add hostname matching and https exception

lib/hostname-redirects.js

@@ -0,0 +1,123 @@
+'use strict';
+
+// TODO detect infinite redirects
+
+module.exports.compile = module.exports.sortOpts = function (opts) {
+  var redirects = opts.redirects;
+  var dups = {};
+  var results = {
+    conflicts: {}
+  , patterns: []
+  , matchesMap: {}
+  };
+
+  redirects.forEach(function (r) {
+    var bare;
+    var www;
+
+    if ('.' === r.id[0]) {
+      // for consistency
+      // TODO this should happen at the database level
+      r.id = '*' + r.id;
+    }
+    if ('*' === r.id[0]) {
+      // TODO check that we are not trying to redirect a tld (.com, .co.uk, .org, etc)
+      // tlds should follow the global policy
+      if (r.id[1] && '.' !== r.id[1]) {
+        // this is not a good place to throw as the consequences of a bug would be
+        // very bad, but errors should never be silent, so we'll compromise
+        console.warn("[NON-FATAL ERROR]: ignoring redirect pattern '" + r.id + "'");
+        results.conflicts[r.id] = r;
+      }
+
+      // nix the '*' for easier matching
+      r.id = r.id.slice(1);
+      if (!r.id) {
+        r.id = '*';
+      }
+      if (dups[r.id]) {
+        results.conflicts[r.id] = r;
+        console.warn("[NON-FATAL ERROR]: duplicate entry for redirect pattern '" + r.id + "'");
+      }
+      dups[r.id] = true;
+      results.patterns.push(r);
+      return;
+    }
+
+    bare = r.id.replace(/^www\./i, '');
+    www = r.id.replace(/^(www\.)?/i, 'www.');
+
+    if (true === r.value) {
+      // implicit add www
+      results.matchesMap[bare] = www;
+      results.matchesMap[www] = www;
+    } else if (false === r.value) {
+      // implicit remove www
+      results.matchesMap[bare] = bare;
+      results.matchesMap[www] = bare;
+    } else if (!r.value) {
+      // (null, '', 0, undefined)
+      // explicitly no change
+      results.matchesMap[r.id] = r.id;
+    } else {
+      // explicit value
+      results.matchesMap[r.id] = r.value;
+    }
+  });
+
+  results.patterns.sort(function (a, b) {
+    return b.id.length - a.id.length;
+  });
+
+  return results;
+};
+
+module.exports.redirectTo = function (hostname, opts) {
+  var redir = opts.matchesMap[hostname];
+
+  if (redir) {
+    if (redir === hostname) {
+      return false;
+    }
+    return redir;
+  }
+
+  // longest to shortest
+  var hasWww = ('www.' === hostname.slice(0, 4));
+  //var noWww = (hasWww && hostname.slice(4)) || hostname;
+  //var yesWww = (hasWww && hostname) || ('www.' + hostname);
+
+  redir = false;
+  opts.patterns.some(function (r) {
+    // r.id begins with a dot, such as '.foo.example.com'
+    if (r.id !== hostname.slice(hostname.length - r.id.length)) {
+      // except for the default, which is an *
+      if ('*' !== r.id) {
+        return false;
+      }
+    }
+
+    if (true === r.value) {
+      // implicit add www
+      redir = hasWww ? hostname : ('www.' + hostname);
+    } else if (false === r.value) {
+      // implicit remove www
+      redir = hasWww ? hostname.slice(4) : hostname;
+    } else if (!r.value) {
+      // (null, '', 0, undefined)
+      // explicitly no change
+      redir = false;
+    } else {
+      // explicit value
+      redir = r.value;
+    }
+
+    return true;
+  });
+
+  if (redir === hostname) {
+    return false;
+  }
+
+  return redir;
+};

lib/no-www.js

@@ -1,14 +1,26 @@
-module.exports.scrubTheDub = function (req, res) {
+'use strict';
+
+module.exports.scrubTheDub = function (req, res, redirectives) {
   // hack for bricked app-cache
   // Also 301 redirects will not work for appcache (must issue html)
   if (require('./unbrick-appcache').unbrick(req, res)) {
-    return;
+    return true;
   }
 
   // TODO port number for non-443
   var escapeHtml = require('escape-html');
-  var newLocation = 'https://' + req.hostname.replace(/^www\./, '') + req.url;
-  var safeLocation = escapeHtml(newLocation);
+  var newLocation;
+  var safeLocation;
+
+  if (redirectives) {
+    newLocation = require('./hostname-redirects').redirectTo(req.hostname, redirectives);
+    if (!newLocation) {
+      return false;
+    }
+  } else {
+    newLocation = 'https://' + req.hostname.replace(/^www\./, '') + req.url;
+  }
+  safeLocation = escapeHtml(newLocation);
 
   var metaRedirect = ''
     + '<html>\n'
@@ -24,4 +36,6 @@ module.exports.scrubTheDub = function (req, res) {
     ;
 
   res.end(metaRedirect);
+
+  return true;
 };

lib/pathname-redirects.js

@@ -0,0 +1,38 @@
+  /*
+  //var escapeRe;
+    //var insecureRedirects;
+    if (require('./unbrick-appcache').unbrick(req, res)) {
+      return;
+    }
+
+    // because I have domains for which I don't want to pay for SSL certs
+    insecureRedirects = (redirects||[]).sort(function (a, b) {
+      var hlen = b.from.hostname.length - a.from.hostname.length;
+      var plen;
+      if (!hlen) {
+        plen = b.from.path.length - a.from.path.length;
+        return plen;
+      }
+      return hlen;
+    }).forEach(function (redirect) {
+      var origHost = host;
+
+      if (!escapeRe) {
+        escapeRe = require('escape-string-regexp');
+      }
+
+      // TODO if '*' === hostname[0], omit '^'
+      host = host.replace(
+        new RegExp('^' + escapeRe(redirect.from.hostname))
+      , redirect.to.hostname
+      );
+      if (host === origHost) {
+        return;
+      }
+      url = url.replace(
+        new RegExp('^' + escapeRe(redirect.from.path))
+      , redirect.to.path
+      );
+    });
+    */
+

tests/hostname-redirects.js

@@ -0,0 +1,72 @@
+'use strict';
+
+var opts = {
+  redirects: [
+    { "id": "*", "value": true }
+  , { "id": "ns2.redirect-www.org", "value": false }
+  , { "id": "hellabit.com", "value": false }
+  , { "id": "*.hellabit.com", "value": false }
+  , { "id": "redirect-www.org", "value": null }
+  , { "id": "www.redirect-www.org", "value": null }
+  , { "id": "no.redirect-www.org", "value": false }
+  , { "id": "*.redirect-www.org", "value": false }
+  , { "id": "*.yes.redirect-www.org", "value": true }
+  , { "id": "yes.redirect-www.org", "value": true }
+  , { "id": "*.maybe.redirect-www.org", "value": null }
+  , { "id": "maybe.redirect-www.org", "value": null }
+  , { "id": "blog.coolaj86.com", "value": 'coolaj86.com' } // TODO pathname
+]
+, matchesMap: null
+, patternsMap: null
+, patterns: null
+};
+
+var redirectTo = require('../lib/hostname-redirects').redirectTo;
+var sortOpts = require('../lib/hostname-redirects').sortOpts;
+
+var domains = {
+// maybewww
+  'redirect-www.org': false
+, 'www.redirect-www.org': false
+, 'maybe.redirect-www.org': false
+, 'www.maybe.redirect-www.org': false
+
+// yeswww
+, 'yes.redirect-www.org': 'www.yes.redirect-www.org'
+, 'foo.yes.redirect-www.org': 'www.foo.yes.redirect-www.org'
+
+// nowww
+, 'www.no.redirect-www.org': 'no.redirect-www.org'
+, 'www.foo.no.redirect-www.org': 'foo.no.redirect-www.org'
+
+, 'ns2.redirect-www.org': false
+, 'www.ns2.redirect-www.org': 'ns2.redirect-www.org'
+
+, 'ns1.redirect-www.org': false
+, 'www.ns1.redirect-www.org': 'ns1.redirect-www.org'
+
+, 'hellabit.com': false
+, 'www.hellabit.com': 'hellabit.com'
+
+// default policy (yeswww)
+, 'ahellabit.com': 'www.ahellabit.com'
+, 'www.ahellabit.com': false
+, 'example.com': 'www.example.com'
+, 'www.example.com': false
+};
+
+var redirects = sortOpts(opts);
+
+console.log(redirects);
+
+Object.keys(domains).forEach(function (domain, i) {
+  var redir = domains[domain];
+  var result = redirectTo(domain, redirects);
+
+  if (redir !== result) {
+    throw new Error("For domain #" + i + " '" + domain + "' expected '" + redir + "' but got '" + result + "'");
+  }
+});
+
+console.log("Didn't throw any errors. Must have worked, eh?");
+console.log("TODO: detect and report infinite redirects");