Implement object_pairs_hook

Author: jnothman

msgpack/_msgpack.pyx

@@ -197,6 +197,7 @@ cdef extern from "unpack.h":
     ctypedef struct msgpack_user:
         int use_list
         PyObject* object_hook
+        bint has_pairs_hook # call object_hook with k-v pairs
         PyObject* list_hook
         char *encoding
         char *unicode_errors
@@ -213,18 +214,32 @@ cdef extern from "unpack.h":
     void template_init(template_context* ctx)
     object template_data(template_context* ctx)
 
-cdef inline init_ctx(template_context *ctx, object object_hook, object list_hook, bint use_list, encoding, unicode_errors):
+cdef inline init_ctx(template_context *ctx, object object_hook, object object_pairs_hook, object list_hook, bint use_list, encoding, unicode_errors):
     template_init(ctx)
     ctx.user.use_list = use_list
     ctx.user.object_hook = ctx.user.list_hook = <PyObject*>NULL
+
+    if object_hook is not None and object_pairs_hook is not None:
+        raise ValueError("object_pairs_hook and object_hook are mutually exclusive.")
+
     if object_hook is not None:
         if not PyCallable_Check(object_hook):
             raise TypeError("object_hook must be a callable.")
         ctx.user.object_hook = <PyObject*>object_hook
+
+    if object_pairs_hook is None:
+        ctx.user.has_pairs_hook = False
+    else:
+        if not PyCallable_Check(object_pairs_hook):
+            raise TypeError("object_pairs_hook must be a callable.")
+        ctx.user.object_hook = <PyObject*>object_pairs_hook
+        ctx.user.has_pairs_hook = True
+
     if list_hook is not None:
         if not PyCallable_Check(list_hook):
             raise TypeError("list_hook must be a callable.")
         ctx.user.list_hook = <PyObject*>list_hook
+
     if encoding is None:
         ctx.user.encoding = NULL
         ctx.user.unicode_errors = NULL
@@ -240,7 +255,7 @@ cdef inline init_ctx(template_context *ctx, object object_hook, object list_hook
             _berrors = unicode_errors
         ctx.user.unicode_errors = PyBytes_AsString(_berrors)
 
-def unpackb(object packed, object object_hook=None, object list_hook=None,
+def unpackb(object packed, object object_hook=None, object object_pairs_hook=None, object list_hook=None,
             bint use_list=0, encoding=None, unicode_errors="strict",
             ):
     """Unpack packed_bytes to object. Returns an unpacked object.
@@ -255,7 +270,7 @@ def unpackb(object packed, object object_hook=None, object list_hook=None,
     cdef Py_ssize_t buf_len
     PyObject_AsReadBuffer(packed, <const_void_ptr*>&buf, &buf_len)
 
-    init_ctx(&ctx, object_hook, list_hook, use_list, encoding, unicode_errors)
+    init_ctx(&ctx, object_hook, object_pairs_hook, list_hook, use_list, encoding, unicode_errors)
     ret = template_execute(&ctx, buf, buf_len, &off, 1)
     if ret == 1:
         obj = template_data(&ctx)
@@ -266,15 +281,15 @@ def unpackb(object packed, object object_hook=None, object list_hook=None,
         return None
 
 
-def unpack(object stream, object object_hook=None, object list_hook=None,
+def unpack(object stream, object object_hook=None, object object_pairs_hook=None, object list_hook=None,
            bint use_list=0, encoding=None, unicode_errors="strict",
            ):
     """Unpack an object from `stream`.
 
     Raises `ValueError` when `stream` has extra bytes.
     """
     return unpackb(stream.read(), use_list=use_list,
-                   object_hook=object_hook, list_hook=list_hook,
+                   object_hook=object_hook, object_pairs_hook=object_pairs_hook, list_hook=list_hook,
                    encoding=encoding, unicode_errors=unicode_errors,
                    )
 
@@ -294,7 +309,10 @@ cdef class Unpacker(object):
     Otherwise, it is deserialized to Python tuple. (default: False)
 
     `object_hook` is same to simplejson. If it is not None, it should be callable
-    and Unpacker calls it when deserializing key-value.
+    and Unpacker calls it with a dict argument after deserializing a map.
+
+    `object_pairs_hook` is same to simplejson. If it is not None, it should be callable
+    and Unpacker calls it with a list of key-value pairs after deserializing a map.
 
     `encoding` is encoding used for decoding msgpack bytes. If it is None (default),
     msgpack bytes is deserialized to Python bytes.
@@ -345,7 +363,7 @@ cdef class Unpacker(object):
         self.buf = NULL
 
     def __init__(self, file_like=None, Py_ssize_t read_size=0, bint use_list=0,
-                 object object_hook=None, object list_hook=None,
+                 object object_hook=None, object object_pairs_hook=None, object list_hook=None,
                  encoding=None, unicode_errors='strict', int max_buffer_size=0,
                  ):
         self.use_list = use_list
@@ -368,7 +386,7 @@ cdef class Unpacker(object):
         self.buf_size = read_size
         self.buf_head = 0
         self.buf_tail = 0
-        init_ctx(&self.ctx, object_hook, list_hook, use_list, encoding, unicode_errors)
+        init_ctx(&self.ctx, object_hook, object_pairs_hook, list_hook, use_list, encoding, unicode_errors)
 
     def feed(self, object next_bytes):
         cdef char* buf

msgpack/unpack.h

@@ -22,6 +22,7 @@
 typedef struct unpack_user {
     int use_list;
     PyObject *object_hook;
+    bool has_pairs_hook;
     PyObject *list_hook;
     const char *encoding;
     const char *unicode_errors;
@@ -160,9 +161,7 @@ static inline int template_callback_array_item(unpack_user* u, unsigned int curr
 static inline int template_callback_array_end(unpack_user* u, msgpack_unpack_object* c)
 {
     if (u->list_hook) {
-        PyObject *arglist = Py_BuildValue("(O)", *c);
-        PyObject *new_c = PyEval_CallObject(u->list_hook, arglist);
-        Py_DECREF(arglist);
+        PyObject *new_c = PyEval_CallFunction(u->list_hook, "(O)", *c);
         Py_DECREF(*c);
         *c = new_c;
     }
@@ -171,16 +170,31 @@ static inline int template_callback_array_end(unpack_user* u, msgpack_unpack_obj
 
 static inline int template_callback_map(unpack_user* u, unsigned int n, msgpack_unpack_object* o)
 {
-    PyObject *p = PyDict_New();
+    PyObject *p;
+    if (u->has_pairs_hook) {
+        p = PyList_New(n); // Or use tuple?
+    }
+    else {
+        p = PyDict_New();
+    }
     if (!p)
         return -1;
     *o = p;
     return 0;
 }
 
-static inline int template_callback_map_item(unpack_user* u, msgpack_unpack_object* c, msgpack_unpack_object k, msgpack_unpack_object v)
+static inline int template_callback_map_item(unpack_user* u, unsigned int current, msgpack_unpack_object* c, msgpack_unpack_object k, msgpack_unpack_object v)
 {
-    if (PyDict_SetItem(*c, k, v) == 0) {
+    if (u->has_pairs_hook) {
+        msgpack_unpack_object item = PyTuple_Pack(2, k, v);
+        if (!item)
+            return -1;
+        Py_DECREF(k);
+        Py_DECREF(v);
+        PyList_SET_ITEM(*c, current, item);
+        return 0;
+    }
+    else if (PyDict_SetItem(*c, k, v) == 0) {
         Py_DECREF(k);
         Py_DECREF(v);
         return 0;
@@ -191,9 +205,7 @@ static inline int template_callback_map_item(unpack_user* u, msgpack_unpack_obje
 static inline int template_callback_map_end(unpack_user* u, msgpack_unpack_object* c)
 {
     if (u->object_hook) {
-        PyObject *arglist = Py_BuildValue("(O)", *c);
-        PyObject *new_c = PyEval_CallObject(u->object_hook, arglist);
-        Py_DECREF(arglist);
+        PyObject *new_c = PyEval_CallFunction(u->object_hook, "(O)", *c);
         Py_DECREF(*c);
         *c = new_c;
     }

msgpack/unpack_template.h

@@ -357,7 +357,7 @@ msgpack_unpack_func(int, _execute)(msgpack_unpack_struct(_context)* ctx, const c
 		c->ct = CT_MAP_VALUE;
 		goto _header_again;
 	case CT_MAP_VALUE:
-		if(construct_cb(_map_item)(user, &c->obj, c->map_key, obj) < 0) { goto _failed; }
+		if(construct_cb(_map_item)(user, c->count, &c->obj, c->map_key, obj) < 0) { goto _failed; }
 		if(++c->count == c->size) {
 			obj = c->obj;
 			construct_cb(_map_end)(user, &obj);

test/test_obj.py

@@ -26,6 +26,16 @@ def test_decode_hook():
     unpacked = unpackb(packed, object_hook=_decode_complex)
     eq_(unpacked[1], 1+2j)
 
+def test_decode_pairs_hook():
+    packed = packb([3, {1: 2, 3: 4}])
+    prod_sum = 1 * 2 + 3 * 4
+    unpacked = unpackb(packed, object_pairs_hook=lambda l: sum(k * v for k, v in l))
+    eq_(unpacked[1], prod_sum)
+
+@raises(ValueError)
+def test_only_one_obj_hook():
+    unpackb('', object_hook=lambda x: x, object_pairs_hook=lambda x: x)
+
 @raises(ValueError)
 def test_bad_hook():
     packed = packb([3, 1+2j], default=lambda o: o)

test/test_pack.py

@@ -111,10 +111,9 @@ def test_odict():
     seq = [(b'one', 1), (b'two', 2), (b'three', 3), (b'four', 4)]
     od = odict(seq)
     assert_equal(unpackb(packb(od)), dict(seq))
-    # After object_pairs_hook is implemented.
-    #def pair_hook(seq):
-    #    return seq
-    #assert_equal(unpackb(packb(od), object_pairs_hook=pair_hook), seq)
+    def pair_hook(seq):
+        return seq
+    assert_equal(unpackb(packb(od), object_pairs_hook=pair_hook), seq)
 
 
 if __name__ == '__main__':