Update Contributing Docs (#1915)

[etc/vim.git] / src / black_primer / lib.py

diff --git a/src/black_primer/lib.py b/src/black_primer/lib.py
index be0618ca5b2f53d3fe5c12a5f9ef51256951a14e..5c5576e1ff334f8b8d22f2d58d704438914c6f25 100644 (file)
--- a/src/black_primer/lib.py
+++ b/src/black_primer/lib.py
@@ -1,24 +1,38 @@
 #!/usr/bin/env python3
 
 #!/usr/bin/env python3
 

-# Module '__future__' has no attribute 'annotations'
-from __future__ import annotations  # type: ignore
-

 import asyncio
 import asyncio

+import errno

 import json
 import logging
 import json
 import logging

+import os
+import stat
+import sys
+from functools import partial

 from pathlib import Path
 from pathlib import Path

+from platform import system

 from shutil import rmtree, which
 from subprocess import CalledProcessError
 from sys import version_info
 from shutil import rmtree, which
 from subprocess import CalledProcessError
 from sys import version_info

-from typing import Any, Dict, NamedTuple, Optional, Sequence, Tuple
+from typing import Any, Callable, Dict, NamedTuple, Optional, Sequence, Tuple

 from urllib.parse import urlparse
 
 import click
 
 
 from urllib.parse import urlparse
 
 import click
 
 

+WINDOWS = system() == "Windows"
+BLACK_BINARY = "black.exe" if WINDOWS else "black"
+GIT_BIANRY = "git.exe" if WINDOWS else "git"

 LOG = logging.getLogger(__name__)
 
 
 LOG = logging.getLogger(__name__)
 
 

+# Windows needs a ProactorEventLoop if you want to exec subprocesses
+# Starting with 3.8 this is the default - can remove when Black >= 3.8
+# mypy only respects sys.platform if directly in the evaluation
+# https://mypy.readthedocs.io/en/latest/common_issues.html#python-version-and-system-platform-checks  # noqa: B950
+if sys.platform == "win32":
+    asyncio.set_event_loop(asyncio.ProactorEventLoop())
+
+

 class Results(NamedTuple):
     stats: Dict[str, int] = {}
     failed_projects: Dict[str, CalledProcessError] = {}
 class Results(NamedTuple):
     stats: Dict[str, int] = {}
     failed_projects: Dict[str, CalledProcessError] = {}


@@ -26,7 +40,7 @@ class Results(NamedTuple):
 
 async def _gen_check_output(
     cmd: Sequence[str],
 
 async def _gen_check_output(
     cmd: Sequence[str],

-    timeout: float = 30,
+    timeout: float = 300,

     env: Optional[Dict[str, str]] = None,
     cwd: Optional[Path] = None,
 ) -> Tuple[bytes, bytes]:
     env: Optional[Dict[str, str]] = None,
     cwd: Optional[Path] = None,
 ) -> Tuple[bytes, bytes]:


@@ -45,15 +59,17 @@ async def _gen_check_output(
         raise
 
     if process.returncode != 0:
         raise
 
     if process.returncode != 0:

+        returncode = process.returncode
+        if returncode is None:
+            returncode = 69
+

         cmd_str = " ".join(cmd)
         cmd_str = " ".join(cmd)

-        raise CalledProcessError(
-            process.returncode, cmd_str, output=stdout, stderr=stderr
-        )
+        raise CalledProcessError(returncode, cmd_str, output=stdout, stderr=stderr)

 
     return (stdout, stderr)
 
 
 
     return (stdout, stderr)
 
 

-async def analyze_results(project_count: int, results: Results) -> int:
+def analyze_results(project_count: int, results: Results) -> int:

     failed_pct = round(((results.stats["failed"] / project_count) * 100), 2)
     success_pct = round(((results.stats["success"] / project_count) * 100), 2)
 
     failed_pct = round(((results.stats["failed"] / project_count) * 100), 2)
     success_pct = round(((results.stats["success"] / project_count) * 100), 2)
 


@@ -68,16 +84,18 @@ async def analyze_results(project_count: int, results: Results) -> int:
         bold=bool(results.stats["failed"]),
         fg="red",
     )
         bold=bool(results.stats["failed"]),
         fg="red",
     )

-    click.echo(f" - {results.stats['disabled']} projects Disabled by config")
+    s = "" if results.stats["disabled"] == 1 else "s"
+    click.echo(f" - {results.stats['disabled']} project{s} disabled by config")
+    s = "" if results.stats["wrong_py_ver"] == 1 else "s"

     click.echo(
     click.echo(

-        f" - {results.stats['wrong_py_ver']} projects skipped due to Python Version"
+        f" - {results.stats['wrong_py_ver']} project{s} skipped due to Python version"

     )
     click.echo(
         f" - {results.stats['skipped_long_checkout']} skipped due to long checkout"
     )
 
     if results.failed_projects:
     )
     click.echo(
         f" - {results.stats['skipped_long_checkout']} skipped due to long checkout"
     )
 
     if results.failed_projects:

-        click.secho("\nFailed Projects:\n", bold=True)
+        click.secho("\nFailed projects:\n", bold=True)

 
     for project_name, project_cpe in results.failed_projects.items():
         print(f"## {project_name}:")
 
     for project_name, project_cpe in results.failed_projects.items():
         print(f"## {project_name}:")


@@ -94,9 +112,9 @@ async def analyze_results(project_count: int, results: Results) -> int:
 async def black_run(
     repo_path: Path, project_config: Dict[str, Any], results: Results
 ) -> None:
 async def black_run(
     repo_path: Path, project_config: Dict[str, Any], results: Results
 ) -> None:

-    """Run black and record failures"""
-    cmd = [str(which("black"))]
-    if project_config["cli_arguments"]:
+    """Run Black and record failures"""
+    cmd = [str(which(BLACK_BINARY))]
+    if "cli_arguments" in project_config and project_config["cli_arguments"]:

         cmd.extend(*project_config["cli_arguments"])
     cmd.extend(["--check", "--diff", "."])
 
         cmd.extend(*project_config["cli_arguments"])
     cmd.extend(["--check", "--diff", "."])
 


@@ -106,12 +124,31 @@ async def black_run(
         results.stats["failed"] += 1
         LOG.error(f"Running black for {repo_path} timed out ({cmd})")
     except CalledProcessError as cpe:
         results.stats["failed"] += 1
         LOG.error(f"Running black for {repo_path} timed out ({cmd})")
     except CalledProcessError as cpe:

-        # TODO: This might need to be tuned and made smarter for higher signal
-        if not project_config["expect_formatting_changes"] and cpe.returncode == 1:
+        # TODO: Tune for smarter for higher signal
+        # If any other return value than 1 we raise - can disable project in config
+        if cpe.returncode == 1:
+            if not project_config["expect_formatting_changes"]:
+                results.stats["failed"] += 1
+                results.failed_projects[repo_path.name] = cpe
+            else:
+                results.stats["success"] += 1
+            return
+        elif cpe.returncode > 1:

             results.stats["failed"] += 1
             results.failed_projects[repo_path.name] = cpe
             return
 
             results.stats["failed"] += 1
             results.failed_projects[repo_path.name] = cpe
             return
 

+        LOG.error(f"Unknown error with {repo_path}")
+        raise
+
+    # If we get here and expect formatting changes something is up
+    if project_config["expect_formatting_changes"]:
+        results.stats["failed"] += 1
+        results.failed_projects[repo_path.name] = CalledProcessError(
+            0, cmd, b"Expected formatting changes but didn't get any!", b""
+        )
+        return
+

     results.stats["success"] += 1
 
 
     results.stats["success"] += 1
 
 


@@ -123,7 +160,7 @@ async def git_checkout_or_rebase(
     depth: int = 1,
 ) -> Optional[Path]:
     """git Clone project or rebase"""
     depth: int = 1,
 ) -> Optional[Path]:
     """git Clone project or rebase"""

-    git_bin = str(which("git"))
+    git_bin = str(which(GIT_BIANRY))

     if not git_bin:
         LOG.error("No git binary found")
         return None
     if not git_bin:
         LOG.error("No git binary found")
         return None


@@ -149,9 +186,33 @@ async def git_checkout_or_rebase(
     return repo_path
 
 
     return repo_path
 
 

+def handle_PermissionError(
+    func: Callable, path: Path, exc: Tuple[Any, Any, Any]
+) -> None:
+    """
+    Handle PermissionError during shutil.rmtree.
+
+    This checks if the erroring function is either 'os.rmdir' or 'os.unlink', and that
+    the error was EACCES (i.e. Permission denied). If true, the path is set writable,
+    readable, and executable by everyone. Finally, it tries the error causing delete
+    operation again.
+
+    If the check is false, then the original error will be reraised as this function
+    can't handle it.
+    """
+    excvalue = exc[1]
+    LOG.debug(f"Handling {excvalue} from {func.__name__}... ")
+    if func in (os.rmdir, os.unlink) and excvalue.errno == errno.EACCES:
+        LOG.debug(f"Setting {path} writable, readable, and executable by everyone... ")
+        os.chmod(path, stat.S_IRWXU | stat.S_IRWXG | stat.S_IRWXO)  # chmod 0777
+        func(path)  # Try the error causing delete operation again
+    else:
+        raise
+
+

 async def load_projects_queue(
     config_path: Path,
 async def load_projects_queue(
     config_path: Path,

-) -> Tuple[Dict[str, Any], asyncio.Queue[str]]:
+) -> Tuple[Dict[str, Any], asyncio.Queue]:

     """Load project config and fill queue with all the project names"""
     with config_path.open("r") as cfp:
         config = json.load(cfp)
     """Load project config and fill queue with all the project names"""
     with config_path.open("r") as cfp:
         config = json.load(cfp)


@@ -159,7 +220,7 @@ async def load_projects_queue(
     # TODO: Offer more options here
     # e.g. Run on X random packages or specific sub list etc.
     project_names = sorted(config["projects"].keys())
     # TODO: Offer more options here
     # e.g. Run on X random packages or specific sub list etc.
     project_names = sorted(config["projects"].keys())

-    queue: asyncio.Queue[str] = asyncio.Queue(maxsize=len(project_names))
+    queue: asyncio.Queue = asyncio.Queue(maxsize=len(project_names))

     for project in project_names:
         await queue.put(project)
 
     for project in project_names:
         await queue.put(project)
 


@@ -169,14 +230,14 @@ async def load_projects_queue(
 async def project_runner(
     idx: int,
     config: Dict[str, Any],
 async def project_runner(
     idx: int,
     config: Dict[str, Any],

-    queue: asyncio.Queue[str],
+    queue: asyncio.Queue,

     work_path: Path,
     results: Results,
     long_checkouts: bool = False,
     rebase: bool = False,
     keep: bool = False,
 ) -> None:
     work_path: Path,
     results: Results,
     long_checkouts: bool = False,
     rebase: bool = False,
     keep: bool = False,
 ) -> None:

-    """Checkout project and run black on it + record result"""
+    """Check out project and run Black on it + record result"""

     loop = asyncio.get_event_loop()
     py_version = f"{version_info[0]}.{version_info[1]}"
     while True:
     loop = asyncio.get_event_loop()
     py_version = f"{version_info[0]}.{version_info[1]}"
     while True:


@@ -185,6 +246,7 @@ async def project_runner(
         except asyncio.QueueEmpty:
             LOG.debug(f"project_runner {idx} exiting")
             return
         except asyncio.QueueEmpty:
             LOG.debug(f"project_runner {idx} exiting")
             return

+        LOG.debug(f"worker {idx} working on {project_name}")

 
         project_config = config["projects"][project_name]
 
 
         project_config = config["projects"][project_name]
 


@@ -216,7 +278,12 @@ async def project_runner(
 
         if not keep:
             LOG.debug(f"Removing {repo_path}")
 
         if not keep:
             LOG.debug(f"Removing {repo_path}")

-            await loop.run_in_executor(None, rmtree, repo_path)
+            rmtree_partial = partial(
+                rmtree, path=repo_path, onerror=handle_PermissionError
+            )
+            await loop.run_in_executor(None, rmtree_partial)
+
+        LOG.info(f"Finished {project_name}")

 
 
 async def process_queue(
 
 
 async def process_queue(


@@ -242,11 +309,13 @@ async def process_queue(
 
     config, queue = await load_projects_queue(Path(config_file))
     project_count = queue.qsize()
 
     config, queue = await load_projects_queue(Path(config_file))
     project_count = queue.qsize()

-    LOG.info(f"{project_count} projects to run black over")
-    if not project_count:
+    s = "" if project_count == 1 else "s"
+    LOG.info(f"{project_count} project{s} to run Black over")
+    if project_count < 1:

         return -1
 
         return -1
 

-    LOG.debug(f"Using {workers} parallel workers to run black")
+    s = "" if workers == 1 else "s"
+    LOG.debug(f"Using {workers} parallel worker{s} to run Black")

     # Wait until we finish running all the projects before analyzing
     await asyncio.gather(
         *[
     # Wait until we finish running all the projects before analyzing
     await asyncio.gather(
         *[


@@ -258,4 +327,8 @@ async def process_queue(
     )
 
     LOG.info("Analyzing results")
     )
 
     LOG.info("Analyzing results")

-    return await analyze_results(project_count, results)
+    return analyze_results(project_count, results)
+
+
+if __name__ == "__main__":  # pragma: nocover
+    raise NotImplementedError("lib is a library, funnily enough.")