Trim URLs in breadcrumbs; add maxUrlLength config opt (#906)

Author: benvinegar

docs/config.rst

@@ -205,6 +205,12 @@ Those configuration options are documented below:
     By default, Raven does not truncate messages. If you need to truncate
     characters for whatever reason, you may set this to limit the length.
 
+.. describe:: maxUrlLength
+
+    By default, Raven will truncate URLs as they appear in breadcrumbs and other meta
+    interfaces to 250 characters in order to minimize bytes over the wire. This does *not*
+    affect URLs in stack traces.
+
 .. describe:: autoBreadcrumbs
 
     Enables/disables automatic collection of breadcrumbs. Possible values are:

src/raven.js

@@ -50,6 +50,9 @@ function Raven() {
         crossOrigin: 'anonymous',
         collectWindowErrors: true,
         maxMessageLength: 0,
+
+        // By default, truncates URL values to 250 chars
+        maxUrlLength: 250,
         stackTraceLimit: 50,
         autoBreadcrumbs: true,
         sampleRate: 1
@@ -1324,9 +1327,46 @@ Raven.prototype = {
             exception.value = truncate(exception.value, max);
         }
 
+        var request = data.request;
+        if (request) {
+            if (request.url) {
+                request.url = truncate(request.url, this._globalOptions.maxUrlLength);
+            }
+            if (request.Referer) {
+                request.Referer = truncate(request.Referer, this._globalOptions.maxUrlLength);
+            }
+        }
+
+        if (data.breadcrumbs && data.breadcrumbs.values)
+            this._trimBreadcrumbs(data.breadcrumbs);
+
         return data;
     },
 
+    /**
+     * Truncate breadcrumb values (right now just URLs)
+     */
+    _trimBreadcrumbs: function (breadcrumbs) {
+        // known breadcrumb properties with urls
+        // TODO: also consider arbitrary prop values that start with (https?)?://
+        var urlprops = {to: 1, from: 1, url: 1},
+            crumb,
+            data;
+
+        for (var i = 0; i < breadcrumbs.values.length; i++) {
+            crumb = breadcrumbs.values[i];
+            if (!crumb.hasOwnProperty('data'))
+                continue;
+
+            data = crumb.data;
+            for (var prop in urlprops) {
+                if (data.hasOwnProperty(prop)) {
+                    data[prop] = truncate(data[prop], this._globalOptions.maxUrlLength);
+                }
+            }
+        }
+    },
+
     _getHttpData: function() {
         if (!this._hasNavigator && !this._hasDocument) return;
         var httpData = {};

test/raven.test.js

@@ -88,6 +88,7 @@ describe('globals', function() {
 
     describe('trimPacket', function() {
         it('should work as advertised', function() {
+            // message/excepion value (maxMessageLength)
             Raven._globalOptions.maxMessageLength = 3;
             assert.deepEqual(
                 Raven._trimPacket({message: 'lol'}),
@@ -105,6 +106,51 @@ describe('globals', function() {
                 Raven._trimPacket({message: 'lolol', exception: {values: [{value: 'lolol'}]}}),
                 {message: 'lol\u2026', exception: {values: [{value: 'lol\u2026'}]}}
             );
+
+            // request URLs (maxUrlLength)
+            Raven._globalOptions.maxUrlLength = 20;
+            assert.deepEqual(
+                Raven._trimPacket({request: {url: 'http://example.com/foo/bar/baz.js', Referer: 'http://example.com/f\u2026'}}),
+                {request: {url: 'http://example.com/f\u2026', Referer: 'http://example.com/f\u2026'}}
+            )
+        });
+    });
+
+    describe('_trimBreadcrumbs', function() {
+        it('should work as expected', function() {
+            Raven._globalOptions.maxUrlLength = 20;
+
+            var breadcrumbs = {
+                values: [{
+                    data: {
+                        donttouch: 'this-is-not-a-url-field-and-should-not-be-touched',
+                        to: 'http://example.com/foo/bar/baz.js',
+                        from: 'http://example.com/foo/bar/baz.js'
+                    }
+                }, {
+                    data: {
+                        donttouch: 'this-is-not-a-url-field-and-should-not-be-touched',
+                        url: 'http://example.com/foo/bar/baz.js'
+                    }
+                }]
+            };
+
+            // NOTE: _trimBreadcrumbs mutates data in-place
+            Raven._trimBreadcrumbs(breadcrumbs);
+            assert.deepEqual(breadcrumbs, {
+                values: [{
+                    data: {
+                        donttouch: 'this-is-not-a-url-field-and-should-not-be-touched',
+                        to: 'http://example.com/f\u2026', // 20 chars
+                        from: 'http://example.com/f\u2026'
+                    }
+                }, {
+                    data: {
+                        donttouch: 'this-is-not-a-url-field-and-should-not-be-touched',
+                        url: 'http://example.com/f\u2026'
+                    }
+                }]
+            });
         });
     });
 

typescript/raven.d.ts

@@ -56,6 +56,9 @@ interface RavenOptions {
     /** By default, Raven does not truncate messages. If you need to truncate characters for whatever reason, you may set this to limit the length. */
     maxMessageLength?: number;
 
+    /** By default, Raven will truncate URLs as they appear in breadcrumbs and other meta interfaces to 250 characters in order to minimize bytes over the wire. This does *not* affect URLs in stack traces. */
+    maxUrlLength?: number;
+
     /** Override the default HTTP data transport handler. */
     transport?: (options: RavenTransportOptions) => void;
 }