From 1557cefbd76257ceb830f65806831b490faf0acc Mon Sep 17 00:00:00 2001
From: ArthurHoaro <arthur@hoa.ro>
Date: Mon, 4 Jan 2016 10:45:54 +0100
Subject: Fixes #410 - Retrieve title fails in multiple cases

  * `get_http_url()` renamed to `get_http_response()`.
  * Use the same HTTP context to retrieve response headers and content.
  * Follow HTTP 301 and 302 redirections to retrieve the title (default max 3 redirections).
  * Add `LinkUtils` to extract titles and charset.
  * Try to retrieve charset from HTTP headers first (new), then HTML content.
  * Use mb_string to re-encode title if necessary.
---
 application/LinkUtils.php | 79 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 79 insertions(+)
 create mode 100755 application/LinkUtils.php

(limited to 'application/LinkUtils.php')

diff --git a/application/LinkUtils.php b/application/LinkUtils.php
new file mode 100755
index 00000000..26dd6b67
--- /dev/null
+++ b/application/LinkUtils.php
@@ -0,0 +1,79 @@
+<?php
+
+/**
+ * Extract title from an HTML document.
+ *
+ * @param string $html HTML content where to look for a title.
+ *
+ * @return bool|string Extracted title if found, false otherwise.
+ */
+function html_extract_title($html)
+{
+    if (preg_match('!<title>(.*)</title>!is', $html, $matches)) {
+        return trim(str_replace("\n", ' ', $matches[1]));
+    }
+    return false;
+}
+
+/**
+ * Determine charset from downloaded page.
+ * Priority:
+ *   1. HTTP headers (Content type).
+ *   2. HTML content page (tag <meta charset>).
+ *   3. Use a default charset (default: UTF-8).
+ *
+ * @param array  $headers           HTTP headers array.
+ * @param string $htmlContent       HTML content where to look for charset.
+ * @param string $defaultCharset    Default charset to apply if other methods failed.
+ *
+ * @return string Determined charset.
+ */
+function get_charset($headers, $htmlContent, $defaultCharset = 'utf-8')
+{
+    if ($charset = headers_extract_charset($headers)) {
+        return $charset;
+    }
+
+    if ($charset = html_extract_charset($htmlContent)) {
+        return $charset;
+    }
+
+    return $defaultCharset;
+}
+
+/**
+ * Extract charset from HTTP headers if it's defined.
+ *
+ * @param array $headers HTTP headers array.
+ *
+ * @return bool|string Charset string if found (lowercase), false otherwise.
+ */
+function headers_extract_charset($headers)
+{
+    if (! empty($headers['Content-Type']) && strpos($headers['Content-Type'], 'charset=') !== false) {
+        preg_match('/charset="?([^; ]+)/i', $headers['Content-Type'], $match);
+        if (! empty($match[1])) {
+            return strtolower(trim($match[1]));
+        }
+    }
+
+    return false;
+}
+
+/**
+ * Extract charset HTML content (tag <meta charset>).
+ *
+ * @param string $html HTML content where to look for charset.
+ *
+ * @return bool|string Charset string if found, false otherwise.
+ */
+function html_extract_charset($html)
+{
+    // Get encoding specified in HTML header.
+    preg_match('#<meta .*charset="?([^">/]+)"? */?>#Usi', $html, $enc);
+    if (!empty($enc[1])) {
+        return strtolower($enc[1]);
+    }
+
+    return false;
+}
-- 
cgit v1.2.3