Merge pull request #54 from karlicoss/updates

core: update warnings, add warn_if_empty decorator fore move defensive data sources

my/core/__init__.py

@@ -1,4 +1,5 @@
 # this file only keeps the most common & critical types/utility functions
 from .common import PathIsh, Paths, Json
 from .common import get_files, LazyLogger
+from .common import warn_if_empty
 from .cfg import make_config

my/core/common.py

@@ -3,7 +3,7 @@
 from datetime import datetime
 import functools
 import types
-from typing import Union, Callable, Dict, Iterable, TypeVar, Sequence, List, Optional, Any, cast, Tuple
+from typing import Union, Callable, Dict, Iterable, TypeVar, Sequence, List, Optional, Any, cast, Tuple, TYPE_CHECKING
 import warnings
 
 # some helper functions
@@ -130,6 +130,11 @@ def get_files(pp: Paths, glob: str=DEFAULT_GLOB, sort: bool=True) -> Tuple[Path,
     else:
         sources.extend(map(Path, pp))
 
+    def caller() -> str:
+        import traceback
+        # TODO ugh. very flaky... -3 because [<this function>, get_files(), <actual caller>]
+        return traceback.extract_stack()[-3].filename
+
     paths: List[Path] = []
     for src in sources:
         if src.parts[0] == '~':
@@ -141,7 +146,7 @@ def get_files(pp: Paths, glob: str=DEFAULT_GLOB, sort: bool=True) -> Tuple[Path,
             ss = str(src)
             if '*' in ss:
                 if glob != DEFAULT_GLOB:
-                    warnings.warn(f"Treating {ss} as glob path. Explicit glob={glob} argument is ignored!")
+                    warnings.warn(f"{caller()}: treating {ss} as glob path. Explicit glob={glob} argument is ignored!")
                 paths.extend(map(Path, do_glob(ss)))
             else:
                 if not src.is_file():
@@ -154,14 +159,15 @@ def get_files(pp: Paths, glob: str=DEFAULT_GLOB, sort: bool=True) -> Tuple[Path,
 
     if len(paths) == 0:
         # todo make it conditionally defensive based on some global settings
-        # todo stacktrace?
-        warnings.warn(f'No paths were matched against {paths}. This might result in missing data.')
+        # TODO not sure about using warnings module for this
+        import traceback
+        warnings.warn(f'{caller()}: no paths were matched against {paths}. This might result in missing data.')
+        traceback.print_stack()
 
     return tuple(paths)
 
 
 # TODO annotate it, perhaps use 'dependent' type (for @doublewrap stuff)
-from typing import TYPE_CHECKING
 if TYPE_CHECKING:
     from typing import Callable, TypeVar
     from typing_extensions import Protocol
@@ -269,3 +275,60 @@ def isoparse(s: str) -> tzdatetime:
 def get_valid_filename(s: str) -> str:
     s = str(s).strip().replace(' ', '_')
     return re.sub(r'(?u)[^-\w.]', '', s)
+
+
+from typing import Generic, Sized, Callable
+
+
+# X = TypeVar('X')
+def _warn_iterator(it, f: Any=None):
+    emitted = False
+    for i in it:
+        yield i
+        emitted = True
+    if not emitted:
+        warnings.warn(f"Function {f} didn't emit any data, make sure your config paths are correct")
+
+
+# TODO ugh, so I want to express something like:
+# X = TypeVar('X')
+# C = TypeVar('C', bound=Iterable[X])
+# _warn_iterable(it: C) -> C
+# but apparently I can't??? ugh.
+# https://github.com/python/typing/issues/548
+# I guess for now overloads are fine...
+
+from typing import overload
+X = TypeVar('X')
+@overload
+def _warn_iterable(it: List[X]    , f: Any=None) -> List[X]    : ...
+@overload
+def _warn_iterable(it: Iterable[X], f: Any=None) -> Iterable[X]: ...
+def _warn_iterable(it, f=None):
+    if isinstance(it, Sized):
+        sz = len(it)
+        if sz == 0:
+            warnings.warn(f"Function {f} returned empty container, make sure your config paths are correct")
+        return it
+    else:
+        return _warn_iterator(it, f=f)
+
+
+# ok, this seems to work...
+# https://github.com/python/mypy/issues/1927#issue-167100413
+FL = TypeVar('FL', bound=Callable[..., List])
+FI = TypeVar('FI', bound=Callable[..., Iterable])
+
+@overload
+def warn_if_empty(f: FL) -> FL: ...
+@overload
+def warn_if_empty(f: FI) -> FI: ...
+
+
+def warn_if_empty(f):
+    from functools import wraps
+    @wraps(f)
+    def wrapped(*args, **kwargs):
+        res = f(*args, **kwargs)
+        return _warn_iterable(res, f=f)
+    return wrapped # type: ignore

my/rss/all.py

@@ -1,11 +1,16 @@
 '''
 Unified RSS data, merged from different services I used historically
 '''
+# NOTE: you can comment out the sources you're not using
+from . import feedbin, feedly
+
 from typing import Iterable
 from .common import Subscription, compute_subscriptions
 
 
 def subscriptions() -> Iterable[Subscription]:
-    from . import feedbin, feedly
     # TODO google reader?
-    yield from compute_subscriptions(feedbin.states(), feedly.states())
+    yield from compute_subscriptions(
+        feedbin.states(),
+        feedly .states(),
+    )

my/rss/common.py

@@ -17,6 +17,8 @@ class Subscription(NamedTuple):
 SubscriptionState = Tuple[datetime, Sequence[Subscription]]
 
 
+from ..core import warn_if_empty
+@warn_if_empty
 def compute_subscriptions(*sources: Iterable[SubscriptionState]) -> List[Subscription]:
     """
     Keeps track of everything I ever subscribed to.
@@ -34,6 +36,9 @@ def compute_subscriptions(*sources: Iterable[SubscriptionState]) -> List[Subscri
             if feed.url not in by_url:
                 by_url[feed.url] = feed
 
+    if len(states) == 0:
+        return []
+
     _, last_state = max(states, key=lambda x: x[0])
     last_urls = {f.url for f in last_state}
 

my/twitter/all.py

@@ -3,18 +3,17 @@
 """
 
 # NOTE: you can comment out the sources you don't need
-
-
 from . import twint, archive
-from .common import merge_tweets
 
+from .common import merge_tweets
 
 def tweets():
     yield from merge_tweets(
         twint  .tweets(),
         archive.tweets(),
     )
 
+from .common import merge_tweets
 
 def likes():
     yield from merge_tweets(

my/twitter/common.py

@@ -2,7 +2,9 @@
 
 from more_itertools import unique_everseen
 
+from ..core import warn_if_empty
 
+@warn_if_empty
 def merge_tweets(*sources):
     yield from unique_everseen(
         chain(*sources),

tests/misc.py

@@ -48,3 +48,54 @@ def prepare(tmp_path: Path):
 
 # meh
 from my.core.error import test_sort_res_by
+
+
+from typing import Iterable, List
+import warnings
+from my.core import warn_if_empty
+def test_warn_if_empty() -> None:
+    @warn_if_empty
+    def nonempty() -> Iterable[str]:
+        yield 'a'
+        yield 'aba'
+
+    @warn_if_empty
+    def empty() -> List[int]:
+        return []
+
+    # should be rejected by mypy!
+    # todo how to actually test it?
+    # @warn_if_empty
+    # def baad() -> float:
+    #     return 0.00
+
+    # reveal_type(nonempty)
+    # reveal_type(empty)
+
+    with warnings.catch_warnings(record=True) as w:
+        assert list(nonempty()) == ['a', 'aba']
+        assert len(w) == 0
+
+        eee = empty()
+        assert eee == []
+        assert len(w) == 1
+
+
+def test_warn_iterable() -> None:
+    from my.core.common import _warn_iterable
+    i1: List[str] = ['a', 'b']
+    i2: Iterable[int] = iter([1, 2, 3])
+    # reveal_type(i1)
+    # reveal_type(i2)
+    x1 = _warn_iterable(i1)
+    x2 = _warn_iterable(i2)
+    # vvvv this should be flagged by mypy
+    # _warn_iterable(123)
+    # reveal_type(x1)
+    # reveal_type(x2)
+    with warnings.catch_warnings(record=True) as w:
+        assert x1 is i1 # should be unchanged!
+        assert len(w) == 0
+
+        assert list(x2) == [1, 2, 3]
+        assert len(w) == 0