From e1b590e1d4fe3f8d31089a103b03efcd25ffc1e4 Mon Sep 17 00:00:00 2001
From: rocky <rocky@dusytfeet.com>
Date: Wed, 24 Dec 2025 21:29:38 -0500
Subject: [PATCH] DRY unmarshal integer reading routines

---
 xdis/unmarsh_graal.py | 43 +++++++++++----------------
 xdis/unmarsh_rust.py  | 13 --------
 xdis/unmarshal.py     | 69 ++++++++++++++++++++++++++-----------------
 3 files changed, 60 insertions(+), 65 deletions(-)

diff --git a/xdis/unmarsh_graal.py b/xdis/unmarsh_graal.py
index 2f860969..54eb1a34 100644
--- a/xdis/unmarsh_graal.py
+++ b/xdis/unmarsh_graal.py
@@ -235,7 +235,7 @@ def graal_readBigInteger(self):
         Reads a marshaled big integer from the input stream.
         """
         negative = False
-        sz = self.graal_readInt() # Get the size in shorts
+        sz = self.read_uint32() # Get the size in shorts
         if sz < 0:
             negative = True
             sz = -sz
@@ -277,7 +277,7 @@ def graal_readBooleanArray(self) -> tuple[bool, ...]:
         Python equivalent of Python Graal's readBooleanArray() from
         MarshalModuleBuiltins.java
         """
-        length: int = int(unpack("<i", self.fp.read(4))[0])
+        length: int = self.read_uint32()
         return tuple([bool(self.graal_readByte()) for _ in range(length)])
 
     def graal_readByte(self) -> int:
@@ -292,7 +292,7 @@ def graal_readBytes(self) -> bytes:
         Python equivalent of Python Graal's readBytes() from
         MarshalModuleBuiltins.java
         """
-        length: int = unpack("<i", self.fp.read(4))[0]
+        length: int = self.read_uint32()
         return bytes([self.graal_readByte() for _ in range(length)])
 
     def graal_readDouble(self) -> float:
@@ -307,23 +307,16 @@ def graal_readDoubleArray(self) -> tuple[float, ...]:
         Python equivalent of Python Graal's readDoubleArray() from
         MarshalModuleBuiltins.java
         """
-        length: int = int(unpack("<i", self.fp.read(4))[0])
+        length: int = self.read_uint32()
         return tuple([self.graal_readDouble() for _ in range(length)])
 
-    def graal_readInt(self) -> int:
-        """
-        Python equivalent of Python Graal's readInt() from
-        MarshalModuleBuiltins.java
-        """
-        return int(unpack("<i", self.fp.read(4))[0])
-
     def graal_readIntArray(self) -> tuple[int, ...]:
         """
         Python equivalent of Python Graal's readIntArray() from
         MarshalModuleBuiltins.java
         """
         length: int = int(unpack("<i", self.fp.read(4))[0])
-        return tuple([self.graal_readInt() for _ in range(length)])
+        return tuple([self.read_int32() for _ in range(length)])
 
     def graal_readLong(self) -> int:
         """
@@ -370,7 +363,7 @@ def graal_readStringArray(self) -> tuple[str, ...]:
         Python equvalent of Python Graal's readObjectArray() from
         MarshalModuleBuiltins.java
         """
-        length: int = self.graal_readInt()
+        length: int = self.read_uint32()
         return tuple([self.graal_readString() for _ in range(length)])
 
     def graal_readSparseTable(self) -> Dict[int, tuple]:
@@ -378,10 +371,10 @@ def graal_readSparseTable(self) -> Dict[int, tuple]:
         Python equvalent of Python Graal's readObjectArray() from
         MarshalModuleBuiltins.java
         """
-        self.graal_readInt()  # the length return value isn't used.
+        self.read_uint32()  # the length return value isn't used.
         table = {}  # new int[length][];
         while True:
-            i = self.graal_readInt()
+            i = self.read_int32()
             if i == -1:
                 return table
             table[i] = self.graal_readIntArray()
@@ -494,15 +487,15 @@ def t_graal_CodeUnit(self, save_ref, bytes_for_s: bool = False):
 
         co_name = self.graal_readString()
         co_qualname = self.graal_readString()
-        co_argcount = self.graal_readInt()
-        co_kwonlyargcount = self.graal_readInt()
-        co_posonlyargcount = self.graal_readInt()
+        co_argcount = self.read_uint32()
+        co_kwonlyargcount = self.read_uint32()
+        co_posonlyargcount = self.read_uint32()
 
-        co_stacksize = self.graal_readInt()
+        co_stacksize = self.read_uint32()
         co_code_offset_in_file = self.fp.tell()
         co_code = self.graal_readBytes()
         other_fields["srcOffsetTable"] = self.graal_readBytes()
-        co_flags = self.graal_readInt()
+        co_flags = self.read_uint32()
 
         # writeStringArray(code.names);
         # writeStringArray(code.varnames);
@@ -542,11 +535,11 @@ def t_graal_CodeUnit(self, save_ref, bytes_for_s: bool = False):
 
         other_fields["primitiveConstants"] = self.graal_readLongArray()
         other_fields["exception_handler_ranges"] = self.graal_readIntArray()
-        other_fields["condition_profileCount"] = self.graal_readInt()
-        other_fields["startLine"] = self.graal_readInt()
-        other_fields["startColumn"] = self.graal_readInt()
-        other_fields["endLine"] = self.graal_readInt()
-        other_fields["endColumn"] = self.graal_readInt()
+        other_fields["condition_profileCount"] = self.read_uint32()
+        other_fields["startLine"] = self.read_uint32()
+        other_fields["startColumn"] = self.read_uint32()
+        other_fields["endLine"] = self.read_uint32()
+        other_fields["endColumn"] = self.read_uint32()
         other_fields["outputCanQuicken"] = self.graal_readBytes()
         other_fields["variableShouldUnbox"] = self.graal_readBytes()
         other_fields["generalizeInputsMap"] = self.graal_readSparseTable()
diff --git a/xdis/unmarsh_rust.py b/xdis/unmarsh_rust.py
index 7add819f..52babac2 100644
--- a/xdis/unmarsh_rust.py
+++ b/xdis/unmarsh_rust.py
@@ -24,7 +24,6 @@
 object.
 """
 
-from struct import unpack
 from typing import Any, Dict, List, Tuple, Union
 
 from xdis.codetype.code313rust import Code313Rust, SourceLocation
@@ -277,18 +276,6 @@ def t_bigint(self, save_ref: bool=False, bytes_for_s: bool=False):
         value = int.from_bytes(byte_data, byteorder='little')
         return value if is_positive else -value
 
-    def read_int16(self):
-        return int(unpack("<h", self.fp.read(2))[0])
-
-    def read_int32(self):
-        return int(unpack("<i", self.fp.read(4))[0])
-
-    def read_slice(self, n: int) -> bytes:
-        return self.fp.read(n)
-
-    def read_uint32(self):
-        return int(unpack("<I", self.fp.read(4))[0])
-
     def read_string(self, n: int, bytes_for_s: bool=False) -> Union[bytes, str]:
         s = self.read_slice(n)
         if not bytes_for_s:
diff --git a/xdis/unmarshal.py b/xdis/unmarshal.py
index fc347218..2f02e579 100644
--- a/xdis/unmarshal.py
+++ b/xdis/unmarshal.py
@@ -195,6 +195,21 @@ def __init__(self, fp, magic_int, bytes_for_s, code_objects={}) -> None:
 
         self.UNMARSHAL_DISPATCH_TABLE = UNMARSHAL_DISPATCH_TABLE
 
+    def read_int16(self) -> int:
+        return int(unpack("<h", self.fp.read(2))[0])
+
+    def read_int32(self) -> int:
+        return int(unpack("<i", self.fp.read(4))[0])
+
+    def read_int64(self) -> int:
+        return int(unpack("<q", self.fp.read(8))[0])
+
+    def read_slice(self, n: int) -> bytes:
+        return self.fp.read(n)
+
+    def read_uint32(self) -> int:
+        return int(unpack("<I", self.fp.read(4))[0])
+
     def load(self):
         """
         ``marshal.load()`` written in Python. When the Python bytecode magic loaded is the
@@ -302,16 +317,16 @@ def t_True(self, save_ref, bytes_for_s: bool = False) -> bool:
         return True
 
     def t_int32(self, save_ref, bytes_for_s: bool = False):
-        return self.r_ref(int(unpack("<i", self.fp.read(4))[0]), save_ref)
+        return self.r_ref(self.read_int32(), save_ref)
 
     def t_long(self, save_ref, bytes_for_s: bool = False):
-        n = unpack("<i", self.fp.read(4))[0]
+        n = self.read_uint32()
         if n == 0:
             return long(0)
         size = abs(n)
         d = long(0)
         for j in range(0, size):
-            md = int(unpack("<h", self.fp.read(2))[0])
+            md = self.read_int16()
             # This operation and turn "d" from a long back
             # into an int.
             d += md << j * 15
@@ -323,7 +338,7 @@ def t_long(self, save_ref, bytes_for_s: bool = False):
 
     # Python 3.4 removed this.
     def t_int64(self, save_ref, bytes_for_s: bool = False):
-        obj = unpack("<q", self.fp.read(8))[0]
+        obj = self.read_int64()
         if save_ref:
             self.intern_objects.append(obj)
         return obj
@@ -342,7 +357,7 @@ def unpack_pre_24() -> float:
             return float(self.fp.read(unpack("B", self.fp.read(1))[0]))
 
         def unpack_newer() -> float:
-            return float(self.fp.read(unpack("<i", self.fp.read(4))[0]))
+            return float(self.fp.read(self.read_int32()))
 
         get_float = unpack_pre_24 if self.magic_int <= 62061 else unpack_newer
 
@@ -363,7 +378,7 @@ def t_string(self, save_ref, bytes_for_s: bool):
         In Python3, this is a ``bytes`` type.  In Python2, it is a string type;
         ``bytes_for_s`` is True when a Python 3 interpreter is reading Python 2 bytecode.
         """
-        strsize = unpack("<i", self.fp.read(4))[0]
+        strsize = self.read_uint32()
         s = self.fp.read(strsize)
         if not bytes_for_s:
             s = compat_str(s)
@@ -377,7 +392,7 @@ def t_ASCII_interned(self, save_ref, bytes_for_s: bool = False):
         the string.
         """
         # FIXME: check
-        strsize = unpack("<i", self.fp.read(4))[0]
+        strsize = self.read_uint32()
         interned = compat_str(self.fp.read(strsize))
         self.intern_strings.append(interned)
         return self.r_ref(interned, save_ref)
@@ -388,7 +403,7 @@ def t_ASCII(self, save_ref, bytes_for_s: bool = False):
         There are true strings in Python3 as opposed to
         bytes.
         """
-        strsize = unpack("<i", self.fp.read(4))[0]
+        strsize = self.read_uint32()
         s = self.fp.read(strsize)
         s = compat_str(s)
         return self.r_ref(s, save_ref)
@@ -407,13 +422,13 @@ def t_short_ASCII_interned(self, save_ref, bytes_for_s: bool = False):
         return self.r_ref(interned, save_ref)
 
     def t_interned(self, save_ref, bytes_for_s: bool = False):
-        strsize = unpack("<i", self.fp.read(4))[0]
+        strsize = self.read_uint32()
         interned = compat_str(self.fp.read(strsize))
         self.intern_strings.append(interned)
         return self.r_ref(interned, save_ref)
 
     def t_unicode(self, save_ref, bytes_for_s: bool = False):
-        strsize = unpack("<i", self.fp.read(4))[0]
+        strsize = self.read_uint32()
         unicodestring = self.fp.read(strsize)
         if self.version_triple < (3, 0):
             string = UnicodeForPython3(unicodestring)
@@ -434,7 +449,7 @@ def t_small_tuple(self, save_ref, bytes_for_s: bool = False):
         return self.r_ref_insert(ret, i)
 
     def t_tuple(self, save_ref, bytes_for_s: bool = False):
-        tuplesize = unpack("<i", self.fp.read(4))[0]
+        tuplesize = self.read_uint32()
         ret = self.r_ref(tuple(), save_ref)
         while tuplesize > 0:
             ret += (self.r_object(bytes_for_s=bytes_for_s),)
@@ -443,7 +458,7 @@ def t_tuple(self, save_ref, bytes_for_s: bool = False):
 
     def t_list(self, save_ref, bytes_for_s: bool = False):
         # FIXME: check me
-        n = unpack("<i", self.fp.read(4))[0]
+        n = self.read_uint32()
         ret = self.r_ref(list(), save_ref)
         while n > 0:
             ret += (self.r_object(bytes_for_s=bytes_for_s),)
@@ -451,7 +466,7 @@ def t_list(self, save_ref, bytes_for_s: bool = False):
         return ret
 
     def t_frozenset(self, save_ref, bytes_for_s: bool = False):
-        setsize = unpack("<i", self.fp.read(4))[0]
+        setsize = self.read_uint32()
         collection, i = self.r_ref_reserve([], save_ref)
         while setsize > 0:
             collection.append(self.r_object(bytes_for_s=bytes_for_s))
@@ -462,7 +477,7 @@ def t_frozenset(self, save_ref, bytes_for_s: bool = False):
         return self.r_ref_insert(final_frozenset, i)
 
     def t_set(self, save_ref, bytes_for_s: bool = False):
-        setsize = unpack("<i", self.fp.read(4))[0]
+        setsize = self.read_uint32()
         ret, i = self.r_ref_reserve(tuple(), save_ref)
         while setsize > 0:
             ret += (self.r_object(bytes_for_s=bytes_for_s),)
@@ -484,7 +499,7 @@ def t_dict(self, save_ref, bytes_for_s: bool = False):
         return ret
 
     def t_python2_string_reference(self, save_ref, bytes_for_s: bool = False):
-        refnum = unpack("<i", self.fp.read(4))[0]
+        refnum = self.read_uint32()
         return self.intern_strings[refnum]
 
     def t_slice(self, save_ref, bytes_for_s: bool = False):
@@ -522,9 +537,9 @@ def t_code(self, save_ref, bytes_for_s: bool = False):
         self.version_triple = magic_int2tuple(self.magic_int)
 
         if self.version_triple >= (2, 3):
-            co_argcount = unpack("<i", self.fp.read(4))[0]
+            co_argcount = self.read_uint32()
         elif self.version_triple >= (1, 3):
-            co_argcount = unpack("<h", self.fp.read(2))[0]
+            co_argcount = self.read_int16()
         else:
             co_argcount = 0
 
@@ -538,7 +553,7 @@ def t_code(self, save_ref, bytes_for_s: bool = False):
             co_posonlyargcount = None
 
         if self.version_triple >= (3, 0):
-            kwonlyargcount = unpack("<i", self.fp.read(4))[0]
+            kwonlyargcount = self.read_uint32()
         else:
             kwonlyargcount = 0
 
@@ -547,21 +562,21 @@ def t_code(self, save_ref, bytes_for_s: bool = False):
             self.version_triple[:2] == (3, 11) and self.is_pypy
         ):
             if self.version_triple >= (2, 3):
-                co_nlocals = unpack("<i", self.fp.read(4))[0]
+                co_nlocals = self.read_uint32()
             elif self.version_triple >= (1, 3):
-                co_nlocals = unpack("<h", self.fp.read(2))[0]
+                co_nlocals = self.read_int16()
 
         if self.version_triple >= (2, 3):
-            co_stacksize = unpack("<i", self.fp.read(4))[0]
+            co_stacksize = self.read_uint32()
         elif self.version_triple >= (1, 5):
-            co_stacksize = unpack("<h", self.fp.read(2))[0]
+            co_stacksize = self.read_int16()
         else:
             co_stacksize = 0
 
         if self.version_triple >= (2, 3):
-            co_flags = unpack("<i", self.fp.read(4))[0]
+            co_flags = self.read_uint32()
         elif self.version_triple >= (1, 3):
-            co_flags = unpack("<h", self.fp.read(2))[0]
+            co_flags = self.read_int16()
         else:
             co_flags = 0
 
@@ -628,9 +643,9 @@ def t_code(self, save_ref, bytes_for_s: bool = False):
 
         if self.version_triple >= (1, 5):
             if self.version_triple >= (2, 3):
-                co_firstlineno = unpack("<i", self.fp.read(4))[0]
+                co_firstlineno = self.read_int32()
             else:
-                co_firstlineno = unpack("<h", self.fp.read(2))[0]
+                co_firstlineno = self.read_int16()
 
             if self.version_triple >= (3, 11) and not self.is_pypy:
                 co_linetable = self.r_object(bytes_for_s=bytes_for_s)
@@ -775,7 +790,7 @@ def t_code_old(self, _, bytes_for_s: bool = False):
 
     # Since Python 3.4
     def t_object_reference(self, save_ref=None, bytes_for_s: bool = False):
-        refnum = unpack("<i", self.fp.read(4))[0]
+        refnum = self.read_uint32()
         return self.intern_objects[refnum]
 
     def t_unknown(self, save_ref=None, bytes_for_s: bool = False):