From 3bc7be3f79e1395f69405401a49c4d33d92e8770 Mon Sep 17 00:00:00 2001
From: Joseph Walton <joe@kafsemo.org>
Date: Mon, 1 Apr 2024 01:38:40 +1100
Subject: [PATCH] Continue to watch files that are replaced. (#1781)

Continue to watch files are replaced, such as by saving in vi,
by watching the parent directory instead.

Add a test, that fails under the previous version, for a modification
to a file after it is deleted and recreated.

Bump the minimum watchfiles version to ensure that ignore_permission_denied,
added in https://github.com/samuelcolvin/watchfiles/pull/224 and released
in https://github.com/samuelcolvin/watchfiles/releases/tag/v0.20.0
is present.

Co-authored-by: Jakob Schnitzer <mail@jakobschnitzer.de>
---
 constraints-old.txt        |  2 +-
 pyproject.toml             |  2 +-
 src/fava/core/watcher.py   | 36 ++++++++++++++++++++++++++++--------
 tests/test_core_watcher.py | 26 ++++++++++++++++++++++++++
 4 files changed, 56 insertions(+), 10 deletions(-)

diff --git a/constraints-old.txt b/constraints-old.txt
index 253b0c689..d5ab3bf26 100644
--- a/constraints-old.txt
+++ b/constraints-old.txt
@@ -115,6 +115,6 @@ uritemplate==4.1.1
     # via google-api-python-client
 urllib3==2.2.1
     # via requests
-watchfiles==0.19.0
+watchfiles==0.20.0
 werkzeug==2.2.0
     # via flask
diff --git a/pyproject.toml b/pyproject.toml
index 3770ff58d..93f2e0812 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -45,7 +45,7 @@ dependencies = [
     "markdown2>=2.3.0,<3",
     "ply",
     "simplejson>=3.16.0,<4",
-    "watchfiles>=0.19.0",
+    "watchfiles>=0.20.0",
 ]
 
 [project.urls]
diff --git a/src/fava/core/watcher.py b/src/fava/core/watcher.py
index dc92c9a63..4d37707b3 100644
--- a/src/fava/core/watcher.py
+++ b/src/fava/core/watcher.py
@@ -6,7 +6,6 @@
 import atexit
 import logging
 import threading
-from itertools import chain
 from os import walk
 from pathlib import Path
 from time import time_ns
@@ -18,28 +17,45 @@
 if TYPE_CHECKING:  # pragma: no cover
     import types
 
+    from watchfiles.main import Change
+
 
 log = logging.getLogger(__name__)
 
 
 class _WatchfilesThread(threading.Thread):
-    def __init__(self, paths: list[Path], mtime: int) -> None:
+    def __init__(
+        self, files: list[Path], folders: list[Path], mtime: int
+    ) -> None:
         super().__init__(daemon=True)
         self._stop_event = threading.Event()
-        self.paths = paths
         self.mtime = mtime
+        self._file_set = {f.absolute() for f in files}
+        self._folder_set = {f.absolute() for f in folders}
 
     def stop(self) -> None:
         """Set the stop event for watchfiles and join the thread."""
         self._stop_event.set()
         self.join()
 
+    def _is_relevant(self, _c: Change, path: str) -> bool:
+        return Path(path) in self._file_set or any(
+            Path(path).is_relative_to(d) for d in self._folder_set
+        )
+
     def run(self) -> None:
         """Watch for changes."""
         atexit.register(self.stop)
 
-        for changes in watch(*self.paths, stop_event=self._stop_event):
-            for _, path in changes:
+        watch_paths = {f.parent for f in self._file_set} | self._folder_set
+
+        for changes in watch(
+            *watch_paths,
+            stop_event=self._stop_event,
+            ignore_permission_denied=True,
+            watch_filter=self._is_relevant,
+        ):
+            for path in {change[1] for change in changes}:
                 try:
                     change_mtime = Path(path).stat().st_mtime_ns
                 except FileNotFoundError:
@@ -98,19 +114,23 @@ class WatchfilesWatcher(WatcherBase):
     def __init__(self) -> None:
         self.last_checked = 0
         self.last_notified = 0
-        self._paths: list[Path] = []
+        self._paths: tuple[list[Path], list[Path]] = ([], [])
         self._thread: _WatchfilesThread | None = None
 
     def update(self, files: Iterable[Path], folders: Iterable[Path]) -> None:
         """Update the folders/files to watch."""
-        new_paths = [p for p in chain(files, folders) if p.exists()]
+        existing_files = [p for p in files if p.exists()]
+        existing_folders = [p for p in folders if p.is_dir()]
+        new_paths = (existing_files, existing_folders)
         if self._thread and new_paths == self._paths:
             self.check()
             return
         self._paths = new_paths
         if self._thread is not None:
             self._thread.stop()
-        self._thread = _WatchfilesThread(self._paths, self.last_checked)
+        self._thread = _WatchfilesThread(
+            existing_files, existing_folders, self.last_checked
+        )
         self._thread.start()
         self.check()
 
diff --git a/tests/test_core_watcher.py b/tests/test_core_watcher.py
index ab52a73b0..c7d1d3424 100644
--- a/tests/test_core_watcher.py
+++ b/tests/test_core_watcher.py
@@ -109,3 +109,29 @@ def test_watchfiles_watcher(watcher_paths: WatcherTestSet) -> None:
         # notify of deleted file
         watcher.notify(watcher_paths.file1)
         assert watcher.check()
+
+
+def test_watchfiles_watcher_recognises_change_to_previously_deleted_file(
+    watcher_paths: WatcherTestSet,
+) -> None:
+    watcher = WatchfilesWatcher()
+    with watcher:
+        assert not watcher.check()  # No thread set up yet.
+
+    with watcher:
+        watcher.update([watcher_paths.file1], [])
+
+        watcher_paths.file1.unlink()
+        assert _watcher_check_within_one_second(watcher)
+        # notify of deleted file
+        watcher.notify(watcher_paths.file1)
+        # True because time_ns() used when file is absent
+        assert watcher.check()
+
+        # Recreate deleted file
+        # sleep to ensure file stamp is greater than time_ns() taken on
+        #  previous FileNotFoundError
+        time.sleep(0.01)
+        watcher_paths.file1.write_text("test-value-2")
+        assert _watcher_check_within_one_second(watcher)
+        assert not watcher.check()