ()
+    {
+        $this->token = $this->getToken();
+    }    
+}
+
+?>