diff --git a/CHANGELOG.md b/CHANGELOG.md
index bcef5eacfe6ad4a642c77b8e36225b0e52d9f683..21677751eef67e5e6218c96247a7b36e3b822679 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -3,4 +3,5 @@ here's only the changelog for the version in development
# v1.4
-* _TBA_
+* improved encoding of small non-all-keys-are-strings dictionaries
+
diff --git a/docs/specification.rst b/docs/specification.rst
index f18f4ad3e66241d1e654cad087302b636df1accd..c789bfdc345691de1671f8a5b8d45608ce2e6dcb 100644
--- a/docs/specification.rst
+++ b/docs/specification.rst
@@ -19,9 +19,11 @@ with len of (value & 0x7F)
* If value's two highest bits are 0100 or 0101, then four lowest bits encode the number of elements,
and the four highest bits encode type of the object:
* 0100 - a list
- * 0101 - an object
- Standard representation for an object or list follows,
- sans the element count.
+ * 0101 - an object whose keys are all strings
+ * 0110 - an object whose keys are not all strings (see value of 19 and 20 to know how it's
+ represented).
+ * Standard representation for a non-key-string object (value 19), string key object (value 11) or list (value 7) follows,
+ sans the element count.
* If value is zero, then next character is the length of the string followed by the string
* If value is 1, then next data is signed int
* If value is 2, then next data is signed short
@@ -51,7 +53,11 @@ with len of (value & 0x7F)
and then a list follows of that many elements
* If value is 17, then next data is a unsigned short,
and then an object follows of that many elements
-* If value is 18, then next data is a unsigned int,
+* If value is 18, then next data is an unsigned int,
and then an object follows of that many elements
-* If value is 19, then next data is a unsigned int,
+* If value is 19, then next data is an unsigned int,
+ and then follow that many pairs of Values (key: value)
+* If value is 20, then next data is an unsigned char,
+ and then follow that many pairs of Values (key: value)
+* If value is 21, then next data is an unsigned short,
and then follow that many pairs of Values (key: value)
diff --git a/minijson/routines.pyx b/minijson/routines.pyx
index 166dc98f8d2fb699b01064bc4ed734ddaf3aeea4..5e108e78357620796a20275b5a81dc45f4a4ea23 100644
--- a/minijson/routines.pyx
+++ b/minijson/routines.pyx
@@ -155,6 +155,11 @@ cpdef tuple parse(bytes data, int starting_position):
elements = value_type & 0xF
offset, e_dict = parse_dict(data, elements, starting_position+1)
return offset+1, e_dict
+ elif value_type & 0xF0 == 0x60:
+ e_dict = {}
+ elements = value_type & 0xF
+ offset, e_dict = parse_sdict(data, elements, starting_position+1)
+ return offset+1, e_dict
elif value_type == 0:
string_length = data[starting_position+1]
offset, b_field_name = parse_cstring(data, starting_position+1)
@@ -228,6 +233,14 @@ cpdef tuple parse(bytes data, int starting_position):
elements, = STRUCT_L.unpack(data[starting_position+1:starting_position+5])
offset, e_dict = parse_sdict(data, elements, starting_position+5)
return offset+5, e_dict
+ elif value_type == 20:
+ elements = data[starting_position+1]
+ offset, e_dict = parse_sdict(data, elements, starting_position+2)
+ return offset+2, e_dict
+ elif value_type == 21:
+ elements, = STRUCT_H.unpack(data[starting_position+1:starting_position+3])
+ offset, e_dict = parse_sdict(data, elements, starting_position+3)
+ return offset+3, e_dict
raise DecodingError('Unknown sequence type %s!' % (value_type, ))
except IndexError as e:
raise DecodingError('String too short!') from e
@@ -362,9 +375,21 @@ cpdef int dump(object data, cio: io.BytesIO) except -1:
raise EncodingError('Keys have to be strings!') from e
return length
else:
- cio.write(b'\x13')
- cio.write(STRUCT_L.pack(length))
- offset = 5
+ if length < 16:
+ cio.write(bytearray([0b01100000 | length]))
+ offset = 1
+ elif length < 256:
+ cio.write(bytearray([20, length]))
+ offset = 2
+ elif length < 0xFFFF:
+ cio.write(b'\x15')
+ cio.write(STRUCT_H.pack(length))
+ offset = 3
+ else:
+ cio.write(b'\x13')
+ cio.write(STRUCT_L.pack(length))
+ offset = 5
+
for key, value in data.items():
offset += dump(key, cio)
offset += dump(value, cio)
diff --git a/setup.py b/setup.py
index 0c5f94595b230aecfda292a2f3239fbcc422179d..0ae1e6569203f857463b6bd7ca429f1c307d7cae 100644
--- a/setup.py
+++ b/setup.py
@@ -14,7 +14,7 @@ if 'DEBUG' in os.environ:
directives['embedsignature'] = True
-setup(version='1.4a1',
+setup(version='1.4',
packages=find_packages(include=['minijson', 'minijson.*']),
ext_modules=build([Multibuild('minijson', find_pyx('minijson'),
dont_snakehouse=dont_snakehouse), ],
diff --git a/tests/test_minijson.py b/tests/test_minijson.py
index 0b2caccc72ed74682d591fcbc244b813043c4099..d8392483957d2489edc2c9ff590b47e87266db43 100644
--- a/tests/test_minijson.py
+++ b/tests/test_minijson.py
@@ -7,6 +7,19 @@ class TestMiniJSON(unittest.TestCase):
def assertSameAfterDumpsAndLoads(self, c):
self.assertEqual(loads(dumps(c)), c)
+ def test_short_nonstring_key_dicts(self):
+ a = {}
+ for i in range(20):
+ a[i] = i
+ self.assertSameAfterDumpsAndLoads(a)
+ a = {}
+ for i in range(300):
+ a[i] = i
+ self.assertSameAfterDumpsAndLoads(a)
+ for i in range(700000):
+ a[i] = i
+ self.assertSameAfterDumpsAndLoads(a)
+
def test_string(self):
a = 'test'
b = 't'*128