From 96152ec7c10dccd2bd66500f0e2f14b5e668ffe7 Mon Sep 17 00:00:00 2001
From: Steven Wittens <steven@acko.net>
Date: Mon, 13 Jun 2011 00:15:41 -0700
Subject: [PATCH] Use hex view for true binary files by default.

---
 Node/shell/meta.js   | 36 +++++++++++++++++++++++------
 Node/shell/reader.js | 55 +++++++++++++++++++++++++++++++++++++++-----
 Node/test.js         |  4 ++--
 todo.txt             |  5 ++--
 4 files changed, 83 insertions(+), 17 deletions(-)

diff --git a/Node/shell/meta.js b/Node/shell/meta.js
index db38e82..7a35319 100644
--- a/Node/shell/meta.js
+++ b/Node/shell/meta.js
@@ -432,18 +432,40 @@ exports.headers.prototype = {
  * Mime type detection.
  */
 exports.sniff = function (file, data) {
-  var parts = file.split('.'),
-      extension = parts.pop();
+  var type = mime.lookup(file),
+      utf8 = true,
+      binary = false;
+
+  // Detect valid UTF-8.
+  // Ignore trailing error due to possible truncation.
+  var attempt = data.toString('utf-8').replace(/�$/, ''), error = attempt.indexOf('�');
+  if (error != -1) {
+    utf8 = false;
+  }
 
-  if ((type = mime.lookup(file)) && (type != 'application/octet-stream')){
-    return type;
+  // Detect binary data.
+  if (/[\u0000]/(attempt)) {
+    binary = true;
   }
 
-  if (/[^\u0001-\uFFFF]/('' + data)) {
-    return 'application/octet-stream';
+  // Plain text.
+  if (utf8 || !binary) {
+    return [ type, { charset: 'utf-8' } ];
   }
 
-  return 'text/plain';
+  // Specific binary type.
+  if (type != 'application/octet-stream') {
+    return type;
+  }
+
+  // If data contains random binary data, then a significant part will be invalid / non-printable.
+  // Use hex-view.
+  if (attempt.replace(/[^\n\r\t\u0020-\uFFFC]/g, '').length < .8 * data.length) {
+    return [ type, { schema: 'termkit.hex' } ];
+  }
+  
+  // Fallback, escaped binary text output.
+  return type;
 };
 
 /**
diff --git a/Node/shell/reader.js b/Node/shell/reader.js
index 04979d4..728fb69 100644
--- a/Node/shell/reader.js
+++ b/Node/shell/reader.js
@@ -8,6 +8,16 @@ var fs = require('fs'),
     composePath = require('misc').composePath,
     expandPath = require('misc').expandPath;
 
+// Is x an object?
+function isObject(x) {
+  return typeof x == 'object';
+}
+
+// Is x an array?
+function isArray(x) {
+  return isObject(x) && x.constructor == [].constructor;
+}
+
 /**
  * Data reader for termkit-style dataIn.
  *
@@ -379,6 +389,32 @@ exports.filesReader.prototype = {
       return types[0];
     }
     
+    // Merge multiple types.
+    var merged = {}, hasParams = false;
+    for (var i in types) {
+      // Process params.
+      if (isArray(types[i])) {
+        var params = types[i][1];
+        
+        // Only merge identical params.
+        for (var j in params) {
+          if (merged[j] === null) continue;
+          if (merged[j] != params[j]) {
+            merged[j] = null;
+          }
+          merged[j] = params[j];
+        }
+
+        // Strip params for merge.
+        types[i] = types[i][0];
+      }
+    }
+    // Check if there are params.
+    for (var j in merged) {
+      hasParams = true;
+      break;
+    }
+    
     function commonPrefix(types) {
       // Sort types, inspect first/last strings.
       types.sort();
@@ -399,6 +435,8 @@ exports.filesReader.prototype = {
       return match;
     }
     
+    var type = 'application/octed-stream';
+    
     prefix = commonPrefix(types);
     if (!(/^[^\/]+\/[^\/]+$/(prefix))) {
       // If we only matched a type category (e.g. text/),
@@ -410,22 +448,27 @@ exports.filesReader.prototype = {
       
       if (!(/^[^\/]+\/[^\/]+$/(prefix))) {
         // Replace with generic type.
-        return meta.default(prefix) || 'application/octet-stream';
+        type = meta.default(prefix) || 'application/octet-stream';
+      }
+      else {
+        type = prefix;
       }
-      
-      return prefix;
     }
     else {
       // Only return a unified type for content types which can be composed safely.
       var composable = meta.composable();
       if (composable[prefix]) {
         if (typeof composable[prefix] == 'string') {
-          return composable[prefix];
+          type = composable[prefix];
+        }
+        else {
+          type = prefix;
         }
-        return prefix;
       }
-      return 'application/octet-stream';
     }
+    
+    // Return with or without merged params.
+    return hasParams ? [ type, merged ] : type;
   },
   
   /**
diff --git a/Node/test.js b/Node/test.js
index 7abf3d6..ecb8e88 100644
--- a/Node/test.js
+++ b/Node/test.js
@@ -43,7 +43,7 @@ function mockClient(flow, callback) {
   var messages = [], success = true, i;
 
   client.send = function (message) {
-    messages.push(JSON.parse(message));
+    messages.push(message);
   };
 
   client.disconnect = function (message) {
@@ -53,7 +53,7 @@ function mockClient(flow, callback) {
   var r = new router.router(client);
   for (i in flow) (function (message) {
     setTimeout(function () {
-      client.emit('message', JSON.stringify(message));
+      client.emit('message', message);
     }, i * 100);
   })(flow[i]);
   
diff --git a/todo.txt b/todo.txt
index 9196b97..5605283 100644
--- a/todo.txt
+++ b/todo.txt
@@ -3,8 +3,9 @@ Tasks:
 [X] redefine ansi colors
 [X] make fileReader with same pattern as reader.js
 [X] cat multiple files -> octet-stream w/ type coercion
-[ ] hex widget
-[ ] hex utility
+[X] hex widget
+[X] hex utility
+[ ] hex detection
 [ ] backgrounding / new command trigger
 [ ] move escapeUnixText into front-end, unixLegacy widget
 [ ] command aliases