zack3d/pylingual
1
0
Fork 0
mirror of https://github.com/syssec-utd/pylingual.git synced 2026-05-10 18:39:03 -07:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #80 from XinlongCS/cflow-refactor

Browse Source 
CDG Dict fix + Misc
This commit is contained in:
Joel-Flores123
2025-08-08 16:35:04 -05:00
committed by GitHub
parent de3fdd6f86 5343a269cf
commit f7ed7a3539
5 changed files with 16 additions and 15 deletions
Download Patch File Download Diff File Expand all files Collapse all files
dev_scripts/cflow.py
+2 -2
View File
@@ -18,7 +18,7 @@ from rich.progress import Progress, BarColumn, TextColumn, TimeRemainingColumn,
from pylingual.control_flow_reconstruction.cfg import CFG from pylingual.control_flow_reconstruction.cfg import CFG
from pylingual.control_flow_reconstruction.structure import bc_to_cft from pylingual.control_flow_reconstruction.structure import bc_to_cft
from pylingual.main import print_result from pylingual.main import print_result
from pylingual.control_flow_reconstruction.source import SourceContext from pylingual.control_flow_reconstruction.source import SourceContext, sanitize_lines
from pylingual.editable_bytecode import PYCFile from pylingual.editable_bytecode import PYCFile
from pylingual.equivalence_check import TestResult, compare_pyc from pylingual.equivalence_check import TestResult, compare_pyc
from pylingual.utils.version import PythonVersion from pylingual.utils.version import PythonVersion
@@ -64,7 +64,7 @@ def run(file: Path, out_dir: Path, version: PythonVersion, print=False):
file = next(file.iterdir()) file = next(file.iterdir())
in_src = normalize_source(file.read_text(), replace_docstrings=True) in_src = normalize_source(file.read_text(), replace_docstrings=True)
src_lines = in_src.split("\n") src_lines = sanitize_lines(in_src.split("\n"))
in_path = out_dir / "a.py" in_path = out_dir / "a.py"
in_path.write_text(in_src, encoding="utf-8") in_path.write_text(in_src, encoding="utf-8")
in_pyc = out_dir / "a.pyc" in_pyc = out_dir / "a.pyc"
pylingual/control_flow_reconstruction/cfg.py
+2 -1
View File
@@ -199,5 +199,6 @@ class CFG(DiGraph_CFT):
cdg = nx.create_empty_copy(self) cdg = nx.create_empty_copy(self)
cdg.add_edges_from((B, A, {"kind": EdgeKind.Fall}) for A, B in itertools.product(self.nodes, self.nodes) if A != B and control_dependent(A, B)) cdg.add_edges_from((B, A, {"kind": EdgeKind.Fall}) for A, B in itertools.product(self.nodes, self.nodes) if A != B and control_dependent(A, B))
cdg.remove_node(self.end) cdg.remove_node(self.end)
cdg.add_edges_from(((self.start, n) for n in cdg.nodes if cdg.in_degree(n) == 0 and n != self.start), kind=EdgeKind.Fall) start_nodes = [n for n in cdg.nodes if cdg.in_degree(n) == 0 and n != self.start]
cdg.add_edges_from(((self.start, n) for n in start_nodes), kind=EdgeKind.Fall)
return cdg return cdg
pylingual/control_flow_reconstruction/templates/Block.py
+3 -2
View File
@@ -54,11 +54,12 @@ class RemoveUnreachable(ControlFlowTemplate):
return node return node
@register_template(0, 0, (3, 13)) @register_template(0, 0, (3, 12), (3, 13))
class JumpTemplate(ControlFlowTemplate): class JumpTemplate(ControlFlowTemplate):
template = T( template = T(
body=~N("jump", None).with_cond(without_instructions("CLEANUP_THROW")), body=~N("jump", None).with_cond(without_instructions("CLEANUP_THROW")),
jump=N("tail", "block?") jump=N("tail", "block?")
.with_cond(has_no_lines)
.with_cond(no_self_edges) .with_cond(no_self_edges)
.with_in_deg(1) .with_in_deg(1)
.with_cond( .with_cond(
@@ -71,7 +72,7 @@ class JumpTemplate(ControlFlowTemplate):
exact_instructions("POP_JUMP_IF_FALSE"), exact_instructions("POP_JUMP_IF_FALSE"),
), ),
block=N.tail(), block=N.tail(),
tail=N.tail(), tail=N.tail().with_cond(without_instructions("NOP")),
) )
try_match = make_try_match( try_match = make_try_match(
pylingual/control_flow_reconstruction/templates/Conditional.py
+2 -2
View File
@@ -50,8 +50,8 @@ class IfElseLoop(ControlFlowTemplate):
@register_template(2, 41) @register_template(2, 41)
class IfThen(ControlFlowTemplate): class IfThen(ControlFlowTemplate):
template = T( template = T(
if_header=~N("if_body", "tail").with_cond(without_top_level_instructions("WITH_EXCEPT_START", "CHECK_EXC_MATCH", "FOR_ITER")), if_header=~N("if_body", "tail").with_cond(without_top_level_instructions("WITH_EXCEPT_START", "CHECK_EXC_MATCH", "FOR_ITER", "JUMP_IF_NOT_EXC_MATCH")),
if_body=~N("tail").with_in_deg(1) | ~N("tail.").with_in_deg(1).with_cond(run_is(2)) | ~N.tail().with_in_deg(1).with_cond(exact_instructions("LOAD_CONST","RETURN_VALUE")), if_body=~N(None).with_in_deg(1).of_type(BreakTemplate, ContinueTemplate) | ~N("tail").with_in_deg(1) | ~N("tail.").with_in_deg(1).with_cond(run_is(2)) | ~N.tail().with_in_deg(1).with_cond(exact_instructions("LOAD_CONST","RETURN_VALUE"), exact_instructions("POP_TOP", "LOAD_CONST","RETURN_VALUE")),
tail=N.tail(), tail=N.tail(),
) )
pylingual/control_flow_reconstruction/templates/Loop.py
+7 -8
View File
@@ -15,7 +15,6 @@ from ..utils import (
exact_instructions, exact_instructions,
ending_instructions, ending_instructions,
has_no_lines, has_no_lines,
has_some_lines,
condense_mapping, condense_mapping,
defer_source_to, defer_source_to,
starting_instructions, starting_instructions,
@@ -180,14 +179,14 @@ class InlinedComprehensionTemplate(ControlFlowTemplate):
class BreakTemplate(ControlFlowTemplate): class BreakTemplate(ControlFlowTemplate):
@classmethod @classmethod
def try_match(cls, cfg, node): def try_match(cls, cfg, node):
if not with_top_level_instructions("POP_TOP", "LOAD_CONST", "RETURN_VALUE", "RETURN_CONST", "JUMP_ABSOLUTE", "JUMP_FORWARD", "JUMP_BACKWARD", "BREAK_LOOP")(cfg, node) or has_no_lines(cfg, node): if not with_top_level_instructions("POP_TOP", "LOAD_FAST", "LOAD_CONST", "RETURN_VALUE", "RETURN_CONST", "JUMP_ABSOLUTE", "JUMP_FORWARD", "JUMP_BACKWARD", "BREAK_LOOP", "POP_BLOCK")(cfg, node) or has_no_lines(cfg, node):
return None return None
i = len(node.get_instructions()) - 1 i = len(node.get_instructions()) - 1
while i >= 0: while i >= 0:
instruction = node.get_instructions()[i].opname instruction = node.get_instructions()[i].opname
if instruction in {"POP_TOP", "LOAD_CONST", "RETURN_VALUE", "RETURN_CONST", "JUMP_ABSOLUTE", "JUMP_FORWARD", "JUMP_BACKWARD", "BREAK_LOOP"}: if instruction in {"POP_TOP", "LOAD_FAST", "LOAD_CONST", "RETURN_VALUE", "RETURN_CONST", "JUMP_ABSOLUTE", "JUMP_FORWARD", "JUMP_BACKWARD", "BREAK_LOOP", "POP_BLOCK"}:
if node.get_instructions()[i].starts_line is not None: if node.get_instructions()[i].starts_line is not None and not any(node.get_instructions()[i].source_line.strip().startswith(word) for word in {"pass", "...", "return"}):
return condense_mapping(cls, cfg, {"child": node}, "child") return condense_mapping(cls, cfg, {"child": node}, "child")
else: else:
i -= 1 i -= 1
@@ -203,14 +202,14 @@ class BreakTemplate(ControlFlowTemplate):
class ContinueTemplate(ControlFlowTemplate): class ContinueTemplate(ControlFlowTemplate):
@classmethod @classmethod
def try_match(cls, cfg, node): def try_match(cls, cfg, node):
if not with_top_level_instructions("JUMP_ABSOLUTE", "JUMP_BACKWARD", "CONTINUE_LOOP", "POP_EXCEPT")(cfg, node) or has_no_lines(cfg, node): if not with_top_level_instructions("JUMP_ABSOLUTE", "JUMP_BACKWARD", "CONTINUE_LOOP", "POP_EXCEPT", "POP_BLOCK")(cfg, node) or has_no_lines(cfg, node):
return None return None
i = len(node.get_instructions()) - 1 i = len(node.get_instructions()) - 1
while i >= 0: while i >= 0:
instruction = node.get_instructions()[i].opname instruction = node.get_instructions()[i].opname
if instruction in {"JUMP_ABSOLUTE", "JUMP_BACKWARD", "CONTINUE_LOOP", "POP_EXCEPT"}: if instruction in {"JUMP_ABSOLUTE", "JUMP_BACKWARD", "CONTINUE_LOOP", "POP_EXCEPT", "POP_BLOCK"}:
if node.get_instructions()[i].starts_line is not None: if node.get_instructions()[i].starts_line is not None and not any(node.get_instructions()[i].source_line.strip().startswith(word) for word in {"pass", "...", "return"}):
return condense_mapping(cls, cfg, {"child": node}, "child") return condense_mapping(cls, cfg, {"child": node}, "child")
else: else:
i -= 1 i -= 1
@@ -244,7 +243,7 @@ class FixLoop(ControlFlowTemplate):
if cfg.dominates(node, predecessor): if cfg.dominates(node, predecessor):
back_edges.append(predecessor) back_edges.append(predecessor)
if not back_edges or with_top_level_instructions("SEND")(cfg, node): if not back_edges or all(n == node for n in back_edges) or with_top_level_instructions("SEND")(cfg, node):
return None return None
# Get all nodes encompassed by the loop excluding source node and initial false jump # Get all nodes encompassed by the loop excluding source node and initial false jump