feat: g2p convert now accepts --file option to read a file

This replaces the formerly hidden feature of heuristically detecting
existing .txt files.

Also fix test_update_schema to:
 - be quiet unless there's an error
 - correctly catch errors and display the problem filename when there is
   an error.

g2p/cli.py

@@ -441,6 +441,13 @@ def generate_mapping(  # noqa: C901
     is_flag=True,
     help="Show all the conversion steps applied.",
 )
+@click.option(
+    "--file",
+    "-f",
+    default=False,
+    is_flag=True,
+    help="Read input from plain-text file INPUT_TEXT.",
+)
 @click.option(
     "--check/--no-check",
     "-c",
@@ -478,6 +485,7 @@ def convert(  # noqa: C901
     input_text,
     tok,
     check,
+    file,
     debugger,
     pretty_edges,
     tok_lang,
@@ -531,20 +539,23 @@ def convert(  # noqa: C901
         raise click.UsageError(
             f"Path between '{in_lang}' and '{out_lang}' does not exist"
         )
-    # Figure if the input text is on the command line or in a file
-    input_text_is_a_file = (
-        os.path.exists(input_text)
-        and input_text.endswith("txt")
-        or re.match(r"/dev/(fd/[0-9]*|stdin)", input_text)
-    )
     to_close = None
     try:
-        if input_text_is_a_file:
-            to_close = lines = open(input_text, encoding="utf8")
-        elif input_text == "-":
-            lines = sys.stdin
+        if file:
+            if input_text == "-":
+                lines = sys.stdin
+            else:
+                try:
+                    to_close = lines = open(input_text, encoding="utf8")
+                except FileNotFoundError as e:
+                    raise click.UsageError(f"Could not open file {input_text}: {e}")
         else:
             lines = [input_text]
+            if os.path.exists(input_text) and input_text.endswith(".txt"):
+                LOGGER.warning(
+                    "The old heuristic for detecting file input is deprecated, "
+                    f"specify --file if you meant to read your input text from file {input_text}."
+                )
         # Determine which tokenizer to use, if any
         if tok is not None and not tok and tok_lang is not None:
             raise click.UsageError(
@@ -572,7 +583,7 @@ def convert(  # noqa: C901
             if len(outputs) > 1:
                 click.echo(pprint.pformat(outputs, indent=4))
             else:
-                click.echo(tg.output_string, nl=not input_text_is_a_file)
+                click.echo(tg.output_string, nl=not file)
     finally:
         if to_close is not None:
             to_close.close()

g2p/tests/test_cli.py

@@ -11,7 +11,6 @@
 import jsonschema
 import yaml
 from click.testing import CliRunner
-from tqdm import tqdm
 
 from g2p._version import VERSION
 from g2p.cli import (
@@ -122,13 +121,14 @@ def test_update_schema(self):
                 encoding="utf8",
             ) as f:
                 schema = json.load(f)
-        for config in tqdm(
-            Path(LANGS_DIR).glob("**/config-g2p.yaml"),
-            desc="Validating all configurations against current schema",
-        ):
+        # Validate all configurations against the current schema, quietly unless there's an error:
+        for config in Path(LANGS_DIR).glob("**/config-g2p.yaml"):
             with open(config, encoding="utf8") as f:
                 config_yaml = yaml.safe_load(f)
-            self.assertIsNone(jsonschema.validate(config_yaml, schema=schema))
+            try:
+                jsonschema.validate(config_yaml, schema=schema)
+            except jsonschema.exceptions.ValidationError:
+                self.fail(f"Error validating {config}")
 
     def test_convert(self):
         langs_to_test = load_public_test_data()
@@ -490,7 +490,7 @@ def test_show_mappings(self):
 
     def test_convert_from_file(self):
         input_file = os.path.join(DATA_DIR, "fra_simple.txt")
-        results = self.runner.invoke(convert, [input_file, "fra", "fra-ipa"])
+        results = self.runner.invoke(convert, [input_file, "fra", "fra-ipa", "--file"])
         self.assertEqual(results.exit_code, 0)
         self.assertIn("fʁɑ̃sɛ", results.output)
         with open(input_file, "r", encoding="utf8") as f:
@@ -501,6 +501,25 @@ def test_convert_from_file(self):
         # The output should have the same number of lines as the input
         self.assertEqual(lines_in, len(results.output.splitlines()))
 
+        # - is stdin
+        results = self.runner.invoke(
+            convert, ["--file", "-", "fra", "fra-ipa"], input="français"
+        )
+        self.assertEqual(results.exit_code, 0)
+        self.assertIn("fʁɑ̃sɛ", results.output)
+
+        # warning about deprecated heuristic file detection
+        with self.assertLogs(LOGGER, "WARNING") as cm:
+            self.runner.invoke(convert, [input_file, "fra", "fra-ipa"])
+        self.assertIn("deprecated", "".join(cm.output))
+
+        # Error for --file with non existent file
+        results = self.runner.invoke(
+            convert, ["does_not_exist.txt", "fra", "fra-ipa", "--file"]
+        )
+        self.assertNotEqual(results.exit_code, 0)
+        self.assertIn("No such file or directory", results.output)
+
     def test_convert_errors(self):
         """Exercise code handling error situations in g2p convert"""
         results = self.runner.invoke(convert, "asdf bad_in_lang eng-ipa")