add JSON helpers

weboob/browser/pages.py

@@ -412,6 +412,18 @@ class JsonPage(Page):
     def data(self):
         return self.response.text
 
+    def get(self, path):
+        node = self.doc
+        for name in filter(None, path.strip('.').split('.')):
+            node = node.get(name)
+            if node is None:
+                break
+        return node
+
+    def path(self, path):
+        from weboob.tools.json import mini_jsonpath
+        return mini_jsonpath(self.doc, path)
+
     def build_doc(self, text):
         from weboob.tools.json import json
         return json.loads(text)

weboob/tools/json.py

@@ -20,7 +20,7 @@
 # because we don't want to import this file by "import json"
 from __future__ import absolute_import
 
-__all__ = ['json']
+__all__ = ['json', 'mini_jsonpath']
 
 try:
     # try simplejson first because it is faster
@@ -28,3 +28,45 @@ try:
 except ImportError:
     # Python 2.6+ has a module similar to simplejson
     import json
+
+
+def mini_jsonpath(node, path):
+    """
+    Evaluates a dot separated path against JSON data. Path can contains
+    star wilcards. Always returns a generator.
+
+    Relates to http://goessner.net/articles/JsonPath/ but in a really basic
+    and simpler form.
+
+    >>> list(mini_jsonpath({"x": 95, "y": 77, "z": 68}, 'y'))
+    [77]
+    >>> list(mini_jsonpath({"x": {"y": {"z": "nested"}}}, 'x.y.z'))
+    ['nested']
+    >>> list(mini_jsonpath('{"data": [{"x": "foo", "y": 13}, {"x": "bar", "y": 42}, {"x": "baz", "y": 128}]}', 'data.*.y'))
+    [13, 42, 128]
+    """
+
+    def iterkeys(i):
+        return range(len(i)) if type(i) is list else i.iterkeys()
+
+    def cut(s):
+        p = s.split('.', 1) if s else [None]
+        return p + [None] if len(p) == 1 else p
+
+    if isinstance(node, basestring):
+        node = json.loads(node)
+
+    queue = [(node, cut(path))]
+    while queue:
+        node, (name, rest) = queue.pop(0)
+        if name is None:
+            yield node
+            continue
+        elif type(node) not in (dict, list):
+            continue
+        if name == '*':
+            keys = iterkeys(node)
+        else:
+            keys = [int(name) if type(node) is list else name]
+        for k in keys:
+            queue.append((node[k], cut(rest)))