Provide alternative backends for caching of extracted content

Implement Flushable for clearing the cache

README.md

@@ -63,6 +63,24 @@ File:
 	- FileTextExtractable
 ```
 
+By default any extracted content will be cached against the database row.
+
+Alternatively, extracted content can be cached using SS_Cache to prevent excessive database growth.
+In order to swap out the cache backend you can use the following yaml configuration.
+
+
+```yaml
+---
+Name: mytextextraction
+After: '#textextraction'
+---
+Injector:
+  FileTextCache: FileTextCache_SSCache
+FileTextCache_SSCache:
+  lifetime: 3600 # Number of seconds to cache content for
+
+```
+
 ### XPDF
 
 PDFs require special handling, for example through the [XPDF](http://www.foolabs.com/xpdf/)

_config/config.yml

@@ -1,2 +1,8 @@
+---
+Name: textextraction
+---
+Injector:
+  FileTextCache: FileTextCache_Database
+
 SolrCellTextExtractor:
-#  base_url: 'http://localhost:8983/solr/update/extract'
+#  base_url: 'http://localhost:8983/solr/update/extract'

code/extensions/FileTextCache.php

@@ -0,0 +1,81 @@
+<?php
+
+interface FileTextCache {
+
+	/**
+	 * Save extracted content for a given File entity
+	 *
+	 * @param File $file
+	 * @param string $content
+	 */
+	public function save(File $file, $content);
+
+	/**
+	 * Return any cached extracted content for a given file entity
+	 *
+	 * @param File $file
+	 */
+	public function load(File $file);
+}
+
+/**
+ * Caches the extracted content on the record for the file
+ */
+class FileTextCache_Database implements FileTextCache {
+	
+	public function load(File $file) {
+		return $file->FileContentCache;
+	}
+
+	public function save(File $file, $content) {
+		$file->FileContentCache = $content;
+		$file->write();
+	}
+
+}
+
+/**
+ * Uses SS_Cache with a lifetime to cache extracted content
+ */
+class FileTextCache_SSCache implements FileTextCache, Flushable {
+
+	/**
+	 * Default cache to 1 hour
+	 *
+	 * @var int
+	 * @config
+	 */
+	private static $lifetime = 3600;
+
+	/**
+	 * @return SS_Cache
+	 */
+	protected static function get_cache() {
+		$lifetime = Config::inst()->get(__CLASS__, 'lifetime');
+		$cache = SS_Cache::factory(__CLASS__);
+		$cache->setLifetime($lifetime);
+		return $cache;
+	}
+
+	protected function getKey(File $file) {
+		return md5($file->getFullPath);
+	}
+
+	public function load(File $file) {
+		$key = $this->getKey($file);
+		$cache = self::get_cache();
+		return $cache->load($key);
+	}
+
+	public function save(File $file, $content) {
+		$key = $this->getKey($file);
+		$cache = self::get_cache();
+		return $cache->save($content, $key);
+	}
+
+	public static function flush() {
+		$cache = self::get_cache();
+		$cache->clean();
+	}
+
+}

code/extensions/FileTextExtractable.php

@@ -19,6 +19,30 @@ class FileTextExtractable extends DataExtension {
 		'FileContent' => 'Text'
 	);
 
+	private static $dependencies = array(
+		'TextCache' => '%$FileTextCache'
+	);
+
+	/**
+	 * @var FileTextCache
+	 */
+	protected $fileTextCache = null;
+
+	/**
+	 *
+	 * @param FileTextCache $cache
+	 */
+	public function setTextCache(FileTextCache $cache) {
+		$this->fileTextCache = $cache;
+	}
+
+	/**
+	 * @return FileTextCache
+	 */
+	public function getTextCache() {
+		return $this->fileTextCache;
+	}
+
 	/**
 	 * Helper function for template
 	 *
@@ -37,17 +61,25 @@ class FileTextExtractable extends DataExtension {
 	 * @return string
 	 */
 	public function extractFileAsText($disableCache = false) {
-		if (!$disableCache && $this->owner->FileContentCache) return $this->owner->FileContentCache;
+		if (!$disableCache) {
+			$text = $this->getTextCache()->load($this->owner);
+			if($text) {
+				return $text;
+			}
+		}
 
 		// Determine which extractor can process this file.
 		$extractor = FileTextExtractor::for_file($this->owner->FullPath);
-		if (!$extractor) return null;
+		if (!$extractor) {
+			return null;
+		}
 
 		$text = $extractor->getContent($this->owner->FullPath);
-		if (!$text) return null;
+		if (!$text) {
+			return null;
+		}
 
-		$this->owner->FileContentCache = $text;
-		$this->owner->write();
+		$this->getTextCache()->save($this->owner, $text);
 
 		return $text;
 	}