Merge pull request #393 from adam-vessey/7.x-datastream-cache-headers

Cache headers for datastreams.

includes/admin.form.inc

@@ -55,6 +55,12 @@ function islandora_repository_admin(array $form, array &$form_state) {
           '#description' => t('The PID of the Root Collection Object'),
           '#required' => TRUE,
         ),
+        'islandora_use_datastream_cache_headers' => array(
+          '#type' => 'checkbox',
+          '#title' => t('Generate/parse datastream HTTP cache headers'),
+          '#description' => t('HTTP caching can reduce network traffic, by allowing clients to used cached copies.'),
+          '#default_value' => variable_get('islandora_use_datastream_cache_headers', TRUE),
+        ),
       ),
       'islandora_namespace' => array(
         '#type' => 'fieldset',

includes/datastream.inc

@@ -46,8 +46,6 @@ function islandora_view_datastream(AbstractDatastream $datastream, $download = F
     }
   }
 
-  header_remove('Cache-Control');
-  header_remove('Expires');
   header('Content-type: ' . $datastream->mimetype);
   if ($datastream->controlGroup == 'M' || $datastream->controlGroup == 'X') {
     header('Content-length: ' . $datastream->size);
@@ -59,13 +57,164 @@ function islandora_view_datastream(AbstractDatastream $datastream, $download = F
     $filename = $datastream->label . '.' . $extension;
     header("Content-Disposition: attachment; filename=\"$filename\"");
   }
+
+  $cache_check = islandora_view_datastream_cache_check($datastream);
+  if ($cache_check !== 200) {
+    if ($cache_check === 304) {
+      header('HTTP/1.1 304 Not Modified');
+    }
+    elseif ($cache_check === 412) {
+      header('HTTP/1.0 412 Precondition Failed');
+    }
+  }
+  islandora_view_datastream_set_cache_headers($datastream);
+
   drupal_page_is_cacheable(FALSE);
   // Try not to load the file into PHP memory!
-  ob_end_flush();
+  // Close and flush ALL the output buffers!
+  while (@ob_end_flush()) {
+  };
+
+  // New content needed.
+  if ($cache_check === 200) {
     $datastream->getContent('php://output');
+  }
   exit();
 }
 
+/**
+ * Parse "etags" from HTTP If-Match or If-None-Match headers.
+ *
+ * Parses from the CSV-like struture supported by HTTP headers into an array,
+ * so `"asdf", "fdsa", W/"2132"` should become an array containing the strings:
+ * - asdf
+ * - fdsa
+ * - 2132
+ *
+ * @see http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.24
+ * @see http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.26
+ *
+ * @param string $header_value
+ *   The value from the headers.
+ *
+ * @return array
+ *   An array containing all the etags present.
+ */
+function islandora_parse_http_match_headers($header_value) {
+  $matches = array();
+  // Match the CSV-like structure supported by the HTTP headers.
+  $count = preg_match_all('/(((W\/)?("?)(\*|.+?)\4)(, +)?)/', $header_value, $matches);
+  // The fifth sub-expression/group is which will contain the etags.
+  return $matches[5];
+}
+
+/**
+ * Validate cache headers.
+ *
+ * @param AbstractDatastream $datastream
+ *   The datastream for which to check the request headers against.
+ *
+ * @return int
+ *   An integer representing the HTTP response code. One of:
+ *   - 200: Proceed as normal. (Full download).
+ *   - 304: Resource hasn't changed; pass cache validation.
+ *   - 412: Resource has changed; fail cache validation.
+ *
+ * @see http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html
+ */
+function islandora_view_datastream_cache_check(AbstractDatastream $datastream) {
+  if (!variable_get('islandora_use_datastream_cache_headers', TRUE)) {
+    return 200;
+  }
+
+  // Let's assume that if we get here, we'll be able to complete the request.
+  $return = 200;
+
+  if (isset($_SERVER['HTTP_IF_MODIFIED_SINCE'])) {
+    $modified_since = DateTime::createFromFormat('D, d M Y H:i:s e', $_SERVER['HTTP_IF_MODIFIED_SINCE']);
+    if ($datastream->createdDate->getTimestamp() - $modified_since->getTimestamp() > 0) {
+      // Changed!
+      return $return;
+    }
+    else {
+      $return = 304;
+    }
+  }
+  if ($return === 200 && isset($_SERVER['HTTP_IF_UNMODIFIED_SINCE'])) {
+    $unmodified_since = DateTime::createFromFormat('D, d M Y H:i:s e', $_SERVER['HTTP_IF_UNMODIFIED_SINCE']);
+    if ($datastream->createdDate->getTimestamp() !== $unmodified_since->getTimestamp()) {
+      // Changed!
+      $return = 412;
+    }
+    else {
+      return $return;
+    }
+  }
+
+  // Only consider Etags we have provided.
+  if (isset($datastream->checksum)) {
+    $tags = array();
+    foreach ($datastream as $offset => $version) {
+      if (isset($version->checksum)) {
+        $tags[$offset] = $version->checksum;
+      }
+    }
+
+    if ($return === 200 && isset($_SERVER['HTTP_IF_MATCH'])) {
+      $request_tags = islandora_parse_http_match_headers($_SERVER['HTTP_IF_MATCH']);
+      if (in_array('*', $request_tags) || count(array_intersect($tags, $request_tags)) > 0) {
+        // There's a match... Let things go ahead.
+        return $return;
+      }
+      else {
+        $return = 412;
+      }
+    }
+    if (in_array($return, array(200, 304), TRUE) && isset($_SERVER['HTTP_IF_NONE_MATCH'])) {
+      $request_tags = islandora_parse_http_match_headers($_SERVER['HTTP_IF_NONE_MATCH']);
+      if (in_array('*', $request_tags) || count(array_intersect($tags, $request_tags)) > 0) {
+        $return = 304;
+      }
+      else {
+        $return = 200;
+      }
+    }
+  }
+
+  return $return;
+}
+
+/**
+ * Set various HTTP headers for caching.
+ *
+ * @param AbstractDatastream $datastream
+ *   The datastream being viewed/downloaded.
+ */
+function islandora_view_datastream_set_cache_headers(AbstractDatastream $datastream) {
+  if (variable_get('islandora_use_datastream_cache_headers', TRUE)) {
+    // Force cache revalidation.
+    header('Expires: Sun, 19 Nov 1978 05:00:00 GMT');
+    $cache_control = array();
+    if ($datastream->parent->repository->api->connection->username == 'anonymous') {
+      $cache_control[] = 'public';
+    }
+    else {
+      $cache_control[] = 'private';
+    }
+    $cache_control[] = 'must-revalidate';
+    $cache_control[] = 'max-age=0';
+    header('Cache-Control: ' . implode(', ', $cache_control));
+    header('Last-Modified: ' . $datastream->createdDate->format('D, d M Y H:i:s \G\M\T'));
+    if (isset($datastream->checksum)) {
+      header("Etag: \"{$datastream->checksum}\"");
+    }
+  }
+  else {
+    header_remove('Cache-Control');
+    header_remove('Expires');
+  }
+}
+
 /**
  * Get the human readable size of the given datastream.
  *

islandora.info

@@ -18,5 +18,6 @@ files[] = tests/ingest.test
 files[] = tests/hooked_access.test
 files[] = tests/islandora_manage_permissions.test
 files[] = tests/datastream_versions.test
+files[] = tests/datastream_cache.test
 files[] = tests/derivatives.test
 php = 5.3

tests/datastream_cache.test

@@ -0,0 +1,160 @@
+<?php
+
+/**
+ * @file
+ * Tests to verify the cache headers we provide.
+ */
+
+class IslandoraDatastreamCacheTestCase extends IslandoraWebTestCase {
+
+  /**
+   * Gets info to display to describe this test.
+   *
+   * @see IslandoraWebTestCase::getInfo()
+   */
+  public static function getInfo() {
+    return array(
+      'name' => 'Datastream Cache Headers',
+      'description' => 'Check our headers work as we expect them to.',
+      'group' => 'Islandora',
+    );
+  }
+
+  /**
+   * Creates an admin user and a connection to a fedora repository.
+   *
+   * @see IslandoraWebTestCase::setUp()
+   */
+  public function setUp() {
+    parent::setUp();
+    $this->repository = $this->admin->repository;
+    $this->purgeTestObjects();
+  }
+
+  /**
+   * Free any objects/resources created for this test.
+   *
+   * @see IslandoraWebTestCase::tearDown()
+   */
+  public function tearDown() {
+    $this->purgeTestObjects();
+    parent::tearDown();
+  }
+
+  /**
+   * Purge any objects created by the test's in this class.
+   */
+  public function purgeTestObjects() {
+    $objects = array(
+      'test:test',
+    );
+    foreach ($objects as $object) {
+      try {
+        $object = $this->repository->getObject($object);
+        $this->repository->purgeObject($object->id);
+      }
+      catch (Exception $e) {
+        // Meh... Either it didn't exist or the purge failed.
+      }
+    }
+  }
+
+  /**
+   * Create our test object.
+   */
+  protected function createTestObject() {
+    $object = $this->repository->constructObject('test:test');
+    $object->label = 'Test object';
+    $object->models = 'test:model';
+    $datastream = $object->constructDatastream('asdf', 'M');
+    $datastream->label = 'datastream of doom';
+    $datastream->mimetype = 'text/plain';
+    $datastream->content = 'And then things happened.';
+    $datastream->checksumType = 'SHA-1';
+    $object->ingestDatastream($datastream);
+    $this->repository->ingestObject($object);
+    return $object;
+  }
+
+  /**
+   * Test HTTP cache headers.
+   */
+  public function testCacheHeaders() {
+    $object = $this->createTestObject();
+    $datastream = $object['asdf'];
+
+    $user = $this->drupalCreateUser(array(ISLANDORA_VIEW_OBJECTS));
+    $this->drupalLogin($user);
+
+    // Test If-Modified-Since.
+    $result = $this->drupalGet("islandora/object/{$object->id}/datastream/{$datastream->id}/view", array(), array(
+      'If-Modified-Since: ' . $datastream->createdDate->format('D, d M Y H:i:s \G\M\T'),
+    ));
+    $this->assertResponse(304);
+    $result = $this->drupalGet("islandora/object/{$object->id}/datastream/{$datastream->id}/view", array(), array(
+      'If-Modified-Since: ' . $datastream->createdDate->sub(new DateInterval('P1M'))->format('D, d M Y H:i:s \G\M\T'),
+    ));
+    $this->assertResponse(200);
+
+    // Test If-Unmodified-Since.
+    $result = $this->drupalGet("islandora/object/{$object->id}/datastream/{$datastream->id}/view", array(), array(
+      'If-Unmodified-Since: ' . $datastream->createdDate->format('D, d M Y H:i:s \G\M\T'),
+    ));
+    $this->assertResponse(200);
+    $result = $this->drupalGet("islandora/object/{$object->id}/datastream/{$datastream->id}/view", array(), array(
+      'If-Unmodified-Since: ' . $datastream->createdDate->sub(new DateInterval('P1M'))->format('D, d M Y H:i:s \G\M\T'),
+    ));
+    $this->assertResponse(412);
+
+    // Test If-Match.
+    $result = $this->drupalGet("islandora/object/{$object->id}/datastream/{$datastream->id}/view", array(), array(
+      format_string('If-Match: "!checksum"', array(
+        '!checksum' => $datastream->checksum,
+      )),
+    ));
+    $this->assertResponse(200);
+    $result = $this->drupalGet("islandora/object/{$object->id}/datastream/{$datastream->id}/view", array(), array(
+      format_string('If-Match: "!checksum"', array(
+        '!checksum' => 'dont-match' . $datastream->checksum,
+      )),
+    ));
+    $this->assertResponse(412);
+
+    // Test If-None-Match.
+    $result = $this->drupalGet("islandora/object/{$object->id}/datastream/{$datastream->id}/view", array(), array(
+      format_string('If-None-Match: "!checksum"', array(
+        '!checksum' => $datastream->checksum,
+      )),
+    ));
+    $this->assertResponse(304);
+    $result = $this->drupalGet("islandora/object/{$object->id}/datastream/{$datastream->id}/view", array(), array(
+      format_string('If-None-Match: "!checksum"', array(
+        '!checksum' => 'dont-match' . $datastream->checksum,
+      )),
+    ));
+    $this->assertResponse(200);
+
+    // Test combination of If-None-Match and If-Modified-Since
+    $result = $this->drupalGet("islandora/object/{$object->id}/datastream/{$datastream->id}/view", array(), array(
+      'If-Modified-Since: ' . $datastream->createdDate->format('D, d M Y H:i:s \G\M\T'),
+      format_string('If-None-Match: "!checksum"', array(
+        '!checksum' => $datastream->checksum,
+      )),
+    ));
+    $this->assertResponse(304);
+    $result = $this->drupalGet("islandora/object/{$object->id}/datastream/{$datastream->id}/view", array(), array(
+      'If-Modified-Since: ' . $datastream->createdDate->format('D, d M Y H:i:s \G\M\T'),
+      format_string('If-None-Match: "!checksum"', array(
+        '!checksum' => 'dont-match' . $datastream->checksum,
+      )),
+    ));
+    $this->assertResponse(200);
+    $result = $this->drupalGet("islandora/object/{$object->id}/datastream/{$datastream->id}/view", array(), array(
+      'If-Modified-Since: ' . $datastream->createdDate->sub(new DateInterval('P1M'))->format('D, d M Y H:i:s \G\M\T'),
+      format_string('If-None-Match: "!checksum"', array(
+        '!checksum' => $datastream->checksum,
+      )),
+    ));
+    $this->assertResponse(200);
+  }
+}