Codebase intelligence for AI and humans: code health scores, auto-generated docs, git analytics, dead code detection, and architectural decisions via MCP.
Python aicode-complexitycode-healthcode-intelligencecode-qualitydead-codedeveloper-toolsdocumentationgit-analyticsmcpopen-sourcepythonrefactoringstatic-analysistechnical-debt
fix(refactoring): stop offering an Extract Method span that is nearly the whole body (#2952)
Raghav Chamadiya2c89a75caea4
diff --git a/packages/core/src/repowise/core/analysis/health/dataflow/slice.py b/packages/core/src/repowise/core/analysis/health/dataflow/slice.pyindex c027262ce..752684e1c 100644--- a/packages/core/src/repowise/core/analysis/health/dataflow/slice.py+++ b/packages/core/src/repowise/core/analysis/health/dataflow/slice.py@@ -15,8 +15,9 @@ it cuts at statement boundaries within a single block (so never a partialbranch or a mid-``try`` split), contains no control-flow jump that leaves theregion (``return`` / ``raise`` / ``break`` / ``continue`` -> single clean exit),removes real complexity (at least one decision point), is substantial enough to-matter, and has at most one return and a small parameter list. Everything else-is suppressed -- ten great extractions, not two hundred maybes.+matter without being nearly the whole body, and has at most one return and a+small parameter list. Everything else is suppressed -- ten great extractions,+not two hundred maybes.Line-based liveness over D2's def/use occurrences realises the IN/OUT inference;the CFG/jump scan realises the single-exit predicate. The jump and nested-scope@@ -50,6 +51,13 @@ _MAX_PARAMS = 5 # too many ins => the span is not cohesive_MAX_RETURNS = 1 # a single clean return (v1); multi-output is future work# Backstop against a pathological function producing too many sub-ranges._MAX_CANDIDATES = 4000+# A span holding this share of the body's code lines or more is the whole+# function under another name, not a split: what stays behind is a guard, a+# wrapper (``try`` / ``with``) or the final ``return``. Measured on 301 plans+# from seven repos (Python, TS, Go, Java, Rust): 52 sat at or above 0.75, and+# nearly all of those at 0.8 or more left only such a shell. The span offered+# instead may be the same block minus a statement, just under the cut.+_MAX_BODY_SHARE = 0.75@dataclass(frozen=True)@@ -97,6 +105,8 @@ def find_extractions(analysis: FunctionAnalysis, lmap: LanguageNodeMap) -> list[# The statement container of the body (Go nests it in a ``statement_list``# inside the ``block``); spans covering it whole are not extractions.body_container = _unwrap_container(body, lmap.block_kinds)+ lines = _function_lines(fn_node)+ max_slice_nloc = _MAX_BODY_SHARE * _stmts_nloc(body_container.named_children, lines)def_lines, use_lines = _var_lines(analysis.def_use)declared_first = _declared_before_read(analysis.def_use)@@ -120,7 +130,6 @@ def find_extractions(analysis: FunctionAnalysis, lmap: LanguageNodeMap) -> list[else None)- lines = _function_lines(fn_node)out: list[Extraction] = []evaluated = 0for block, loop in _all_blocks(fn_node, lmap.block_kinds, scope_kinds, lmap.loop_kinds):@@ -147,9 +156,7 @@ def find_extractions(analysis: FunctionAnalysis, lmap: LanguageNodeMap) -> list[nested_prefix.append(nested_prefix[-1] + (1 if _holds_a_named_nested_function([st], lmap) else 0))- code_prefix.append(- code_prefix[-1] + len(_code_line_numbers(st, lines, drop_docstrings=True))- )+ code_prefix.append(code_prefix[-1] + _stmts_nloc([st], lines))for i in range(n):for j in range(i, n):evaluated += 1@@ -172,7 +179,7 @@ def find_extractions(analysis: FunctionAnalysis, lmap: LanguageNodeMap) -> list[if has_jump or decisions < _MIN_CCN_REMOVED:continueslice_nloc = code_prefix[j + 1] - code_prefix[i]- if slice_nloc < _MIN_SLICE_NLOC:+ if not _MIN_SLICE_NLOC <= slice_nloc < max_slice_nloc:continuespan = stmts[i : j + 1]s = span[0].start_point[0] + 1@@ -213,6 +220,11 @@ def _function_lines(fn_node: Node) -> list[str]:return [""] * fn_node.start_point[0] + text.splitlines()+def _stmts_nloc(stmts: list[Node], lines: list[str]) -> int:+ """Code lines of *stmts* by the walker's NLOC rule, summed per statement."""+ return sum(len(_code_line_numbers(st, lines, drop_docstrings=True)) for st in stmts)++def _sorted(candidates: list[Extraction]) -> list[Extraction]:return sorted(candidates,diff --git a/tests/unit/health/refactoring_corpus_fixture.py b/tests/unit/health/refactoring_corpus_fixture.pyindex a366334ef..81953e6ae 100644--- a/tests/unit/health/refactoring_corpus_fixture.py+++ b/tests/unit/health/refactoring_corpus_fixture.py@@ -52,26 +52,34 @@ def extractions_for(language: str, filename: str) -> dict[str, list[dict[str, An"""Every span the slicer offers, per function, for one corpus file.Keyed by function name so a golden diff names the archetype that moved.++ The archetypes are small on purpose, so most of their spans hold nearly the+ whole body. The body-share gate is a worth rule, not a soundness one, and+ left on it would hide the soundness gates this corpus exists to pin; it is+ lifted here and tested on its own."""+ from unittest import mock+from repowise.core.analysis.health.complexity.languages import get_language_map+ from repowise.core.analysis.health.dataflow import slice as slicerfrom repowise.core.analysis.health.dataflow.analyze import analyze_file- from repowise.core.analysis.health.dataflow.slice import find_extractionslmap = get_language_map(language)result = analyze_file(filename, language, source_for(filename), flagged_only=False)out: dict[str, list[dict[str, Any]]] = {}- for analysis in result.functions:- out[analysis.name] = [- {- "start_line": item.start_line,- "end_line": item.end_line,- "params": list(item.params),- "returns": list(item.returns),- "slice_nloc": item.slice_nloc,- "ccn_removed": item.ccn_removed,- }- for item in find_extractions(analysis, lmap)- ]+ with mock.patch.object(slicer, "_MAX_BODY_SHARE", float("inf")):+ for analysis in result.functions:+ out[analysis.name] = [+ {+ "start_line": item.start_line,+ "end_line": item.end_line,+ "params": list(item.params),+ "returns": list(item.returns),+ "slice_nloc": item.slice_nloc,+ "ccn_removed": item.ccn_removed,+ }+ for item in slicer.find_extractions(analysis, lmap)+ ]return outdiff --git a/tests/unit/health/test_refactoring_extract_method.py b/tests/unit/health/test_refactoring_extract_method.pyindex 4908e65d6..33de182eb 100644--- a/tests/unit/health/test_refactoring_extract_method.py+++ b/tests/unit/health/test_refactoring_extract_method.py@@ -150,6 +150,7 @@ def test_extractions_are_deterministic():def _find_extractions_reference(analysis, lmap):from repowise.core.analysis.health.dataflow.slice import (+ _MAX_BODY_SHARE,_MAX_CANDIDATES,_MAX_PARAMS,_MAX_RETURNS,@@ -159,11 +160,13 @@ def _find_extractions_reference(analysis, lmap):Extraction,_all_blocks,_declared_before_read,+ _function_lines,_infer_in_out,_loop_carry_free,_outs_definitely_assigned,_sorted,_span_metrics,+ _stmts_nloc,_unwrap_container,_var_lines,)@@ -175,6 +178,8 @@ def _find_extractions_reference(analysis, lmap):if body is None:return []body_container = _unwrap_container(body, lmap.block_kinds)+ lines = _function_lines(fn_node)+ body_nloc = _stmts_nloc(body_container.named_children, lines)def_lines, use_lines = _var_lines(analysis.def_use)declared_first = _declared_before_read(analysis.def_use)decision_kinds = (@@ -218,8 +223,8 @@ def _find_extractions_reference(analysis, lmap):decisions, has_jump = _span_metrics(span, decision_kinds, jump_kinds, scope_kinds)if has_jump or decisions < _MIN_CCN_REMOVED:continue- slice_nloc = sum(st.end_point[0] - st.start_point[0] + 1 for st in span)- if slice_nloc < _MIN_SLICE_NLOC:+ slice_nloc = _stmts_nloc(span, lines)+ if slice_nloc < _MIN_SLICE_NLOC or slice_nloc >= _MAX_BODY_SHARE * body_nloc:continues = span[0].start_point[0] + 1e = span[-1].end_point[0] + 1@@ -524,11 +529,17 @@ def test_slice_nloc_counts_code_lines_not_comments():total += limitelse:total += x+ total = min(total, limit * 10)+ total = max(total, 0)+ print(total)return total"""fn = _first(src)lmap = get_language_map("python")- loop = [c for c in find_extractions(fn, lmap) if c.end_line - c.start_line >= 7]+ # Spans ending with the loop (line 11), the ones holding the comments.+ loop = [+ c for c in find_extractions(fn, lmap) if c.end_line == 11 and c.end_line - c.start_line >= 5+ ]assert loopassert all(c.slice_nloc <= c.end_line - c.start_line + 1 - 2 for c in loop)@@ -695,3 +706,30 @@ def test_plan_params_leave_out_a_loop_counter_the_span_declares():plan = suggestions[0].planassert plan["params"] == ["a", "n"]assert plan["returns"] == ["total"]+++def test_a_span_holding_nearly_the_whole_body_is_not_a_split():+ # Everything but the closing return: lifting it would leave a function that+ # only calls the helper, the smell moved under a new name.+ lmap = get_language_map("python")+ src = """+ def tally(items, limit):+ total = 0+ seen = 0+ for x in items:+ if x > limit:+ total += limit+ else:+ total += x+ seen += 1+ return total+ """+ assert find_extractions(_first(src), lmap) == []+++def test_a_span_leaving_real_work_behind_is_still_offered():+ lmap = get_language_map("python")+ fn = _first(_PROCESS)+ (best, *_) = find_extractions(fn, lmap)+ assert "average" in best.returns+ assert best.end_line < fn.end_line