diff --git a/controllers/Parse.php b/controllers/Parse.php
index 8540dbc..49f00af 100644
--- a/controllers/Parse.php
+++ b/controllers/Parse.php
@@ -67,6 +67,10 @@ class Parse {
$this->_pretty = true;
}
+ if($request->get('include-mf1')) {
+ $opts['include-mf1'] = $request->get('include-mf1') == 'false' ? false : true;
+ }
+
$url = $request->get('url');
$html = $request->get('html') ?: $request->get('body');
diff --git a/lib/XRay/Formats/HTML.php b/lib/XRay/Formats/HTML.php
index 5846266..eb53c7e 100644
--- a/lib/XRay/Formats/HTML.php
+++ b/lib/XRay/Formats/HTML.php
@@ -58,7 +58,11 @@ class HTML extends Format {
}
}
- $mf2 = \mf2\Parse($html, $url);
+ $includeMF1 = true;
+ if(isset($opts['include-mf1']) && $opts['include-mf1'] == false)
+ $includeMF1 = false;
+
+ $mf2 = \Mf2\parse($html, $url, $includeMF1);
$canonical = false;
diff --git a/lib/XRay/Parser.php b/lib/XRay/Parser.php
index e2900db..28886b8 100644
--- a/lib/XRay/Parser.php
+++ b/lib/XRay/Parser.php
@@ -29,6 +29,25 @@ class Parser {
} else {
$found = $this->_findLinkInTree($opts['target'], $document['data']);
$error_description = 'The Microformats at the source URL do not contain a link to the target URL. Check the source URL in a Microformats parser such as php.microformats.io';
+
+ if(!$found && isset($document['html'])) {
+ // If no link was found in the parsed mf2 tree, check for a link in the HTML
+ $found = $this->_findLinkInHTML($opts['target'], $document['html']);
+ // If there is a link, and if the HTML document has no mf2, then downgrade to a regular mention
+ if($found) {
+ $mf2Data = Formats\HTML::parse($this->http, $http_response, ['include-mf1'=>false]);
+ if(isset($mf2Data['data']['type']) && $mf2Data['data']['type'] == 'unknown') {
+ // Since the link was found in the HTML, but not in the parsed tree, it shouldn't return the parsed document
+ $document['data'] = [
+ 'type' => 'unknown'
+ ];
+ } else {
+ // Otherwise, the document did have mf2, but the link wasn't in it (checked earlier), so set found=false
+ $found = false;
+ }
+ }
+ }
+
}
if(!$found) {
diff --git a/tests/ParseTest.php b/tests/ParseTest.php
index d9afeb2..61a2f68 100644
--- a/tests/ParseTest.php
+++ b/tests/ParseTest.php
@@ -1135,4 +1135,73 @@ class ParseTest extends PHPUnit_Framework_TestCase {
$this->assertEquals('https://aaronparecki.com/2019/12/01/10/homeautomation', $data['data']['url']);
$this->assertEquals('https://aaronparecki.com/2019/12/01/10/homeautomation', $data['data']['rels']['canonical']);
}
+
+ public function testTargetLinkOutsideHEntry() {
+ $url = 'http://source.example.com/target-test-link-outside-h-entry';
+ $response = $this->parse(['url' => $url, 'target' => 'https://target.example.com/']);
+
+ $body = $response->getContent();
+ $this->assertEquals(200, $response->getStatusCode());
+ $data = json_decode($body, true);
+
+ $this->assertEquals('no_link_found', $data['error']);
+ }
+
+ public function testTargetLinkWithBadMf1() {
+ $url = 'http://source.example.com/target-test-only-bad-mf1';
+ $response = $this->parse(['url' => $url, 'target' => 'https://target.example.com/']);
+
+ $body = $response->getContent();
+ $this->assertEquals(200, $response->getStatusCode());
+ $data = json_decode($body, true);
+
+ $this->assertEquals('unknown', $data['data']['type']);
+ }
+
+ public function testTargetLinkWithValidMf1() {
+ $url = 'http://source.example.com/target-test-only-good-mf1';
+ $response = $this->parse(['url' => $url, 'target' => 'https://target.example.com/']);
+
+ $body = $response->getContent();
+ $this->assertEquals(200, $response->getStatusCode());
+ $data = json_decode($body, true);
+
+ $this->assertEquals('entry', $data['data']['type']);
+ $this->assertEquals('target', $data['data']['content']['html']);
+ }
+
+ public function testTargetLinkOutsideValidMf1() {
+ $url = 'http://source.example.com/target-test-link-outside-valid-mf1';
+ $response = $this->parse(['url' => $url, 'target' => 'https://target.example.com/']);
+
+ $body = $response->getContent();
+ $this->assertEquals(200, $response->getStatusCode());
+ $data = json_decode($body, true);
+
+ // Since the link was found in the HTML, but not in the parsed tree, it shouldn't return the parsed document
+ $this->assertEquals('unknown', $data['data']['type']);
+ }
+
+ public function testDisableMf1Parsing() {
+ $url = 'http://source.example.com/target-test-only-good-mf1';
+ $response = $this->parse(['url' => $url, 'include-mf1' => 'false']);
+
+ $body = $response->getContent();
+ $this->assertEquals(200, $response->getStatusCode());
+ $data = json_decode($body, true);
+
+ $this->assertEquals('unknown', $data['data']['type']);
+ }
+
+ public function testEnableMf1Parsing() {
+ $url = 'http://source.example.com/target-test-only-good-mf1';
+ $response = $this->parse(['url' => $url, 'include-mf1' => 'true']);
+
+ $body = $response->getContent();
+ $this->assertEquals(200, $response->getStatusCode());
+ $data = json_decode($body, true);
+
+ $this->assertEquals('entry', $data['data']['type']);
+ }
+
}
diff --git a/tests/data/source.example.com/target-test-link-outside-h-entry b/tests/data/source.example.com/target-test-link-outside-h-entry
new file mode 100644
index 0000000..3d719fd
--- /dev/null
+++ b/tests/data/source.example.com/target-test-link-outside-h-entry
@@ -0,0 +1,17 @@
+HTTP/1.1 200 OK
+Server: Apache
+Date: Wed, 09 Dec 2015 03:29:14 GMT
+Content-Type: text/html; charset=utf-8
+Connection: keep-alive
+
+
+
+ Test
+
+
+
+
+
+
diff --git a/tests/data/source.example.com/target-test-link-outside-valid-mf1 b/tests/data/source.example.com/target-test-link-outside-valid-mf1
new file mode 100644
index 0000000..046e433
--- /dev/null
+++ b/tests/data/source.example.com/target-test-link-outside-valid-mf1
@@ -0,0 +1,17 @@
+HTTP/1.1 200 OK
+Server: Apache
+Date: Wed, 09 Dec 2015 03:29:14 GMT
+Content-Type: text/html; charset=utf-8
+Connection: keep-alive
+
+
+
+ Test
+
+
+
+
+
+
diff --git a/tests/data/source.example.com/target-test-only-bad-mf1 b/tests/data/source.example.com/target-test-only-bad-mf1
new file mode 100644
index 0000000..6de94d1
--- /dev/null
+++ b/tests/data/source.example.com/target-test-only-bad-mf1
@@ -0,0 +1,16 @@
+HTTP/1.1 200 OK
+Server: Apache
+Date: Wed, 09 Dec 2015 03:29:14 GMT
+Content-Type: text/html; charset=utf-8
+Connection: keep-alive
+
+
+
+ Test
+
+
+
+
+
diff --git a/tests/data/source.example.com/target-test-only-good-mf1 b/tests/data/source.example.com/target-test-only-good-mf1
new file mode 100644
index 0000000..97ee246
--- /dev/null
+++ b/tests/data/source.example.com/target-test-only-good-mf1
@@ -0,0 +1,16 @@
+HTTP/1.1 200 OK
+Server: Apache
+Date: Wed, 09 Dec 2015 03:29:14 GMT
+Content-Type: text/html; charset=utf-8
+Connection: keep-alive
+
+
+
+ Test
+
+
+
+
+