3 files changed, 286 insertions, 26 deletions

diff --git a/application/http/HttpAccess.php b/application/http/HttpAccess.php
new file mode 100644
index 00000000..646a5264
--- /dev/null
+++ b/application/http/HttpAccess.php
@@ -0,0 +1,47 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Shaarli\Http;
+
+/**
+ * Class HttpAccess
+ *
+ * This is mostly an OOP wrapper for HTTP functions defined in `HttpUtils`.
+ * It is used as dependency injection in Shaarli's container.
+ *
+ * @package Shaarli\Http
+ */
+class HttpAccess
+{
+    public function getHttpResponse(
+        $url,
+        $timeout = 30,
+        $maxBytes = 4194304,
+        $curlHeaderFunction = null,
+        $curlWriteFunction = null
+    ) {
+        return get_http_response($url, $timeout, $maxBytes, $curlHeaderFunction, $curlWriteFunction);
+    }
+
+    public function getCurlDownloadCallback(
+        &$charset,
+        &$title,
+        &$description,
+        &$keywords,
+        $retrieveDescription
+    ) {
+        return get_curl_download_callback(
+            $charset,
+            $title,
+            $description,
+            $keywords,
+            $retrieveDescription
+        );
+    }
+
+    public function getCurlHeaderCallback(&$charset, $curlGetInfo = 'curl_getinfo')
+    {
+        return get_curl_header_callback($charset, $curlGetInfo);
+    }
+}
diff --git a/application/http/HttpUtils.php b/application/http/HttpUtils.php
index 2ea9195d..28c12969 100644
--- a/application/http/HttpUtils.php
+++ b/application/http/HttpUtils.php
@@ -6,12 +6,14 @@ use Shaarli\Http\Url;
  * GET an HTTP URL to retrieve its content
  * Uses the cURL library or a fallback method
  *
- * @param string          $url               URL to get (http://...)
- * @param int             $timeout           network timeout (in seconds)
- * @param int             $maxBytes          maximum downloaded bytes (default: 4 MiB)
- * @param callable|string $curlWriteFunction Optional callback called during the download (cURL CURLOPT_WRITEFUNCTION).
- *                                           Can be used to add download conditions on the
- *                                           headers (response code, content type, etc.).
+ * @param string          $url                URL to get (http://...)
+ * @param int             $timeout            network timeout (in seconds)
+ * @param int             $maxBytes           maximum downloaded bytes (default: 4 MiB)
+ * @param callable|string $curlHeaderFunction Optional callback called during the download of headers
+ *                                            (CURLOPT_HEADERFUNCTION)
+ * @param callable|string $curlWriteFunction  Optional callback called during the download (cURL CURLOPT_WRITEFUNCTION).
+ *                                            Can be used to add download conditions on the
+ *                                            headers (response code, content type, etc.).
  *
  * @return array HTTP response headers, downloaded content
  *
@@ -35,8 +37,13 @@ use Shaarli\Http\Url;
  * @see http://stackoverflow.com/q/9183178
  * @see http://stackoverflow.com/q/1462720
  */
-function get_http_response($url, $timeout = 30, $maxBytes = 4194304, $curlWriteFunction = null)
-{
+function get_http_response(
+    $url,
+    $timeout = 30,
+    $maxBytes = 4194304,
+    $curlHeaderFunction = null,
+    $curlWriteFunction = null
+) {
     $urlObj = new Url($url);
     $cleanUrl = $urlObj->idnToAscii();
 
@@ -70,7 +77,8 @@ function get_http_response($url, $timeout = 30, $maxBytes = 4194304, $curlWriteF
     // General cURL settings
     curl_setopt($ch, CURLOPT_AUTOREFERER, true);
     curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
-    curl_setopt($ch, CURLOPT_HEADER, true);
+    // Default header download if the $curlHeaderFunction is not defined
+    curl_setopt($ch, CURLOPT_HEADER, !is_callable($curlHeaderFunction));
     curl_setopt(
         $ch,
         CURLOPT_HTTPHEADER,
@@ -81,25 +89,21 @@ function get_http_response($url, $timeout = 30, $maxBytes = 4194304, $curlWriteF
     curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
     curl_setopt($ch, CURLOPT_USERAGENT, $userAgent);
 
-    if (is_callable($curlWriteFunction)) {
-        curl_setopt($ch, CURLOPT_WRITEFUNCTION, $curlWriteFunction);
-    }
-
     // Max download size management
     curl_setopt($ch, CURLOPT_BUFFERSIZE, 1024*16);
     curl_setopt($ch, CURLOPT_NOPROGRESS, false);
+    if (is_callable($curlHeaderFunction)) {
+        curl_setopt($ch, CURLOPT_HEADERFUNCTION, $curlHeaderFunction);
+    }
+    if (is_callable($curlWriteFunction)) {
+        curl_setopt($ch, CURLOPT_WRITEFUNCTION, $curlWriteFunction);
+    }
     curl_setopt(
         $ch,
         CURLOPT_PROGRESSFUNCTION,
-        function ($arg0, $arg1, $arg2, $arg3, $arg4 = 0) use ($maxBytes) {
-            if (version_compare(phpversion(), '5.5', '<')) {
-                // PHP version lower than 5.5
-                // Callback has 4 arguments
-                $downloaded = $arg1;
-            } else {
-                // Callback has 5 arguments
-                $downloaded = $arg2;
-            }
+        function ($arg0, $arg1, $arg2, $arg3, $arg4) use ($maxBytes) {
+            $downloaded = $arg2;
+
             // Non-zero return stops downloading
             return ($downloaded > $maxBytes) ? 1 : 0;
         }
@@ -369,7 +373,11 @@ function server_url($server)
  */
 function index_url($server)
 {
-    $scriptname = $server['SCRIPT_NAME'];
+    if (defined('SHAARLI_ROOT_URL') && null !== SHAARLI_ROOT_URL) {
+        return rtrim(SHAARLI_ROOT_URL, '/') . '/';
+    }
+
+    $scriptname = !empty($server['SCRIPT_NAME']) ? $server['SCRIPT_NAME'] : '/';
     if (endsWith($scriptname, 'index.php')) {
         $scriptname = substr($scriptname, 0, -9);
     }
@@ -377,7 +385,7 @@ function index_url($server)
 }
 
 /**
- * Returns the absolute URL of the current script, with the query
+ * Returns the absolute URL of the current script, with current route and query
  *
  * If the resource is "index.php", then it is removed (for better-looking URLs)
  *
@@ -387,10 +395,17 @@ function index_url($server)
  */
 function page_url($server)
 {
+    $scriptname = $server['SCRIPT_NAME'] ?? '';
+    if (endsWith($scriptname, 'index.php')) {
+        $scriptname = substr($scriptname, 0, -9);
+    }
+
+    $route = preg_replace('@^' . $scriptname . '@', '', $server['REQUEST_URI'] ?? '');
     if (! empty($server['QUERY_STRING'])) {
-        return index_url($server).'?'.$server['QUERY_STRING'];
+        return index_url($server) . $route . '?' . $server['QUERY_STRING'];
     }
-    return index_url($server);
+
+    return index_url($server) . $route;
 }
 
 /**
@@ -477,3 +492,132 @@ function is_https($server)
 
     return ! empty($server['HTTPS']);
 }
+
+/**
+ * Get cURL callback function for CURLOPT_WRITEFUNCTION
+ *
+ * @param string $charset     to extract from the downloaded page (reference)
+ * @param string $curlGetInfo Optionally overrides curl_getinfo function
+ *
+ * @return Closure
+ */
+function get_curl_header_callback(
+    &$charset,
+    $curlGetInfo = 'curl_getinfo'
+) {
+    $isRedirected = false;
+
+    return function ($ch, $data) use ($curlGetInfo, &$charset, &$isRedirected) {
+        $responseCode = $curlGetInfo($ch, CURLINFO_RESPONSE_CODE);
+        $chunkLength = strlen($data);
+        if (!empty($responseCode) && in_array($responseCode, [301, 302])) {
+            $isRedirected = true;
+            return $chunkLength;
+        }
+        if (!empty($responseCode) && $responseCode !== 200) {
+            return false;
+        }
+        // After a redirection, the content type will keep the previous request value
+        // until it finds the next content-type header.
+        if (! $isRedirected || strpos(strtolower($data), 'content-type') !== false) {
+            $contentType = $curlGetInfo($ch, CURLINFO_CONTENT_TYPE);
+        }
+        if (!empty($contentType) && strpos($contentType, 'text/html') === false) {
+            return false;
+        }
+        if (!empty($contentType) && empty($charset)) {
+            $charset = header_extract_charset($contentType);
+        }
+
+        return $chunkLength;
+    };
+}
+
+/**
+ * Get cURL callback function for CURLOPT_WRITEFUNCTION
+ *
+ * @param string $charset     to extract from the downloaded page (reference)
+ * @param string $title       to extract from the downloaded page (reference)
+ * @param string $description to extract from the downloaded page (reference)
+ * @param string $keywords    to extract from the downloaded page (reference)
+ * @param bool   $retrieveDescription Automatically tries to retrieve description and keywords from HTML content
+ * @param string $curlGetInfo Optionally overrides curl_getinfo function
+ *
+ * @return Closure
+ */
+function get_curl_download_callback(
+    &$charset,
+    &$title,
+    &$description,
+    &$keywords,
+    $retrieveDescription
+) {
+    $currentChunk = 0;
+    $foundChunk = null;
+
+    /**
+     * cURL callback function for CURLOPT_WRITEFUNCTION (called during the download).
+     *
+     * While downloading the remote page, we check that the HTTP code is 200 and content type is 'html/text'
+     * Then we extract the title and the charset and stop the download when it's done.
+     *
+     * @param resource $ch   cURL resource
+     * @param string   $data chunk of data being downloaded
+     *
+     * @return int|bool length of $data or false if we need to stop the download
+     */
+    return function ($ch, $data) use (
+        $retrieveDescription,
+        &$charset,
+        &$title,
+        &$description,
+        &$keywords,
+        &$currentChunk,
+        &$foundChunk
+    ) {
+        $chunkLength = strlen($data);
+        $currentChunk++;
+
+        if (empty($charset)) {
+            $charset = html_extract_charset($data);
+        }
+        if (empty($title)) {
+            $title = html_extract_title($data);
+            $foundChunk = ! empty($title) ? $currentChunk : $foundChunk;
+        }
+        if (empty($title)) {
+            $title = html_extract_tag('title', $data);
+            $foundChunk = ! empty($title) ? $currentChunk : $foundChunk;
+        }
+        if ($retrieveDescription && empty($description)) {
+            $description = html_extract_tag('description', $data);
+            $foundChunk = ! empty($description) ? $currentChunk : $foundChunk;
+        }
+        if ($retrieveDescription && empty($keywords)) {
+            $keywords = html_extract_tag('keywords', $data);
+            if (! empty($keywords)) {
+                $foundChunk = $currentChunk;
+                // Keywords use the format tag1, tag2 multiple words, tag
+                // So we format them to match Shaarli's separator and glue multiple words with '-'
+                $keywords = implode(' ', array_map(function($keyword) {
+                    return implode('-', preg_split('/\s+/', trim($keyword)));
+                }, explode(',', $keywords)));
+            }
+        }
+
+        // We got everything we want, stop the download.
+        // If we already found either the title, description or keywords,
+        // it's highly unlikely that we'll found the other metas further than
+        // in the same chunk of data or the next one. So we also stop the download after that.
+        if ((!empty($responseCode) && !empty($contentType) && !empty($charset)) && $foundChunk !== null
+            && (! $retrieveDescription
+                || $foundChunk < $currentChunk
+                || (!empty($title) && !empty($description) && !empty($keywords))
+            )
+        ) {
+            return false;
+        }
+
+        return $chunkLength;
+    };
+}
diff --git a/application/http/MetadataRetriever.php b/application/http/MetadataRetriever.php
new file mode 100644
index 00000000..ba9bd40c
--- /dev/null
+++ b/application/http/MetadataRetriever.php
@@ -0,0 +1,69 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Shaarli\Http;
+
+use Shaarli\Config\ConfigManager;
+
+/**
+ * HTTP Tool used to extract metadata from external URL (title, description, etc.).
+ */
+class MetadataRetriever
+{
+    /** @var ConfigManager */
+    protected $conf;
+
+    /** @var HttpAccess */
+    protected $httpAccess;
+
+    public function __construct(ConfigManager $conf, HttpAccess $httpAccess)
+    {
+        $this->conf = $conf;
+        $this->httpAccess = $httpAccess;
+    }
+
+    /**
+     * Retrieve metadata for given URL.
+     *
+     * @return array [
+     *                  'title' => <remote title>,
+     *                  'description' => <remote description>,
+     *                  'tags' => <remote keywords>,
+     *               ]
+     */
+    public function retrieve(string $url): array
+    {
+        $charset = null;
+        $title = null;
+        $description = null;
+        $tags = null;
+        $retrieveDescription = $this->conf->get('general.retrieve_description');
+
+        // Short timeout to keep the application responsive
+        // The callback will fill $charset and $title with data from the downloaded page.
+        $this->httpAccess->getHttpResponse(
+            $url,
+            $this->conf->get('general.download_timeout', 30),
+            $this->conf->get('general.download_max_size', 4194304),
+            $this->httpAccess->getCurlHeaderCallback($charset),
+            $this->httpAccess->getCurlDownloadCallback(
+                $charset,
+                $title,
+                $description,
+                $tags,
+                $retrieveDescription
+            )
+        );
+
+        if (!empty($title) && strtolower($charset) !== 'utf-8') {
+            $title = mb_convert_encoding($title, 'utf-8', $charset);
+        }
+
+        return [
+            'title' => $title,
+            'description' => $description,
+            'tags' => $tags,
+        ];
+    }
+}