diff --git a/feeds.info b/feeds.info index 4991435..e9ef158 100644 --- a/feeds.info +++ b/feeds.info @@ -73,6 +73,7 @@ files[] = tests/feeds_mapper_link.test files[] = tests/feeds_mapper_summary.test files[] = tests/feeds_mapper_taxonomy.test files[] = tests/feeds_tokens.test +files[] = tests/FeedsEnclosureTest.test files[] = tests/http_request.test files[] = tests/parser_csv.test diff --git a/libraries/http_request.inc b/libraries/http_request.inc index 39cd0ed..84a9513 100644 --- a/libraries/http_request.inc +++ b/libraries/http_request.inc @@ -9,6 +9,29 @@ */ /** + * Error code for when the URL could not be parsed. + * + * @var int + */ +define('FEEDS_ERROR_PARSE_ERROR', -1001); + +/** + * Error code for when the scheme of an URL could not be determined. + * + * An example of a scheme is 'http'. + * + * @var int + */ +define('FEEDS_ERROR_NO_SCHEME', -1002); + +/** + * Error code for when the scheme of an URL is not supported. + * + * @var int + */ +define('FEEDS_ERROR_INVALID_SCHEME', -1003); + +/** * PCRE for finding the link tags in html. */ define('HTTP_REQUEST_PCRE_LINK_TAG', '/(.*)<\/link>|(\/)?>)/si'); @@ -24,6 +47,11 @@ define('HTTP_REQUEST_PCRE_TAG_ATTRIBUTES', '/[\x09\x0A\x0B\x0C\x0D\x20]+([^\x09\ class HRCurlException extends Exception {} /** + * For HTTP requests that do not return a 2xx code. + */ +class FeedsHTTPRequestException extends Exception {} + +/** * Discovers RSS or atom feeds at the given URL. * * If document in given URL is an HTML document, function attempts to discover @@ -147,9 +175,13 @@ function http_request_get($url, $username = NULL, $password = NULL, $accept_inva // cURL can only support either http:// or https://. // CURLOPT_PROTOCOLS is only supported with cURL 7.19.4 $uri = parse_url($url); - if (!isset($uri['scheme'])) { + if ($uri === FALSE) { + $result->error = 'unable to parse URL'; + $result->code = FEEDS_ERROR_PARSE_ERROR; + } + elseif (!isset($uri['scheme'])) { $result->error = 'missing schema'; - $result->code = -1002; + $result->code = FEEDS_ERROR_NO_SCHEME; } else { switch ($uri['scheme']) { @@ -160,7 +192,7 @@ function http_request_get($url, $username = NULL, $password = NULL, $accept_inva default: $result->error = 'invalid schema ' . $uri['scheme']; - $result->code = -1003; + $result->code = FEEDS_ERROR_INVALID_SCHEME; break; } } @@ -284,6 +316,56 @@ function http_request_get($url, $username = NULL, $password = NULL, $accept_inva } /** + * Checks the result of the HTTP Request. + * + * @param string $url + * The URL that was requested. + * @param object $result + * The HTTP Request result. + * + * @throws FeedsHTTPRequestException + * In case the result code of the HTTP request is not in the 2xx series. + */ +function http_request_check_result($url, $result) { + if (!in_array($result->code, array(200, 201, 202, 203, 204, 205, 206))) { + $vars = array( + '@url' => $url, + '@code' => $result->code, + '@error' => isset($result->error) ? $result->error : 'Unknown error', + ); + + switch ($result->code) { + case FEEDS_ERROR_PARSE_ERROR: + $message = t('Download of @url failed because it could not be parsed.', $vars); + break; + + case FEEDS_ERROR_NO_SCHEME: + $message = t("Download of @url failed because its scheme could not be determined. The URL is expected to start with something like '@example'.", $vars + array( + '@example' => 'http://', + )); + break; + + case FEEDS_ERROR_INVALID_SCHEME: + $message = t('Download of @url failed because its scheme is not supported: @error. Examples of supported schemes are: @supported.', $vars + array( + '@supported' => implode(', ', array('http', 'https')), + )); + break; + + default: + if (isset($result->error)) { + $message = t('Download of @url failed with code @code and the following error: @error.', $vars); + } + else { + $message = t('Download of @url failed with code @code.', $vars); + } + break; + } + + throw new FeedsHTTPRequestException($message); + } +} + +/** * Decides if it's possible to use cURL or not. * * @return bool diff --git a/plugins/FeedsHTTPFetcher.inc b/plugins/FeedsHTTPFetcher.inc index cd91b63..9f7d0d1 100644 --- a/plugins/FeedsHTTPFetcher.inc +++ b/plugins/FeedsHTTPFetcher.inc @@ -42,15 +42,16 @@ class FeedsHTTPFetcherResult extends FeedsFetcherResult { } /** - * Overrides FeedsFetcherResult::getRaw(); + * Overrides FeedsFetcherResult::getRaw(). + * + * @throws FeedsHTTPRequestException + * In case the result code of the HTTP request is not in the 2xx series. */ public function getRaw() { if (!isset($this->raw)) { feeds_include_library('http_request.inc', 'http_request'); $result = http_request_get($this->url, NULL, NULL, $this->acceptInvalidCert, $this->timeout); - if (!in_array($result->code, array(200, 201, 202, 203, 204, 205, 206))) { - throw new Exception(t('Download of @url failed with code !code.', array('@url' => $this->url, '!code' => $result->code))); - } + http_request_check_result($this->url, $result); $this->raw = $result->data; } diff --git a/plugins/FeedsParser.inc b/plugins/FeedsParser.inc index e4fb66c..5de55ee 100644 --- a/plugins/FeedsParser.inc +++ b/plugins/FeedsParser.inc @@ -430,13 +430,14 @@ class FeedsEnclosure extends FeedsElement { * * @return string * The content of the referenced resource. + * + * @throws FeedsHTTPRequestException + * In case the result code of the HTTP request is not in the 2xx series. */ public function getContent() { feeds_include_library('http_request.inc', 'http_request'); $result = http_request_get($this->getUrlEncodedValue()); - if ($result->code != 200) { - throw new Exception(t('Download of @url failed with code !code.', array('@url' => $this->getUrlEncodedValue(), '!code' => $result->code))); - } + http_request_check_result($this->getUrlEncodedValue(), $result); return $result->data; } diff --git a/tests/FeedsEnclosureTest.test b/tests/FeedsEnclosureTest.test new file mode 100644 index 0000000..ef64651 --- /dev/null +++ b/tests/FeedsEnclosureTest.test @@ -0,0 +1,75 @@ + 'FeedsEnclosure class test', + 'description' => 'Covers class FeedsEnclosure.', + 'group' => 'Feeds', + ); + } + + /** + * Tests getting content from an URL that failed to parse. + * + * @see ::getUrlParseErrorDataProvider() + */ + public function testUrlParseError() { + foreach ($this->getUrlParseErrorDataProvider() as $testdata) { + $e = NULL; + $vars = $testdata['vars'] + array( + '@url' => $testdata['url'], + ); + + // Create a new enclosure. + $enclosure = new FeedsEnclosure($testdata['url'], 'text/plain'); + try { + // The enclosure will try to download the content via HTTP. + $enclosure->getContent(); + } + catch (Exception $e) { + $this->pass(format_string("Exception thrown on url '@url'.", $vars)); + $this->assertEqual(format_string($testdata['error'], $vars), $e->getMessage()); + } + if (!isset($e)) { + $this->fail(format_string("Exception thrown on url '@url'.", $vars)); + } + } + } + + /** + * Data provider for ::testUrlParseError(). + */ + protected function getUrlParseErrorDataProvider() { + return array( + array( + 'url' => 'http://', + 'error' => 'Download of @url failed because it could not be parsed.', + 'vars' => array(), + ), + array( + 'url' => '//www.example.com/path?foo=bar', + 'error' => "Download of @url failed because its scheme could not be determined. The URL is expected to start with something like '@example'.", + 'vars' => array( + '@example' => 'http://', + ), + ), + array( + 'url' => 'nonsupportedscheme://www.example.com', + 'error' => 'Download of @url failed because its scheme is not supported: @error. Examples of supported schemes are: @supported.', + 'vars' => array( + '@error' => 'invalid schema nonsupportedscheme', + '@supported' => implode(', ', array('http', 'https')), + ), + ), + ); + } + +}