feat: ✨ Include pathspec library (#49)

Co-authored-by: robvanderleek <[email protected]>

codelimit/__main__.py

@@ -26,13 +26,13 @@ def list_commands(self, ctx: Context):
 
 @cli.command(help="Check file(s)")
 def check(
-        paths: Annotated[List[Path], typer.Argument(exists=True)],
-        exclude: Annotated[
-            Optional[list[str]], typer.Option(help="Glob patterns for exclusion")
-        ] = None,
-        quiet: Annotated[
-            bool, typer.Option("--quiet", help="No output when successful")
-        ] = False,
+    paths: Annotated[List[Path], typer.Argument(exists=True)],
+    exclude: Annotated[
+        Optional[list[str]], typer.Option(help="Glob patterns for exclusion")
+    ] = None,
+    quiet: Annotated[
+        bool, typer.Option("--quiet", help="No output when successful")
+    ] = False,
 ):
     if exclude:
         Configuration.excludes.extend(exclude)
@@ -41,12 +41,12 @@ def check(
 
 @cli.command(help="Scan a codebase")
 def scan(
-        path: Annotated[
-            Path, typer.Argument(exists=True, file_okay=False, help="Codebase root")
-        ] = Path("."),
-        exclude: Annotated[
-            Optional[list[str]], typer.Option(help="Glob patterns for exclusion")
-        ] = None
+    path: Annotated[
+        Path, typer.Argument(exists=True, file_okay=False, help="Codebase root")
+    ] = Path("."),
+    exclude: Annotated[
+        Optional[list[str]], typer.Option(help="Glob patterns for exclusion")
+    ] = None,
 ):
     if exclude:
         Configuration.excludes.extend(exclude)
@@ -55,14 +55,14 @@ def scan(
 
 @cli.command(help="Show report for codebase")
 def report(
-        path: Annotated[
-            Path, typer.Argument(exists=True, file_okay=False, help="Codebase root")
-        ] = Path("."),
-        full: Annotated[bool, typer.Option("--full", help="Show full report")] = False,
-        totals: Annotated[bool, typer.Option("--totals", help="Only show totals")] = False,
-        fmt: Annotated[
-            ReportFormat, typer.Option("--format", help="Output format")
-        ] = ReportFormat.text,
+    path: Annotated[
+        Path, typer.Argument(exists=True, file_okay=False, help="Codebase root")
+    ] = Path("."),
+    full: Annotated[bool, typer.Option("--full", help="Show full report")] = False,
+    totals: Annotated[bool, typer.Option("--totals", help="Only show totals")] = False,
+    fmt: Annotated[
+        ReportFormat, typer.Option("--format", help="Output format")
+    ] = ReportFormat.text,
 ):
     report_command(path, full, totals, fmt)
 
@@ -75,15 +75,15 @@ def _version_callback(show: bool):
 
 @cli.callback()
 def main(
-        verbose: Annotated[
-            Optional[bool], typer.Option("--verbose", "-v", help="Verbose output")
-        ] = False,
-        version: Annotated[
-            Optional[bool],
-            typer.Option(
-                "--version", "-V", help="Show version", callback=_version_callback
-            ),
-        ] = None,
+    verbose: Annotated[
+        Optional[bool], typer.Option("--verbose", "-v", help="Verbose output")
+    ] = False,
+    version: Annotated[
+        Optional[bool],
+        typer.Option(
+            "--version", "-V", help="Show version", callback=_version_callback
+        ),
+    ] = None,
 ):
     """Code Limit: Your refactoring alarm."""
     if verbose:

codelimit/commands/check.py

@@ -2,46 +2,49 @@
 from pathlib import Path
 
 import typer
+from pathspec import PathSpec
 from pygments.lexers import get_lexer_for_filename
 from pygments.util import ClassNotFound
 
 from codelimit.common.CheckResult import CheckResult
+from codelimit.common.Configuration import Configuration
 from codelimit.common.Scanner import is_excluded, scan_file
 from codelimit.common.lexer_utils import lex
 from codelimit.languages import Languages
 
 
 def check_command(paths: list[Path], quiet: bool):
     check_result = CheckResult()
+    excludes_spec = PathSpec.from_lines("gitignore", Configuration.excludes)
     for path in paths:
         if path.is_file():
-            _handle_file_path(path, check_result)
+            _handle_file_path(path, check_result, excludes_spec)
         elif path.is_dir():
             for root, dirs, files in os.walk(path.absolute()):
                 files = [f for f in files if not f[0] == "."]
                 dirs[:] = [d for d in dirs if not d[0] == "."]
                 for file in files:
                     abs_path = Path(os.path.join(root, file))
                     rel_path = abs_path.relative_to(path.absolute())
-                    if is_excluded(rel_path):
+                    if is_excluded(rel_path, excludes_spec):
                         continue
                     check_file(abs_path, check_result)
     exit_code = 1 if check_result.unmaintainable > 0 else 0
     if (
-            not quiet
-            or check_result.hard_to_maintain > 0
-            or check_result.unmaintainable > 0
+        not quiet
+        or check_result.hard_to_maintain > 0
+        or check_result.unmaintainable > 0
     ):
         check_result.report()
     raise typer.Exit(code=exit_code)
 
 
-def _handle_file_path(path: Path, check_result: CheckResult):
+def _handle_file_path(path: Path, check_result: CheckResult, excludes_spec: PathSpec):
     if not path.is_absolute():
         abs_path = path.absolute().resolve()
         try:
             rel_path = abs_path.relative_to(Path.cwd())
-            if is_excluded(rel_path):
+            if is_excluded(rel_path, excludes_spec):
                 return
         except ValueError:
             pass

codelimit/commands/report.py

@@ -77,7 +77,9 @@ def _report_functions(report: Report, path: Path, full: bool, fmt, console: Cons
     if fmt == ReportFormat.markdown:
         console.print(_report_functions_markdown(root, report_units), soft_wrap=True)
     else:
-        console.print(_report_functions_text(root, units, report_units, full), soft_wrap=True)
+        console.print(
+            _report_functions_text(root, units, report_units, full), soft_wrap=True
+        )
 
 
 def get_root(path: Path) -> Path | None:
@@ -109,17 +111,21 @@ def _report_functions_text(root, units, report_units, full) -> Text:
         file_path = unit.file if root is None else root.joinpath(unit.file)
         result.append(format_measurement(str(file_path), unit.measurement).append("\n"))
     if not full and len(units) > REPORT_LENGTH:
-        result.append(f"[bold]{len(units) - REPORT_LENGTH} more rows, use --full option to get all rows[/bold]\n")
+        result.append(
+            f"[bold]{len(units) - REPORT_LENGTH} more rows, use --full option to get all rows[/bold]\n"
+        )
     return result
 
 
-def _report_functions_markdown(root: Path | None, report_units: list[ReportUnit]) -> str:
+def _report_functions_markdown(
+    root: Path | None, report_units: list[ReportUnit]
+) -> str:
     result = ""
     result += "| **File** | **Line** | **Column** | **Length** | **Function** |\n"
     result += "| --- | ---: | ---: | ---: | --- |\n"
     for unit in report_units:
         file_path = unit.file if root is None else root.joinpath(unit.file)
-        type = '✖' if unit.measurement.value > 60 else '⚠'
+        type = "✖" if unit.measurement.value > 60 else "⚠"
         result += (
             f"| {str(file_path)} | {unit.measurement.start.line} | {unit.measurement.start.column} | "
             f"{unit.measurement.value} | {type} {unit.measurement.unit_name} |\n"

codelimit/commands/upload.py

@@ -5,11 +5,11 @@
 from codelimit.common.report.Report import Report
 from codelimit.common.report.ReportReader import ReportReader
 from codelimit.github_auth import get_github_token
-from codelimit.utils import read_cached_report, upload_report, make_report_path
+from codelimit.utils import upload_report, make_report_path
 
 
 def upload_command(
-        repository: str, branch: str, report_file: Path, token: str, url: str
+    repository: str, branch: str, report_file: Path, token: str, url: str
 ):
     if report_file:
         report = ReportReader.from_json(report_file.read_text())

codelimit/common/Scanner.py

@@ -1,4 +1,3 @@
-import fnmatch
 import locale
 import os
 from datetime import datetime
@@ -86,15 +85,17 @@ def _scan_folder(
     cached_report: Union[Report, None] = None,
     add_file_entry: Union[Callable[[SourceFileEntry], None], None] = None,
 ):
-    gitignore = _read_gitignore(folder)
+    excludes = Configuration.excludes.copy()
+    gitignore_excludes = _read_gitignore(folder)
+    if gitignore_excludes:
+        excludes.extend(gitignore_excludes)
+    excludes_spec = PathSpec.from_lines("gitignore", excludes)
     for root, dirs, files in os.walk(folder.absolute()):
         files = [f for f in files if not f[0] == "."]
         dirs[:] = [d for d in dirs if not d[0] == "."]
         for file in files:
             rel_path = Path(os.path.join(root, file)).relative_to(folder.absolute())
-            if is_excluded(rel_path) or (
-                gitignore is not None and is_excluded_by_gitignore(rel_path, gitignore)
-            ):
+            if is_excluded(rel_path, excludes_spec):
                 continue
             try:
                 lexer = get_lexer_for_filename(rel_path)
@@ -178,25 +179,12 @@ def scan_file(tokens: list[Token], language: Language) -> list[Measurement]:
     return measurements
 
 
-def is_excluded(path: Path):
-    for exclude in Configuration.excludes:
-        exclude_parts = exclude.split(os.sep)
-        if len(exclude_parts) == 1:
-            for part in path.parts:
-                if fnmatch.fnmatch(part, exclude):
-                    return True
-        else:
-            if fnmatch.fnmatch(str(path), exclude):
-                return True
-    return False
-
-
-def _read_gitignore(path: Path) -> PathSpec | None:
+def _read_gitignore(path: Path) -> list[str] | None:
     gitignore_path = path.joinpath(".gitignore")
     if gitignore_path.exists():
-        return PathSpec.from_lines("gitignore", gitignore_path.read_text().splitlines())
+        return gitignore_path.read_text().splitlines()
     return None
 
 
-def is_excluded_by_gitignore(path: Path, gitignore: PathSpec):
-    return gitignore.match_file(path)
+def is_excluded(path: Path, spec: PathSpec):
+    return spec.match_file(path)

codelimit/common/gsm/Pattern.py

@@ -17,7 +17,7 @@ def __init__(self, start: int, automata: DFA):
     def consume(self, item) -> State | None:
         for transition in self.state.transition:
             predicate_id = id(transition[0])
-            if not predicate_id in self.predicate_map:
+            if predicate_id not in self.predicate_map:
                 self.predicate_map[predicate_id] = deepcopy(transition[0])
             predicate = self.predicate_map[predicate_id]
             if predicate.accept(item):

codelimit/common/token_matching/predicate/Balanced.py

@@ -43,4 +43,4 @@ def __hash__(self):
         return hash((self.left, self.right, self.depth))
 
     def __str__(self):
-        return f"<Balanced {self.left} {self.right} {id(self)}>"
+        return f"<Balanced {self.left} {self.right} {id(self)}>"

codelimit/common/token_matching/predicate/Keyword.py

@@ -22,4 +22,4 @@ def __hash__(self):
         return hash(self.keyword)
 
     def __str__(self):
-        return f'<Keyword {self.keyword}>'
+        return f"<Keyword {self.keyword}>"

codelimit/common/token_matching/predicate/Symbol.py

@@ -22,4 +22,4 @@ def __hash__(self):
         return hash(self.symbol)
 
     def __str__(self):
-        return f'<Symbol {self.symbol}>'
+        return f"<Symbol {self.symbol}>"

codelimit/common/token_matching/predicate/TokenValue.py

@@ -21,6 +21,5 @@ def __eq__(self, other: object) -> bool:
     def __hash__(self):
         return hash(self.value)
 
-
     def __str__(self):
-        return f'<TokenValue {self.value}>'
+        return f"<TokenValue {self.value}>"

codelimit/utils.py

@@ -24,7 +24,7 @@ def read_cached_report(path: Path) -> Optional[Report]:
 
 
 def upload_report(
-        report: Report, repository: str, branch: str, url: str, token: str
+    report: Report, repository: str, branch: str, url: str, token: str
 ) -> None:
     result = api_post_report(report, branch, repository, url, token)
     if result.ok:
@@ -44,9 +44,9 @@ def api_post_report(report, branch, repository, url, token):
         f'{{{{"repository": "{repository}", "branch": "{branch}", "report":{{}}}}}}'
     )
     with Progress(
-            SpinnerColumn(),
-            TextColumn("[progress.description]{task.description}"),
-            transient=True,
+        SpinnerColumn(),
+        TextColumn("[progress.description]{task.description}"),
+        transient=True,
     ) as progress:
         progress.add_task(description=f"Uploading report to {url}", total=None)
         result = requests.post(

tests/commands/test_report.py

@@ -1,4 +1,7 @@
-from codelimit.commands.report import _report_totals_markdown, _report_functions_markdown
+from codelimit.commands.report import (
+    _report_totals_markdown,
+    _report_functions_markdown,
+)
 from codelimit.common.LanguageTotals import LanguageTotals
 from codelimit.common.Location import Location
 from codelimit.common.Measurement import Measurement

tests/common/test_Scanner.py

@@ -5,11 +5,7 @@
 from pathspec import PathSpec
 
 from codelimit.common.Configuration import Configuration
-from codelimit.common.Scanner import (
-    scan_codebase,
-    is_excluded,
-    is_excluded_by_gitignore,
-)
+from codelimit.common.Scanner import scan_codebase, is_excluded
 from codelimit.common.source_utils import get_location_range
 
 
@@ -88,30 +84,27 @@ def test_skip_hidden_files():
 
 
 def test_is_excluded():
-    assert is_excluded(Path("venv/foo/bar.py"))
-    assert not is_excluded(Path("foo/bar.py"))
+    excludes_spec = PathSpec.from_lines("gitignore", Configuration.excludes)
 
-    Configuration.excludes = ["output"]
+    assert is_excluded(Path("venv/foo/bar.py"), excludes_spec)
+    assert not is_excluded(Path("foo/bar.py"), excludes_spec)
 
-    assert is_excluded(Path("output/foo/bar.py"))
-    assert not is_excluded(Path("venv/foo/bar.py"))
-    assert not is_excluded(Path("foo/bar.py"))
+    excludes_spec = PathSpec.from_lines("gitignore", ["output"])
 
-    Configuration.excludes = ["foo/bar/*"]
+    assert is_excluded(Path("output/foo/bar.py"), excludes_spec)
+    assert not is_excluded(Path("venv/foo/bar.py"), excludes_spec)
+    assert not is_excluded(Path("foo/bar.py"), excludes_spec)
 
-    assert is_excluded(Path("foo/bar/foobar.py"))
+    excludes_spec = PathSpec.from_lines("gitignore", ["foo/bar/*"])
 
+    assert is_excluded(Path("foo/bar/foobar.py"), excludes_spec)
 
-def test_is_excluded_by_gitignore():
-    Configuration.excludes = ["site/"]
-    gitignore = PathSpec.from_lines("gitwildmatch", ["site/"])
+    excludes_spec = PathSpec.from_lines("gitignore", ["site/"])
 
-    assert is_excluded_by_gitignore(
-        Path("site/assets/javascripts/lunr/wordcut.js"), gitignore
-    )
+    assert is_excluded(Path("site/assets/javascripts/lunr/wordcut.js"), excludes_spec)
 
-    gitignore = PathSpec.from_lines("gitwildmatch", ["!site/"])
+    excludes_spec = PathSpec.from_lines("gitignore", ["!site/"])
 
-    assert not is_excluded_by_gitignore(
-        Path("site/assets/javascripts/lunr/wordcut.js"), gitignore
+    assert not is_excluded(
+        Path("site/assets/javascripts/lunr/wordcut.js"), excludes_spec
     )