Support object_pairs_hook

Merge remote-tracking branch 'jnothman/object_pairs_hook' into 0.2-maint
Conflicts:
	msgpack/_msgpack.pyx
	test/test_pack.py
	test/test_sequnpack.py

7 files changed, 143 insertions, 90 deletions

diff --git a/ChangeLog.rst b/ChangeLog.rst
index f49b577..6a4d27b 100644
--- a/ChangeLog.rst
+++ b/ChangeLog.rst
@@ -1,3 +1,12 @@
+0.3.0
+=====
+:release date: in development
+
+Changes
+-------
+* Add ``.skip()`` method to ``Unpacker`` (thanks to jnothman)
+
+
 0.2.3
 =======
 :release date: in development
diff --git a/msgpack/_msgpack.pyx b/msgpack/_msgpack.pyx
index 44a50ae..dca6237 100644
--- a/msgpack/_msgpack.pyx
+++ b/msgpack/_msgpack.pyx
@@ -201,6 +201,7 @@ cdef extern from "unpack.h":
     ctypedef struct msgpack_user:
         bint use_list
         PyObject* object_hook
+        bint has_pairs_hook # call object_hook with k-v pairs
         PyObject* list_hook
         char *encoding
         char *unicode_errors
@@ -213,13 +214,54 @@ cdef extern from "unpack.h":
         PyObject* key
 
     int template_execute(template_context* ctx, const_char_ptr data,
-                         size_t len, size_t* off) except -1
+                         size_t len, size_t* off, bint construct) except -1
     void template_init(template_context* ctx)
     object template_data(template_context* ctx)
 
+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
+    else:
+        if isinstance(encoding, unicode):
+            _bencoding = encoding.encode('ascii')
+        else:
+            _bencoding = encoding
+        ctx.user.encoding = PyBytes_AsString(_bencoding)
+        if isinstance(unicode_errors, unicode):
+            _berrors = unicode_errors.encode('ascii')
+        else:
+            _berrors = unicode_errors
+        ctx.user.unicode_errors = PyBytes_AsString(_berrors)
 
 def unpackb(object packed, object object_hook=None, object list_hook=None,
             use_list=None, encoding=None, unicode_errors="strict",
+            object_pairs_hook=None,
             ):
     """Unpack packed_bytes to object. Returns an unpacked object.
 
@@ -234,39 +276,11 @@ def unpackb(object packed, object object_hook=None, object list_hook=None,
 
     PyObject_AsReadBuffer(packed, <const_void_ptr*>&buf, &buf_len)
 
-    if encoding is None:
-        enc = NULL
-        err = NULL
-    else:
-        if isinstance(encoding, unicode):
-            bencoding = encoding.encode('ascii')
-        else:
-            bencoding = encoding
-        if isinstance(unicode_errors, unicode):
-            berrors = unicode_errors.encode('ascii')
-        else:
-            berrors = unicode_errors
-        enc = PyBytes_AsString(bencoding)
-        err = PyBytes_AsString(berrors)
-
-    template_init(&ctx)
     if use_list is None:
         warnings.warn("Set use_list explicitly.", category=DeprecationWarning, stacklevel=1)
-        ctx.user.use_list = 0
-    else:
-        ctx.user.use_list = use_list
-    ctx.user.object_hook = ctx.user.list_hook = NULL
-    ctx.user.encoding = <const_char_ptr>enc
-    ctx.user.unicode_errors = <const_char_ptr>err
-    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 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
-    ret = template_execute(&ctx, buf, buf_len, &off)
+        use_list = 0
+    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)
         if off < buf_len:
@@ -278,6 +292,7 @@ def unpackb(object packed, object object_hook=None, object list_hook=None,
 
 def unpack(object stream, object object_hook=None, object list_hook=None,
            use_list=None, encoding=None, unicode_errors="strict",
+            object_pairs_hook=None,
            ):
     """Unpack an object from `stream`.
 
@@ -287,7 +302,7 @@ def unpack(object stream, object object_hook=None, object list_hook=None,
         warnings.warn("Set use_list explicitly.", category=DeprecationWarning, stacklevel=1)
         use_list = 0
     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,
                    )
 
@@ -307,7 +322,10 @@ cdef class Unpacker(object):
     Otherwise, it is deserialized to Python tuple.
 
     `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.
@@ -357,9 +375,8 @@ cdef class Unpacker(object):
         self.buf = NULL
 
     def __init__(self, file_like=None, Py_ssize_t read_size=0, use_list=None,
-                 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,
-                 object object_pairs_hook=None,
                  ):
         if use_list is None:
             warnings.warn("Set use_list explicitly.", category=DeprecationWarning, stacklevel=1)
@@ -384,31 +401,7 @@ cdef class Unpacker(object):
         self.buf_size = read_size
         self.buf_head = 0
         self.buf_tail = 0
-        template_init(&self.ctx)
-        self.ctx.user.use_list = use_list
-        self.ctx.user.object_hook = self.ctx.user.list_hook = <PyObject*>NULL
-        if object_hook is not None:
-            if not PyCallable_Check(object_hook):
-                raise TypeError("object_hook must be a callable.")
-            self.ctx.user.object_hook = <PyObject*>object_hook
-        if list_hook is not None:
-            if not PyCallable_Check(list_hook):
-                raise TypeError("list_hook must be a callable.")
-            self.ctx.user.list_hook = <PyObject*>list_hook
-        if encoding is None:
-            self.ctx.user.encoding = NULL
-            self.ctx.user.unicode_errors = NULL
-        else:
-            if isinstance(encoding, unicode):
-                self._bencoding = encoding.encode('ascii')
-            else:
-                self._bencoding = encoding
-            self.ctx.user.encoding = PyBytes_AsString(self._bencoding)
-            if isinstance(unicode_errors, unicode):
-                self._berrors = unicode_errors.encode('ascii')
-            else:
-                self._berrors = unicode_errors
-            self.ctx.user.unicode_errors = PyBytes_AsString(self._berrors)
+        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
@@ -469,15 +462,18 @@ cdef class Unpacker(object):
         else:
             self.file_like = None
 
-    cpdef unpack(self):
-        """unpack one object"""
+    cdef object _unpack(self, bint construct):
         cdef int ret
+        cdef object obj
         while 1:
-            ret = template_execute(&self.ctx, self.buf, self.buf_tail, &self.buf_head)
+            ret = template_execute(&self.ctx, self.buf, self.buf_tail, &self.buf_head, construct)
             if ret == 1:
-                o = template_data(&self.ctx)
+                if construct:
+                    obj = template_data(&self.ctx)
+                else:
+                    obj = None
                 template_init(&self.ctx)
-                return o
+                return obj
             elif ret == 0:
                 if self.file_like is not None:
                     self.read_from_file()
@@ -486,11 +482,19 @@ cdef class Unpacker(object):
             else:
                 raise ValueError("Unpack failed: error = %d" % (ret,))
 
+    def unpack(self):
+        """unpack one object"""
+        return self._unpack(1)
+
+    def skip(self):
+        """read and ignore one object, returning None"""
+        return self._unpack(0)
+
     def __iter__(self):
         return self
 
     def __next__(self):
-        return self.unpack()
+        return self._unpack(1)
 
     # for debug.
     #def _buf(self):
diff --git a/msgpack/unpack.h b/msgpack/unpack.h
index a106f9c..7064a1b 100644
--- a/msgpack/unpack.h
+++ b/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;
     }
diff --git a/msgpack/unpack_template.h b/msgpack/unpack_template.h
index b844a24..6080a51 100644
--- a/msgpack/unpack_template.h
+++ b/msgpack/unpack_template.h
@@ -95,7 +95,7 @@ msgpack_unpack_func(msgpack_unpack_object, _data)(msgpack_unpack_struct(_context
 }
 
 
-msgpack_unpack_func(int, _execute)(msgpack_unpack_struct(_context)* ctx, const char* data, size_t len, size_t* off)
+msgpack_unpack_func(int, _execute)(msgpack_unpack_struct(_context)* ctx, const char* data, size_t len, size_t* off, int construct)
 {
 	assert(len >= *off);
 
@@ -117,14 +117,17 @@ msgpack_unpack_func(int, _execute)(msgpack_unpack_struct(_context)* ctx, const c
 
 	int ret;
 
+#define construct_cb(name) \
+    construct && msgpack_unpack_callback(name)
+
 #define push_simple_value(func) \
-	if(msgpack_unpack_callback(func)(user, &obj) < 0) { goto _failed; } \
+	if(construct_cb(func)(user, &obj) < 0) { goto _failed; } \
 	goto _push
 #define push_fixed_value(func, arg) \
-	if(msgpack_unpack_callback(func)(user, arg, &obj) < 0) { goto _failed; } \
+	if(construct_cb(func)(user, arg, &obj) < 0) { goto _failed; } \
 	goto _push
 #define push_variable_value(func, base, pos, len) \
-	if(msgpack_unpack_callback(func)(user, \
+	if(construct_cb(func)(user, \
 		(const char*)base, (const char*)pos, len, &obj) < 0) { goto _failed; } \
 	goto _push
 
@@ -140,9 +143,9 @@ msgpack_unpack_func(int, _execute)(msgpack_unpack_struct(_context)* ctx, const c
 
 #define start_container(func, count_, ct_) \
 	if(top >= MSGPACK_EMBED_STACK_SIZE) { goto _failed; } /* FIXME */ \
-	if(msgpack_unpack_callback(func)(user, count_, &stack[top].obj) < 0) { goto _failed; } \
+	if(construct_cb(func)(user, count_, &stack[top].obj) < 0) { goto _failed; } \
 	if((count_) == 0) { obj = stack[top].obj; \
-		msgpack_unpack_callback(func##_end)(user, &obj); \
+		construct_cb(func##_end)(user, &obj); \
 		goto _push; } \
 	stack[top].ct = ct_; \
 	stack[top].size  = count_; \
@@ -340,10 +343,10 @@ _push:
 	c = &stack[top-1];
 	switch(c->ct) {
 	case CT_ARRAY_ITEM:
-		if(msgpack_unpack_callback(_array_item)(user, c->count, &c->obj, obj) < 0) { goto _failed; }
+		if(construct_cb(_array_item)(user, c->count, &c->obj, obj) < 0) { goto _failed; }
 		if(++c->count == c->size) {
 			obj = c->obj;
-			msgpack_unpack_callback(_array_end)(user, &obj);
+			construct_cb(_array_end)(user, &obj);
 			--top;
 			/*printf("stack pop %d\n", top);*/
 			goto _push;
@@ -354,10 +357,10 @@ _push:
 		c->ct = CT_MAP_VALUE;
 		goto _header_again;
 	case CT_MAP_VALUE:
-		if(msgpack_unpack_callback(_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;
-			msgpack_unpack_callback(_map_end)(user, &obj);
+			construct_cb(_map_end)(user, &obj);
 			--top;
 			/*printf("stack pop %d\n", top);*/
 			goto _push;
@@ -399,6 +402,7 @@ _end:
 	*off = p - (const unsigned char*)data;
 
 	return ret;
+#undef construct_cb
 }
 
 
diff --git a/test/test_obj.py b/test/test_obj.py
index d809093..12a149f 100644
--- a/test/test_obj.py
+++ b/test/test_obj.py
@@ -26,6 +26,16 @@ def test_decode_hook():
     unpacked = unpackb(packed, object_hook=_decode_complex, use_list=1)
     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(b'', 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)
diff --git a/test/test_pack.py b/test/test_pack.py
index 9bd2b32..9009d35 100644
--- a/test/test_pack.py
+++ b/test/test_pack.py
@@ -110,10 +110,9 @@ def test_odict():
     seq = [(b'one', 1), (b'two', 2), (b'three', 3), (b'four', 4)]
     od = odict(seq)
     assert_equal(unpackb(packb(od), use_list=1), 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__':
diff --git a/test/test_sequnpack.py b/test/test_sequnpack.py
index 21fc3be..dac36a8 100644
--- a/test/test_sequnpack.py
+++ b/test/test_sequnpack.py
@@ -28,6 +28,21 @@ def test_foobar():
         k += 1
     assert k == len(b'foobar')
 
+def test_foobar_skip():
+    unpacker = Unpacker(read_size=3, use_list=1)
+    unpacker.feed(b'foobar')
+    assert unpacker.unpack() == ord(b'f')
+    unpacker.skip()
+    assert unpacker.unpack() == ord(b'o')
+    unpacker.skip()
+    assert unpacker.unpack() == ord(b'a')
+    unpacker.skip()
+    try:
+        o = unpacker.unpack()
+        assert 0, "should raise exception"
+    except StopIteration:
+        assert 1, "ok"
+
 def test_maxbuffersize():
     nose.tools.assert_raises(ValueError, Unpacker, read_size=5, max_buffer_size=3)
     unpacker = Unpacker(read_size=3, max_buffer_size=3, use_list=1)