First pass on type hints + some fixes

Some fixes were added to avoid some possible errors
found with mypy.

See #39

Author: tcalmant

javaobj/utils.py

@@ -30,6 +30,7 @@
 from __future__ import absolute_import
 
 # Standard library
+from typing import Any, Tuple
 import logging
 import struct
 import sys
@@ -76,7 +77,7 @@ def log_error(message, ident=0):
 
 
 def read_struct(data, fmt_str):
-    # type: (bytes, str) -> list
+    # type: (bytes, str) -> Tuple
     """
     Reads input bytes and extract the given structure. Returns both the read
     elements and the remaining data
@@ -90,7 +91,7 @@ def read_struct(data, fmt_str):
 
 
 def read_string(data, length_fmt="H"):
-    # type: (bytes, str) -> UNICODE_TYPE
+    # type: (bytes, str) -> Tuple[UNICODE_TYPE, bytes]
     """
     Reads a serialized string
 

javaobj/v1/beans.py

@@ -198,6 +198,8 @@ def __init__(self, classdesc=None):
         JavaObject.__init__(self)
         self.classdesc = classdesc
 
+    def __hash__(self):
+        return list.__hash__(self)
 
 class JavaByteArray(JavaObject):
     """

javaobj/v1/transformers.py

@@ -30,6 +30,8 @@
 import functools
 
 from .beans import JavaObject
+from .unmarshaller import JavaObjectUnmarshaller
+from ..constants import ClassDescFlags, TerminalCode, TypeCode
 from ..utils import (
     log_debug,
     log_error,
@@ -59,6 +61,9 @@ def __init__(self, unmarshaller):
             list.__init__(self)
             JavaObject.__init__(self)
 
+        def __hash__(self):
+                return list.__hash__(self)
+
         def __extra_loading__(self, unmarshaller, ident=0):
             # type: (JavaObjectUnmarshaller, int) -> None
             """
@@ -110,6 +115,9 @@ def __init__(self, unmarshaller):
             dict.__init__(self)
             JavaObject.__init__(self)
 
+        def __hash__(self):
+            return dict.__hash__(self)
+
         def __extra_loading__(self, unmarshaller, ident=0):
             # type: (JavaObjectUnmarshaller, int) -> None
             """
@@ -128,15 +136,15 @@ def __extra_loading__(self, unmarshaller, ident=0):
             """
             # Ignore the blockdata opid
             (opid,) = unmarshaller._readStruct(">B")
-            if opid != unmarshaller.SC_BLOCK_DATA:
+            if opid != ClassDescFlags.SC_BLOCK_DATA:
                 raise ValueError("Start of block data not found")
 
             # Read HashMap fields
             self.buckets = unmarshaller._read_value(
-                unmarshaller.TYPE_INTEGER, ident
+                TypeCode.TYPE_INTEGER, ident
             )
             self.size = unmarshaller._read_value(
-                unmarshaller.TYPE_INTEGER, ident
+                TypeCode.TYPE_INTEGER, ident
             )
 
             # Read entries
@@ -147,7 +155,7 @@ def __extra_loading__(self, unmarshaller, ident=0):
 
             # Ignore the end of the blockdata
             unmarshaller._read_and_exec_opcode(
-                ident, [unmarshaller.TC_ENDBLOCKDATA]
+                ident, [TerminalCode.TC_ENDBLOCKDATA]
             )
 
             # Ignore the trailing 0
@@ -165,6 +173,9 @@ def __init__(self, unmarshaller):
             set.__init__(self)
             JavaObject.__init__(self)
 
+        def __hash__(self):
+                return set.__hash__(self)
+
         def __extra_loading__(self, unmarshaller, ident=0):
             # type: (JavaObjectUnmarshaller, int) -> None
             """

javaobj/v1/unmarshaller.py

@@ -36,6 +36,7 @@
 from __future__ import absolute_import
 
 # Standard library
+from typing import Any, Union
 import os
 import struct
 
@@ -719,7 +720,7 @@ def do_enum(self, parent=None, ident=0):
         return enum
 
     def _read_value(self, raw_field_type, ident, name=""):
-        # type: (bytes, int, str) -> Any
+        # type: (Union[bytes, int, TypeCode], int, str) -> Any
         """
         Reads the next value, of the given type
 
@@ -729,15 +730,21 @@ def _read_value(self, raw_field_type, ident, name=""):
         :return: The read value
         :raise RuntimeError: Unknown field type
         """
-        if isinstance(raw_field_type, (TypeCode, int)):
+        if isinstance(raw_field_type, TypeCode):
             field_type = raw_field_type
+        elif isinstance(raw_field_type, int):
+            field_type = TypeCode(raw_field_type)
         else:
             # We don't need details for arrays and objects
-            field_type = TypeCode(ord(raw_field_type[0]))
+            raw_code = raw_field_type[0]
+            if isinstance(raw_code, int):
+                field_type = TypeCode(raw_code)
+            else:
+                field_type = TypeCode(ord(raw_code))
 
         if field_type == TypeCode.TYPE_BOOLEAN:
             (val,) = self._readStruct(">B")
-            res = bool(val)
+            res = bool(val)  # type: Any
         elif field_type == TypeCode.TYPE_BYTE:
             (res,) = self._readStruct(">b")
         elif field_type == TypeCode.TYPE_CHAR:

javaobj/v2/beans.py

@@ -202,7 +202,7 @@ def __init__(self, field_type, name, class_name=None):
         # type: (FieldType, str, Optional[JavaString]) -> None
         self.type = field_type
         self.name = name
-        self.class_name = class_name  # type: JavaString
+        self.class_name = class_name
         self.is_inner_class_reference = False
 
         if self.class_name:

javaobj/v2/core.py

@@ -216,15 +216,15 @@ def _dump_instance(self, instance):
 
         if instance.annotations:
             lines.append("\tobject annotations:")
-            for cd, content in instance.annotations.items():
-                lines.append("\t" + cd.name)
-                for c in content:
+            for cd, annotation in instance.annotations.items():
+                lines.append("\t" + (cd.name or "null"))
+                for c in annotation:
                     lines.append("\t\t" + str(c))
 
         if instance.field_data:
             lines.append("\tfield data:")
             for field, obj in instance.field_data.items():
-                line = "\t\t" + field.name + ": "
+                line = "\t\t" + (field.name or "null") + ": "
                 if isinstance(obj, ParsedJavaContent):
                     content = obj  # type: ParsedJavaContent
                     h = content.handle
@@ -280,7 +280,7 @@ def _do_null(self, _):
         return None
 
     def _read_content(self, type_code, block_data, class_desc=None):
-        # type: (int, bool) -> ParsedJavaContent
+        # type: (int, bool, Optional[JavaClassDesc]) -> ParsedJavaContent
         """
         Parses the next content
         """
@@ -345,7 +345,7 @@ def _read_classdesc(self):
         return self._do_classdesc(type_code)
 
     def _do_classdesc(self, type_code):
-        # type: (int, bool) -> JavaClassDesc
+        # type: (int) -> JavaClassDesc
         """
         Parses a class description
         """
@@ -432,7 +432,7 @@ def _custom_readObject(self, class_name):
         raise ValueError("Custom readObject can not be processed")
 
     def _read_class_annotations(self, class_desc=None):
-        # type: () -> List[ParsedJavaContent]
+        # type: (Optional[JavaClassDesc]) -> List[ParsedJavaContent]
         """
         Reads the annotations associated to a class
         """
@@ -649,7 +649,7 @@ def _do_array(self, type_code):
         """
         cd = self._read_classdesc()
         handle = self._new_handle()
-        if len(cd.name) < 2:
+        if not cd.name or len(cd.name) < 2:
             raise ValueError("Invalid name in array class description")
 
         # ParsedJavaContent type

javaobj/v2/main.py

@@ -31,7 +31,7 @@
 
 
 def load(file_object, *transformers, **kwargs):
-    # type: (IO[bytes], ObjectTransformer) -> Any
+    # type: (IO[bytes], ObjectTransformer, Any) -> Any
     """
     Deserializes Java primitive data and objects serialized using
     ObjectOutputStream from a file-like object.
@@ -68,7 +68,7 @@ def load(file_object, *transformers, **kwargs):
 
 
 def loads(data, *transformers, **kwargs):
-    # type: (bytes, ObjectTransformer) -> Any
+    # type: (bytes, ObjectTransformer, Any) -> Any
     """
     Deserializes Java objects and primitive data serialized using
     ObjectOutputStream from bytes.

javaobj/v2/stream.py

@@ -26,11 +26,11 @@
 
 from __future__ import absolute_import
 
-from typing import Any, IO, List
+from typing import Any, IO, List, Tuple
 import struct
 
 from ..modifiedutf8 import decode_modified_utf8
-from ..utils import unicode_char
+from ..utils import unicode_char, UNICODE_TYPE
 
 # ------------------------------------------------------------------------------
 
@@ -65,7 +65,7 @@ def file_descriptor(self):
         return self.__fd
 
     def read(self, struct_format):
-        # type: (str) -> List[Any]
+        # type: (str) -> Tuple[Any, ...]
         """
         Reads from the input stream, using struct
 
@@ -103,7 +103,7 @@ def read_ubyte(self):
         return self.read(">B")[0]
 
     def read_char(self):
-        # type: () -> chr
+        # type: () -> UNICODE_TYPE
         """
         Shortcut to read a single `char` (2 bytes)
         """

javaobj/v2/transformers.py

@@ -25,7 +25,7 @@
 """
 
 # Standard library
-from typing import List, Optional
+from typing import List, Optional, Tuple
 import functools
 
 # Numpy (optional)
@@ -37,7 +37,8 @@
 
 # Javaobj
 from .api import ObjectTransformer
-from .beans import JavaInstance
+from .beans import JavaInstance, JavaClassDesc
+from .core import JavaStreamParser, DataStreamReader
 from ..constants import TerminalCode, TypeCode
 from ..utils import to_bytes, log_error, log_debug, read_struct, read_string
 
@@ -137,7 +138,7 @@ class JavaMap(dict, JavaInstance):
     Python-Java dictionary/map bridge type
     """
 
-    HANDLED_CLASSES = ("java.util.HashMap", "java.util.TreeMap")
+    HANDLED_CLASSES = ("java.util.HashMap", "java.util.TreeMap")  # type: Tuple[str, ...]
 
     def __init__(self):
         dict.__init__(self)
@@ -166,7 +167,7 @@ class JavaLinkedHashMap(JavaMap):
     Linked has map are handled with a specific block data
     """
 
-    HANDLED_CLASSES = "java.util.LinkedHashMap"
+    HANDLED_CLASSES = ("java.util.LinkedHashMap",)
 
     def load_from_blockdata(self, parser, reader, indent=0):
         # type: (JavaStreamParser, DataStreamReader, int) -> bool
@@ -204,7 +205,7 @@ class JavaSet(set, JavaInstance):
     Python-Java set bridge type
     """
 
-    HANDLED_CLASSES = ("java.util.HashSet", "java.util.LinkedHashSet")
+    HANDLED_CLASSES = ("java.util.HashSet", "java.util.LinkedHashSet")  # type: Tuple[str, ...]
 
     def __init__(self):
         set.__init__(self)
@@ -229,7 +230,7 @@ class JavaTreeSet(JavaSet):
     Tree sets are handled a bit differently
     """
 
-    HANDLED_CLASSES = "java.util.TreeSet"
+    HANDLED_CLASSES = ("java.util.TreeSet",)
 
     def load_from_instance(self, indent=0):
         # type: (int) -> bool