From: David Kerkeslager <kerkeslager@gmail.com>
Date: Thu, 13 Apr 2017 19:38:58 +0000 (-0400)
Subject: Added string and list parsing
X-Git-Url: https://code.kerkeslager.com/?a=commitdiff_plain;ds=inline;h=355a12ff139812658f6775ff0bef4265f8687997;p=ton

Added string and list parsing
---

diff --git a/don/string.py b/don/string.py
index ee925f6..53551df 100644
--- a/don/string.py
+++ b/don/string.py
@@ -98,7 +98,7 @@ def _binary32_parser(s):
     match = _BINARY32_MATCHER.match(s)
 
     if match:
-        # TODO Validate that the double is in range
+        # TODO Validate that the float is in range
         return _shared.ParseResult(
             success = True,
             value = float(match.group(1)),
@@ -126,7 +126,6 @@ def _binary_parser(s):
     match = _BINARY_MATCHER.match(s)
 
     if match:
-        # TODO Validate that the double is in range
         return _shared.ParseResult(
             success = True,
             value = binascii.unhexlify(match.group(1)),
@@ -135,6 +134,67 @@ def _binary_parser(s):
 
     return _shared._FAILED_PARSE_RESULT
 
+def _make_utf_parser(encoding):
+    matcher = re.compile(r'"(.*?)"' + encoding)
+
+    def parser(s):
+        match = matcher.match(s)
+
+        if match:
+            return _shared.ParseResult(
+                success = True,
+                value = match.group(1),
+                remaining = s[match.end():],
+            )
+
+        return _shared._FAILED_PARSE_RESULT
+
+    return parser
+
+def _prefix_with_comma(parser):
+    def wrapped(s):
+        if s.startswith(','):
+            s = s[1:]
+
+            result = parser(s)
+            if not result.success:
+                raise Exception('Trailing comma before "{}"'.format(s))
+
+            return result
+
+        return _shared._FAILED_PARSE_RESULT
+
+    return wrapped
+
+def _list_parser(s):
+    # TODO Assert they are all the same type
+    if not s.startswith('['):
+        return _shared._FAILED_PARSE_RESULT
+    s = s[1:]
+
+    value = []
+
+    first = True
+    parse_result = _object_parser(s)
+
+    while parse_result.success:
+        value.append(parse_result.value)
+        s = parse_result.remaining
+        parse_result = _prefix_with_comma(_object_parser)(s)
+
+    if not s.startswith(']'):
+        return _shared._FAILED_PARSE_RESULT
+
+    return _shared.ParseResult(
+        success = True,
+        value = value,
+        remaining = s[1:],
+    )
+
+
+
+def _dictionary_parser(s):
+    return _shared._FAILED_PARSE_RESULT
 
 
 _PARSERS = [
@@ -148,6 +208,11 @@ _PARSERS = [
     _binary32_parser,
     _binary64_parser,
     _binary_parser,
+    _make_utf_parser('utf8'),
+    _make_utf_parser('utf16'),
+    _make_utf_parser('utf32'),
+    _list_parser,
+    _dictionary_parser,
 ]
 
 def _object_parser(source):
diff --git a/test_don.py b/test_don.py
index 1b68059..fe46bbc 100644
--- a/test_don.py
+++ b/test_don.py
@@ -214,16 +214,34 @@ class TestStringDeserialize(unittest.TestCase):
     def test_deserializes_double(self):
         self.assertEqual(1.0, string.deserialize('1.0d'))
 
-    def test_serializes_binary(self):
+    def test_deserializes_binary(self):
         self.assertEqual(
             b'\xde\xad\xbe\xef',
-            string.deserialize(b'\x30\x00\x00\x00\x04\xde\xad\xbe\xef'),
+            string.deserialize('"deadbeef"b'),
         )
 
-    def test_serializes_binary(self):
+    def test_deserializes_utf8(self):
         self.assertEqual(
-            b'\xde\xad\xbe\xef',
-            string.deserialize('"deadbeef"b'),
+            "Hello, world",
+            string.deserialize('"Hello, world"utf8'),
+        )
+
+    def test_deserializes_utf16(self):
+        self.assertEqual(
+            "Hello, world",
+            string.deserialize('"Hello, world"utf16'),
+        )
+
+    def test_deserializes_utf32(self):
+        self.assertEqual(
+            "Hello, world",
+            string.deserialize('"Hello, world"utf32'),
+        )
+
+    def test_deserializes_list(self):
+        self.assertEqual(
+            [1,2,3,4,5],
+            string.deserialize("[1i8,2i8,3i8,4i8,5i8]"),
         )
 
 unittest.main()