From d83344dcceb4f06e9910d591c966c0f57885369c Mon Sep 17 00:00:00 2001
From: nbm <nbm@c91229c3-7414-0410-bfa2-8a42b809f60b>
Date: Mon, 20 Feb 2006 14:33:42 +0000
Subject: [PATCH] Add OpenDocument indexing support (requires unzip)

---
 plugins/ktstandard/KTIndexer.php                    |  2 ++
 plugins/ktstandard/contents/OpenDocumentIndexer.php | 79 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 81 insertions(+), 0 deletions(-)
 create mode 100644 plugins/ktstandard/contents/OpenDocumentIndexer.php

diff --git a/plugins/ktstandard/KTIndexer.php b/plugins/ktstandard/KTIndexer.php
index 3704647..f3cfdb6 100644
--- a/plugins/ktstandard/KTIndexer.php
+++ b/plugins/ktstandard/KTIndexer.php
@@ -16,6 +16,8 @@ class KTIndexerPlugin extends KTPlugin {
                 'ktstandard.indexer.triggers.pdf', 'contents/PdfIndexer.php');
         $this->registerTrigger('content', 'transform', 'KTPostscriptIndexerTrigger',
                 'ktstandard.indexer.triggers.ps', 'contents/PsIndexer.php');
+        $this->registerTrigger('content', 'transform', 'KTOpenDocumentIndexerTrigger',
+                'ktstandard.indexer.triggers.opendocument', 'contents/OpenDocumentIndexer.php');
     }
 }
 
diff --git a/plugins/ktstandard/contents/OpenDocumentIndexer.php b/plugins/ktstandard/contents/OpenDocumentIndexer.php
new file mode 100644
index 0000000..540478f
--- /dev/null
+++ b/plugins/ktstandard/contents/OpenDocumentIndexer.php
@@ -0,0 +1,79 @@
+<?php
+
+require_once(KT_DIR . '/plugins/ktstandard/contents/BaseIndexer.php');
+
+class KTOpenDocumentIndexerTrigger extends KTBaseIndexerTrigger {
+    var $mimetypes = array(
+       'application/octet-stream' => true,
+       'application/zip' => true,
+       'application/x-zip' => true,
+       'application/vnd.oasis.opendocument.text' => true,
+       'application/vnd.oasis.opendocument.text-template' => true,
+       'application/vnd.oasis.opendocument.presentation' => true,
+       'application/vnd.oasis.opendocument.presentation-template' => true,
+       'application/vnd.oasis.opendocument.spreadsheet' => true,
+       'application/vnd.oasis.opendocument.spreadsheet-template' => true,
+    );
+
+    function transform() {
+        $iMimeTypeId = $this->oDocument->getMimeTypeId();
+        $sMimeType = KTMime::getMimeTypeName($iMimeTypeId);
+        $sFileName = $this->oDocument->getFileName();
+        if (in_array($sMimeType, array('application/octet-stream', 'application/zip', 'application/x-zip'))) {
+            $sExtension = KTMime::stripAllButExtension($sFileName);
+            $sTable = KTUtil::getTableName('mimetypes');
+            $sQuery = "SELECT id FROM $sTable WHERE LOWER(filetypes) = ?";
+            $aParams = array($sExtension);
+            $id = DBUtil::getOneResultKey(array($sQuery, $aParams), 'id');
+            if ($id) {
+                $this->oDocument->setMimeTypeId($id);
+                $this->oDocument->update();
+            }
+        }
+        parent::transform();
+    }
+
+    function extract_contents($sFilename, $sTmpFilename) {
+        $sUnzipCommand = KTUtil::findCommand("import/unzip", "unzip");
+        if (empty($sUnzipCommand)) {
+            return;
+        }
+        $this->sTmpPath = tempnam('/tmp', 'opendocumentextract');
+        if ($this->sTmpPath === false) {
+            return;
+        }
+        unlink($this->sTmpPath);
+        mkdir($this->sTmpPath, 0700);
+
+        $sCmd = array(
+            $sUnzipCommand,
+            "-q", "-n",
+            "-d", $this->sTmpPath,
+            $sFilename,
+        );
+        KTUtil::pexec($sCmd);
+
+        $sManifest = sprintf("%s/%s", $this->sTmpPath, "META-INF/manifest.xml");
+        if (!file_exists($sManifest)) {
+            $this->cleanup();
+            return;
+        }
+        $sContentFile = sprintf("%s/%s", $this->sTmpPath, "content.xml");
+        if (!file_exists($sContentFile)) {
+            $this->cleanup();
+            return;
+        }
+
+        $sContent = file_get_contents($sContentFile);
+        $sContent = preg_replace ("@(</?[^>]*>)+@", " ", $sContent);
+
+        $this->cleanup();
+        return $sContent;
+    }
+
+    function cleanup() {
+        KTUtil::deleteDirectory($this->sTmpPath);
+    }
+}
+
+?>
--
libgit2 0.21.4