Fix em-dash prose punctuation across messages and docs

CHANGELOG.md

@@ -17,7 +17,7 @@ A fix release. The first one is the reason to upgrade.
 
 - **The review screen pinned a CPU core while it waited for you.** The
   progress screens poll for a keypress so they can notice `q` mid-run,
-  and that non-blocking mode stayed on the window afterwards —
+  and that non-blocking mode stayed on the window afterwards:
   `curses.wrapper` does not reset it between screens. The review list,
   which expects to wait for input, read "no key pressed" instead and
   redrew immediately: about 17,000 times a second, for as long as the
@@ -29,8 +29,8 @@ A fix release. The first one is the reason to upgrade.
   decision only otherwise, so a row you skipped that later matched on a
   `--rematch` went to Lidarr anyway, and a re-pick lost to the candidate
   the matcher had led with. A decision now outranks the match result. A
-  matched row carrying no release group id — possible in a state file
-  from an older version — is left out instead of failing at Lidarr's
+  matched row carrying no release group id (possible in a state file
+  from an older version) is left out instead of failing at Lidarr's
   lookup.
 - **One failed search request stranded the albums behind it.** Searches
   go out in batches of a hundred and a failing batch stopped the rest, so
@@ -41,7 +41,7 @@ A fix release. The first one is the reason to upgrade.
 - **A Lidarr URL with credentials in it appeared in error messages.**
   `http://user:pw@host` is how you get through a reverse proxy that asks
   for basic auth, and the whole URL was quoted back in connection errors,
-  timeouts, non-JSON replies and proxy error pages — the kind of text
+  timeouts, non-JSON replies and proxy error pages, the kind of text
   that ends up pasted into a bug report. The password is stripped from
   what's printed; the request still sends it.
 
@@ -51,8 +51,8 @@ A fix release. The first one is the reason to upgrade.
   be permanent for that run: the state file has always understood a
   cleared decision, but nothing could write one.
 - **A push without `--search` now says how many albums are monitored but
-  idle.** Monitoring is not downloading — Lidarr picks monitored albums
-  up on its own schedule — and a run ending "added 40" while nothing
+  idle.** Monitoring is not downloading (Lidarr picks monitored albums
+  up on its own schedule), and a run ending "added 40" while nothing
   downloads reads like a finished job.
 
 ### Changed
@@ -79,15 +79,15 @@ Match a CSV of albums against MusicBrainz, resolve the uncertain ones in a
 review screen, and add the results to Lidarr as monitored albums. Runs in
 three resumable stages and can be stopped and restarted at any point.
 
-Notes for anyone who ran this from git before the release — three fixes
+Notes for anyone who ran this from git before the release. Three fixes
 changed behaviour you may have been affected by:
 
 - **Albums pushed to Lidarr stayed unmonitored.** The artist was added with
   `addOptions.monitor = "none"`, which makes Lidarr unmonitor the artist and
   then, once its background scan finishes, every album of theirs including
   the one just pushed. Verified against Lidarr 3.1.0.4875: a fresh push
   reported "added" and fifteen seconds later nothing was monitored. If you
-  pushed a chart with an earlier build, check Lidarr — those albums are
+  pushed a chart with an earlier build, check Lidarr: those albums are
   probably sitting there unmonitored, and re-running chartarr will fix them.
 - **Editing the CSV between runs mixed up the matches.** Rows were keyed by
   position, so inserting a line at the top shifted every key and each album

chartarr/__init__.py

@@ -1,3 +1,3 @@
-"""chartarr — feed your album charts to Lidarr."""
+"""chartarr - feed your album charts to Lidarr."""
 
 __version__ = "0.1.1"

chartarr/cli.py

@@ -67,7 +67,7 @@ def status(line: str) -> None:
     if not sys.stdout.isatty():
         return
     width = shutil.get_terminal_size().columns - 1
-    # truncate and pad by display cells, not code points — a cjk label is
+    # truncate and pad by display cells, not code points: a cjk label is
     # wider than len() says, and overflowing the row garbles the redraw
     line = screen._fit(line, width)
     sys.stdout.write("\r" + line + " " * max(0, width - screen.cells(line)))
@@ -119,13 +119,13 @@ def _ask(prompt: str, secret: bool = False) -> str:
         return (getpass.getpass(prompt) if secret else input(prompt)).strip()
     except (EOFError, KeyboardInterrupt):
         print(file=sys.stderr)  # step off the interrupted prompt line
-        fail("setup cancelled — nothing was saved")
+        fail("setup cancelled; nothing was saved")
         return ""  # unreachable; fail() exits
 
 
 def setup_wizard(existing: dict) -> dict:
     if not sys.stdin.isatty():
-        fail("lidarr isn't set up and there's no terminal to ask on — set "
+        fail("lidarr isn't set up and there's no terminal to ask on; set "
              "LIDARR_URL and LIDARR_API_KEY, or run chartarr --setup in a terminal")
     default = existing.get("lidarr_url", "http://localhost:8686")
     url = _ask(f"lidarr url [{default}]: ") or default
@@ -233,19 +233,19 @@ def load_csv(path: Path):
     # python 3.12 stopped rejecting those in the csv reader, so the file
     # parses into gibberish column names instead of failing; say what it is.
     if b"\x00" in head:
-        fail(f'{path} looks like utf-16 (excel\'s "unicode text" export) — '
+        fail(f'{path} looks like utf-16 (excel\'s "unicode text" export); '
              're-save it as "csv utf-8" and rerun')
     try:
         with path.open(newline="", encoding="utf-8-sig") as f:
             reader = csv.DictReader(f)
             rows = list(reader)
             cols = reader.fieldnames or []
     except UnicodeDecodeError:
-        fail(f"{path} isn't utf-8 — re-save it as utf-8 "
+        fail(f"{path} isn't utf-8; re-save it as utf-8 "
              '(in excel: "csv utf-8") and rerun')
     except csv.Error as e:
-        hint = (' — excel\'s "unicode text" export is utf-16, which does '
-                'this; re-save as "csv utf-8" and rerun'
+        hint = (' (excel\'s "unicode text" export is utf-16, which does '
+                'this; re-save as "csv utf-8" and rerun)'
                 if "NUL" in str(e) else "")
         fail(f"can't parse {path}: {e}{hint}")
     except OSError as e:
@@ -321,12 +321,12 @@ def events():
         # don't promise saved progress when the first lookup was the one
         # that failed; there is nothing to resume and saying so is a lie
         done = sum(counts.values()) - sum(base.values())
-        fail("musicbrainz stopped answering — "
+        fail("musicbrainz stopped answering; "
              + (f"the {_n(done, 'row')} matched so far {'is' if done == 1 else 'are'} "
                 "saved, rerun to pick up the rest" if done
                 else "nothing was matched, so nothing was saved; try again later"))
     if stopped:
-        print(dim("stopped — progress is saved, rerun to resume"))
+        print(dim("stopped; progress is saved, rerun to resume"))
         sys.exit(0)
 
 
@@ -338,7 +338,7 @@ def stage_review(rows, artist_col, title_col, state: State) -> None:
     if not pending:
         return
     if not (sys.stdin.isatty() and sys.stdout.isatty()):
-        print(dim(f"{len(pending)} rows need review — rerun in a terminal, "
+        print(dim(f"{len(pending)} rows need review; rerun in a terminal, "
                   f"or use --yes to push without them"))
         return
     review.run(pending, artist_col, title_col, state.add_decision)
@@ -356,7 +356,7 @@ def import_set(rows, state: State) -> list[dict]:
 
     a decision outranks the match result. review only offers uncertain
     rows, but a row can be decided and then match cleanly on a --rematch,
-    and the state file is editable — either way the answer the user gave
+    and the state file is editable; either way the answer the user gave
     is the one they meant, so an explicit skip is honoured on a matched
     row and a re-pick replaces the automatic choice.
     """
@@ -472,14 +472,14 @@ def events():
     elif touched and not args.dry_run:
         # without --search the albums sit there monitored and idle, which
         # looks like a finished job that downloaded nothing
-        print(dim(f"{_n(len(touched), 'album')} monitored but not searched — "
+        print(dim(f"{_n(len(touched), 'album')} monitored but not searched; "
                   f"rerun with --search, or hit Search in lidarr"))
     for f_ in failures[:8]:
         print(dim(f"  {f_}"))
     if len(failures) > 8:
         print(dim(f"  … and {len(failures) - 8} more"))
     if stopped:
-        print(dim("stopped — the push is safe to rerun"))
+        print(dim("stopped; the push is safe to rerun"))
         sys.exit(0)
 
 
@@ -542,10 +542,10 @@ def main(argv=None) -> None:
         try:
             _main(build_parser().parse_args(argv))
         except KeyboardInterrupt:
-            # covers ctrl-c anywhere — csv loading, state replay, a wizard
-            # network call — not just the matching/push loops
+            # covers ctrl-c anywhere (csv loading, state replay, a wizard
+            # network call), not just the matching/push loops
             status_end()
-            print(dim("stopped — progress is saved, rerun to resume"))
+            print(dim("stopped; progress is saved, rerun to resume"))
             sys.exit(130)
         finally:
             # flush inside the try so a closed pipe surfaces here as a
@@ -568,7 +568,7 @@ def _main(args) -> None:
         if p.exists():
             fail("sample.csv already exists here")
         p.write_text(EXAMPLE_CSV, encoding="utf-8")
-        print("wrote sample.csv — try: chartarr sample.csv --dry-run")
+        print("wrote sample.csv; try: chartarr sample.csv --dry-run")
         return
 
     if args.demo:

chartarr/demo.py

@@ -124,7 +124,7 @@ def match_events():
               f"review {cli.accent(counts.get('review', 0))} · "
               f"not found {cli.accent(counts.get('not_found', 0))}")
         if stopped:
-            print(cli.dim("demo over — nothing was saved or sent"))
+            print(cli.dim("demo over; nothing was saved or sent"))
             return
 
         # review, for real
@@ -163,4 +163,4 @@ def push_events():
     except KeyboardInterrupt:
         print()
 
-    print(cli.dim("demo over — nothing was saved or sent"))
+    print(cli.dim("demo over; nothing was saved or sent"))

chartarr/lidarr.py

@@ -30,7 +30,7 @@ def __init__(self, message: str, status: int = 0, body: str = ""):
 def _safe_url(url: str) -> str:
     """the url with any user:password@ removed, for showing in messages.
 
-    a lidarr url can carry basic-auth credentials — behind a reverse proxy
+    a lidarr url can carry basic-auth credentials: behind a reverse proxy
     that asks for them, http://user:pw@host is how you get through. those
     end up in every connection error otherwise, and errors get pasted into
     bug reports.
@@ -98,15 +98,15 @@ def _call(self, path: str, method: str = "GET", **kw):
             r = self.s.request(method, url, timeout=60, **kw)
         except requests.ConnectionError as e:
             raise LidarrError(
-                f"Can't reach Lidarr at {self.shown} — is it running, and is the "
+                f"Can't reach Lidarr at {self.shown}; is it running, and is the "
                 f"URL right? (the address you use in your browser)") from e
         except requests.Timeout as e:
             raise LidarrError(f"Lidarr at {self.shown} timed out.") from e
         except (requests.exceptions.InvalidSchema,
                 requests.exceptions.MissingSchema,
                 requests.exceptions.InvalidURL) as e:
             raise LidarrError(
-                f"{self.shown} is not a URL Lidarr can be reached at — it needs "
+                f"{self.shown} is not a URL Lidarr can be reached at; it needs "
                 f"to start with http:// or https://") from e
         except requests.RequestException as e:
             # requests quotes the url it was given, credentials and all
@@ -128,7 +128,7 @@ def _call(self, path: str, method: str = "GET", **kw):
             return r.json()
         except ValueError as e:
             raise LidarrError(
-                f"Lidarr returned something that isn't JSON — is {self.shown} "
+                f"Lidarr returned something that isn't JSON; is {self.shown} "
                 f"really Lidarr, and not a login page or another service?") from e
 
     def status(self) -> dict:
@@ -179,8 +179,8 @@ def add_album(self, rgid: str, quality_profile_id: int,
 
         adding does not search. lidarr's addOptions.searchForNewAlbum is
         read by SearchForRecentlyAdded, which ArtistScannedHandler only
-        reaches for an artist with no AddOptions — never the artist this
-        add just created — and the handler clears AddOptions on its way
+        reaches for an artist with no AddOptions (never the artist this
+        add just created), and the handler clears AddOptions on its way
         out, so the flag is stored and then dropped (Lidarr#5012). the
         caller searches explicitly with search_albums instead.
         """
@@ -194,7 +194,7 @@ def add_album(self, rgid: str, quality_profile_id: int,
         album = self.lookup(rgid)
         if album is None:
             raise LidarrError(
-                "Lidarr's lookup didn't find this MusicBrainz ID — it needs a "
+                "Lidarr's lookup didn't find this MusicBrainz ID; it needs a "
                 "release group ID, not a release ID")
         wanted = [rgid] + [r for r in (also_monitor or []) if r != rgid]
         artist = album["artist"]

chartarr/matcher.py

@@ -163,7 +163,7 @@ def _first_artist(rg: dict) -> tuple[str | None, str]:
 # same-titled "Fleetwood Mac" rgs). each secondary type costs clearly more
 # than the Album bump (0.03) so a clean studio album always outranks a
 # same-titled flavoured one; ties otherwise favour the fewest secondary
-# types. the penalty lives in the sort key ONLY — confidence/title_sim are
+# types. the penalty lives in the sort key ONLY: confidence/title_sim are
 # untouched, so a legitimately live/soundtrack chart entry that wins its row
 # anyway ("The Last Waltz", "Purple Rain") still clears the auto-match gate.
 _SECONDARY_PENALTY = 0.05
@@ -289,7 +289,7 @@ def absorb(data: dict, from_alias: bool = False) -> None:
     # alias pass: when the winner is either unproven (fails the auto-match
     # gates) or a flavoured release group the row did not ask for, spend a
     # few lookups confirming aliases for candidates an alias could still
-    # save — right artist, unproven title. this is what lets "Blackstar"
+    # save: right artist, unproven title. this is what lets "Blackstar"
     # auto-match ★ instead of stalling on "Blackstar Radio Edits".
     hinted = _hinted_types(t_vars)
 

chartarr/review.py

@@ -21,7 +21,7 @@ def run(items, artist_col, title_col, on_decision):
         return
     if not available():
         print(f"this terminal can't draw the review screen "
-              f"(TERM={os.environ.get('TERM', '')!r}) — rerun in a working "
+              f"(TERM={os.environ.get('TERM', '')!r}); rerun in a working "
               f"terminal, or use --yes to push just the confident matches")
         return
     _run(_loop, items, artist_col, title_col, on_decision)

chartarr/screen.py

@@ -22,7 +22,7 @@ def available() -> bool:
     back to the plain-line output instead of crashing.
 
     note: cpython caches setupterm's first success process-wide, so this
-    answers for the TERM the process started with — which is the one that
+    answers for the TERM the process started with, which is the one that
     matters. tests that flip TERM must probe in a subprocess.
     """
     if curses is None:
@@ -43,7 +43,7 @@ def _run(func, *args):
 
     # a utf-8 locale makes ncurses draw wide characters correctly, but an
     # LC_ALL this box doesn't know (classic ssh-forwarded locale) must not
-    # kill the screen — degrade toward the C locale instead
+    # kill the screen; degrade toward the C locale instead
     try:
         locale.setlocale(locale.LC_ALL, "")
     except locale.Error:
@@ -56,7 +56,7 @@ def entry(scr, *inner):
         # every session starts blocking. cpython keeps nodelay on the window
         # itself and curses.wrapper does not reset it between sessions, so
         # the nodelay(True) _progress needs would carry into the review loop
-        # that follows it — whose getch() would then return -1 forever and
+        # that follows it, whose getch() would then return -1 forever and
         # redraw at ~17k frames a second on a pinned core. a screen that
         # wants non-blocking input asks for it, as _progress does.
         try:
@@ -75,7 +75,7 @@ def _cell(ch: str) -> int:
 
 
 def cells(s: str) -> int:
-    """display cells s occupies — the terminal lays out by cell, not code
+    """display cells s occupies: the terminal lays out by cell, not code
     point, and cjk take two, so len() undercounts exactly on the
     dual-script charts this tool is for."""
     return sum(_cell(ch) for ch in s)
@@ -128,7 +128,7 @@ def _dark_background() -> bool:
 
     COLORFGBG is what terminals that care about this set (rxvt, konsole,
     some terminfo-aware setups); "15;0" means light text on dark. absent
-    it, assume dark — the common case, and both shades are readable there.
+    it, assume dark: the common case, and both shades are readable there.
     """
     fgbg = os.environ.get("COLORFGBG", "")
     if ";" in fgbg:

tests/test_cli.py

@@ -335,7 +335,7 @@ def half(rows, artist_col, title_col):
 
 def test_review_choice_beats_the_auto_match_and_skips_are_dropped(tmp_path):
     # the reviewed pick has to reach lidarr, not the candidate the matcher
-    # led with — that's the whole point of the review screen
+    # led with; that's the whole point of the review screen
     p = _csv(tmp_path, "title,artist\nZiggy Stardust,David Bowie\n"
                        "Kid A,Radiohead\nDummy,Portishead\n")
     rows, _, _ = load_csv(p)

tests/test_lidarr.py

@@ -108,7 +108,7 @@ def test_also_monitor_names_every_sibling_once():
 def test_adding_never_asks_lidarr_to_search():
     # searchForNewAlbum is read by SearchForRecentlyAdded, which
     # ArtistScannedHandler only reaches when the artist has no AddOptions
-    # — never the artist this add just created — and the handler clears
+    # (never the artist this add just created), and the handler clears
     # AddOptions on the way out, so the flag is stored and dropped
     # (Lidarr#5012). searching is the caller's job, via AlbumSearch.
     stage_lookup()
@@ -182,7 +182,7 @@ def test_duplicate_row_already_monitored_is_skipped():
 @responses.activate
 def test_config_400_is_an_error_not_a_duplicate():
     # the original client sniffed the body for "exist", so "Quality Profile
-    # does not exist" — a config mistake — took the duplicate path and died
+    # does not exist" (a config mistake) took the duplicate path and died
     # with a baffling "conflict but album not found afterwards". it must
     # surface as the validation error it is.
     stage_lookup()
@@ -191,7 +191,7 @@ def test_config_400_is_an_error_not_a_duplicate():
                          "errorMessage": "Quality Profile does not exist"}])
     with pytest.raises(LidarrError, match="Quality Profile"):
         add()
-    assert len(responses.calls) == 3  # find, lookup, post — no duplicate dance
+    assert len(responses.calls) == 3  # find, lookup, post; no duplicate dance
 
 
 @responses.activate

tests/test_matcher.py

@@ -151,7 +151,7 @@ def test_hint_waives_only_the_named_type():
 def test_alias_counts_as_title():
     # mb titles bowie's Blackstar "★"; the name every chart uses is only an
     # alias, and releasegroup:"Blackstar" returns just "Blackstar Radio
-    # Edits" — the aliased album must outscore it
+    # Edits"; the aliased album must outscore it
     cands = score_rgs(
         [rg("Blackstar Radio Edits", "David Bowie", "radio-id", ptype="Single"),
          rg("★", "David Bowie", "star-id", aliases=["Blackstar", "★ (Blackstar)"])],