From: Christian Weiske <cweiske@cweiske.de>
Date: Thu, 5 Nov 2015 06:40:03 +0000 (+0100)
Subject: Detect text files for unknown file types
X-Git-Tag: v0.7.0~1
X-Git-Url: https://git.cweiske.de/phorkie.git/commitdiff_plain/0b24764b8d1065fc57e219c431112860c7147dd0

Detect text files for unknown file types
---

diff --git a/README.rst b/README.rst
index 2e99e6c..dde1bda 100644
--- a/README.rst
+++ b/README.rst
@@ -42,6 +42,7 @@ Features
 - webhook support - get notified when pastes are created, edited or deleted
 - atom feed for new and updated pastes
 - notifies remote instances via linkbacks when a paste has been forked
+- text file detection for unknown file types
 
 
 ============
@@ -117,6 +118,8 @@ on your machine:
 
 - Git v1.7.5 or later
 - PHP v5.3.0 or later
+
+  - optionally the ``mbstring`` extension
 - PEAR v1.9.2 or later
 
 ::
diff --git a/src/phorkie/File.php b/src/phorkie/File.php
index 300e810..cf5daae 100644
--- a/src/phorkie/File.php
+++ b/src/phorkie/File.php
@@ -126,15 +126,21 @@ class File
     }
 
     /**
-     * @return string Mime type of file
+     * @return string Mime type of file, NULL if no type detected
      */
     public function getMimeType()
     {
         $ext = $this->getExt();
-        if (!isset($GLOBALS['phorkie']['languages'][$ext])) {
-            return null;
+        if (isset($GLOBALS['phorkie']['languages'][$ext])) {
+            return $GLOBALS['phorkie']['languages'][$ext]['mime'];
         }
-        return $GLOBALS['phorkie']['languages'][$ext]['mime'];
+
+        $mte = new \MIME_Type_Extension();
+        $type = $mte->getMIMEType($this->getFilename());
+        if (!\PEAR::isError($type)) {
+            return $type;
+        }
+        return null;
     }
 
     /**
@@ -159,22 +165,53 @@ class File
     {
         $ext = $this->getExt();
         if ($ext == '') {
-            //no file extension? then consider the size
-            $size = filesize($this->getFullPath());
-            //files <= 4kiB are considered to be text
-            return $size <= 4096;
+            return $this->isNonBinary();
         }
 
-        if (!isset($GLOBALS['phorkie']['languages'][$ext]['mime'])) {
-            return false;
+        $type = $this->getMimeType();
+        if ($type === null) {
+            return $this->isNonBinary();
         }
-
-        $type = $GLOBALS['phorkie']['languages'][$ext]['mime'];
         return substr($type, 0, 5) === 'text/'
             || $type == 'application/javascript'
             || substr($type, -4) == '+xml'
             || substr($type, -5) == '+json';
     }
+
+    /**
+     * Look at the file's bytes and guess if it's binary or not.
+     *
+     * @return boolean True if it's most likely plain text
+     */
+    public function isNonBinary()
+    {
+        $fp = fopen($this->getFullPath(), 'r');
+        if (!$fp) {
+            return false;
+        }
+
+        //When multibyte extension is not installed,
+        // we only allow files with ASCII characters.
+        // Files with UTF-8 characters will not be detected as text.
+        $hasMb = function_exists('mb_detect_encoding');
+
+        $pos = 0;
+        $data = '';
+        while (false !== ($char = fgetc($fp)) && ++$pos < 100) {
+            $data .= $char;
+            if (!$hasMb && ord($char) > 128) {
+                return false;
+            }
+        }
+        if (!$hasMb) {
+            return true;
+        }
+
+        if (mb_detect_encoding($data) === false) {
+            return false;
+        }
+        return true;
+    }
 }
 
 ?>