general: work around the issue when warnings module sometimes logs th…

…e same warning multiple times

see python/cpython#73858 (comment)

src/promnesia/common.py

@@ -76,13 +76,26 @@ def file(cls, path: PathIsh, line: Optional[int]=None, relative_to: Optional[Pat
     # but generally, it will be
     # (url|file)(linenumber|json_path|anchor)
 
+
+@lru_cache(None)
+def warn_once(message: str) -> None:
+    # you'd think that warnings module already logs warnings only once per line..
+    # but sadly it's not the case
+    # see https://github.com/karlicoss/python_duplicate_warnings_investigation/blob/master/test.py
+    warnings.warn(message, stacklevel=2)
+
+
+def _warn_no_xdg_mime() -> None:
+    warn_once("No xdg-mime on your OS! If you're on OSX, perhaps you can help me! https://github.com/karlicoss/open-in-editor/issues/1")
+
+
 @lru_cache(1)
 def _detect_mime_handler() -> str:
     def exists(what: str) -> bool:
         try:
             r = run(f'xdg-mime query default x-scheme-handler/{what}'.split(), stdout=PIPE)
         except (FileNotFoundError, NotADirectoryError):  # ugh seems that osx might throw NotADirectory for some reason
-            warnings.warn("No xdg-mime on your OS! If you're on OSX, perhaps you can help me! https://github.com/karlicoss/open-in-editor/issues/1")
+            _warn_no_xdg_mime()
             return False
         if r.returncode > 0:
             warnings.warn('xdg-mime failed') # hopefully rest is in stderr
@@ -102,6 +115,7 @@ def exists(what: str) -> bool:
         result = 'emacs:'
 
     # 2. now try to use newer editor:// thing
+    # TODO flip order here? should rely on editor:// first?
 
     # TODO would be nice to collect warnings and display at the end
     if not exists('editor'):

src/promnesia/sources/auto.py

@@ -22,6 +22,7 @@
 import pytz
 
 from ..common import Visit, Url, PathIsh, get_logger, Loc, get_tmpdir, extract_urls, Extraction, Result, Results, mime, traverse, file_mtime, echain, logger
+from ..common import warn_once
 from ..config import use_cores
 
 
@@ -167,7 +168,7 @@ def _org(path: Path) -> Results:
 Replacer = Optional[Callable[[str, str], str]]
 
 def index(
-        *paths: Union[PathIsh],
+        *paths: PathIsh,
         ignored: Union[Sequence[str], str]=(),
         follow: bool=True,
         replacer: Replacer=None,
@@ -282,6 +283,8 @@ def by_path(pp: Path) -> Tuple[Optional[Ex], Optional[Mime]]:
 
 def _index_file(pp: Path, opts: Options) -> Results:
     logger = get_logger()
+    # TODO need to keep debug logs here...
+    # logger.info(f"indexing {pp}")
     # TODO use kompress?
     # TODO not even sure if it's used...
     suf = pp.suffix.lower()
@@ -307,10 +310,9 @@ def _index_file(pp: Path, opts: Options) -> Results:
 
     ip, pm = by_path(pp)
     if ip is None:
-        # TODO use warning (with mime/ext as key?)
-        # TODO only log once? # hmm..
+        # todo not really sure about using warnings vs yielding error here?
         msg = f'No extractor for suffix {suf}, mime {pm}'
-        warnings.warn(msg)
+        warn_once(msg)
         yield echain(ex, RuntimeError(msg))
         return
 

src/promnesia/sources/plaintext.py

@@ -98,8 +98,10 @@ def extract_from_path(path: PathIsh) -> Command:
                 '.gz',
                 '.zip',
         )):
-            logger.info(f"Extracting from compressed file {path}")
+            # todo should be debug?
+            # or should delete it completely, feels like unpacking archives here is a bit too much
             raise RuntimeError(f"Archives aren't supported yet: {path}")
+            logger.info(f"Extracting from compressed file {path}")
             import lzma
             from tempfile import NamedTemporaryFile
             # TODO hopefully, no collisions