Updated for consistency with php-mf2 v0.2.0

mf2/Shim/Instagram.php → Mf2/Shim/Instagram.php

@@ -1,9 +1,9 @@
 <?php
-namespace mf2\Shim;
+namespace Mf2\Shim;
 
-use mf2\Shim;
+use Mf2\Parser;
 
-class Instagram extends Shim {
+class Instagram extends Parser {
 
   // TODO: parse() method that returns an 'items' array that corresponds to if instagram.com
   // had proper microformats markup
@@ -18,6 +18,4 @@ public function parse() {
 
     return array('items' => array_values(array_filter($mfs)));
   }
-
-
 }

Mf2/Shim/Twitter.php

@@ -0,0 +1,87 @@
+<?php
+namespace Mf2\Shim;
+
+use Mf2;
+use DateTime;
+use DOMElement;
+use Exception;
+
+function parseTwitter($html, $url=null) {
+	$parser = new Twitter($html, $url, false);
+	return array_merge($parser->parseRelsAndAlternates(), $parser->parse());
+}
+
+class Twitter extends Mf2\Parser {
+	public function parseTweet(DOMElement $el, $parseReplies=true) {
+		$tweetTextEl = $this->query('.//p' . Mf2\xpcs('tweet-text'), $el)->item(0);
+
+		$authorNameEl = $this->query('.//*' . Mf2\xpcs('fullname'), $el)->item(0);
+		$authorNickEl = $this->query('.//*' . Mf2\xpcs('username'), $el)->item(0);
+		$authorPhotoEl = $this->query('.//*' . Mf2\xpcs('avatar'), $el)->item(0);
+
+		$publishedEl = $this->query('.//*' . Mf2\xpcs('_timestamp'), $el)->item(0);
+		$publishedTimestamp = $publishedEl->getAttribute('data-time');
+		try {
+			$publishedDateTime = DateTime::createFromFormat('U', $publishedTimestamp)->format(DateTime::W3C);
+		} catch (Exception $e) {
+			$publishedDateTime = '';
+		}
+
+		$urlEl = $this->query('.//*' . Mf2\xpcs('tweet-timestamp'), $el)->item(0);
+
+		$htmlTweetContent = '';
+		foreach ($tweetTextEl->childNodes as $node) {
+			$htmlTweetContent .= $node->C14N();
+		}
+
+		$tweet = array(
+			'type' => array('h-entry'),
+			'properties' => array(
+				'uid' => array(),
+				'name' => array($tweetTextEl->nodeValue),
+				'content' => array(array(
+					'value' => $tweetTextEl->nodeValue,
+					'html' =>  $htmlTweetContent
+				)),
+				'summary' => array($tweetTextEl->nodeValue),
+				'url' => array($urlEl->getAttribute('href')),
+				'published' => array($publishedDateTime),
+				'author' => array(
+						array(
+						'type' => array('h-card'),
+						'properties' => array(
+							'uid' => array(),
+							'name' => array($authorNameEl->nodeValue),
+							'nickname' => array($authorNickEl->nodeValue),
+							'photo' => array($authorPhotoEl->getAttribute('src')),
+							'url' => array('https://twitter.com/' . $authorNickEl->nodeValue)
+						)
+					)
+				)
+			)
+		);
+
+		if ($parseReplies) {
+			foreach ($this->query('//*' . Mf2\xpcs('permalink-replies') . '//*' . Mf2\xpcs('tweet')) as $reply) {
+				$tweet['properties']['comment'][] = $this->parseTweet($reply, false);
+			}
+		}
+
+		return $tweet;
+	}
+
+	/**
+   * Parse
+   * 
+   * @return array
+   */
+  public function parse() {
+    $items = array();
+
+    foreach($this->query('//div' . Mf2\xpcs('tweet', 'permalink-tweet')) as $node) {
+      $items[] = $this->parseTweet($node);
+    }
+
+    return array('items' => array_values(array_filter($items)));
+  }
+}

Mf2/functions.php

@@ -0,0 +1,23 @@
+<?php
+
+namespace Mf2;
+
+function spaceSeparatedAttributeXpathSelector($needle, $haystack="@class") {
+	return 'contains(concat(" ", ' . $haystack . ', " "), " ' . $needle .' ")';
+}
+
+/**
+ * XPath Class Selector
+ * 
+ * @return string
+ */
+function xpcs() {
+	$classnames = func_get_args();
+	if (count($classnames) == 1) {
+		return '[' . spaceSeparatedAttributeXpathSelector($classnames[0]) . ']';
+	} else {
+		return '[' . implode(' and ', array_map(function ($item) {
+			return spaceSeparatedAttributeXpathSelector($item);
+		}, $classnames)) . ']';
+	}
+}

README.md

@@ -1,9 +1,44 @@
 php-mf2-shim
 ============
 
-A parsing library for parsing pages from a few common non-mf2 sites into the mf2 structure
+A parsing library for parsing pages from a few common non-mf2 sites into the mf2 structure.
 
-## Warning!
+## Installation
 
-This code is very much in progress at the moment! Feel free to contribute to development, or feel free to stay away from it while I keep working on it.
+Install php-mf2-shim with Composer by adding "mf2/shim": "0.2.*" to the require object in your composer.json and running `php composer.phar update`.
 
+You could install it by just downloading /Mf2/functions.php, /Mf2/Shim/*.php and including those, but please use Composer. Seriously, it’s amazing.
+
+## Usage
+
+mf2-shim is PSR-0 autoloadable, so all you have to do to load it is:
+
+* Include Composer’s auto-generated autoload file (/vendor/autoload.php)
+* Call Mf2\parseTwitter() with the HTML (or a DOMDocument), and optionally the URL to resolve relative URLs against.
+
+## Examples
+
+```php
+<?php
+
+require 'vendor/autoload.php';
+
+use Mf2;
+
+$output = Mf2\Shim\parseTwitter($html, $url);
+```
+
+## Changelog
+
+### v0.2.0 (BREAKING)
+
+* Restructured things for consistency with php-mf2 v0.2.0
+* Removed dependencies, now only depends on php-mf2
+* Twitter parsing code tweet content is now an e-* dict containing html and value keys with raw and plaintext values, respectively
+
+### v0.1.0 (2013-06-07)
+
+* Initial tagged release
+* MIT Licenesed
+* Twitter tests passing, Instagram test failing
+* Tweet permalink page parsing working including authors and reply tweets

composer.json

@@ -16,17 +16,12 @@
   ],
   "require": {
     "php": ">=5.3.0",
-    "mf2/mf2": "*",
-    "symfony/css-selector": "*"
+		"mf2/mf2": "0.2.*"
   },
   "require-dev": {
     "phpunit/phpunit": "3.7.*"
   },
   "autoload": {
-    "psr-0": {
-      "mf2\\Shim": "",
-      "mf2\\Shim\\Twitter": "",
-      "mf2\\Shim\\Instagram": ""
-    }
+		"files": ["Mf2/functions.php", "Mf2/Shim/Twitter.php", "Mf2/Shim/Instagram.php"]
   }
-}
+}