Iteration over JavaSript objects in Python.

You can now do the 'for val in obj' syntax when obj is a wrapped JavaScript
object.

Thanks Mark Lee for the patch showing how to do this for arrays.

[#10 state:resolved]

Author: Paul Davis

THANKS

@@ -2,3 +2,4 @@
 Thanks For Patches and Pointers
 ===============================
 benoitc - build system tweaks, compiling on OpenBSD
+Mark Lee - Patch for iterating JS arrays in Python

spidermonkey/jsarray.c

@@ -33,6 +33,12 @@ Array_get_item(Object* self, Py_ssize_t idx)
     jsval rval;
     jsint pos = (jsint) idx;
 
+    if(idx >= Array_length(self))
+    {
+        PyErr_SetString(PyExc_IndexError, "List index out of range.");
+        return NULL;
+    }
+
     if(!JS_GetElement(self->cx->cx, self->obj, pos, &rval))
     {
         PyErr_SetString(PyExc_AttributeError, "Failed to get array item.");
@@ -60,6 +66,12 @@ Array_set_item(Object* self, Py_ssize_t idx, PyObject* val)
     return 0;
 }
 
+PyObject*
+Array_iterator(Object* self)
+{
+    return PySeqIter_New(self);
+}
+
 static PyMemberDef Array_members[] = {
     {0, 0, 0, 0}
 };
@@ -108,7 +120,7 @@ PyTypeObject _ArrayType = {
     0,		                                    /*tp_clear*/
     0,		                                    /*tp_richcompare*/
     0,		                                    /*tp_weaklistoffset*/
-    0,		                                    /*tp_iter*/
+    (getiterfunc)Array_iterator,		        /*tp_iter*/
     0,		                                    /*tp_iternext*/
     Array_methods,                              /*tp_methods*/
     Array_members,                              /*tp_members*/

spidermonkey/jsiterator.c

@@ -0,0 +1,166 @@
+/*
+ * Copyright 2009 Paul J. Davis <paul.joseph.davis@gmail.com>
+ *
+ * This file is part of the python-spidermonkey package released
+ * under the MIT license.
+ *
+ */
+
+#include "spidermonkey.h"
+
+PyObject*
+Iterator_Wrap(Context* cx, JSObject* obj)
+{
+    Iterator* self = NULL;
+    PyObject* tpl = NULL;
+    PyObject* ret = NULL;
+    JSObject* iter = NULL;
+    jsval priv;
+
+    // Build our new python object.
+    tpl = Py_BuildValue("(O)", cx);
+    if(tpl == NULL) goto error;
+    
+    self = (Iterator*) PyObject_CallObject((PyObject*) IteratorType, tpl);
+    if(self == NULL) goto error;
+    
+    // Attach a JS property iterator.
+    self->iter = JS_NewPropertyIterator(cx->cx, obj);
+    if(self->iter == NULL) goto error;
+
+    self->root = OBJECT_TO_JSVAL(self->iter);
+    if(!JS_AddRoot(cx->cx, &(self->root)))
+    {
+        PyErr_SetString(PyExc_RuntimeError, "Failed to set GC root.");
+        goto error;
+    }
+
+    ret = (PyObject*) self;
+    goto success;
+
+error:
+    Py_XDECREF(self);
+    ret = NULL; // In case it was AddRoot
+success:
+    Py_XDECREF(tpl);
+    return (PyObject*) ret;
+}
+
+PyObject*
+Iterator_new(PyTypeObject* type, PyObject* args, PyObject* kwargs)
+{
+    Context* cx = NULL;
+    Iterator* self = NULL;
+
+    if(!PyArg_ParseTuple(args, "O!", ContextType, &cx)) goto error;
+
+    self = (Iterator*) type->tp_alloc(type, 0);
+    if(self == NULL) goto error;
+    
+    Py_INCREF(cx);
+    self->cx = cx;
+    self->iter = NULL;
+    goto success;
+
+error:
+    ERROR("spidermonkey.Iterator.new");
+success:
+    return (PyObject*) self;
+}
+
+int
+Iterator_init(Iterator* self, PyObject* args, PyObject* kwargs)
+{
+    return 0;
+}
+
+void
+Iterator_dealloc(Iterator* self)
+{
+    if(self->iter != NULL)
+    {
+        JS_RemoveRoot(self->cx->cx, &(self->root));
+    }
+   
+    Py_XDECREF(self->cx);
+}
+
+PyObject*
+Iterator_next(Iterator* self)
+{
+    PyObject* ret = NULL;
+    jsid propid;
+    jsval propname;
+
+    if(!JS_NextProperty(self->cx->cx, self->iter, &propid))
+    {
+        PyErr_SetString(PyExc_RuntimeError, "Failed to iterate next JS prop.");
+        goto done;
+    }
+
+    if(!JS_IdToValue(self->cx->cx, propid, &propname))
+    {
+        PyErr_SetString(PyExc_RuntimeError, "Failed to convert property id.");
+        goto done;
+    }
+
+    if(propname != JSVAL_VOID)
+    {
+        ret = js2py(self->cx->cx, propname);
+    }
+
+    // We return NULL with no error to signal completion.
+
+done:
+    return ret;
+}
+
+static PyMemberDef Iterator_members[] = {
+    {NULL}
+};
+
+static PyMethodDef Iterator_methods[] = {
+    {NULL}
+};
+
+PyTypeObject _IteratorType = {
+    PyObject_HEAD_INIT(NULL)
+    0,                                          /*ob_size*/
+    "spidermonkey.Iterator",                    /*tp_name*/
+    sizeof(Iterator),                           /*tp_basicsize*/
+    0,                                          /*tp_itemsize*/
+    (destructor)Iterator_dealloc,               /*tp_dealloc*/
+    0,                                          /*tp_print*/
+    0,                                          /*tp_getattr*/
+    0,                                          /*tp_setattr*/
+    0,                                          /*tp_compare*/
+    0,                                          /*tp_repr*/
+    0,                                          /*tp_as_number*/
+    0,                                          /*tp_as_sequence*/
+    0,                                          /*tp_as_mapping*/
+    0,                                          /*tp_hash*/
+    0,                                          /*tp_call*/
+    0,                                          /*tp_str*/
+    0,                                          /*tp_getattro*/
+    0,                                          /*tp_setattro*/
+    0,                                          /*tp_as_buffer*/
+    Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE,   /*tp_flags*/
+    "JavaScript Iterator",                      /*tp_doc*/
+    0,		                                    /*tp_traverse*/
+    0,		                                    /*tp_clear*/
+    0,                                          /*tp_richcompare*/
+    0,		                                    /*tp_weaklistoffset*/
+    0,		                                    /*tp_iter*/
+    (iternextfunc)Iterator_next,		        /*tp_iternext*/
+    Iterator_methods,                           /*tp_methods*/
+    Iterator_members,                           /*tp_members*/
+    0,                                          /*tp_getset*/
+    0,                                          /*tp_base*/
+    0,                                          /*tp_dict*/
+    0,                                          /*tp_descr_get*/
+    0,                                          /*tp_descr_set*/
+    0,                                          /*tp_dictoffset*/
+    (initproc)Iterator_init,                    /*tp_init*/
+    0,                                          /*tp_alloc*/
+    Iterator_new,                               /*tp_new*/
+};

spidermonkey/jsiterator.h

@@ -0,0 +1,28 @@
+/*
+ * Copyright 2009 Paul J. Davis <paul.joseph.davis@gmail.com>
+ *
+ * This file is part of the python-spidermonkey package released
+ * under the MIT license.
+ *
+ */
+
+#ifndef PYSM_JSITERATOR_H
+#define PYSM_JSITERATOR_H
+
+/*
+    This is a representation of a JavaScript
+    object in Python land.
+*/
+
+typedef struct {
+    PyObject_HEAD
+    Context* cx;
+    JSObject* iter;
+    jsval root;
+} Iterator;
+
+extern PyTypeObject _IteratorType;
+
+PyObject* Iterator_Wrap(Context* cx, JSObject* obj);
+
+#endif

spidermonkey/jsobject.c

@@ -338,6 +338,12 @@ Object_rich_cmp(Object* self, PyObject* other, int op)
     return ret;
 }
 
+PyObject*
+Object_iterator(Object* self, PyObject* args, PyObject* kwargs)
+{
+    return Iterator_Wrap(self->cx, self->obj);
+}
+
 static PyMemberDef Object_members[] = {
     {NULL}
 };
@@ -379,7 +385,7 @@ PyTypeObject _ObjectType = {
     0,		                                    /*tp_clear*/
     (richcmpfunc)Object_rich_cmp,		        /*tp_richcompare*/
     0,		                                    /*tp_weaklistoffset*/
-    0,		                                    /*tp_iter*/
+    (getiterfunc)Object_iterator,		        /*tp_iter*/
     0,		                                    /*tp_iternext*/
     Object_methods,                             /*tp_methods*/
     Object_members,                             /*tp_members*/

spidermonkey/spidermonkey.c

@@ -18,6 +18,7 @@ PyTypeObject* ContextType = NULL;
 PyTypeObject* ObjectType = NULL;
 PyTypeObject* ArrayType = NULL;
 PyTypeObject* FunctionType = NULL;
+PyTypeObject* IteratorType = NULL;
 PyTypeObject* HashCObjType = NULL;
 PyObject* JSError = NULL;
 
@@ -40,6 +41,8 @@ initspidermonkey(void)
     _FunctionType.tp_base = &_ObjectType;
     if(PyType_Ready(&_FunctionType) < 0) return;
 
+    if(PyType_Ready(&_IteratorType) < 0) return;
+
     if(PyType_Ready(&_HashCObjType) < 0) return;
 
     m = Py_InitModule3("spidermonkey", spidermonkey_methods,
@@ -70,6 +73,10 @@ initspidermonkey(void)
     Py_INCREF(FunctionType);
     PyModule_AddObject(m, "Function", (PyObject*) FunctionType);
 
+    IteratorType = &_IteratorType;
+    Py_INCREF(IteratorType);
+    // No module access on purpose.
+
     HashCObjType = &_HashCObjType;
     Py_INCREF(HashCObjType);
     // Don't add access from the module on purpose.

spidermonkey/spidermonkey.h

@@ -27,6 +27,7 @@
 #include "jsobject.h"
 #include "jsarray.h"
 #include "jsfunction.h"
+#include "jsiterator.h"
 
 #include "convert.h"
 #include "error.h"
@@ -40,6 +41,7 @@ extern PyTypeObject* ClassType;
 extern PyTypeObject* ObjectType;
 extern PyTypeObject* ArrayType;
 extern PyTypeObject* FunctionType;
+extern PyTypeObject* IteratorType;
 extern PyTypeObject* HashCObjType;
 extern PyObject* JSError;
 

tests/t.py

@@ -53,6 +53,12 @@ def lt(a, b):
 def gt(a, b):
     assert a > b, "%r <= %r" % (a, b)
 
+def isin(a, b):
+    assert a in b, "%r is not in %r" % (a, b)
+
+def isnotin(a, b):
+    assert a not in b, "%r is in %r" % (a, b)
+
 def has(a, b):
     assert hasattr(a, b), "%r has no attribute %r" % (a, b)
 

tests/test-iterate.py

@@ -4,33 +4,29 @@
 # under the MIT license.
 import t
 
-@t.rt()
-def test_iter_py(rt):
-    pairs = [
-        ({"foo": "bar", "baz": "bam"}, ["foo", "baz"]),
-        (["a", "b", "c"], [0, 1, 2])
-    ]
-    def check(a, b):
-        cx = rt.new_context()
-        cx.add_global("zip", a)
-        js = """
-            var ret = [];
-            for(var v in zip) {ret.push(v);}
-            ret;
-        """
-        t.eq(cx.execute(js), b)
-    map(lambda x: check(*x), pairs)
+js_script = """
+var ret = [];
+for(var v in data) {ret.push(v);}
+ret;
+"""
 
-@t.rt()
-def test_iter_js(rt):
-    pairs = [
-        ('var f = {"foo": 1, "domino": "daily"}; f;', ["domino", "foo"]),
-        ('["foo", 1, "bing"]', [0, 1, 2])
-    ]
-    def check(a, b):
-        cx = rt.new_context()
-        ret = cx.execute(a)
-        data = [k for k in ret]
-        data.sort()
-        t.eq(data, b)
-    map(lambda x: check(*x), pairs)
+@t.glbl("data", {"foo": "bar", "baz": "bam"})
+def test_iter_py_map(cx, glbl):
+    t.eq(cx.execute(js_script), ["foo", "baz"])
+
+@t.glbl("data", ["a", "b", "c"])
+def test_iter_py_array(cx, glbl):
+    t.eq(cx.execute(js_script), [0, 1, 2])
+
+@t.cx()
+def test_iter_js_object(cx):
+    ret = cx.execute('var f = {"foo": 1, "domino": "daily"}; f;')
+    items = set(["domino", "foo"])
+    for k in ret:
+        t.isin(k, items)
+        items.remove(k)
+
+@t.cx()
+def test_iter_js_array(cx):
+    ret = cx.execute('["foo", 1, "bing", [3, 6]]')
+    t.eq([k for k in ret], ["foo", 1, "bing", [3, 6]])