Handle invalid source encodings

src/sentry/lang/javascript/processor.py

@@ -344,6 +344,20 @@ def fetch_file(url, project=None, release=None, allow_scraping=True):
         }
         raise CannotFetchSource(error)
 
+    # Make sure the file we're getting back is unicode, if it's not,
+    # it's either some encoding that we don't understand, or it's binary
+    # data which we can't process.
+    if not isinstance(result[1], unicode):
+        try:
+            result[1] = result[1].decode('utf8')
+        except UnicodeDecodeError:
+            error = {
+                'type': EventError.JS_INVALID_SOURCE_ENCODING,
+                'value': 'utf8',
+                'url': url,
+            }
+            raise CannotFetchSource(error)
+
     return UrlResult(url, result[0], result[1])
 
 

src/sentry/models/eventerror.py

@@ -15,6 +15,7 @@ class EventError(object):
     JS_MISSING_SOURCE = 'js_no_source'
     JS_INVALID_SOURCEMAP = 'js_invalid_source'
     JS_TOO_MANY_REMOTE_SOURCES = 'js_too_many_sources'
+    JS_INVALID_SOURCE_ENCODING = 'js_invalid_source_encoding'
 
     _messages = {
         INVALID_DATA: 'Discarded invalid value for parameter \'{name}\'',
@@ -29,6 +30,7 @@ class EventError(object):
         JS_MISSING_SOURCE: 'Source code was not found for {url}',
         JS_INVALID_SOURCEMAP: 'Sourcemap was invalid or not parseable: {url}',
         JS_TOO_MANY_REMOTE_SOURCES: 'The maximum number of remote source requests was made',
+        JS_INVALID_SOURCE_ENCODING: 'Source file was not \'{value}\' encoding: {url}'
     }
 
     @classmethod