Detect Python 2.7 weird "for-block" bytecode

It has an unnecessary JUMP_ABSOLUTE to a JUMP_LOOP

Fixes #408

uncompyle6/parsers/parse27.py

@@ -9,6 +9,7 @@ from uncompyle6.parsers.parse2 import Python2Parser
 from uncompyle6.parsers.reducecheck import (
     aug_assign1_check,
     ifelsestmt,
+    for_block_check,
     or_check,
     tryelsestmt,
     except_handler,
@@ -87,6 +88,11 @@ class Python27Parser(Python2Parser):
 
         for_block    ::= l_stmts_opt JUMP_BACK
 
+        # In 2.7 there is occasionally a for_block has an unusual
+        # form: there is a JUMP_ABSOLUTE which jumps to the second JUMP_BACK
+        # listed below. Both JUMP_BACKS go to the same position so the
+        # the JUMP_ABSOLUTE and JUMP_BACK not necessary
+        for_block    ::= l_stmts_opt JUMP_ABSOLUTE JUMP_BACK JUMP_BACK
         """
 
     def p_jump27(self, args):
@@ -234,6 +240,7 @@ class Python27Parser(Python2Parser):
         self.reduce_check_table = {
             "aug_assign1": aug_assign1_check,
             "except_handler": except_handler,
+            "for_block": for_block_check.for_block_invalid,
             "ifelsestmt": ifelsestmt,
             "or": or_check,
             "tryelsestmt": tryelsestmt,
@@ -247,6 +254,8 @@ class Python27Parser(Python2Parser):
         self.check_reduce["except_handler"] = "tokens"
         self.check_reduce["except_handler_else"] = "tokens"
 
+        self.check_reduce["for_block"] = "tokens"
+
         self.check_reduce["or"] = "AST"
         self.check_reduce["raise_stmt1"] = "AST"
         self.check_reduce["iflaststmtl"] = "AST"

uncompyle6/parsers/reducecheck/__init__.py

@@ -8,6 +8,7 @@ from uncompyle6.parsers.reducecheck.iflaststmt import *
 from uncompyle6.parsers.reducecheck.ifstmt import *
 from uncompyle6.parsers.reducecheck.ifstmt2 import *
 from uncompyle6.parsers.reducecheck.ifstmts_jump import *
+from uncompyle6.parsers.reducecheck.for_block_check import *
 from uncompyle6.parsers.reducecheck.or_check import *
 from uncompyle6.parsers.reducecheck.testtrue import *
 from uncompyle6.parsers.reducecheck.tryelsestmt import *

uncompyle6/parsers/reducecheck/for_block_check.py

@@ -0,0 +1,73 @@
+#  Copyright (c) 2022 Rocky Bernstein
+
+from uncompyle6.scanners.tok import Token
+
+
+def for_block_invalid(self, lhs, n, rule, tree, tokens, first: int, last: int) -> bool:
+
+    # print("XXX", first, last)
+    # for t in range(first, last):
+    #     print(tokens[t])
+    # print("=" * 30)
+
+    if rule == (
+        "for_block",
+        ("l_stmts_opt", "JUMP_ABSOLUTE", "JUMP_BACK", "JUMP_BACK"),
+    ):
+        # Check that the two JUMP_BACK's go to the same place.
+        jump_back1 = tokens[last - 2]
+        jump_back2 = tokens[last - 1]
+        if jump_back1.attr != jump_back2.attr:
+            return True
+
+        # Also check that JUMP_ABSOLUTE jumps to the JUMP_BACK.
+        # In this situation the JUMP_ABSOLUTE and a JUMP_BACK
+        # is not needed, but it seems to be there anyway.
+
+        jump_absolute = tokens[last - 3]
+        if jump_absolute.attr != jump_back2.offset:
+            return True
+
+        # Right now all of this is known to happen only in Python 2.7.
+        if self.version[:2] == (2, 7):
+            return False
+
+    if len(rule[1]) <= 1 or not tree:
+        return False
+
+    come_froms = tree[-1]
+    # This is complicated, but note that the JUMP_IF instruction comes immediately
+    # *before* _ifstmts_jump so that's what we have to test
+    # the COME_FROM against. This can be complicated by intervening
+    # POP_TOP, and pseudo COME_FROM, ELSE instructions
+    #
+    pop_jump_index = first - 1
+    while pop_jump_index > 0 and tokens[pop_jump_index] in (
+        "ELSE",
+        "POP_TOP",
+        "JUMP_FORWARD",
+        "COME_FROM",
+    ):
+        pop_jump_index -= 1
+
+    # FIXME: something is fishy when and EXTENDED ARG is needed before the
+    # pop_jump_index instruction to get the argment. In this case, the
+    # _ifsmtst_jump can jump to a spot beyond the come_froms.
+    # That is going on in the non-EXTENDED_ARG case is that the POP_JUMP_IF
+    # jumps to a JUMP_(FORWARD) which is changed into an EXTENDED_ARG POP_JUMP_IF
+    # to the jumped forwarded address
+    if tokens[pop_jump_index].attr > 256:
+        return False
+
+    pop_jump_offset = tokens[pop_jump_index].off2int(prefer_last=False)
+    if isinstance(come_froms, Token):
+        if tokens[pop_jump_index].attr < pop_jump_offset and tree[0] != "pass":
+            # This is a jump backwards to a loop. All bets are off here when there the
+            # unless statement is "pass" which has no instructions associated with it.
+            return False
+        return come_froms.attr is not None and pop_jump_offset > come_froms.attr
+
+    elif len(come_froms) == 0:
+        return False
+    else:
+        return pop_jump_offset > come_froms[-1].attr