handle duplicate files in file packager

2 files changed, 12 insertions, 0 deletions

diff --git a/tests/runner.py b/tests/runner.py
index c8e3de37..e847f727 100755
--- a/tests/runner.py
+++ b/tests/runner.py
@@ -6837,6 +6837,10 @@ f.close()
       Popen(['python', EMCC, os.path.join(self.get_dir(), 'main.cpp'), '--embed-file', 'somefile.txt']).communicate()
       self.assertContained('|hello from a file wi|', run_js(os.path.join(self.get_dir(), 'a.out.js')))
 
+      # preload twice, should not err
+      Popen(['python', EMCC, os.path.join(self.get_dir(), 'main.cpp'), '--embed-file', 'somefile.txt', '--embed-file', 'somefile.txt']).communicate()
+      self.assertContained('|hello from a file wi|', run_js(os.path.join(self.get_dir(), 'a.out.js')))
+
     def test_multidynamic_link(self):
       # Linking the same dynamic library in will error, normally, since we statically link it, causing dupe symbols
       # A workaround is to use --ignore-dynamic-linking, see emcc --help for details
diff --git a/tools/file_packager.py b/tools/file_packager.py
index 9e98320a..4f24af28 100644
--- a/tools/file_packager.py
+++ b/tools/file_packager.py
@@ -101,6 +101,14 @@ for file_ in data_files:
   file_['name'] = file_['name'].replace(os.path.sep, '/')
   file_['net_name'] = file_['name']
 
+# remove duplicates (can occur naively, for example preload dir/, preload dir/subdir/)
+seen = {}
+def was_seen(name):
+  if seen.get(name): return True
+  seen[name] = 1
+  return False
+data_files = filter(lambda file_: not was_seen(file_['name']), data_files)
+
 data_target = sys.argv[1]
 
 # Set up folders