* initial commit

author: emkael <emkael@tlen.pl> 2015-01-27 16:31:12 +0100
committer: emkael <emkael@tlen.pl> 2015-01-27 16:31:12 +0100
commit: cab5dcbafb0f82ba742ac014741dfc4208be9fbd (patch)
tree: 3081fc9150ac7d15280c099a4699085a2a27087f
8 files changed, 155 insertions, 0 deletions
diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..bb0d1d4
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,2 @@
+_log
+_stats
diff --git a/cache/.gitignore b/cache/.gitignore
new file mode 100644
index 0000000..72e8ffc
--- /dev/null
+++ b/cache/.gitignore
@@ -0,0 +1 @@
+*
diff --git a/http/.htaccess b/http/.htaccess
new file mode 100644
index 0000000..28a620e
--- /dev/null
+++ b/http/.htaccess
@@ -0,0 +1,8 @@
+RewriteEngine On
+
+RewriteBase /
+
+RewriteCond %{SCRIPT_FILENAME} !-f
+RewriteCond %{SCRIPT_FILENAME} !-d
+RewriteRule ^[^\.]*$ index.php [QSA,L]
+
diff --git a/http/foto/.gitignore b/http/foto/.gitignore
new file mode 100644
index 0000000..72e8ffc
--- /dev/null
+++ b/http/foto/.gitignore
@@ -0,0 +1 @@
+*
diff --git a/http/index.php b/http/index.php
new file mode 100644
index 0000000..42a2731
--- /dev/null
+++ b/http/index.php
@@ -0,0 +1,121 @@
+<?php
+
+define('CACHE_PATH', '../cache');
+define('CEZAR_URL', 'http://msc.com.pl/cezar/');
+
+// fetch item from content cache
+function get_cache_content($cacheKey, $url, $force = FALSE) {
+  $cacheFile = realpath(dirname(__FILE__) . '/' . CACHE_PATH) . '/' . $cacheKey;
+  if ($force || !(file_exists($cacheFile) && filemtime($cacheFile) > strtotime('-1 day'))) {
+    fetch_with_user_agent_spoof($cacheFile, $url);
+  }
+  return file_get_contents($cacheFile);
+}
+
+// save remote URL locally, forwarding browser's User Agent String
+function fetch_with_user_agent_spoof($targetFile, $sourceUrl) {
+    file_put_contents(
+                      $targetFile,
+                      file_get_contents($sourceUrl,
+                                        FALSE,
+                                        stream_context_create(['http' => [
+                                                                          'user_agent' => $_SERVER['HTTP_USER_AGENT']
+                                                                          ]
+                                                               ]))
+                      );
+}
+
+// parse requested path (after rewrite)
+$url = parse_url($_SERVER['REQUEST_URI']);
+$path = array_values(array_filter(explode('/', $url['path'])));
+
+if ($path) {
+  // /[ANYTHING]/refresh disables cache (forces cache refresh)
+  $cache = !(count($path) > 1 && $path[1] == 'refresh');
+
+  // build Cezar URL for requested path
+  $path[0] = urldecode($path[0]);  
+  $searchUrl = new http\Url(CEZAR_URL,
+                            ['query' => http_build_query(
+                                                         ['pid_search' => $path[0],
+                                                          'p' => '21']
+                                                         )]);
+  
+  $mscUrl = $searchUrl->toString();
+
+  $content = get_cache_content($path[0], $mscUrl, !$cache); // requested content
+  $contentLines = explode(PHP_EOL, $content);
+
+  // if the comment delimiters are present, we're possibly dealing with the content we want, slice it and wrap it
+  $delimiters = array_keys(preg_grep('/---- page content /', $contentLines));
+  if ($delimiters) {
+    $content = '<html><head><meta http-equiv="Content-Type" content="text/html; charset=utf-8" /><base href="/" /><style>body {width: 580px; font-family: Tahoma, Geneva, Arial, Helvetica, "sans-serif";}</style></head><body>'
+      . implode(PHP_EOL, array_slice($contentLines, $delimiters[0]+1, $delimiters[1]-$delimiters[0]-1))
+      . '</body></html>';
+  }
+  else {
+    die('Malformed (even more than usually) content :(');
+  }
+
+  require_once('/usr/share/php/QueryPath/QueryPath.php');
+
+  $html = htmlqp($content, NULL, ['convert_to_encoding' => 'utf-8']);
+
+  // for search list pages, replace links with internal links to player IDs
+  $links = $html->find('a[href^="?p=21&pid="]');
+  if ($links->size()) {
+    foreach ($links as $link) {
+      $href = [];
+      parse_str($link->attr('href'), $href);
+      $link->attr('href', $href['pid']);
+    }
+  }
+
+  $html->top();
+
+  // remove general crap
+  $html->find('script, table.msc_noprint, center>p')->remove();
+  $html->top();
+
+  // leave only first-ish table of the content
+  $html->find('table > tr > td')->eq(1)->remove();
+  $html->top();
+  $html->find('table > tr > td > table')->eq(2)->remove();
+  $html->top();
+  $html->find('table > tr > td > table')->eq(2)->remove();
+  $html->top();
+
+  // remove internal Cezar links
+  $innerLinks = $html->find('table > tr > td > table a');
+  foreach ($innerLinks as $innerLink) {
+    $innerLink->removeAttr('href');
+  }
+  // get rid of Cezar link icons (right green arrows)
+  $html->find('img[src*="ico_link_8.gif"]')->remove();
+  $html->top();
+
+  // proxy all external images, by resolving them relatively to the original server
+  // and cache them locally
+  // internal images are left untouched in the markup and are proxied through pic/fetch.php handler
+  // (if they're not present/overwritten locally)
+  $images = $html->find('img')->not('[src^="pic/"]');
+  foreach ($images as $image) {
+    $src = $image->attr('src');
+    $url = new http\Url(CEZAR_URL, $src, http\Url::FROM_ENV | http\Url::SANITIZE_PATH | http\Url::JOIN_PATH | http\Url::REPLACE);
+    $imageUrl = $url->toString();
+    $cachedImageUrl = 'foto/' . md5($imageUrl) . '.' . array_pop(explode('.', $imageUrl));
+    if (!file_exists($cachedImageUrl) || !$cache) {
+      fetch_with_user_agent_spoof($cachedImageUrl, $imageUrl);
+    }
+    $image->attr('src', $cachedImageUrl);
+  }
+  $html->top();
+
+  // all done
+  print $html->html();
+}
+else {
+  die('Nothing to see here, move along.');
+}
+
+?>
diff --git a/http/pic/.gitignore b/http/pic/.gitignore
new file mode 100644
index 0000000..ef5c65e
--- /dev/null
+++ b/http/pic/.gitignore
@@ -0,0 +1,3 @@
+*.jpg
+*.png
+*.gif
diff --git a/http/pic/.htaccess b/http/pic/.htaccess
new file mode 100644
index 0000000..615bc19
--- /dev/null
+++ b/http/pic/.htaccess
@@ -0,0 +1,6 @@
+RewriteEngine On
+
+RewriteCond %{SCRIPT_FILENAME} !-f
+RewriteCond %{SCRIPT_FILENAME} !-d
+RewriteRule .* fetch.php [QSA,L]
+
diff --git a/http/pic/fetch.php b/http/pic/fetch.php
new file mode 100644
index 0000000..1fb94cd
--- /dev/null
+++ b/http/pic/fetch.php
@@ -0,0 +1,13 @@
+<?php
+
+$resource = @file_get_contents('http://msc.com.pl/cezar' . $_SERVER['REQUEST_URI']);
+if ($resource) {
+  $filename = array_pop(explode('/', $_SERVER['REQUEST_URI']));
+  file_put_contents($filename, $resource);
+  foreach ($http_response_header as $header) {
+    header($header);
+  }
+  readfile($filename);
+}
+
+?>
author	emkael <emkael@tlen.pl>	2015-01-27 16:31:12 +0100
committer	emkael <emkael@tlen.pl>	2015-01-27 16:31:12 +0100
commit	cab5dcbafb0f82ba742ac014741dfc4208be9fbd (patch)
tree	3081fc9150ac7d15280c099a4699085a2a27087f