From 966023be17f2d288188d351cd42f159717a4d090 Mon Sep 17 00:00:00 2001
From: rubenwardy <rw@rubenwardy.com>
Date: Mon, 26 Feb 2024 01:18:38 +0000
Subject: [PATCH] Fix textdomain validation in .tr parser

---
 app/tasks/minetestcheck/translation.py |  8 +++++++-
 app/tests/unit/bad_args.fr.tr          |  1 +
 app/tests/unit/bad_escape.fr.tr        |  1 +
 app/tests/unit/test_translation.py     | 17 ++++++++---------
 app/tests/unit/unknown_arg.fr.tr       |  1 +
 5 files changed, 18 insertions(+), 10 deletions(-)

diff --git a/app/tasks/minetestcheck/translation.py b/app/tasks/minetestcheck/translation.py
index 86448a8e..5199f60c 100644
--- a/app/tasks/minetestcheck/translation.py
+++ b/app/tasks/minetestcheck/translation.py
@@ -38,6 +38,7 @@ def parse_tr(filepath: str) -> Translation:
 	assert filename_parts[-1] == "tr"
 	language = filename_parts[-2]
 	textdomain = ".".join(filename_parts[0:-2])
+	had_textdomain_comment = False
 
 	with open(filepath, "r", encoding="utf-8") as existing_file:
 		lines = existing_file.readlines()
@@ -54,9 +55,14 @@ def parse_tr(filepath: str) -> Translation:
 				# discard all subsequent textdomain lines
 				if line.startswith("# textdomain:"):
 					line_textdomain = line[13:].strip()
+					had_textdomain_comment = True
 					if line_textdomain != textdomain:
 						raise SyntaxError(
 							f"Line {line_index + 1}: The filename's textdomain ({textdomain}) should match the comment ({line_textdomain})")
+
+			elif not had_textdomain_comment:
+				raise SyntaxError(f"Missing `# textdomain: {textdomain}` at the top of the file")
+
 			else:
 				i = 0
 				had_equals = False
@@ -84,7 +90,7 @@ def parse_tr(filepath: str) -> Translation:
 									if int(code) != next_variable:
 										raise SyntaxError(
 											f"Line {line_index + 1}: Arguments out of order in source, found @{code} and expected @{next_variable}." +
-											"Arguments in source must be in increasing order, without gaps or repetitions, starting from 1")
+											" Arguments in source must be in increasing order, without gaps or repetitions, starting from 1")
 
 									next_variable += 1
 							else:
diff --git a/app/tests/unit/bad_args.fr.tr b/app/tests/unit/bad_args.fr.tr
index 358f6d65..56a9a082 100644
--- a/app/tests/unit/bad_args.fr.tr
+++ b/app/tests/unit/bad_args.fr.tr
@@ -1 +1,2 @@
+# textdomain: bad_args
 Some @1 args @5=Some @1 args @5
diff --git a/app/tests/unit/bad_escape.fr.tr b/app/tests/unit/bad_escape.fr.tr
index 4fcedc07..f6e54fb7 100644
--- a/app/tests/unit/bad_escape.fr.tr
+++ b/app/tests/unit/bad_escape.fr.tr
@@ -1 +1,2 @@
+# textdomain: bad_escape
 Bad @x escape = Bad @x escape
diff --git a/app/tests/unit/test_translation.py b/app/tests/unit/test_translation.py
index c34f58bd..665eb5f3 100644
--- a/app/tests/unit/test_translation.py
+++ b/app/tests/unit/test_translation.py
@@ -20,15 +20,14 @@ def test_parses_tr():
 	assert out.entries["Maybe @\n@=@"] == "Peut être @\n@=@"
 
 
-def test_parses_tr_infers_textdomain():
+def test_parses_tr_error_on_missing_textdomain():
 	dirname = os.path.dirname(__file__)
 	filepath = os.path.join(dirname, "no_textdomain_comment.fr.tr")
-	out = parse_tr(filepath)
 
-	assert out.language == "fr"
-	assert out.textdomain == "no_textdomain_comment"
-	assert len(out.entries) == 1
-	assert out.entries["Hello, World!"] == "Bonjour, Monde!"
+	with pytest.raises(SyntaxError) as e:
+		parse_tr(filepath)
+
+	assert str(e.value) == "Missing `# textdomain: no_textdomain_comment` at the top of the file"
 
 
 def test_parses_tr_error_on_textdomain_mismatch():
@@ -58,7 +57,7 @@ def test_parses_tr_error_on_bad_escape():
 	with pytest.raises(SyntaxError) as e:
 		parse_tr(filepath)
 
-	assert str(e.value) == "Line 1: Unknown escape character: x"
+	assert str(e.value) == "Line 2: Unknown escape character: x"
 
 
 def test_parses_tr_error_on_bad_args():
@@ -68,7 +67,7 @@ def test_parses_tr_error_on_bad_args():
 	with pytest.raises(SyntaxError) as e:
 		parse_tr(filepath)
 
-	assert "Line 1: Arguments out of order in source, found @5 and expected @2." in str(e.value)
+	assert "Line 2: Arguments out of order in source, found @5 and expected @2." in str(e.value)
 
 
 def test_parses_tr_error_on_unknown_arg():
@@ -78,4 +77,4 @@ def test_parses_tr_error_on_unknown_arg():
 	with pytest.raises(SyntaxError) as e:
 		parse_tr(filepath)
 
-	assert str(e.value) == "Line 1: Unknown argument @2 in translated string"
+	assert str(e.value) == "Line 2: Unknown argument @2 in translated string"
diff --git a/app/tests/unit/unknown_arg.fr.tr b/app/tests/unit/unknown_arg.fr.tr
index efe6fcdb..c403030b 100644
--- a/app/tests/unit/unknown_arg.fr.tr
+++ b/app/tests/unit/unknown_arg.fr.tr
@@ -1 +1,2 @@
+# textdomain: unknown_arg
 Some @1 arg=Some @2 arg