Optimize type checking range lookup with binary search

src/archunitpython/common/extraction/extract_graph.py

@@ -4,6 +4,7 @@
 
 import ast
 import os
+from bisect import bisect_right
 
 from archunitpython.common.extraction.graph import Edge, Graph, ImportKind
 from archunitpython.common.fluentapi.checkable import CheckOptions
@@ -59,12 +60,8 @@ def extract_graph(
 
     project_path = os.path.abspath(project_path)
     excludes = list(exclude_patterns) if exclude_patterns is not None else list(_DEFAULT_EXCLUDE)
-    ignore_type_checking_imports = bool(
-        options and options.ignore_type_checking_imports
-    )
-    cache_key = _build_cache_key(
-        project_path, excludes, ignore_type_checking_imports
-    )
+    ignore_type_checking_imports = bool(options and options.ignore_type_checking_imports)
+    cache_key = _build_cache_key(project_path, excludes, ignore_type_checking_imports)
 
     if options and options.clear_cache:
         _graph_cache.pop(cache_key, None)
@@ -105,6 +102,7 @@ def _extract_graph_uncached(
 
     edges: list[Edge] = []
     py_files_set = set(py_files)
+    normalized_py_file_set = {_normalize(f) for f in py_files_set}
 
     for file_path in py_files:
         # Add self-referencing edge (ensures the file appears as a node)
@@ -119,19 +117,14 @@ def _extract_graph_uncached(
         # Extract and resolve imports
         imports = _extract_imports(file_path)
         for module_name, import_kind in imports:
-            if (
-                ignore_type_checking_imports
-                and import_kind == ImportKind.TYPE_IMPORT
-            ):
+            if ignore_type_checking_imports and import_kind == ImportKind.TYPE_IMPORT:
                 continue
             resolved, is_external = _resolve_import(
                 module_name, file_path, project_path, import_kind
             )
             if resolved and resolved != _normalize(file_path):
                 # Check if the resolved path is in our project
-                if not is_external and resolved not in {
-                    _normalize(f) for f in py_files_set
-                }:
+                if not is_external and resolved not in normalized_py_file_set:
                     is_external = True
 
                 edges.append(
@@ -156,11 +149,7 @@ def _find_python_files(root: str, exclude: list[str]) -> list[str]:
     py_files: list[str] = []
     for dirpath, dirnames, filenames in os.walk(root):
         # Filter out excluded directories in-place
-        dirnames[:] = [
-            d
-            for d in dirnames
-            if not _should_exclude(d, exclude)
-        ]
+        dirnames[:] = [d for d in dirnames if not _should_exclude(d, exclude)]
 
         for filename in filenames:
             if filename.endswith(".py") and not _should_exclude(filename, exclude):
@@ -240,23 +229,27 @@ def _find_type_checking_ranges(tree: ast.Module) -> list[tuple[int, int]]:
             if is_type_checking and node.body:
                 start = node.body[0].lineno
                 end = max(
-                    getattr(n, "end_lineno", n.lineno)
-                    for n in node.body
-                    if hasattr(n, "lineno")
+                    getattr(n, "end_lineno", n.lineno) for n in node.body if hasattr(n, "lineno")
                 )
                 ranges.append((start, end))
 
-    return ranges
+    return sorted(ranges, key=lambda ele: ele[0])
 
 
-def _in_type_checking(
-    node: ast.AST, ranges: list[tuple[int, int]]
-) -> bool:
+def _in_type_checking(node: ast.AST, ranges: list[tuple[int, int]]) -> bool:
     """Check if a node is inside a TYPE_CHECKING block."""
     if not hasattr(node, "lineno"):
         return False
     lineno = node.lineno
-    return any(start <= lineno <= end for start, end in ranges)
+
+    if not isinstance(lineno,int):
+        return False
+
+    matched_index = bisect_right(ranges, lineno, key=lambda ele: ele[0]) - 1
+    if matched_index < 0:
+        return False
+    start, end = ranges[matched_index]
+    return start <= lineno <= end
 
 
 def _resolve_import(

src/archunitpython/common/pattern_matching.py

@@ -50,9 +50,7 @@ def matches_pattern(file_path: str, filter_: Filter) -> bool:
     return bool(filter_.regexp.search(target_string))
 
 
-def matches_pattern_classname(
-    class_name: str, file_path: str, filter_: Filter
-) -> bool:
+def matches_pattern_classname(class_name: str, file_path: str, filter_: Filter) -> bool:
     """Check if a class/file matches a filter, supporting classname target."""
     target = filter_.options.target
 

src/archunitpython/common/projection/cycles/johnsons_apsp.py

@@ -53,13 +53,9 @@ def _explore_neighbours(self, current_node: NumberNode) -> None:
         if self._is_part_of_current_start_cycle(current_node):
             self._unblock(current_node)
         else:
-            for neighbour in CycleUtils.get_outgoing_neighbours(
-                current_node, self._graph
-            ):
+            for neighbour in CycleUtils.get_outgoing_neighbours(current_node, self._graph):
                 if self._is_blocked(neighbour):
-                    self._blocked_map.append(
-                        _BlockedBy(blocked=current_node, by=neighbour)
-                    )
+                    self._blocked_map.append(_BlockedBy(blocked=current_node, by=neighbour))
 
     def _unblock(self, node: NumberNode) -> None:
         self._blocked = [n for n in self._blocked if n is not node]
@@ -74,9 +70,8 @@ def _is_part_of_current_start_cycle(self, current_node: NumberNode) -> bool:
         if self._start is None:
             return False
         for cycle in self._cycles:
-            if (
-                cycle[0].from_node == self._start.node
-                and any(e.from_node == current_node.node for e in cycle)
+            if cycle[0].from_node == self._start.node and any(
+                e.from_node == current_node.node for e in cycle
             ):
                 return True
         return False

src/archunitpython/common/projection/cycles/tarjan_scc.py

@@ -18,9 +18,7 @@ def __init__(self, node_id: int) -> None:
 class TarjanSCC:
     """Tarjan's algorithm for finding strongly connected components."""
 
-    def find_strongly_connected_components(
-        self, edges: list[NumberEdge]
-    ) -> list[list[NumberEdge]]:
+    def find_strongly_connected_components(self, edges: list[NumberEdge]) -> list[list[NumberEdge]]:
         """Find all strongly connected components in the graph.
 
         Returns a list of edge lists, where each inner list contains
@@ -78,9 +76,7 @@ def _visit(self, vertex: _Vertex) -> None:
             if scc_vertices:
                 scc_ids = {v.id for v in scc_vertices}
                 scc_edges = [
-                    e
-                    for e in self._edges
-                    if e.from_node in scc_ids and e.to_node in scc_ids
+                    e for e in self._edges if e.from_node in scc_ids and e.to_node in scc_ids
                 ]
                 if scc_edges:
                     self._sccs.append(scc_edges)

src/archunitpython/common/projection/project_cycles.py

@@ -72,8 +72,7 @@ def _from_domain(self, cycles: list[list[NumberEdge]]) -> ProjectedCycles:
                     (
                         se
                         for se in self._source_edges
-                        if se.source_label == source_label
-                        and se.target_label == target_label
+                        if se.source_label == source_label and se.target_label == target_label
                     ),
                     None,
                 )

src/archunitpython/common/util/logger.py

@@ -44,9 +44,7 @@ def _ensure_file_handler(self, options: LoggingOptions) -> None:
 
         mode = "a" if options.append_to_log_file else "w"
         self._file_handler = logging.FileHandler(str(log_path), mode=mode)
-        self._file_handler.setFormatter(
-            logging.Formatter("[%(levelname)s] %(message)s")
-        )
+        self._file_handler.setFormatter(logging.Formatter("[%(levelname)s] %(message)s"))
         self._logger.addHandler(self._file_handler)
 
     def _log(self, level: str, options: LoggingOptions | None, message: str) -> None:

src/archunitpython/files/assertion/custom_file_logic.py

@@ -83,9 +83,7 @@ def gather_custom_file_violations(
 
     for node in nodes:
         # Check if node matches all pre-filters
-        if pre_filters and not all(
-            matches_pattern(node.label, f) for f in pre_filters
-        ):
+        if pre_filters and not all(matches_pattern(node.label, f) for f in pre_filters):
             continue
 
         file_info = _build_file_info(node.label)
@@ -94,14 +92,10 @@ def gather_custom_file_violations(
         if is_negated:
             # shouldNot: violation if condition IS True
             if result:
-                violations.append(
-                    CustomFileViolation(message=message, file_info=file_info)
-                )
+                violations.append(CustomFileViolation(message=message, file_info=file_info))
         else:
             # should: violation if condition is NOT True
             if not result:
-                violations.append(
-                    CustomFileViolation(message=message, file_info=file_info)
-                )
+                violations.append(CustomFileViolation(message=message, file_info=file_info))
 
     return violations

src/archunitpython/files/assertion/depend_on_external_modules.py

@@ -34,8 +34,7 @@ def gather_depend_on_external_module_violations(
 
     for edge in edges:
         source_matches = all(
-            matches_pattern(edge.source_label, filter_)
-            for filter_ in subject_filters
+            matches_pattern(edge.source_label, filter_) for filter_ in subject_filters
         )
         if not source_matches:
             continue
@@ -49,16 +48,12 @@ def gather_depend_on_external_module_violations(
         if is_negated:
             if target_matches:
                 violations.append(
-                    ViolatingExternalModuleDependency(
-                        dependency=edge, is_negated=True
-                    )
+                    ViolatingExternalModuleDependency(dependency=edge, is_negated=True)
                 )
         else:
             if not target_matches:
                 violations.append(
-                    ViolatingExternalModuleDependency(
-                        dependency=edge, is_negated=False
-                    )
+                    ViolatingExternalModuleDependency(dependency=edge, is_negated=False)
                 )
 
     return violations

src/archunitpython/files/assertion/depend_on_files.py

@@ -41,27 +41,19 @@ def gather_depend_on_file_violations(
     violations: list[Violation] = []
 
     for edge in edges:
-        source_matches = all(
-            matches_pattern(edge.source_label, f) for f in subject_filters
-        )
+        source_matches = all(matches_pattern(edge.source_label, f) for f in subject_filters)
         if not source_matches:
             continue
 
-        target_matches = all(
-            matches_pattern(edge.target_label, f) for f in object_filters
-        )
+        target_matches = all(matches_pattern(edge.target_label, f) for f in object_filters)
 
         if is_negated:
             # shouldNot: violation if dependency EXISTS
             if target_matches:
-                violations.append(
-                    ViolatingFileDependency(dependency=edge, is_negated=True)
-                )
+                violations.append(ViolatingFileDependency(dependency=edge, is_negated=True))
         else:
             # should: violation if dependency does NOT match
             if not target_matches:
-                violations.append(
-                    ViolatingFileDependency(dependency=edge, is_negated=False)
-                )
+                violations.append(ViolatingFileDependency(dependency=edge, is_negated=False))
 
     return violations

src/archunitpython/files/fluentapi/files.py

@@ -76,15 +76,11 @@ def in_path(self, path: Pattern) -> "FilesShouldCondition":
 
     def should(self) -> "PositiveMatchPatternFileConditionBuilder":
         """Begin positive assertion (files SHOULD ...)."""
-        return PositiveMatchPatternFileConditionBuilder(
-            self._project_path, list(self._filters)
-        )
+        return PositiveMatchPatternFileConditionBuilder(self._project_path, list(self._filters))
 
     def should_not(self) -> "NegatedMatchPatternFileConditionBuilder":
         """Begin negative assertion (files SHOULD NOT ...)."""
-        return NegatedMatchPatternFileConditionBuilder(
-            self._project_path, list(self._filters)
-        )
+        return NegatedMatchPatternFileConditionBuilder(self._project_path, list(self._filters))
 
 
 class FilesShouldCondition:
@@ -111,15 +107,11 @@ def in_path(self, path: Pattern) -> "FilesShouldCondition":
 
     def should(self) -> "PositiveMatchPatternFileConditionBuilder":
         """Begin positive assertion (files SHOULD ...)."""
-        return PositiveMatchPatternFileConditionBuilder(
-            self._project_path, list(self._filters)
-        )
+        return PositiveMatchPatternFileConditionBuilder(self._project_path, list(self._filters))
 
     def should_not(self) -> "NegatedMatchPatternFileConditionBuilder":
         """Begin negative assertion (files SHOULD NOT ...)."""
-        return NegatedMatchPatternFileConditionBuilder(
-            self._project_path, list(self._filters)
-        )
+        return NegatedMatchPatternFileConditionBuilder(self._project_path, list(self._filters))
 
 
 class PositiveMatchPatternFileConditionBuilder:
@@ -135,9 +127,7 @@ def have_no_cycles(self) -> "CycleFreeFileCondition":
 
     def depend_on_files(self) -> "DependOnFileConditionBuilder":
         """Begin dependency assertion - files SHOULD depend on ..."""
-        return DependOnFileConditionBuilder(
-            self._project_path, self._filters, is_negated=False
-        )
+        return DependOnFileConditionBuilder(self._project_path, self._filters, is_negated=False)
 
     def depend_on_external_modules(
         self,
@@ -192,9 +182,7 @@ def __init__(self, project_path: str | None, filters: list[Filter]) -> None:
 
     def depend_on_files(self) -> "DependOnFileConditionBuilder":
         """Begin dependency assertion - files SHOULD NOT depend on ..."""
-        return DependOnFileConditionBuilder(
-            self._project_path, self._filters, is_negated=True
-        )
+        return DependOnFileConditionBuilder(self._project_path, self._filters, is_negated=True)
 
     def depend_on_external_modules(
         self,
@@ -243,9 +231,7 @@ def adhere_to(
 class DependOnFileConditionBuilder:
     """Configure dependency target patterns."""
 
-    def __init__(
-        self, project_path: str | None, filters: list[Filter], is_negated: bool
-    ) -> None:
+    def __init__(self, project_path: str | None, filters: list[Filter], is_negated: bool) -> None:
         self._project_path = project_path
         self._filters = filters
         self._is_negated = is_negated
@@ -285,9 +271,7 @@ def in_path(self, path: Pattern) -> "DependOnFileCondition":
 class DependOnExternalModuleConditionBuilder:
     """Configure external module dependency target patterns."""
 
-    def __init__(
-        self, project_path: str | None, filters: list[Filter], is_negated: bool
-    ) -> None:
+    def __init__(self, project_path: str | None, filters: list[Filter], is_negated: bool) -> None:
         self._project_path = project_path
         self._filters = filters
         self._is_negated = is_negated
@@ -447,9 +431,7 @@ def check(self, options: CheckOptions | None = None) -> list[Violation]:
         if empty is not None:
             return empty
 
-        return gather_regex_matching_violations(
-            nodes, self._check_filters, self._is_negated
-        )
+        return gather_regex_matching_violations(nodes, self._check_filters, self._is_negated)
 
 
 class CustomFileCheckableCondition:

src/archunitpython/metrics/assertion/metric_thresholds.py

@@ -31,9 +31,7 @@ class FileCountViolation(Violation):
     comparison: MetricComparison
 
 
-def check_threshold(
-    value: float, threshold: float, comparison: MetricComparison
-) -> bool:
+def check_threshold(value: float, threshold: float, comparison: MetricComparison) -> bool:
     """Check if a value violates a threshold.
 
     Returns True if the value is a VIOLATION.

src/archunitpython/metrics/calculation/distance.py

@@ -103,8 +103,6 @@ def calculate_distance_metrics_for_project(
         average_instability=sum(m.instability for m in metrics) / len(metrics),
         average_distance=sum(m.distance for m in metrics) / len(metrics),
         files_in_zone_of_pain=sum(1 for m in metrics if m.in_zone_of_pain),
-        files_in_zone_of_uselessness=sum(
-            1 for m in metrics if m.in_zone_of_uselessness
-        ),
+        files_in_zone_of_uselessness=sum(1 for m in metrics if m.in_zone_of_uselessness),
         total_files=len(files),
     )