Merge pull request #157 from 2elli/python-3.13

Python 3.10-3.13 line number and opcode argrepr fixes

Author: rocky

xdis/codetype/__init__.py

@@ -28,7 +28,6 @@
 from xdis.codetype.code38 import Code38
 from xdis.codetype.code310 import Code310
 from xdis.codetype.code311 import Code311, Code311FieldNames
-from xdis.codetype.code312 import Code312
 from xdis.version_info import PYTHON_VERSION_TRIPLE
 
 
@@ -101,7 +100,7 @@ def codeType2Portable(code, version_tuple=PYTHON_VERSION_TRIPLE):
                 co_firstlineno=code.co_firstlineno,
                 co_linetable=line_table,
             )
-        elif version_tuple[:2] == (3,11):
+        elif version_tuple[:2] >= (3,11):
             return Code311(
                 co_argcount=code.co_argcount,
                 co_posonlyargcount=code.co_posonlyargcount,
@@ -122,27 +121,6 @@ def codeType2Portable(code, version_tuple=PYTHON_VERSION_TRIPLE):
                 co_linetable=line_table,
                 co_exceptiontable=code.co_exceptiontable,
             )
-        elif version_tuple[:2] >= (3,12):
-            return Code312(
-                co_argcount=code.co_argcount,
-                co_posonlyargcount=code.co_posonlyargcount,
-                co_kwonlyargcount=code.co_kwonlyargcount,
-                co_nlocals=code.co_nlocals,
-                co_stacksize=code.co_stacksize,
-                co_flags=code.co_flags,
-                co_consts=code.co_consts,
-                co_code=code.co_code,
-                co_names=code.co_names,
-                co_varnames=code.co_varnames,
-                co_freevars=code.co_freevars,
-                co_cellvars=code.co_cellvars,
-                co_filename=code.co_filename,
-                co_name=code.co_name,
-                co_qualname=code.co_qualname,
-                co_firstlineno=code.co_firstlineno,
-                co_linetable=line_table,
-                co_exceptiontable=code.co_exceptiontable,
-            )
     elif version_tuple > (2, 0):
         # 2.0 .. 2.7
         return Code2(
@@ -209,11 +187,9 @@ def portableCodeType(version_tuple=PYTHON_VERSION_TRIPLE):
         elif version_tuple[:2] == (3, 10):
             # 3.10
             return Code310
-        elif version_tuple[:2] == (3,11):
+        elif version_tuple[:2] >= (3,11):
             # 3.11 ...
             return Code311
-        elif version_tuple[:2] >= (3,12):
-            return Code312
     elif version_tuple > (2, 0):
         # 2.0 .. 2.7
         return Code2

xdis/codetype/code310.py

@@ -17,6 +17,8 @@
 import types
 from copy import deepcopy
 
+import struct
+
 from xdis.codetype.code38 import Code38
 from xdis.cross_types import UnicodeForPython3
 from xdis.version_info import PYTHON_VERSION_TRIPLE, version_tuple_to_str
@@ -160,41 +162,25 @@ def co_lines(self):
                      equal to the size of the bytecode.  line will
                      either be a positive integer, or None
 
+        Parsing implementation adapted from: https://github.com/python/cpython/blob/3.10/Objects/lnotab_notes.txt
         """
-        line_number = self.co_firstlineno
-        start_offset = 0
-        byte_increments = [c for c in tuple(self.co_linetable[0::2])]
-        line_deltas = [c for c in tuple(self.co_linetable[1::2])]
-
-        if len(byte_increments) == 0:
-            yield start_offset, len(self.co_code), line_number
-            return
-
-        byte_incr = byte_increments[0]
-        line_delta = line_deltas[0]
-        assert line_delta != -128, "the first line delta can't logically be -128"
-        assert isinstance(line_delta, int)
-        if line_delta > 127:
-            line_delta = 256 - line_delta
-        else:
-            line_number += line_delta
-
-        for byte_incr, line_delta in zip(byte_increments[1:], line_deltas[1:]):
+        line = self.co_firstlineno
+        end_offset = 0
+        # co_linetable is pairs of (offset_delta: unsigned byte, line_delta: signed byte)
+        for offset_delta, line_delta in struct.iter_unpack('=Bb', self.co_linetable):
             assert isinstance(line_delta, int)
-            assert isinstance(byte_incr, int)
-            end_offset = start_offset + byte_incr
-            if line_delta > 127:
-                line_delta = 256 - line_delta
-
-            if line_delta == -128:
-                line_delta = 0
-            yield start_offset, end_offset, line_number
-            line_number += line_delta
+            assert isinstance(offset_delta, int)
+            if line_delta == 0: # No change to line number, just accumulate changes to end
+                end_offset += offset_delta
+                continue
             start_offset = end_offset
-
-        end_offset = len(self.co_code)
-        yield start_offset, end_offset, line_number
-        return
+            end_offset = start_offset + offset_delta
+            if line_delta == -128: # No valid line number -- skip entry
+                continue
+            line += line_delta
+            if end_offset == start_offset: # Empty range, omit.
+                continue
+            yield start_offset, end_offset, line
 
     def encode_lineno_tab(self):
         """

xdis/codetype/code311.py

@@ -23,6 +23,37 @@
 from xdis.version_info import PYTHON_VERSION_TRIPLE, version_tuple_to_str
 
 
+# Note: order is the positional order given in the Python docs for
+# 3.11 types.Codetype.
+# "posonlyargcount" is not used, but it is in other Python versions, so it
+# has to be included since this structure is used as the Union type
+# for all code types.
+Code311FieldNames = """
+        co_argcount
+        co_posonlyargcount
+        co_kwonlyargcount
+        co_nlocals
+        co_stacksize
+        co_flags
+        co_consts
+        co_code
+        co_names
+        co_varnames
+        co_freevars
+        co_cellvars
+        co_filename
+        co_name
+        co_qualname
+        co_firstlineno
+        co_linetable
+        co_exceptiontable
+"""
+
+Code311FieldTypes = deepcopy(Code310FieldTypes)
+Code311FieldTypes.update({"co_qualname": str, "co_exceptiontable": bytes})
+
+
+##### Parse location table #####
 def parse_location_entries(location_bytes, first_line):
     """
     Parses the locations table described in: https://github.com/python/cpython/blob/3.11/Objects/locations.md
@@ -131,36 +162,6 @@ def decode_signed_varint(s):
     return entries
 
 
-# Note: order is the positional order given in the Python docs for
-# 3.11 types.Codetype.
-# "posonlyargcount" is not used, but it is in other Python versions, so it
-# has to be included since this structure is used as the Union type
-# for all code types.
-Code311FieldNames = """
-        co_argcount
-        co_posonlyargcount
-        co_kwonlyargcount
-        co_nlocals
-        co_stacksize
-        co_flags
-        co_consts
-        co_code
-        co_names
-        co_varnames
-        co_freevars
-        co_cellvars
-        co_filename
-        co_name
-        co_qualname
-        co_firstlineno
-        co_linetable
-        co_exceptiontable
-"""
-
-Code311FieldTypes = deepcopy(Code310FieldTypes)
-Code311FieldTypes.update({"co_qualname": str, "co_exceptiontable": bytes})
-
-
 ##### NEW "OPAQUE" LINE TABLE PARSING #####
 # See: https://github.com/python/cpython/blob/aaed91cabcedc16c089c4b1c9abb1114659a83d3/Objects/codeobject.c#L1245C1-L1245C17
 PY_CODE_LOCATION_INFO_SHORT0 = 0

xdis/codetype/code312.py

@@ -49,15 +49,10 @@
 def_op(loc, "RERAISE",                 119,   3, 0)
 def_op(loc, "GEN_START",               129,   1, 0)
 def_op(loc, "MATCH_CLASS",             152,   2, 1)
-# fmt: on
-
 
+# fmt: on
 opcode_arg_fmt = opcode_arg_fmt310 = opcode_arg_fmt39.copy()
 opcode_extended_fmt = opcode_extended_fmt310 = opcode_extended_fmt39.copy()
 
-# fmt: on
-
-opcode_arg_fmt = opcode_arg_fmt10 = opcode_arg_fmt39.copy()
-
 update_pj3(globals(), loc)
 finalize_opcodes(loc)

xdis/opcodes/opcode_310.py

@@ -297,10 +297,6 @@ def format_BINARY_OP(arg) -> str:
 del opcode_extended_fmt311["INPLACE_TRUE_DIVIDE"]
 del opcode_extended_fmt311["INPLACE_XOR"]
 
-opcode_extended_fmt = opcode_extended_fmt311
-
-opcode_arg_fmt = opcode_arg_fmt11 = opcode_arg_fmt310.copy()
-
 from xdis.opcodes.opcode_310 import findlinestarts
 
 update_pj3(globals(), loc)

xdis/opcodes/opcode_311.py

@@ -177,6 +177,10 @@ def format_CALL_INTRINSIC_1(arg) -> str:
 def format_CALL_INTRINSIC_2(arg) -> str:
     return _intrinsic_2_descs[arg]
 
+
+opcode_extended_fmt = opcode_extended_fmt312 = opcode_extended_fmt311.copy()
+opcode_arg_fmt = opcode_arg_fmt12 = opcode_arg_fmt311.copy()
+
 ### update arg formatting
 opcode_arg_fmt312 = {
     **opcode_arg_fmt311,
@@ -185,9 +189,6 @@ def format_CALL_INTRINSIC_2(arg) -> str:
         "CALL_INTRINSIC_2": format_CALL_INTRINSIC_2,
     },
 }
-opcode_extended_fmt = opcode_extended_fmt312 = opcode_extended_fmt311.copy()
-
-opcode_arg_fmt = opcode_arg_fmt12 = opcode_arg_fmt311.copy()
 
 from xdis.opcodes.opcode_311 import findlinestarts
 

xdis/opcodes/opcode_312.py

@@ -450,8 +450,8 @@
 opcode_extended_fmt = opcode_312.opcode_extended_fmt312.copy()
 for fmt_table in (opcode_arg_fmt, opcode_extended_fmt):
     fmt_table.pop("MAKE_FUNCTION")  # MAKE_FUNCTION formatting not in 3.13
-opcode_arg_fmt13 = opcode_arg_fmt
-opcode_extended_fmt13 = opcode_extended_fmt
+opcode_arg_fmt313 = opcode_arg_fmt
+opcode_extended_fmt313 = opcode_extended_fmt
 
 
 # update any calls to findlinestarts to include the version tuple