Added support for Numpy arrays loading

javaobj loads a JavaArray bean which content is a
numpy array instead of a list.
This adds an indirection compared to the previous API.

Fixes #33

Author: tcalmant

javaobj/v2/api.py

@@ -27,10 +27,8 @@
 from typing import Optional
 
 from .beans import JavaClassDesc, JavaInstance
-
-
-class JavaStreamParser:
-    pass
+from .stream import DataStreamReader
+from ..constants import TypeCode
 
 
 class ObjectTransformer:
@@ -49,4 +47,21 @@ def create_instance(self, classdesc):
         :param classdesc: The description of a Java class
         :return: The Python form of the object, or the original JavaObject
         """
-        raise NotImplementedError
+        return None
+
+    def load_array(self, reader, field_type, size):
+        # type: (DataStreamReader, TypeCode, int) -> Optional[list]
+        """
+        Loads and returns the content of a Java array, if possible.
+
+        The result of this method must be the content of the array, i.e. a list
+        or an array. It will be stored in a JavaArray bean created by the
+        parser.
+
+        This method must return None if it can't handle the array.
+
+        :param reader: The data stream reader
+        :param field_type: Type of the elements of the array
+        :param size: Number of elements in the array
+        """
+        return None

javaobj/v2/beans.py

@@ -26,11 +26,10 @@
 
 from __future__ import absolute_import
 
-from enum import Enum, IntEnum
+from enum import IntEnum
 from typing import Any, Dict, List, Optional, Set
 import logging
 
-from .stream import DataStreamReader
 from ..constants import ClassDescFlags, TypeCode
 from ..modifiedutf8 import decode_modified_utf8, byte_to_int
 from ..utils import UNICODE_TYPE
@@ -518,7 +517,9 @@ def dump(self, indent=0):
         prefix = "\t" * indent
         sub_prefix = "\t" * (indent + 1)
         dump = [
-            prefix + "[array 0x{0:x}: {1} items]".format(self.handle, len(self))
+            "{0}[array 0x{1:x}: {2} items - stored as {3}]".format(
+                prefix, self.handle, len(self), type(self.data).__name__
+            )
         ]
         for x in self:
             if isinstance(x, ParsedJavaContent):

javaobj/v2/core.py

@@ -636,7 +636,13 @@ def _do_array(self, type_code):
             raise ValueError("Invalid array size")
 
         # Array content
-        content = [self._read_field_value(field_type) for _ in range(size)]
+        for transformer in self.__transformers:
+            content = transformer.load_array(self.__reader, field_type, size)
+            if content is not None:
+                break
+        else:
+            content = [self._read_field_value(field_type) for _ in range(size)]
+
         return JavaArray(handle, cd, field_type, content)
 
     def _do_exception(self, type_code):

javaobj/v2/main.py

@@ -8,7 +8,7 @@
 
 from .api import ObjectTransformer
 from .core import JavaStreamParser
-from .transformers import DefaultObjectTransformer
+from .transformers import DefaultObjectTransformer, NumpyArrayTransformer
 
 # ------------------------------------------------------------------------------
 
@@ -31,6 +31,10 @@ def load(file_object, *transformers, **kwargs):
     else:
         all_transformers.append(DefaultObjectTransformer())
 
+    if kwargs.get("use_numpy_arrays", False):
+        # Use the numpy array transformer if requested
+        all_transformers.append(NumpyArrayTransformer())
+
     # Parse the object(s)
     parser = JavaStreamParser(file_object, all_transformers)
     contents = parser.run()

javaobj/v2/transformers.py

@@ -24,13 +24,23 @@
     limitations under the License.
 """
 
+# Standard library
 from typing import List, Optional
 import functools
 
-from .beans import BlockData, JavaClassDesc, JavaInstance
+# Numpy (optional)
+try:
+    import numpy
+except ImportError:
+    numpy = None
+
+
+# Javaobj
+from .api import ObjectTransformer
+from .beans import JavaClassDesc, JavaInstance
 from .core import JavaStreamParser
 from .stream import DataStreamReader
-from ..constants import TerminalCode
+from ..constants import TerminalCode, TypeCode
 from ..utils import to_bytes, log_error, log_debug, read_struct, read_string
 
 
@@ -405,7 +415,7 @@ def do_period(self, data):
         return data
 
 
-class DefaultObjectTransformer:
+class DefaultObjectTransformer(ObjectTransformer):
 
     KNOWN_TRANSFORMERS = (
         JavaBool,
@@ -454,3 +464,39 @@ def create_instance(self, classdesc):
 
             log_debug(">>> java_object: {0}".format(java_object))
             return java_object
+
+
+class NumpyArrayTransformer(ObjectTransformer):
+    """
+    Loads arrays as numpy arrays if possible
+    """
+
+    # Convertion of a Java type char to its NumPy equivalent
+    NUMPY_TYPE_MAP = {
+        TypeCode.TYPE_BYTE: "B",
+        TypeCode.TYPE_CHAR: "b",
+        TypeCode.TYPE_DOUBLE: ">d",
+        TypeCode.TYPE_FLOAT: ">f",
+        TypeCode.TYPE_INTEGER: ">i",
+        TypeCode.TYPE_LONG: ">l",
+        TypeCode.TYPE_SHORT: ">h",
+        TypeCode.TYPE_BOOLEAN: ">B",
+    }
+
+    def load_array(self, reader, field_type, size):
+        # type: (DataStreamReader, TypeCode, int) -> Optional[list]
+        """
+        Loads a Java array, if possible
+        """
+        if numpy is not None:
+            try:
+                dtype = self.NUMPY_TYPE_MAP[field_type]
+            except KeyError:
+                # Unhandled data type
+                return None
+            else:
+                return numpy.fromfile(
+                    reader.file_descriptor, dtype=dtype, count=size,
+                )
+
+        return None