construct: Recursive normalization of construct parse result

If we want to use construct parse results to generate JSON serializable
dicts, we need to

* apply the filter_dict() operation recursively, and
* simplify the construct Container and ListContainer classes to
  a simple dict and/or list.

We introduce a pySim.construct.parse_construct() helper which is
subsequently used from all pySim.filesystem caller sites.

Change-Id: I319414eb69808ef65895293832bb30519f45949d

pySim/construct.py

@@ -1,3 +1,4 @@
+import typing
 from construct import *
 from pySim.utils import b2h, h2b, swap_nibbles
 import gsm0338
@@ -84,6 +85,34 @@ def filter_dict(d, exclude_prefix='_'):
             res[key] = value
     return res
 
+from construct.lib.containers import Container, ListContainer
+from construct.core import EnumIntegerString
+
+def normalize_construct(c):
+    """Convert a construct specific type to a related base type, mostly useful
+    so we can serialize it."""
+    # we need to include the filter_dict as we otherwise get elements like this
+    # in the dict: '_io': <_io.BytesIO object at 0x7fdb64e05860> which we cannot json-serialize
+    c = filter_dict(c)
+    if isinstance(c, Container) or isinstance(c, dict):
+        r = {k : normalize_construct(v) for (k, v) in c.items()}
+    elif isinstance(c, ListContainer):
+        r = [normalize_construct(x) for x in c]
+    elif isinstance(c, list):
+        r = [normalize_construct(x) for x in c]
+    elif isinstance(c, EnumIntegerString):
+        r = str(c)
+    else:
+        r = c
+    return r
+
+def parse_construct(c, raw_bin_data:bytes, length:typing.Optional[int]=None, exclude_prefix:str='_'):
+    """Helper function to wrap around normalize_construct() and filter_dict()."""
+    if not length:
+        length = len(raw_bin_data)
+    parsed = c.parse(raw_bin_data, total_len=length)
+    return normalize_construct(parsed)
+
 # here we collect some shared / common definitions of data types
 LV = Prefixed(Int8ub, HexAdapter(GreedyBytes))
 

pySim/filesystem.py

@@ -35,7 +35,7 @@ import argparse
 from typing import cast, Optional, Iterable, List, Any, Dict, Tuple
 
 from pySim.utils import sw_match, h2b, b2h, i2h, is_hex, auto_int, bertlv_parse_one, Hexstr
-from pySim.construct import filter_dict
+from pySim.construct import filter_dict, parse_construct
 from pySim.exceptions import *
 from pySim.jsonpath import js_path_find, js_path_modify
 
@@ -490,7 +490,7 @@ class TransparentEF(CardEF):
         if callable(method):
             return method(b2h(raw_bin_data))
         if self._construct:
-            return filter_dict(self._construct.parse(raw_bin_data, total_len=len(raw_bin_data)))
+            return parse_construct(self._construct, raw_bin_data)
         return {'raw': raw_bin_data.hex()}
 
     def decode_hex(self, raw_hex_data:str) -> dict:
@@ -513,7 +513,7 @@ class TransparentEF(CardEF):
         if callable(method):
             return method(raw_bin_data)
         if self._construct:
-            return filter_dict(self._construct.parse(raw_bin_data, total_len=len(raw_bin_data)))
+            return parse_construct(self._construct, raw_bin_data)
         return {'raw': raw_bin_data.hex()}
 
     def encode_bin(self, abstract_data:dict) -> bytearray:
@@ -712,7 +712,7 @@ class LinFixedEF(CardEF):
         if callable(method):
             return method(raw_bin_data)
         if self._construct:
-            return filter_dict(self._construct.parse(raw_bin_data, total_len=len(raw_bin_data)))
+            return parse_construct(self._construct, raw_bin_data)
         return {'raw': raw_bin_data.hex()}
 
     def decode_record_bin(self, raw_bin_data:bytearray) -> dict:
@@ -735,7 +735,7 @@ class LinFixedEF(CardEF):
         if callable(method):
             return method(raw_hex_data)
         if self._construct:
-            return filter_dict(self._construct.parse(raw_bin_data, total_len=len(raw_bin_data)))
+            return parse_construct(self._construct, raw_bin_data)
         return {'raw': raw_hex_data}
 
     def encode_record_hex(self, abstract_data:dict) -> str:
@@ -834,7 +834,7 @@ class TransRecEF(TransparentEF):
         if callable(method):
             return method(raw_bin_data)
         if self._construct:
-            return filter_dict(self._construct.parse(raw_bin_data, total_len=len(raw_bin_data)))
+            return parse_construct(self._construct, raw_bin_data)
         return {'raw': raw_hex_data}
 
     def decode_record_bin(self, raw_bin_data:bytearray) -> dict:
@@ -857,7 +857,7 @@ class TransRecEF(TransparentEF):
         if callable(method):
             return method(raw_hex_data)
         if self._construct:
-            return filter_dict(self._construct.parse(raw_bin_data, total_len=len(raw_bin_data)))
+            return parse_construct(self._construct, raw_bin_data)
         return {'raw': raw_hex_data}
 
     def encode_record_hex(self, abstract_data:dict) -> str: