From: Vincent Vanwaelscappel <vincent@cubedesigners.com>
Date: Wed, 26 Jan 2022 17:38:13 +0000 (+0100)
Subject: wip #5058 @0.5
X-Git-Url: http://git.cubedesigners.com/?a=commitdiff_plain;h=b7e66d0bac3865bbdc8280cbd8a012fb85e7ef5c;p=fluidbook_tools.git

wip #5058 @0.5
---

diff --git a/src/Search/SearchIndex.php b/src/Search/SearchIndex.php
index b900662..a068e3d 100644
--- a/src/Search/SearchIndex.php
+++ b/src/Search/SearchIndex.php
@@ -17,54 +17,95 @@ class SearchIndex
         $this->_pages[$page] = new Page($documentPage, $path);
     }
 
-    /**
-     * @return array
-     */
-    public function compileIndex()
+    protected function fillIndexWithWords(&$index, $page, $ipage)
     {
-        $index = [];
-        foreach ($this->_pages as $pageNumber => $page) {
-            $twords = explode("\n", $page->getIndex());
+        $twords = explode("\n", trim($ipage));
 
-            foreach ($twords as $woadata) {
-                $w1 = explode(',', trim($woadata));
-                if (count($w1) <= 1) {
-                    continue;
+        foreach ($twords as $woadata) {
+            $w1 = explode(',', trim($woadata));
+            if (count($w1) <= 1) {
+                continue;
+            }
+            list($woa, $worddata) = $w1;
+            $e = explode("\t", $worddata, 2);
+            if (count($e) < 2) {
+                continue;
+            }
+            list($total, $wordslist) = $e;
+
+            if ($woa == '') {
+                continue;
+            }
+
+            if (!isset($index[$woa])) {
+                $index[$woa] = array('t' => 0, 'w' => array());
+            }
+            $index[$woa]['t'] += $total;
+
+            $words = explode("\t", $wordslist);
+
+            foreach ($words as $word) {
+                list($wordwa, $count) = explode('$', $word, 2);
+                if (!isset($index[$woa]['w'][$wordwa])) {
+                    $index[$woa]['w'][$wordwa] = array('t' => 0, 'p' => array());
                 }
-                list($woa, $worddata) = $w1;
-                $e = explode("\t", $worddata, 2);
-                if (count($e) < 2) {
-                    continue;
+                if (!isset($index[$woa]['w'][$wordwa]['p'][$page])) {
+                    $index[$woa]['w'][$wordwa]['p'][$page] = 0;
                 }
-                list($total, $wordslist) = $e;
+                $index[$woa]['w'][$wordwa]['t'] += $count;
+                $index[$woa]['w'][$wordwa]['p'][$page] += $count;
+            }
+        }
+    }
 
-                if ($woa === '') {
-                    continue;
-                }
 
-                if (!isset($index[$woa])) {
-                    $index[$woa] = array('t' => 0, 'w' => array());
-                }
-                $index[$woa]['t'] += (int)$total;
-
-                $words = explode("\t", $wordslist);
-
-                foreach ($words as $word) {
-                    list($wordwa, $count) = explode('$', $word, 2);
-                    if (!isset($index[$woa]['w'][$wordwa])) {
-                        $index[$woa]['w'][$wordwa] = array('t' => 0, 'p' => [$pageNumber => 0]);
-                    }
-                    if (!isset($index[$woa]['w'][$wordwa]['p'][$pageNumber])) {
-                        $index[$woa]['w'][$wordwa]['p'][$pageNumber] = 0;
-                    }
-                    $index[$woa]['w'][$wordwa]['t'] += (int)$count;
-                    $index[$woa]['w'][$wordwa]['p'][$pageNumber] += (int)$count;
+    protected function fillIndexWithWordsSimple(&$index, $page, $ipage)
+    {
+        $twords = explode("\n", trim($ipage));
+
+        foreach ($twords as $woadata) {
+            $w1 = explode(',', trim($woadata));
+            if (count($w1) <= 1) {
+                continue;
+            }
+            list($woa, $worddata) = $w1;
+            $e = explode("\t", $worddata, 2);
+            if (count($e) < 2) {
+                continue;
+            }
+            list($total, $wordslist) = $e;
+
+            if ($woa == '') {
+                continue;
+            }
+
+            if (!isset($index[$woa])) {
+                $index[$woa] = array('t' => 0, 'p' => array());
+            }
+            $index[$woa]['t'] += $total;
+
+            $words = explode("\t", $wordslist);
+            foreach ($words as $word) {
+                list($wordwa, $count) = explode('$', $word, 2);
+                if (!isset($index[$woa]['p'][$page])) {
+                    $index[$woa]['p'][$page] = 0;
                 }
+                $index[$woa]['p'][$page] += $count;
             }
         }
-        return $index;
     }
 
+    /**
+     * @return array
+     */
+    public function compileIndex()
+    {
+        $index = [];
+        foreach ($this->_pages as $pageNumber => $page) {
+            $this->fillIndexWithWordsSimple($index, $pageNumber, $page->getIndex());
+        }
+        return $index;
+    }
 
     /**
      * @return stdClass