Add Clear() for python Builder (#8186)

Co-authored-by: Derek Bailey <derekbailey@google.com>

python/flatbuffers/builder.py

@@ -146,6 +146,20 @@ class Builder(object):
         ## @endcond
         self.finished = False
 
+    def Clear(self) -> None:
+        ## @cond FLATBUFFERS_INTERNAL
+        self.current_vtable = None
+        self.head = UOffsetTFlags.py_type(len(self.Bytes))
+        self.minalign = 1
+        self.objectEnd = None
+        self.vtables = {}
+        self.nested = False
+        self.forceDefaults = False
+        self.sharedStrings = {}
+        self.vectorNumElems = None
+        ## @endcond
+        self.finished = False
+
     def Output(self):
         """Return the portion of the buffer that has been used for writing data.
 

tests/PythonTest.sh

@@ -40,7 +40,7 @@ function run_tests() {
     JYTHONPATH=${runtime_library_dir}:${gen_code_path} \
     COMPARE_GENERATED_TO_GO=0 \
     COMPARE_GENERATED_TO_JAVA=0 \
-    $1 py_test.py $2 $3 $4 $5
+    $1 py_test.py $2 $3 $4 $5 $6
     if [ $1 = python3 ]; then
       PYTHONDONTWRITEBYTECODE=1 \
       PYTHONPATH=${runtime_library_dir}:${gen_code_path} \
@@ -52,12 +52,12 @@ function run_tests() {
 }
 
 # Run test suite with these interpreters. The arguments are benchmark counts.
-run_tests python2.6 100 100 100 false
-run_tests python2.7 100 100 100 false
-run_tests python2.7 100 100 100 true
-run_tests python3 100 100 100 false
-run_tests python3 100 100 100 true
-run_tests pypy 100 100 100 false
+run_tests python2.6 100 100 100 100 false
+run_tests python2.7 100 100 100 100 false
+run_tests python2.7 100 100 100 100 true
+run_tests python3 100 100 100 100 false
+run_tests python3 100 100 100 100 true
+run_tests pypy 100 100 100 100 false
 
 # NOTE: We'd like to support python2.5 in the future.
 
@@ -77,7 +77,7 @@ if $(which coverage >/dev/null); then
 
   PYTHONDONTWRITEBYTECODE=1 \
   PYTHONPATH=${runtime_library_dir}:${gen_code_path} \
-  coverage run --source=flatbuffers,MyGame py_test.py 0 0 0 false > /dev/null
+  coverage run --source=flatbuffers,MyGame py_test.py 0 0 0 0 false > /dev/null
 
   echo
   cov_result=`coverage report --omit="*flatbuffers/vendor*,*py_test*" \

tests/py_test.py

@@ -2017,9 +2017,9 @@ class TestByteLayout(unittest.TestCase):
         ])
 
 
-def make_monster_from_generated_code(sizePrefix=False, file_identifier=None):
+def make_monster_from_generated_code(b=None, sizePrefix=False, file_identifier=None):
   """ Use generated code to build the example Monster. """
-
+  if b is None:
     b = flatbuffers.Builder(0)
   string = b.CreateString('MyMonster')
   test1 = b.CreateString('test1')
@@ -2767,6 +2767,43 @@ class TestNestedUnionTables(unittest.TestCase):
     self.assertEqual(nestUnionDecodeTFromBuf2.data.test3.b, nestUnion.data.test3.b)
 
 
+class TestBuilderClear(unittest.TestCase):
+
+  def test_consistency(self):
+    """ Checks if clear resets the state of the builder. """
+    b = flatbuffers.Builder(0)
+
+    # Add some data to the buffer
+    off1 = b.CreateString('a' * 1024)
+    want = b.Bytes[b.Head():]
+
+    # Reset the builder
+    b.Clear()
+
+    # Readd the same data into the buffer
+    off2 = b.CreateString('a' * 1024)
+    got = b.Bytes[b.Head():]
+
+    # Expect to get the same data into the buffer at the same offset
+    self.assertEqual(off1, off2)
+    self.assertEqual(want, got)
+
+  def test_repeated_clear_after_builder_reuse(self):
+    init_buf = None
+    init_off = None
+    b = flatbuffers.Builder(0)
+
+    for i in range(5):
+      buf, off = make_monster_from_generated_code(b)
+      b.Clear()
+
+      if i > 0:
+        self.assertEqual(init_buf, buf)
+        self.assertEqual(init_off, off)
+      else:
+        init_buf = buf
+        init_off = off
+
 def CheckAgainstGoldDataGo():
   try:
     gen_buf, gen_off = make_monster_from_generated_code()
@@ -2912,6 +2949,18 @@ def BenchmarkMakeMonsterFromGeneratedCode(count, length):
          (count, length, duration, rate, data_rate)))
 
 
+def BenchmarkBuilderClear(count, length):
+  b = flatbuffers.Builder(length)
+  duration = timeit.timeit(stmt=lambda: make_monster_from_generated_code(b),
+                           number=count)
+  rate = float(count) / duration
+  data = float(length * count) / float(1024 * 1024)
+  data_rate = data / float(duration)
+
+  print(('built %d %d-byte flatbuffers (reused buffer) in %.2fsec:'
+         ' %.2f/sec, %.2fMB/sec' % (count, length, duration, rate, data_rate)))
+
+
 def backward_compatible_run_tests(**kwargs):
   if PY_VERSION < (2, 6):
     sys.stderr.write('Python version less than 2.6 are not supported')
@@ -2940,10 +2989,10 @@ def backward_compatible_run_tests(**kwargs):
 def main():
   import os
   import sys
-  if not len(sys.argv) == 5:
+  if not len(sys.argv) == 6:
     sys.stderr.write('Usage: %s <benchmark vtable count> '
                      '<benchmark read count> <benchmark build count> '
-                     '<is_onefile>\n' % sys.argv[0])
+                     '<benchmark clear builder> <is_onefile>\n' % sys.argv[0])
     sys.stderr.write('       Provide COMPARE_GENERATED_TO_GO=1   to check'
                      'for bytewise comparison to Go data.\n')
     sys.stderr.write('       Provide COMPARE_GENERATED_TO_JAVA=1 to check'
@@ -2951,9 +3000,9 @@ def main():
     sys.stderr.flush()
     sys.exit(1)
 
-  kwargs = dict(argv=sys.argv[:-4])
+  kwargs = dict(argv=sys.argv[:-5])
 
-  create_namespace_shortcut(sys.argv[4].lower() == 'true')
+  create_namespace_shortcut(sys.argv[5].lower() == 'true')
 
   # show whether numpy is present, as it changes the test logic:
   try:
@@ -2978,6 +3027,7 @@ def main():
   bench_vtable = int(sys.argv[1])
   bench_traverse = int(sys.argv[2])
   bench_build = int(sys.argv[3])
+  bench_clear = int(sys.argv[4])
   if bench_vtable:
     BenchmarkVtableDeduplication(bench_vtable)
   if bench_traverse:
@@ -2986,7 +3036,9 @@ def main():
   if bench_build:
     buf, off = make_monster_from_generated_code()
     BenchmarkMakeMonsterFromGeneratedCode(bench_build, len(buf))
-
+  if bench_clear:
+    buf, off = make_monster_from_generated_code()
+    BenchmarkBuilderClear(bench_build, len(buf))
 
 if __name__ == '__main__':
   main()