From 940239c516fe0fb2d7804cf3297a91aebcec1f9d Mon Sep 17 00:00:00 2001
From: Gunnar Kreitz <gkreitz@kattis.com>
Date: Wed, 2 Sep 2026 13:52:46 +0200
Subject: [PATCH 1/6] Use fixture Diagnostics in verify_hello, and add a few
 more asserts

---
 tests/test_verify_hello.py | 26 ++++++++++++--------------
 1 file changed, 12 insertions(+), 14 deletions(-)

diff --git a/tests/test_verify_hello.py b/tests/test_verify_hello.py
index 95a48935..393f9074 100644
--- a/tests/test_verify_hello.py
+++ b/tests/test_verify_hello.py
@@ -1,18 +1,12 @@
-import logging
-import pathlib
+from pathlib import Path
 
 from problemtools import checks, model
-from problemtools.diagnostics import LoggingDiagnostics
+from problemtools.diagnostics import Diagnostics
 
 
-def _make_diag(shortname: str) -> LoggingDiagnostics:
-    return LoggingDiagnostics.create(shortname, log_level=logging.WARNING)
+def test_load_hello(diag: Diagnostics) -> None:
+    probdir = (Path(__file__).parent / 'hello').resolve()
 
-
-def test_load_hello():
-    probdir = (pathlib.Path(__file__).parent / 'hello').resolve()
-
-    diag = _make_diag('hello')
     problem = model.load_problem(probdir, diag)
     assert problem.shortname == 'hello'
 
@@ -27,9 +21,13 @@ def test_load_hello():
     assert not problem.metadata.is_multi_pass()
     assert not problem.metadata.is_submit_answer()
 
+    assert len(problem.testdata.get_all_testcases()) == 1, 'Hello should have exactly 1 test case'
+    assert problem.graders.grader is None, 'Hello uses the default grader'
+    assert problem.output_validators.uses_default(problem.format_version, problem.metadata), 'Hello uses the default validator'
+
 
-def test_load_twice():
-    probdir = (pathlib.Path(__file__).parent / 'hello').resolve()
+def test_load_twice(diag: Diagnostics):
+    probdir = (Path(__file__).parent / 'hello').resolve()
 
-    model.load_problem(probdir, _make_diag('hello'))
-    model.load_problem(probdir, _make_diag('hello'))
+    model.load_problem(probdir, diag)
+    model.load_problem(probdir, diag)

From f990f9f9d66f23fab26af413be11ecac892f0721 Mon Sep 17 00:00:00 2001
From: Gunnar Kreitz <gkreitz@kattis.com>
Date: Wed, 2 Sep 2026 14:03:33 +0200
Subject: [PATCH 2/6] Extract directory computation in tests to pepare to move
 data

---
 tests/conftest.py       | 14 ++++++++++++++
 tests/test_config.py    |  5 ++---
 tests/test_languages.py |  3 ++-
 tests/test_latex.py     | 17 +++++++++--------
 tests/test_markdown.py  | 17 ++++++++---------
 tests/test_xss.py       |  8 ++++----
 6 files changed, 39 insertions(+), 25 deletions(-)

diff --git a/tests/conftest.py b/tests/conftest.py
index 1ea67d38..4762dcc1 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -1,10 +1,24 @@
 """Shared pytest fixtures and test doubles."""
 
+from pathlib import Path
+
 import pytest
 
 from problemtools.diagnostics import Diagnostics
 
 
+def datadir() -> Path:
+    """Root directory holding static test fixture data."""
+    # TODO: move config1, config2, languages_examples, and problems under a
+    # tests/data directory, and change this to Path(__file__).parent / 'data'.
+    return Path(__file__).parent.resolve()
+
+
+def example_directory(problem_name: str) -> Path:
+    """Path to one of the example problems shipped in the repo's top-level examples/ directory."""
+    return (Path(__file__).parent.parent / 'examples' / problem_name).resolve()
+
+
 class RecordingDiagnostics(Diagnostics):
     """A Diagnostics that records messages instead of emitting them, for asserting on in tests."""
 
diff --git a/tests/test_config.py b/tests/test_config.py
index c9d60e5d..49a40979 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -1,12 +1,11 @@
 import pytest
 
 from problemtools import config
+from tests.conftest import datadir
 
 
 def config_paths_mock():
-    from pathlib import Path
-
-    return [Path(__file__).parent / 'config1', Path(__file__).parent / 'config2']
+    return [datadir() / 'config1', datadir() / 'config2']
 
 
 def test_load_basic_config(monkeypatch):
diff --git a/tests/test_languages.py b/tests/test_languages.py
index e0255562..80c6a4c0 100644
--- a/tests/test_languages.py
+++ b/tests/test_languages.py
@@ -6,6 +6,7 @@
 import pytest
 
 from problemtools import languages
+from tests.conftest import datadir
 
 
 class Language_test(TestCase):
@@ -295,7 +296,7 @@ def test_get_compile_command_splits_multiple_files(self):
         assert command == [shutil.which('echo'), subs.path, 'main.foo', 'helper1.bar', 'helper2.bar', subs.binary]
 
 
-__EXAMPLES_PATH = os.path.join(os.path.dirname(__file__), 'languages_examples')
+__EXAMPLES_PATH = datadir() / 'languages_examples'
 
 
 def examples_path(test_file):
diff --git a/tests/test_latex.py b/tests/test_latex.py
index 41484a12..6f687dec 100644
--- a/tests/test_latex.py
+++ b/tests/test_latex.py
@@ -3,13 +3,14 @@
 from pathlib import Path
 
 from problemtools import problem2html, problem2pdf
+from tests.conftest import example_directory
 
 
 def test_pdf_render_verifyproblem():
     # Same options as in verifyproblem
     options = problem2pdf.get_parser().parse_args([''])
-    problem_path = Path(__file__).parent / '..' / 'examples' / 'guess'
-    options.problem = str(problem_path.resolve())
+    problem_path = example_directory('guess')
+    options.problem = str(problem_path)
     options.language = 'en'
     options.nopdf = True
     options.quiet = True
@@ -20,9 +21,9 @@ def test_pdf_render_verifyproblem():
 def test_pdf_render_problem2pdf():
     # Same options as typical problem2pdf usage
     with tempfile.TemporaryDirectory() as temp_dir:
-        problem_path = Path(__file__).parent / '..' / 'examples' / 'guess'
+        problem_path = example_directory('guess')
         temp_filename = Path(temp_dir) / 'guess.pdf'
-        options = problem2pdf.get_parser().parse_args(['-o', str(temp_filename), '-l', 'en', '-q', str(problem_path.resolve())])
+        options = problem2pdf.get_parser().parse_args(['-o', str(temp_filename), '-l', 'en', '-q', str(problem_path)])
         if not problem2pdf.convert(options):
             assert False, 'PDF conversion failed'
         with open(temp_filename, 'rb') as temp_file:
@@ -32,9 +33,9 @@ def test_pdf_render_problem2pdf():
 def test_html_render_different(diag):
     # Same options as typical problem2html usage
     with tempfile.TemporaryDirectory() as temp_dir:
-        problem_path = Path(__file__).parent / '..' / 'examples' / 'different'
+        problem_path = example_directory('different')
         temp_dir = Path(temp_dir) / 'different_html'
-        options = problem2html.get_parser().parse_args(['-d', str(temp_dir), '-l', 'en', '-q', str(problem_path.resolve())])
+        options = problem2html.get_parser().parse_args(['-d', str(temp_dir), '-l', 'en', '-q', str(problem_path)])
         problem2html.convert(options, diag)
         with open(temp_dir / 'index.html', 'r') as temp_file:
             full_html = temp_file.read()
@@ -48,9 +49,9 @@ def test_html_render_different(diag):
 def test_html_render_guess(diag):
     # Same options as typical problem2html usage
     with tempfile.TemporaryDirectory() as temp_dir:
-        problem_path = Path(__file__).parent / '..' / 'examples' / 'guess'
+        problem_path = example_directory('guess')
         temp_dir = Path(temp_dir) / 'guess_html'
-        options = problem2html.get_parser().parse_args(['-d', str(temp_dir), '-l', 'en', '-q', str(problem_path.resolve())])
+        options = problem2html.get_parser().parse_args(['-d', str(temp_dir), '-l', 'en', '-q', str(problem_path)])
         problem2html.convert(options, diag)
         with open(temp_dir / 'index.html', 'r') as temp_file:
             full_html = temp_file.read()
diff --git a/tests/test_markdown.py b/tests/test_markdown.py
index 71e48521..2357858f 100644
--- a/tests/test_markdown.py
+++ b/tests/test_markdown.py
@@ -1,14 +1,13 @@
-from pathlib import Path
-
 import pytest
 
 from problemtools.statement_util import find_footnotes
+from tests.conftest import datadir
 from tests.test_xss import render, renderpdf
 
 # TODO: add when guess is updated to 2023-07
 # def test_pdf_render():
 #     with tempfile.TemporaryDirectory() as temp_dir:
-#         problem_path = Path(__file__).parent / '..' / 'examples' / 'guess'
+#         problem_path = example_directory('guess')
 #         args, _unknown = problem2pdf.get_parser().parse_known_args(
 #                 ['--problem', str(problem_path.resolve()), '-l', 'sv', '--dest-dir', str(temp_dir)]
 #             )
@@ -16,7 +15,7 @@
 
 
 def test_sample_escaping(diag):
-    problem_path = Path(__file__).parent / 'problems' / 'specialcharacterssample'
+    problem_path = datadir() / 'problems' / 'specialcharacterssample'
     html = render(problem_path, diag)
     all_printable = r"""0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ!"#$%&'()*+,-./:;<=>?@[\]^_`{|}~"""
     # We escape &, < and >
@@ -30,18 +29,18 @@ def test_footnotes(diag):
     # We always want footnotes to be at the bottom
     # When we insert samples, we need to insert them right above the first footnote
     # To do this, we search for a string (very fragile)
-    problem_path = Path(__file__).parent / 'problems' / 'footnote'
+    problem_path = datadir() / 'problems' / 'footnote'
     html = render(problem_path, diag)
     assert find_footnotes(html) is not None
 
-    problem_path = Path(__file__).parent / 'problems' / 'twofootnotes'
+    problem_path = datadir() / 'problems' / 'twofootnotes'
     html = render(problem_path, diag)
     assert find_footnotes(html) is not None
 
 
 def test_footnotes_href(diag):
     # We use allowlist-based id values for footnotes. Ensure they have not changed
-    problem_path = Path(__file__).parent / 'problems' / 'footnote'
+    problem_path = datadir() / 'problems' / 'footnote'
     html = render(problem_path, diag)
     assert 'fn1' in html and 'fnref1' in html
 
@@ -49,10 +48,10 @@ def test_footnotes_href(diag):
 def test_invalid_image_throws(diag):
     # If images can point to img that doesn't exist, it's arbitrary web request
     for problem in ('imgrequest', 'imgrequest2'):
-        problem_path = Path(__file__).parent / 'problems' / problem
+        problem_path = datadir() / 'problems' / problem
         with pytest.raises(ValueError):
             render(problem_path, diag)
 
     # Pandoc won't make a web request for imgrequest2
     with pytest.raises(ValueError):
-        renderpdf(Path(__file__).parent / 'problems' / 'imgrequest')
+        renderpdf(datadir() / 'problems' / 'imgrequest')
diff --git a/tests/test_xss.py b/tests/test_xss.py
index 597ab468..821a91b8 100644
--- a/tests/test_xss.py
+++ b/tests/test_xss.py
@@ -1,8 +1,8 @@
 import os
 import tempfile
-from pathlib import Path
 
 from problemtools import problem2html, problem2pdf
+from tests.conftest import datadir
 
 
 def render(problem_path, diag):
@@ -24,18 +24,18 @@ def renderpdf(problem_path):
 
 
 def test_no_xss_statement(diag):
-    problem_path = Path(__file__).parent / 'problems' / 'statementxss'
+    problem_path = datadir() / 'problems' / 'statementxss'
     html = render(problem_path, diag)
     assert 'alert' not in html
 
 
 def test_no_xss_problemname(diag):
-    problem_path = Path(__file__).parent / 'problems' / 'problemnamexss'
+    problem_path = datadir() / 'problems' / 'problemnamexss'
     html = render(problem_path, diag)
     assert '<script>' not in html
 
 
 def test_no_xss_sample(diag):
-    problem_path = Path(__file__).parent / 'problems' / 'samplexss'
+    problem_path = datadir() / 'problems' / 'samplexss'
     html = render(problem_path, diag)
     assert '<script>' not in html

From 609f3542465ffd04506426c364fdede26af65e35 Mon Sep 17 00:00:00 2001
From: Gunnar Kreitz <gkreitz@kattis.com>
Date: Wed, 2 Sep 2026 14:07:24 +0200
Subject: [PATCH 3/6] Use hello from examples in test suite, instead of
 duplicating it

---
 tests/hello/data/secret/hello.ans             |  1 -
 tests/hello/data/secret/hello.in              |  0
 tests/hello/input_validators/validate.py      |  8 -----
 tests/hello/problem.yaml                      |  9 -----
 tests/hello/problem_statement/problem.en.tex  |  9 -----
 tests/hello/problem_statement/problem.sv.tex  | 10 ------
 tests/hello/submissions/accepted/hello.cc     |  6 ----
 tests/hello/submissions/accepted/hello.java   |  5 ---
 tests/hello/submissions/accepted/hello.kt     |  4 ---
 tests/hello/submissions/accepted/hello.py     |  3 --
 .../hello/submissions/accepted/hello_alarm.c  | 35 -------------------
 .../run_time_error/memory_limit.cc            | 11 ------
 tests/hello/submissions/wrong_answer/hello.cc |  6 ----
 tests/test_metadata.py                        |  5 ++-
 tests/test_verify_hello.py                    |  7 ++--
 15 files changed, 5 insertions(+), 114 deletions(-)
 delete mode 100644 tests/hello/data/secret/hello.ans
 delete mode 100644 tests/hello/data/secret/hello.in
 delete mode 100755 tests/hello/input_validators/validate.py
 delete mode 100644 tests/hello/problem.yaml
 delete mode 100644 tests/hello/problem_statement/problem.en.tex
 delete mode 100644 tests/hello/problem_statement/problem.sv.tex
 delete mode 100644 tests/hello/submissions/accepted/hello.cc
 delete mode 100644 tests/hello/submissions/accepted/hello.java
 delete mode 100644 tests/hello/submissions/accepted/hello.kt
 delete mode 100644 tests/hello/submissions/accepted/hello.py
 delete mode 100644 tests/hello/submissions/accepted/hello_alarm.c
 delete mode 100644 tests/hello/submissions/run_time_error/memory_limit.cc
 delete mode 100644 tests/hello/submissions/wrong_answer/hello.cc

diff --git a/tests/hello/data/secret/hello.ans b/tests/hello/data/secret/hello.ans
deleted file mode 100644
index 980a0d5f..00000000
--- a/tests/hello/data/secret/hello.ans
+++ /dev/null
@@ -1 +0,0 @@
-Hello World!
diff --git a/tests/hello/data/secret/hello.in b/tests/hello/data/secret/hello.in
deleted file mode 100644
index e69de29b..00000000
diff --git a/tests/hello/input_validators/validate.py b/tests/hello/input_validators/validate.py
deleted file mode 100755
index d98eacf9..00000000
--- a/tests/hello/input_validators/validate.py
+++ /dev/null
@@ -1,8 +0,0 @@
-#!/usr/bin/env python3
-import sys
-from sys import stdin
-
-# There shouldn't be any input
-assert len(stdin.readline()) == 0
-
-sys.exit(42)
diff --git a/tests/hello/problem.yaml b/tests/hello/problem.yaml
deleted file mode 100644
index c69bfdc4..00000000
--- a/tests/hello/problem.yaml
+++ /dev/null
@@ -1,9 +0,0 @@
-source: Kattis
-license: public domain
-name: Hello World!
-
-# Fix memory limit at 512 MiB.  (Note that for most problems, this
-# should not be done.  It is only done in this case because we include
-# a test submission that goes over this limit.)
-limits:
-  memory: 512
diff --git a/tests/hello/problem_statement/problem.en.tex b/tests/hello/problem_statement/problem.en.tex
deleted file mode 100644
index 3df9596c..00000000
--- a/tests/hello/problem_statement/problem.en.tex
+++ /dev/null
@@ -1,9 +0,0 @@
-\problemname{Hello World!}
-
-\section*{Input}
-
-There is no input for this problem.
-
-\section*{Output}
-
-Output should contain one line, containing the string ``Hello World!''. 
diff --git a/tests/hello/problem_statement/problem.sv.tex b/tests/hello/problem_statement/problem.sv.tex
deleted file mode 100644
index d021c237..00000000
--- a/tests/hello/problem_statement/problem.sv.tex
+++ /dev/null
@@ -1,10 +0,0 @@
-\problemname{Hej V\"arlden!} % Silly use of LaTeX just to test plainproblemname
-%% plainproblemname: Hej Världen!
-
-\section*{Indata}
-
-Detta problem har inget indata.
-
-\section*{Output}
-
-Utdata ska bestå av en rad, innehållandes strängen ``Hello World!''.
diff --git a/tests/hello/submissions/accepted/hello.cc b/tests/hello/submissions/accepted/hello.cc
deleted file mode 100644
index 9feeee8e..00000000
--- a/tests/hello/submissions/accepted/hello.cc
+++ /dev/null
@@ -1,6 +0,0 @@
-#include <cstdio>
-
-int main(void) {
-  printf("Hello World!\n");
-  return 0;
-}
diff --git a/tests/hello/submissions/accepted/hello.java b/tests/hello/submissions/accepted/hello.java
deleted file mode 100644
index 49243411..00000000
--- a/tests/hello/submissions/accepted/hello.java
+++ /dev/null
@@ -1,5 +0,0 @@
-public class hello {
-    public static void main(String args[]) {
-	System.out.println("Hello World!");
-    }
-}
\ No newline at end of file
diff --git a/tests/hello/submissions/accepted/hello.kt b/tests/hello/submissions/accepted/hello.kt
deleted file mode 100644
index 3dcd543f..00000000
--- a/tests/hello/submissions/accepted/hello.kt
+++ /dev/null
@@ -1,4 +0,0 @@
-fun main(args: Array<String>) {
-  val words = if (args.size == 0) arrayOf("Hello", "World!") else args
-  System.`out`.println(words.joinToString(separator = " "))
-}
diff --git a/tests/hello/submissions/accepted/hello.py b/tests/hello/submissions/accepted/hello.py
deleted file mode 100644
index 9016c9b6..00000000
--- a/tests/hello/submissions/accepted/hello.py
+++ /dev/null
@@ -1,3 +0,0 @@
-#!/usr/bin/env python3
-
-print('Hello World!')
diff --git a/tests/hello/submissions/accepted/hello_alarm.c b/tests/hello/submissions/accepted/hello_alarm.c
deleted file mode 100644
index ba104baa..00000000
--- a/tests/hello/submissions/accepted/hello_alarm.c
+++ /dev/null
@@ -1,35 +0,0 @@
-#include <signal.h>
-#include <stdio.h>
-#include <stdlib.h>
-
-/* Based on the libc manual*/   
-
-/* This flag controls termination of the main loop. */
-volatile sig_atomic_t keep_going = 1;
-     
-/* The signal handler just clears the flag and re-enables itself. */
-void catch_alarm (int sig)
-{
-  keep_going = 0;
-  signal (sig, catch_alarm);
-}
-
-void do_nothing (void)
-{
-  int i=0;
-  for (i=0;i<1000;i+=1);
-}
-
-int main (void)
-{
-  /* Establish a handler for SIGALRM signals. */
-  signal (SIGALRM, catch_alarm);
-  /* Set an alarm to go off in a little while. */
-  alarm (1);
-  /* Check the flag once in a while to see when to quit. */
-  while (keep_going)
-    do_nothing();
-  
-  printf("Hello World!\n");
-  return EXIT_SUCCESS;
-}
diff --git a/tests/hello/submissions/run_time_error/memory_limit.cc b/tests/hello/submissions/run_time_error/memory_limit.cc
deleted file mode 100644
index 6d858755..00000000
--- a/tests/hello/submissions/run_time_error/memory_limit.cc
+++ /dev/null
@@ -1,11 +0,0 @@
-#include <iostream>
-#include <algorithm>
-
-int main(void) {
-  char *buf = new char[512*1024*1024]; // 512MB
-  buf[0] = 0;
-  for (int i = 1; i < 512*1024*1024; ++i)
-      buf[i] = 23*buf[i-1]+42;
-  std::cout << "Hello World!\n" << std::endl;
-  return 0;
-}
diff --git a/tests/hello/submissions/wrong_answer/hello.cc b/tests/hello/submissions/wrong_answer/hello.cc
deleted file mode 100644
index 80ab5c4f..00000000
--- a/tests/hello/submissions/wrong_answer/hello.cc
+++ /dev/null
@@ -1,6 +0,0 @@
-#include <cstdio>
-
-int main(void) {
-  printf("Hello!");
-  return 0;
-}
diff --git a/tests/test_metadata.py b/tests/test_metadata.py
index 70a90e6c..2fd7e58f 100644
--- a/tests/test_metadata.py
+++ b/tests/test_metadata.py
@@ -1,10 +1,9 @@
-from pathlib import Path
-
 import pytest
 from pydantic import ValidationError
 
 from problemtools import metadata
 from problemtools.formatversion import FormatVersion
+from tests.conftest import example_directory
 
 # A few quick tests of config parsing. pytest structure isn't great here, so code gets repetitive, but I wanted *something* basic in place at least.
 
@@ -101,7 +100,7 @@ def test_parse_complex_type(minimal_2023_conf):
 
 
 def test_load_hello():
-    m = metadata.load_metadata(Path(__file__).parent / 'hello')
+    m = metadata.load_metadata(example_directory('hello'))
     assert m.name['en'] == 'Hello World!'
     assert m.name['sv'] == 'Hej Världen!'
     assert len(m.source) == 1
diff --git a/tests/test_verify_hello.py b/tests/test_verify_hello.py
index 393f9074..e0c63768 100644
--- a/tests/test_verify_hello.py
+++ b/tests/test_verify_hello.py
@@ -1,11 +1,10 @@
-from pathlib import Path
-
 from problemtools import checks, model
 from problemtools.diagnostics import Diagnostics
+from tests.conftest import example_directory
 
 
 def test_load_hello(diag: Diagnostics) -> None:
-    probdir = (Path(__file__).parent / 'hello').resolve()
+    probdir = example_directory('hello')
 
     problem = model.load_problem(probdir, diag)
     assert problem.shortname == 'hello'
@@ -27,7 +26,7 @@ def test_load_hello(diag: Diagnostics) -> None:
 
 
 def test_load_twice(diag: Diagnostics):
-    probdir = (Path(__file__).parent / 'hello').resolve()
+    probdir = example_directory('hello')
 
     model.load_problem(probdir, diag)
     model.load_problem(probdir, diag)

From 69698b3c415a3f28e65482eb5fcd2c4f1a75fce5 Mon Sep 17 00:00:00 2001
From: Gunnar Kreitz <gkreitz@kattis.com>
Date: Wed, 2 Sep 2026 14:08:50 +0200
Subject: [PATCH 4/6] Move data used in tests to a data directory

---
 tests/conftest.py                                             | 4 +---
 tests/{ => data}/config1/broken.yaml                          | 0
 tests/{ => data}/config1/test.yaml                            | 0
 tests/{ => data}/config1/test2.yaml                           | 0
 tests/{ => data}/config2/test2.yaml                           | 0
 tests/{ => data}/languages_examples/py2helper.py2             | 0
 tests/{ => data}/languages_examples/py2main.py                | 0
 tests/{ => data}/languages_examples/py2nosheb.py              | 0
 tests/{ => data}/languages_examples/src1.zoo                  | 0
 tests/{ => data}/languages_examples/src2.zoo                  | 0
 tests/{ => data}/languages_examples/src3.zpp                  | 0
 tests/{ => data}/problems/footnote/problem.yaml               | 0
 tests/{ => data}/problems/footnote/statement/problem.en.md    | 0
 tests/{ => data}/problems/imgrequest/problem.yaml             | 0
 tests/{ => data}/problems/imgrequest/statement/problem.en.md  | 0
 tests/{ => data}/problems/imgrequest2/problem.yaml            | 0
 tests/{ => data}/problems/imgrequest2/statement/problem.en.md | 0
 tests/{ => data}/problems/problemnamexss/problem.yaml         | 0
 .../problems/problemnamexss/statement/problem.en.md           | 0
 tests/{ => data}/problems/samplexss/data/sample/1.ans         | 0
 tests/{ => data}/problems/samplexss/data/sample/1.in          | 0
 tests/{ => data}/problems/samplexss/data/sample/testdata.yaml | 0
 tests/{ => data}/problems/samplexss/data/testdata.yaml        | 0
 tests/{ => data}/problems/samplexss/problem.yaml              | 0
 tests/{ => data}/problems/samplexss/statement/problem.en.md   | 0
 .../problems/specialcharacterssample/data/sample/1.ans        | 0
 .../problems/specialcharacterssample/data/sample/1.in         | 0
 .../specialcharacterssample/data/sample/testdata.yaml         | 0
 .../problems/specialcharacterssample/data/testdata.yaml       | 0
 .../{ => data}/problems/specialcharacterssample/problem.yaml  | 0
 .../problems/specialcharacterssample/statement/problem.en.md  | 0
 tests/{ => data}/problems/statementxss/problem.yaml           | 0
 .../{ => data}/problems/statementxss/statement/problem.en.md  | 0
 tests/{ => data}/problems/twofootnotes/problem.yaml           | 0
 .../{ => data}/problems/twofootnotes/statement/problem.en.md  | 0
 35 files changed, 1 insertion(+), 3 deletions(-)
 rename tests/{ => data}/config1/broken.yaml (100%)
 rename tests/{ => data}/config1/test.yaml (100%)
 rename tests/{ => data}/config1/test2.yaml (100%)
 rename tests/{ => data}/config2/test2.yaml (100%)
 rename tests/{ => data}/languages_examples/py2helper.py2 (100%)
 rename tests/{ => data}/languages_examples/py2main.py (100%)
 rename tests/{ => data}/languages_examples/py2nosheb.py (100%)
 rename tests/{ => data}/languages_examples/src1.zoo (100%)
 rename tests/{ => data}/languages_examples/src2.zoo (100%)
 rename tests/{ => data}/languages_examples/src3.zpp (100%)
 rename tests/{ => data}/problems/footnote/problem.yaml (100%)
 rename tests/{ => data}/problems/footnote/statement/problem.en.md (100%)
 rename tests/{ => data}/problems/imgrequest/problem.yaml (100%)
 rename tests/{ => data}/problems/imgrequest/statement/problem.en.md (100%)
 rename tests/{ => data}/problems/imgrequest2/problem.yaml (100%)
 rename tests/{ => data}/problems/imgrequest2/statement/problem.en.md (100%)
 rename tests/{ => data}/problems/problemnamexss/problem.yaml (100%)
 rename tests/{ => data}/problems/problemnamexss/statement/problem.en.md (100%)
 rename tests/{ => data}/problems/samplexss/data/sample/1.ans (100%)
 rename tests/{ => data}/problems/samplexss/data/sample/1.in (100%)
 rename tests/{ => data}/problems/samplexss/data/sample/testdata.yaml (100%)
 rename tests/{ => data}/problems/samplexss/data/testdata.yaml (100%)
 rename tests/{ => data}/problems/samplexss/problem.yaml (100%)
 rename tests/{ => data}/problems/samplexss/statement/problem.en.md (100%)
 rename tests/{ => data}/problems/specialcharacterssample/data/sample/1.ans (100%)
 rename tests/{ => data}/problems/specialcharacterssample/data/sample/1.in (100%)
 rename tests/{ => data}/problems/specialcharacterssample/data/sample/testdata.yaml (100%)
 rename tests/{ => data}/problems/specialcharacterssample/data/testdata.yaml (100%)
 rename tests/{ => data}/problems/specialcharacterssample/problem.yaml (100%)
 rename tests/{ => data}/problems/specialcharacterssample/statement/problem.en.md (100%)
 rename tests/{ => data}/problems/statementxss/problem.yaml (100%)
 rename tests/{ => data}/problems/statementxss/statement/problem.en.md (100%)
 rename tests/{ => data}/problems/twofootnotes/problem.yaml (100%)
 rename tests/{ => data}/problems/twofootnotes/statement/problem.en.md (100%)

diff --git a/tests/conftest.py b/tests/conftest.py
index 4762dcc1..148ef966 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -9,9 +9,7 @@
 
 def datadir() -> Path:
     """Root directory holding static test fixture data."""
-    # TODO: move config1, config2, languages_examples, and problems under a
-    # tests/data directory, and change this to Path(__file__).parent / 'data'.
-    return Path(__file__).parent.resolve()
+    return (Path(__file__).parent / 'data').resolve()
 
 
 def example_directory(problem_name: str) -> Path:
diff --git a/tests/config1/broken.yaml b/tests/data/config1/broken.yaml
similarity index 100%
rename from tests/config1/broken.yaml
rename to tests/data/config1/broken.yaml
diff --git a/tests/config1/test.yaml b/tests/data/config1/test.yaml
similarity index 100%
rename from tests/config1/test.yaml
rename to tests/data/config1/test.yaml
diff --git a/tests/config1/test2.yaml b/tests/data/config1/test2.yaml
similarity index 100%
rename from tests/config1/test2.yaml
rename to tests/data/config1/test2.yaml
diff --git a/tests/config2/test2.yaml b/tests/data/config2/test2.yaml
similarity index 100%
rename from tests/config2/test2.yaml
rename to tests/data/config2/test2.yaml
diff --git a/tests/languages_examples/py2helper.py2 b/tests/data/languages_examples/py2helper.py2
similarity index 100%
rename from tests/languages_examples/py2helper.py2
rename to tests/data/languages_examples/py2helper.py2
diff --git a/tests/languages_examples/py2main.py b/tests/data/languages_examples/py2main.py
similarity index 100%
rename from tests/languages_examples/py2main.py
rename to tests/data/languages_examples/py2main.py
diff --git a/tests/languages_examples/py2nosheb.py b/tests/data/languages_examples/py2nosheb.py
similarity index 100%
rename from tests/languages_examples/py2nosheb.py
rename to tests/data/languages_examples/py2nosheb.py
diff --git a/tests/languages_examples/src1.zoo b/tests/data/languages_examples/src1.zoo
similarity index 100%
rename from tests/languages_examples/src1.zoo
rename to tests/data/languages_examples/src1.zoo
diff --git a/tests/languages_examples/src2.zoo b/tests/data/languages_examples/src2.zoo
similarity index 100%
rename from tests/languages_examples/src2.zoo
rename to tests/data/languages_examples/src2.zoo
diff --git a/tests/languages_examples/src3.zpp b/tests/data/languages_examples/src3.zpp
similarity index 100%
rename from tests/languages_examples/src3.zpp
rename to tests/data/languages_examples/src3.zpp
diff --git a/tests/problems/footnote/problem.yaml b/tests/data/problems/footnote/problem.yaml
similarity index 100%
rename from tests/problems/footnote/problem.yaml
rename to tests/data/problems/footnote/problem.yaml
diff --git a/tests/problems/footnote/statement/problem.en.md b/tests/data/problems/footnote/statement/problem.en.md
similarity index 100%
rename from tests/problems/footnote/statement/problem.en.md
rename to tests/data/problems/footnote/statement/problem.en.md
diff --git a/tests/problems/imgrequest/problem.yaml b/tests/data/problems/imgrequest/problem.yaml
similarity index 100%
rename from tests/problems/imgrequest/problem.yaml
rename to tests/data/problems/imgrequest/problem.yaml
diff --git a/tests/problems/imgrequest/statement/problem.en.md b/tests/data/problems/imgrequest/statement/problem.en.md
similarity index 100%
rename from tests/problems/imgrequest/statement/problem.en.md
rename to tests/data/problems/imgrequest/statement/problem.en.md
diff --git a/tests/problems/imgrequest2/problem.yaml b/tests/data/problems/imgrequest2/problem.yaml
similarity index 100%
rename from tests/problems/imgrequest2/problem.yaml
rename to tests/data/problems/imgrequest2/problem.yaml
diff --git a/tests/problems/imgrequest2/statement/problem.en.md b/tests/data/problems/imgrequest2/statement/problem.en.md
similarity index 100%
rename from tests/problems/imgrequest2/statement/problem.en.md
rename to tests/data/problems/imgrequest2/statement/problem.en.md
diff --git a/tests/problems/problemnamexss/problem.yaml b/tests/data/problems/problemnamexss/problem.yaml
similarity index 100%
rename from tests/problems/problemnamexss/problem.yaml
rename to tests/data/problems/problemnamexss/problem.yaml
diff --git a/tests/problems/problemnamexss/statement/problem.en.md b/tests/data/problems/problemnamexss/statement/problem.en.md
similarity index 100%
rename from tests/problems/problemnamexss/statement/problem.en.md
rename to tests/data/problems/problemnamexss/statement/problem.en.md
diff --git a/tests/problems/samplexss/data/sample/1.ans b/tests/data/problems/samplexss/data/sample/1.ans
similarity index 100%
rename from tests/problems/samplexss/data/sample/1.ans
rename to tests/data/problems/samplexss/data/sample/1.ans
diff --git a/tests/problems/samplexss/data/sample/1.in b/tests/data/problems/samplexss/data/sample/1.in
similarity index 100%
rename from tests/problems/samplexss/data/sample/1.in
rename to tests/data/problems/samplexss/data/sample/1.in
diff --git a/tests/problems/samplexss/data/sample/testdata.yaml b/tests/data/problems/samplexss/data/sample/testdata.yaml
similarity index 100%
rename from tests/problems/samplexss/data/sample/testdata.yaml
rename to tests/data/problems/samplexss/data/sample/testdata.yaml
diff --git a/tests/problems/samplexss/data/testdata.yaml b/tests/data/problems/samplexss/data/testdata.yaml
similarity index 100%
rename from tests/problems/samplexss/data/testdata.yaml
rename to tests/data/problems/samplexss/data/testdata.yaml
diff --git a/tests/problems/samplexss/problem.yaml b/tests/data/problems/samplexss/problem.yaml
similarity index 100%
rename from tests/problems/samplexss/problem.yaml
rename to tests/data/problems/samplexss/problem.yaml
diff --git a/tests/problems/samplexss/statement/problem.en.md b/tests/data/problems/samplexss/statement/problem.en.md
similarity index 100%
rename from tests/problems/samplexss/statement/problem.en.md
rename to tests/data/problems/samplexss/statement/problem.en.md
diff --git a/tests/problems/specialcharacterssample/data/sample/1.ans b/tests/data/problems/specialcharacterssample/data/sample/1.ans
similarity index 100%
rename from tests/problems/specialcharacterssample/data/sample/1.ans
rename to tests/data/problems/specialcharacterssample/data/sample/1.ans
diff --git a/tests/problems/specialcharacterssample/data/sample/1.in b/tests/data/problems/specialcharacterssample/data/sample/1.in
similarity index 100%
rename from tests/problems/specialcharacterssample/data/sample/1.in
rename to tests/data/problems/specialcharacterssample/data/sample/1.in
diff --git a/tests/problems/specialcharacterssample/data/sample/testdata.yaml b/tests/data/problems/specialcharacterssample/data/sample/testdata.yaml
similarity index 100%
rename from tests/problems/specialcharacterssample/data/sample/testdata.yaml
rename to tests/data/problems/specialcharacterssample/data/sample/testdata.yaml
diff --git a/tests/problems/specialcharacterssample/data/testdata.yaml b/tests/data/problems/specialcharacterssample/data/testdata.yaml
similarity index 100%
rename from tests/problems/specialcharacterssample/data/testdata.yaml
rename to tests/data/problems/specialcharacterssample/data/testdata.yaml
diff --git a/tests/problems/specialcharacterssample/problem.yaml b/tests/data/problems/specialcharacterssample/problem.yaml
similarity index 100%
rename from tests/problems/specialcharacterssample/problem.yaml
rename to tests/data/problems/specialcharacterssample/problem.yaml
diff --git a/tests/problems/specialcharacterssample/statement/problem.en.md b/tests/data/problems/specialcharacterssample/statement/problem.en.md
similarity index 100%
rename from tests/problems/specialcharacterssample/statement/problem.en.md
rename to tests/data/problems/specialcharacterssample/statement/problem.en.md
diff --git a/tests/problems/statementxss/problem.yaml b/tests/data/problems/statementxss/problem.yaml
similarity index 100%
rename from tests/problems/statementxss/problem.yaml
rename to tests/data/problems/statementxss/problem.yaml
diff --git a/tests/problems/statementxss/statement/problem.en.md b/tests/data/problems/statementxss/statement/problem.en.md
similarity index 100%
rename from tests/problems/statementxss/statement/problem.en.md
rename to tests/data/problems/statementxss/statement/problem.en.md
diff --git a/tests/problems/twofootnotes/problem.yaml b/tests/data/problems/twofootnotes/problem.yaml
similarity index 100%
rename from tests/problems/twofootnotes/problem.yaml
rename to tests/data/problems/twofootnotes/problem.yaml
diff --git a/tests/problems/twofootnotes/statement/problem.en.md b/tests/data/problems/twofootnotes/statement/problem.en.md
similarity index 100%
rename from tests/problems/twofootnotes/statement/problem.en.md
rename to tests/data/problems/twofootnotes/statement/problem.en.md

From ca75e7deeb752350af5198d92ea4d2542c2d64d8 Mon Sep 17 00:00:00 2001
From: Gunnar Kreitz <gkreitz@kattis.com>
Date: Wed, 2 Sep 2026 14:22:23 +0200
Subject: [PATCH 5/6] Fix mypy errors in tests (basically just mechanically
 adding missing types)

---
 .../run_and_generate_expected.py              |  2 +-
 tests/test_config.py                          | 14 +--
 tests/test_default_validator.py               |  6 +-
 tests/test_languages.py                       | 96 +++++++++----------
 tests/test_latex.py                           | 21 ++--
 tests/test_markdown.py                        |  9 +-
 tests/test_metadata.py                        | 24 ++---
 tests/test_output_validator.py                |  4 +-
 tests/test_run_limit.py                       |  2 +-
 tests/test_score_range.py                     | 55 +++++------
 tests/test_verify_hello.py                    |  2 +-
 tests/test_xss.py                             | 12 ++-
 12 files changed, 128 insertions(+), 119 deletions(-)

diff --git a/tests/default_validator_tests/run_and_generate_expected.py b/tests/default_validator_tests/run_and_generate_expected.py
index 6b4a54ce..716c3d19 100755
--- a/tests/default_validator_tests/run_and_generate_expected.py
+++ b/tests/default_validator_tests/run_and_generate_expected.py
@@ -21,7 +21,7 @@
 from pathlib import Path
 
 
-def main():
+def main() -> None:
     """Main function"""
     parser = argparse.ArgumentParser(description='Run default_validator and generate expected output files.')
     parser.add_argument('test_dir', type=Path, help='Path to the test directory.')
diff --git a/tests/test_config.py b/tests/test_config.py
index 49a40979..976dabaa 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -1,42 +1,44 @@
+from pathlib import Path
+
 import pytest
 
 from problemtools import config
 from tests.conftest import datadir
 
 
-def config_paths_mock():
+def config_paths_mock() -> list[Path]:
     return [datadir() / 'config1', datadir() / 'config2']
 
 
-def test_load_basic_config(monkeypatch):
+def test_load_basic_config(monkeypatch: pytest.MonkeyPatch) -> None:
     monkeypatch.setattr(config, '__config_file_paths', config_paths_mock)
 
     conf = config.load_config('test.yaml')
     assert conf == {'prop1': 'hello', 'prop2': 5}
 
 
-def test_load_updated_config(monkeypatch):
+def test_load_updated_config(monkeypatch: pytest.MonkeyPatch) -> None:
     monkeypatch.setattr(config, '__config_file_paths', config_paths_mock)
 
     conf = config.load_config('test2.yaml')
     assert conf == {'prop1': 'abc', 'prop2': 23, 'prop3': ['hello', 'world']}
 
 
-def test_load_missing_config(monkeypatch):
+def test_load_missing_config(monkeypatch: pytest.MonkeyPatch) -> None:
     monkeypatch.setattr(config, '__config_file_paths', config_paths_mock)
 
     with pytest.raises(config.ConfigError):
         config.load_config('non_existent_file')
 
 
-def test_load_broken_config(monkeypatch):
+def test_load_broken_config(monkeypatch: pytest.MonkeyPatch) -> None:
     monkeypatch.setattr(config, '__config_file_paths', config_paths_mock)
 
     with pytest.raises(config.ConfigError):
         config.load_config('broken.yaml')
 
 
-def test_update_dict():
+def test_update_dict() -> None:
     update_dict = config.__dict__['__update_dict']
 
     dict1 = {'a': 1, 'b': {'sub1': 1, 'sub2': False}, 'c': 3}
diff --git a/tests/test_default_validator.py b/tests/test_default_validator.py
index 80e206a9..5500ff9e 100644
--- a/tests/test_default_validator.py
+++ b/tests/test_default_validator.py
@@ -35,7 +35,7 @@ def validator() -> Path:
     return VALIDATOR_PATH
 
 
-def discover_test_cases():
+def discover_test_cases() -> list[Path]:
     """
     Finds and returns a list of all test case directories.
     A test case directory is expected to start with 'test_'.
@@ -46,7 +46,7 @@ def discover_test_cases():
 
 
 @pytest.mark.parametrize('test_dir', discover_test_cases(), ids=lambda d: d.name)
-def test_default_validator(validator: Path, test_dir: Path):
+def test_default_validator(validator: Path, test_dir: Path) -> None:
     """
     Runs a single validator test case.
     The test is parametrized to run for each directory discovered by `discover_test_cases`.
@@ -90,4 +90,4 @@ def test_default_validator(validator: Path, test_dir: Path):
                 # If no message is expected, assert that no message was generated.
                 if judgemessage_path.is_file():
                     actual_message = judgemessage_path.read_bytes()
-                    assert not actual_message, f'A validation message was generated but none was expected: {actual_message}'
+                    assert not actual_message, f'A validation message was generated but none was expected: {actual_message!r}'
diff --git a/tests/test_languages.py b/tests/test_languages.py
index 80c6a4c0..a99eb993 100644
--- a/tests/test_languages.py
+++ b/tests/test_languages.py
@@ -11,7 +11,7 @@
 
 class Language_test(TestCase):
     @staticmethod
-    def __language_dict():
+    def __language_dict() -> dict:
         return {
             'name': 'A Language',
             'priority': 100,
@@ -22,10 +22,10 @@ def __language_dict():
             'run': '{binary} {memlim}',
         }
 
-    def test_create(self):
+    def test_create(self) -> None:
         languages.Language('langid', self.__language_dict())
 
-    def test_update(self):
+    def test_update(self) -> None:
         lang = languages.Language('langid', self.__language_dict())
 
         lang.update({'priority': -1})
@@ -52,7 +52,7 @@ def test_update(self):
         lang.update({'run': 'newrun {mainclass}'})
         assert lang.run == 'newrun {mainclass}'
 
-    def test_invalid_id(self):
+    def test_invalid_id(self) -> None:
         vals = self.__language_dict()
         with pytest.raises(TypeError):
             languages.Language(None, vals)  # type: ignore
@@ -65,25 +65,25 @@ def test_invalid_id(self):
         with pytest.raises(languages.LanguageConfigError):
             languages.Language('Capital', vals)
 
-    def test_missing_name(self):
+    def test_missing_name(self) -> None:
         vals = self.__language_dict()
         del vals['name']
         with pytest.raises(languages.LanguageConfigError):
             languages.Language('id', vals)
 
-    def test_invalid_name(self):
+    def test_invalid_name(self) -> None:
         vals = self.__language_dict()
         vals['name'] = ['A List']
         with pytest.raises(languages.LanguageConfigError):
             languages.Language('id', vals)
 
-    def test_missing_priority(self):
+    def test_missing_priority(self) -> None:
         vals = self.__language_dict()
         del vals['priority']
         with pytest.raises(languages.LanguageConfigError):
             languages.Language('id', vals)
 
-    def test_invalid_priority(self):
+    def test_invalid_priority(self) -> None:
         vals = self.__language_dict()
         vals['priority'] = 2.3
         with pytest.raises(languages.LanguageConfigError):
@@ -92,49 +92,49 @@ def test_invalid_priority(self):
         with pytest.raises(languages.LanguageConfigError):
             languages.Language('id', vals)
 
-    def test_missing_files(self):
+    def test_missing_files(self) -> None:
         vals = self.__language_dict()
         del vals['files']
         with pytest.raises(languages.LanguageConfigError):
             languages.Language('id', vals)
 
-    def test_invalid_files(self):
+    def test_invalid_files(self) -> None:
         vals = self.__language_dict()
         vals['files'] = ['*.cc', '*.cpp']
         with pytest.raises(languages.LanguageConfigError):
             languages.Language('id', vals)
 
-    def test_without_shebang(self):
+    def test_without_shebang(self) -> None:
         vals = self.__language_dict()
         del vals['shebang']
         del vals['shebang_files']
         languages.Language('id', vals)
 
-    def test_invalid_shebang(self):
+    def test_invalid_shebang(self) -> None:
         vals = self.__language_dict()
         vals['shebang'] = '(Not an RE'
         with pytest.raises(re.error):
             languages.Language('id', vals)
 
-    def test_shebang_requires_shebang_files(self):
+    def test_shebang_requires_shebang_files(self) -> None:
         vals = self.__language_dict()
         del vals['shebang_files']
         with pytest.raises(languages.LanguageConfigError):
             languages.Language('id', vals)
 
-    def test_shebang_files_requires_shebang(self):
+    def test_shebang_files_requires_shebang(self) -> None:
         vals = self.__language_dict()
         del vals['shebang']
         with pytest.raises(languages.LanguageConfigError):
             languages.Language('id', vals)
 
-    def test_invalid_shebang_files(self):
+    def test_invalid_shebang_files(self) -> None:
         vals = self.__language_dict()
         vals['shebang_files'] = ['*.foo', '*.bar']
         with pytest.raises(languages.LanguageConfigError):
             languages.Language('id', vals)
 
-    def test_get_source_files_does_not_require_readable_files(self):
+    def test_get_source_files_does_not_require_readable_files(self) -> None:
         """get_source_files only inspects file names -- unlike
         get_source_files_for_detection, it must work even for files that
         don't exist, since it's used on a program whose language is
@@ -146,12 +146,12 @@ def test_get_source_files_does_not_require_readable_files(self):
         with pytest.raises(OSError):
             lang.get_source_files_for_detection([missing])
 
-    def test_without_compile(self):
+    def test_without_compile(self) -> None:
         vals = self.__language_dict()
         del vals['compile']
         languages.Language('id', vals)
 
-    def test_invalid_compile(self):
+    def test_invalid_compile(self) -> None:
         vals = self.__language_dict()
         vals['compile'] = ['gcc', '{files}']
         with pytest.raises(languages.LanguageConfigError):
@@ -160,13 +160,13 @@ def test_invalid_compile(self):
         with pytest.raises(languages.LanguageConfigError):
             languages.Language('id', vals)
 
-    def test_missing_run(self):
+    def test_missing_run(self) -> None:
         vals = self.__language_dict()
         del vals['run']
         with pytest.raises(languages.LanguageConfigError):
             languages.Language('id', vals)
 
-    def test_invalid_run(self):
+    def test_invalid_run(self) -> None:
         vals = self.__language_dict()
         vals['run'] = ['python3', '{mainfile}']
         with pytest.raises(languages.LanguageConfigError):
@@ -175,7 +175,7 @@ def test_invalid_run(self):
         with pytest.raises(languages.LanguageConfigError):
             languages.Language('id', vals)
 
-    def test_good_entrypoints(self):
+    def test_good_entrypoints(self) -> None:
         vals = self.__language_dict()
 
         vals['compile'] = 'echo {binary}'
@@ -190,7 +190,7 @@ def test_good_entrypoints(self):
         vals['run'] = 'echo {mainclass}'
         languages.Language('id', vals)
 
-    def test_bad_entrypoints(self):
+    def test_bad_entrypoints(self) -> None:
         vals = self.__language_dict()
 
         # Two different entry points
@@ -205,8 +205,8 @@ def test_bad_entrypoints(self):
             languages.Language('id', vals)
 
     @staticmethod
-    def __subs(**overrides):
-        values = {
+    def __subs(**overrides: str | int) -> languages.CommandSubstitution:
+        values: dict[str, str | int] = {
             'path': '/tmp/prog',
             'files': 'main.foo',
             'binary': '/tmp/prog/run',
@@ -216,16 +216,16 @@ def __subs(**overrides):
             'memlim': 256,
         }
         values.update(overrides)
-        return languages.CommandSubstitution(**values)
+        return languages.CommandSubstitution(**values)  # type: ignore[arg-type]
 
-    def test_check_installed_ok(self):
+    def test_check_installed_ok(self) -> None:
         # compile is 'echo ...' (found on PATH); run is '{binary} {memlim}',
         # i.e. the program we just produced, so there's nothing external to
         # check there.
         lang = languages.Language('id', self.__language_dict())
         assert lang.check_installed() is None
 
-    def test_check_installed_does_not_check_produced_entry_point(self):
+    def test_check_installed_does_not_check_produced_entry_point(self) -> None:
         # No compile step, and run is just the (self-contained) source file
         # -- no external program is needed at all.
         vals = self.__language_dict()
@@ -234,7 +234,7 @@ def test_check_installed_does_not_check_produced_entry_point(self):
         lang = languages.Language('id', vals)
         assert lang.check_installed() is None
 
-    def test_check_installed_missing_compiler(self):
+    def test_check_installed_missing_compiler(self) -> None:
         vals = self.__language_dict()
         vals['compile'] = 'definitely_not_a_real_compiler_xyz {files} {binary}'
         lang = languages.Language('id', vals)
@@ -243,13 +243,13 @@ def test_check_installed_missing_compiler(self):
         assert 'compiler' in msg
         assert 'definitely_not_a_real_compiler_xyz' in msg
 
-    def test_check_installed_missing_absolute_path(self):
+    def test_check_installed_missing_absolute_path(self) -> None:
         vals = self.__language_dict()
         vals['compile'] = '/nonexistent/path/to/compiler {files} {binary}'
         lang = languages.Language('id', vals)
         assert lang.check_installed() is not None
 
-    def test_check_installed_missing_runtime(self):
+    def test_check_installed_missing_runtime(self) -> None:
         # No compile step -- the runtime must still be checked.
         vals = self.__language_dict()
         del vals['compile']
@@ -260,19 +260,19 @@ def test_check_installed_missing_runtime(self):
         assert 'runtime' in msg
         assert 'definitely_not_a_real_runtime_xyz' in msg
 
-    def test_get_compile_command_none_without_compile_step(self):
+    def test_get_compile_command_none_without_compile_step(self) -> None:
         vals = self.__language_dict()
         del vals['compile']
         lang = languages.Language('id', vals)
         assert lang.get_compile_command(self.__subs()) is None
 
-    def test_get_compile_command_resolves_executable(self):
+    def test_get_compile_command_resolves_executable(self) -> None:
         lang = languages.Language('id', self.__language_dict())
         subs = self.__subs()
         command = lang.get_compile_command(subs)
         assert command == [shutil.which('echo'), subs.path, subs.files, subs.binary]
 
-    def test_get_run_command_resolves_executable(self):
+    def test_get_run_command_resolves_executable(self) -> None:
         vals = self.__language_dict()
         vals['compile'] = 'echo {mainfile}'
         vals['run'] = 'echo {mainfile}'
@@ -280,14 +280,14 @@ def test_get_run_command_resolves_executable(self):
         subs = self.__subs()
         assert lang.get_run_command(subs) == [shutil.which('echo'), subs.mainfile]
 
-    def test_get_run_command_leaves_produced_binary_untouched(self):
+    def test_get_run_command_leaves_produced_binary_untouched(self) -> None:
         # run is '{binary} {memlim}' in the base dict -- the first token is
         # the program we just compiled, not something to resolve via PATH.
         lang = languages.Language('id', self.__language_dict())
         subs = self.__subs()
         assert lang.get_run_command(subs) == [subs.binary, str(subs.memlim)]
 
-    def test_get_compile_command_splits_multiple_files(self):
+    def test_get_compile_command_splits_multiple_files(self) -> None:
         # {files} is a single space-separated string, but each file must
         # end up as its own argument rather than one combined string.
         lang = languages.Language('id', self.__language_dict())
@@ -299,17 +299,17 @@ def test_get_compile_command_splits_multiple_files(self):
 __EXAMPLES_PATH = datadir() / 'languages_examples'
 
 
-def examples_path(test_file):
+def examples_path(test_file: str) -> str:
     return os.path.join(__EXAMPLES_PATH, test_file)
 
 
 class Languages_test(TestCase):
-    def test_empty_languages(self):
+    def test_empty_languages(self) -> None:
         lang = languages.Languages()
         assert lang.languages == {}
         assert lang.detect_language(['foo.cpp', 'foo.c', 'foo.py', 'foo.java']) is None
 
-    def test_duplicate_prio(self):
+    def test_duplicate_prio(self) -> None:
         lang = languages.Languages()
         config = {
             'c': {
@@ -331,7 +331,7 @@ def test_duplicate_prio(self):
         with pytest.raises(languages.LanguageConfigError):
             lang.update(config)
 
-    def test_invalid_format(self):
+    def test_invalid_format(self) -> None:
         lang = languages.Languages()
         # Dict of strings instead of dict of dict
         conf1 = {'c': 'C'}
@@ -356,16 +356,16 @@ def test_invalid_format(self):
         with pytest.raises(languages.LanguageConfigError):
             lang.update(conf1)
         with pytest.raises(languages.LanguageConfigError):
-            lang.update(conf2)
+            lang.update(conf2)  # type: ignore[arg-type]
         with pytest.raises(languages.LanguageConfigError):
-            lang.update(conf3)
+            lang.update(conf3)  # type: ignore[arg-type]
 
-    def test_empty(self):
+    def test_empty(self) -> None:
         lang = languages.Languages()
         lang.update({})
         assert lang.languages == {}
 
-    def test_zoo(self):
+    def test_zoo(self) -> None:
         langs = languages.Languages()
 
         zoo = {
@@ -384,15 +384,15 @@ def test_zoo(self):
         langs.update(zoo)
 
         lang = langs.detect_language([examples_path(x) for x in ['src1.zoo']])
-        assert lang.lang_id == 'zoo'
+        assert lang is not None and lang.lang_id == 'zoo'
 
         lang = langs.detect_language([examples_path(x) for x in ['src2.zoo']])
-        assert lang.lang_id == 'zoork'
+        assert lang is not None and lang.lang_id == 'zoork'
 
         lang = langs.detect_language([examples_path(x) for x in ['src2.zoo', 'src3.zpp']])
-        assert lang.lang_id == 'zoopp'
+        assert lang is not None and lang.lang_id == 'zoopp'
 
-    def test_shebang_gate_only_affects_detection_not_actual_source_files(self):
+    def test_shebang_gate_only_affects_detection_not_actual_source_files(self) -> None:
         """Regression test: once a language has won detection, a helper
         file matching its "files" glob must not be dropped just because it
         individually lacks the shebang that broke the tie against another
@@ -423,7 +423,7 @@ def test_shebang_gate_only_affects_detection_not_actual_source_files(self):
         # (unconditional, not gated) = 2.  py3 evidence: py2main.py +
         # py2nosheb.py (both unconditional) = 2.  Tie -> priority decides.
         lang = langs.detect_language(files)
-        assert lang.lang_id == 'py2'
+        assert lang is not None and lang.lang_id == 'py2'
 
         # All three files -- including the shebang-less helper -- belong
         # to the now-settled py2 program.
diff --git a/tests/test_latex.py b/tests/test_latex.py
index 6f687dec..94e7a4fd 100644
--- a/tests/test_latex.py
+++ b/tests/test_latex.py
@@ -3,10 +3,11 @@
 from pathlib import Path
 
 from problemtools import problem2html, problem2pdf
+from problemtools.diagnostics import Diagnostics
 from tests.conftest import example_directory
 
 
-def test_pdf_render_verifyproblem():
+def test_pdf_render_verifyproblem() -> None:
     # Same options as in verifyproblem
     options = problem2pdf.get_parser().parse_args([''])
     problem_path = example_directory('guess')
@@ -18,7 +19,7 @@ def test_pdf_render_verifyproblem():
         assert False, 'PDF conversion failed'
 
 
-def test_pdf_render_problem2pdf():
+def test_pdf_render_problem2pdf() -> None:
     # Same options as typical problem2pdf usage
     with tempfile.TemporaryDirectory() as temp_dir:
         problem_path = example_directory('guess')
@@ -30,14 +31,14 @@ def test_pdf_render_problem2pdf():
             assert temp_file.read(5) == b'%PDF-', 'Output header does not look like a PDF.'
 
 
-def test_html_render_different(diag):
+def test_html_render_different(diag: Diagnostics) -> None:
     # Same options as typical problem2html usage
     with tempfile.TemporaryDirectory() as temp_dir:
         problem_path = example_directory('different')
-        temp_dir = Path(temp_dir) / 'different_html'
-        options = problem2html.get_parser().parse_args(['-d', str(temp_dir), '-l', 'en', '-q', str(problem_path)])
+        out_dir = Path(temp_dir) / 'different_html'
+        options = problem2html.get_parser().parse_args(['-d', str(out_dir), '-l', 'en', '-q', str(problem_path)])
         problem2html.convert(options, diag)
-        with open(temp_dir / 'index.html', 'r') as temp_file:
+        with open(out_dir / 'index.html', 'r') as temp_file:
             full_html = temp_file.read()
             assert re.search('<html>', full_html)
             assert re.search('A Different Problem', full_html)
@@ -46,14 +47,14 @@ def test_html_render_different(diag):
             assert re.search('71293781758123 72784', full_html)  # part of sample
 
 
-def test_html_render_guess(diag):
+def test_html_render_guess(diag: Diagnostics) -> None:
     # Same options as typical problem2html usage
     with tempfile.TemporaryDirectory() as temp_dir:
         problem_path = example_directory('guess')
-        temp_dir = Path(temp_dir) / 'guess_html'
-        options = problem2html.get_parser().parse_args(['-d', str(temp_dir), '-l', 'en', '-q', str(problem_path)])
+        out_dir = Path(temp_dir) / 'guess_html'
+        options = problem2html.get_parser().parse_args(['-d', str(out_dir), '-l', 'en', '-q', str(problem_path)])
         problem2html.convert(options, diag)
-        with open(temp_dir / 'index.html', 'r') as temp_file:
+        with open(out_dir / 'index.html', 'r') as temp_file:
             full_html = temp_file.read()
             assert re.search('<html>', full_html)
             assert re.search('Guess the Number', full_html)
diff --git a/tests/test_markdown.py b/tests/test_markdown.py
index 2357858f..d795e32b 100644
--- a/tests/test_markdown.py
+++ b/tests/test_markdown.py
@@ -1,5 +1,6 @@
 import pytest
 
+from problemtools.diagnostics import Diagnostics
 from problemtools.statement_util import find_footnotes
 from tests.conftest import datadir
 from tests.test_xss import render, renderpdf
@@ -14,7 +15,7 @@
 #         problem2pdf.convert(args)
 
 
-def test_sample_escaping(diag):
+def test_sample_escaping(diag: Diagnostics) -> None:
     problem_path = datadir() / 'problems' / 'specialcharacterssample'
     html = render(problem_path, diag)
     all_printable = r"""0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ!"#$%&'()*+,-./:;<=>?@[\]^_`{|}~"""
@@ -25,7 +26,7 @@ def test_sample_escaping(diag):
     assert all_printable in html
 
 
-def test_footnotes(diag):
+def test_footnotes(diag: Diagnostics) -> None:
     # We always want footnotes to be at the bottom
     # When we insert samples, we need to insert them right above the first footnote
     # To do this, we search for a string (very fragile)
@@ -38,14 +39,14 @@ def test_footnotes(diag):
     assert find_footnotes(html) is not None
 
 
-def test_footnotes_href(diag):
+def test_footnotes_href(diag: Diagnostics) -> None:
     # We use allowlist-based id values for footnotes. Ensure they have not changed
     problem_path = datadir() / 'problems' / 'footnote'
     html = render(problem_path, diag)
     assert 'fn1' in html and 'fnref1' in html
 
 
-def test_invalid_image_throws(diag):
+def test_invalid_image_throws(diag: Diagnostics) -> None:
     # If images can point to img that doesn't exist, it's arbitrary web request
     for problem in ('imgrequest', 'imgrequest2'):
         problem_path = datadir() / 'problems' / problem
diff --git a/tests/test_metadata.py b/tests/test_metadata.py
index 2fd7e58f..7c12f70b 100644
--- a/tests/test_metadata.py
+++ b/tests/test_metadata.py
@@ -1,3 +1,5 @@
+from typing import Any
+
 import pytest
 from pydantic import ValidationError
 
@@ -8,7 +10,7 @@
 # A few quick tests of config parsing. pytest structure isn't great here, so code gets repetitive, but I wanted *something* basic in place at least.
 
 
-def test_parse_empty_legacy():
+def test_parse_empty_legacy() -> None:
     m = metadata.parse_metadata(FormatVersion.LEGACY, {})
     # Just check off a few random things
     assert not m.name
@@ -16,18 +18,18 @@ def test_parse_empty_legacy():
     assert not m.credits.authors
 
 
-def test_parse_legacy_with_problem_names():
+def test_parse_legacy_with_problem_names() -> None:
     m = metadata.parse_metadata(FormatVersion.LEGACY, {}, {'en': 'Hello World!'})
     assert m.name['en'] == 'Hello World!'
 
 
-def test_parse_empty_2023_fails():
+def test_parse_empty_2023_fails() -> None:
     with pytest.raises(ValidationError):
         metadata.parse_metadata(FormatVersion.V_2023_07, {}, {'en': 'Hello World!'})
 
 
 @pytest.fixture
-def minimal_2023_conf():
+def minimal_2023_conf() -> dict[str, Any]:
     return {
         'problem_format_version': '2023-07-draft',
         'uuid': '46fa942f-44c3-46c0-8ddc-22e02d2e5d2b',
@@ -35,21 +37,21 @@ def minimal_2023_conf():
     }
 
 
-def test_parse_minimal_2023(minimal_2023_conf):
+def test_parse_minimal_2023(minimal_2023_conf: dict[str, Any]) -> None:
     m = metadata.parse_metadata(FormatVersion.V_2023_07, minimal_2023_conf, {'en': 'Hello World!'})
     assert m.name['en'] == 'Hello World!'
     assert not m.source
     assert not m.credits.authors
 
 
-def test_parse_typo_fails(minimal_2023_conf):
+def test_parse_typo_fails(minimal_2023_conf: dict[str, Any]) -> None:
     c = minimal_2023_conf
     c['limits'] = {'typo': 1}
     with pytest.raises(ValidationError):
         metadata.parse_metadata(FormatVersion.V_2023_07, c, {'en': 'Hello World!'})
 
 
-def test_parse_single_author_2023(minimal_2023_conf):
+def test_parse_single_author_2023(minimal_2023_conf: dict[str, Any]) -> None:
     c = minimal_2023_conf
     c['credits'] = ' \t  Authy McAuth \t <authy@mcauth.example> \t\t  '  # Add some extra whitespace to check that we strip
     m = metadata.parse_metadata(FormatVersion.V_2023_07, c, {'en': 'Hello World!'})
@@ -58,7 +60,7 @@ def test_parse_single_author_2023(minimal_2023_conf):
     assert m.credits.authors[0].email == 'authy@mcauth.example'
 
 
-def test_parse_single_source_2023(minimal_2023_conf):
+def test_parse_single_source_2023(minimal_2023_conf: dict[str, Any]) -> None:
     c = minimal_2023_conf
     c['source'] = 'NWERC 2024'
     m = metadata.parse_metadata(FormatVersion.V_2023_07, c, {'en': 'Hello World!'})
@@ -67,7 +69,7 @@ def test_parse_single_source_2023(minimal_2023_conf):
     assert m.source[0].url is None
 
 
-def test_parse_multi_source(minimal_2023_conf):
+def test_parse_multi_source(minimal_2023_conf: dict[str, Any]) -> None:
     c = minimal_2023_conf
     c['source'] = [
         {'name': 'NWERC 2024', 'url': 'https://2024.nwerc.example/contest'},
@@ -84,7 +86,7 @@ def test_parse_multi_source(minimal_2023_conf):
     assert m.source[2].url is None
 
 
-def test_parse_complex_type(minimal_2023_conf):
+def test_parse_complex_type(minimal_2023_conf: dict[str, Any]) -> None:
     c = minimal_2023_conf
     c['type'] = ['scoring', 'multi-pass', 'interactive']
     m = metadata.parse_metadata(FormatVersion.V_2023_07, c, {'en': 'Hello World!'})
@@ -99,7 +101,7 @@ def test_parse_complex_type(minimal_2023_conf):
     assert not m.is_submit_answer()
 
 
-def test_load_hello():
+def test_load_hello() -> None:
     m = metadata.load_metadata(example_directory('hello'))
     assert m.name['en'] == 'Hello World!'
     assert m.name['sv'] == 'Hej Världen!'
diff --git a/tests/test_output_validator.py b/tests/test_output_validator.py
index 3562eb7f..17140d4d 100644
--- a/tests/test_output_validator.py
+++ b/tests/test_output_validator.py
@@ -6,7 +6,7 @@
 from problemtools.judge.validate import _get_feedback
 
 
-def test_output_validator_feedback():
+def test_output_validator_feedback() -> None:
     r = random.Random(0)
     with tempfile.TemporaryDirectory() as directory:
         feedback = pathlib.Path(directory) / 'feedback.txt'
@@ -16,7 +16,7 @@ def test_output_validator_feedback():
         assert data is not None and text in data
 
 
-def test_output_validator_feedback_non_unicode():
+def test_output_validator_feedback_non_unicode() -> None:
     r = random.Random(0)
     with tempfile.TemporaryDirectory() as directory:
         feedback = pathlib.Path(directory) / 'feedback.txt'
diff --git a/tests/test_run_limit.py b/tests/test_run_limit.py
index 73985b7d..e99a019e 100644
--- a/tests/test_run_limit.py
+++ b/tests/test_run_limit.py
@@ -5,7 +5,7 @@
 
 
 class Limit_test(TestCase):
-    def test_less(self):
+    def test_less(self) -> None:
         less = limit.__dict__['__limit_less']
         assert less(42, 42)
         assert not less(42, 41)
diff --git a/tests/test_score_range.py b/tests/test_score_range.py
index a3e832e5..28ad4da2 100644
--- a/tests/test_score_range.py
+++ b/tests/test_score_range.py
@@ -3,6 +3,7 @@
 from problemtools.checks.testdata import _check_score_range
 from problemtools.model.paths import abspath, relpath
 from problemtools.model.testdata import TestCase, TestDataGroup
+from tests.conftest import RecordingDiagnostics
 
 # Not test classes -- just named that way by the model. Tell pytest not to collect them.
 TestCase.__test__ = False  # type: ignore[attr-defined]
@@ -49,7 +50,7 @@ def make_group(
     )
 
 
-def test_group_sum_is_default_aggregator(diag):
+def test_group_sum_is_default_aggregator(diag: RecordingDiagnostics) -> None:
     group = make_group(
         'g', [make_testcase('a'), make_testcase('b'), make_testcase('c')], range_='0 30', accept_score=10, reject_score=0
     )
@@ -57,12 +58,12 @@ def test_group_sum_is_default_aggregator(diag):
     assert diag.messages == []
 
 
-def test_group_avg_aggregator(diag):
+def test_group_avg_aggregator(diag: RecordingDiagnostics) -> None:
     group = make_group('g', [make_testcase('a')], grader_flags='avg', accept_score=10, reject_score=0)
     assert _check_score_range(group, False, diag) == (0, 10)
 
 
-def test_avg_aggregator_with_on_reject_break_widens_range(diag):
+def test_avg_aggregator_with_on_reject_break_widens_range(diag: RecordingDiagnostics) -> None:
     # subA (graded first) has a much wider range than subB. With on_reject: break, a non-AC subA
     # stops grading before subB is ever run, so the group's score is just subA's -- as high as
     # 100 -- even though grading both would pull the maximum average down to 55.
@@ -73,7 +74,7 @@ def test_avg_aggregator_with_on_reject_break_widens_range(diag):
     assert diag.messages == []
 
 
-def test_avg_aggregator_with_on_reject_continue_is_tighter(diag):
+def test_avg_aggregator_with_on_reject_continue_is_tighter(diag: RecordingDiagnostics) -> None:
     # Same as above, but without break: subB is always graded too, so avg is always over both.
     subA = make_group('g.subA', [make_testcase('a')], range_='0 100', accept_score=100, reject_score=0)
     subB = make_group('g.subB', [make_testcase('b')], range_='0 10', accept_score=10, reject_score=0)
@@ -82,18 +83,18 @@ def test_avg_aggregator_with_on_reject_continue_is_tighter(diag):
     assert diag.messages == []
 
 
-def test_reject_above_accept(diag):
+def test_reject_above_accept(diag: RecordingDiagnostics) -> None:
     # Corner case: check we don't end up with a broken range if reject_score is above accept_score
     group = make_group('g', [make_testcase('a')], accept_score=0, reject_score=5)
     assert _check_score_range(group, False, diag) == (0, 5)
 
 
-def test_custom_scoring_possible_is_unbounded(diag):
+def test_custom_scoring_possible_is_unbounded(diag: RecordingDiagnostics) -> None:
     group = make_group('g', [make_testcase('a')], accept_score=10, reject_score=0)
     assert _check_score_range(group, True, diag) == (-INF, INF)
 
 
-def test_group_min_aggregator(diag):
+def test_group_min_aggregator(diag: RecordingDiagnostics) -> None:
     group = make_group(
         'g',
         [make_testcase('a'), make_testcase('b')],
@@ -103,27 +104,27 @@ def test_group_min_aggregator(diag):
     assert _check_score_range(group, False, diag) == (0, 1)
 
 
-def test_group_max_aggregator(diag):
+def test_group_max_aggregator(diag: RecordingDiagnostics) -> None:
     group = make_group('g', [make_testcase('a'), make_testcase('b')], grader_flags='max')
     assert _check_score_range(group, False, diag) == (0, 1)
 
 
-def test_last_aggregator_flag_wins(diag):
+def test_last_aggregator_flag_wins(diag: RecordingDiagnostics) -> None:
     group = make_group('g', [make_testcase('a')], grader_flags='max min avg', accept_score=10, reject_score=0)
     assert _check_score_range(group, False, diag) == (0, 10)
 
 
-def test_custom_grading_is_unbounded(diag):
+def test_custom_grading_is_unbounded(diag: RecordingDiagnostics) -> None:
     group = make_group('g', [make_testcase('a')], grading='custom')
     assert _check_score_range(group, False, diag) == (-INF, INF)
 
 
-def test_empty_group_scores_zero(diag):
+def test_empty_group_scores_zero(diag: RecordingDiagnostics) -> None:
     group = make_group('g', [])
     assert _check_score_range(group, False, diag) == (0, 0)
 
 
-def test_nested_groups_compose(diag):
+def test_nested_groups_compose(diag: RecordingDiagnostics) -> None:
     subtask1 = make_group(
         'g.subtask1',
         [make_testcase('a'), make_testcase('b')],
@@ -138,14 +139,14 @@ def test_nested_groups_compose(diag):
     assert diag.messages == []
 
 
-def test_ignore_sample_at_root_skips_sample_group(diag):
+def test_ignore_sample_at_root_skips_sample_group(diag: RecordingDiagnostics) -> None:
     sample = make_group('sample', [make_testcase('s')], accept_score=1000, reject_score=0)
     secret = make_group('secret', [make_testcase('a')], accept_score=100, reject_score=0)
     root = make_group('data', [sample, secret], grader_flags='ignore_sample', is_root=True)
     assert _check_score_range(root, False, diag) == (0, 100)
 
 
-def test_ignore_sample_is_a_no_op_below_root(diag):
+def test_ignore_sample_is_a_no_op_below_root(diag: RecordingDiagnostics) -> None:
     sample = make_group('sample', [make_testcase('s')], accept_score=1000, reject_score=0)
     secret = make_group('secret', [make_testcase('a')], accept_score=100, reject_score=0)
     # Misconfigured (checks._check_group flags this separately). We just aggregate all children
@@ -156,13 +157,13 @@ def test_ignore_sample_is_a_no_op_below_root(diag):
 # --- Declared range vs. what can be inferred ---
 
 
-def test_no_warning_when_declared_matches_computed(diag):
+def test_no_warning_when_declared_matches_computed(diag: RecordingDiagnostics) -> None:
     group = make_group('g', [make_testcase('a')], range_='0 10', accept_score=10, reject_score=0)
     assert _check_score_range(group, False, diag) == (0, 10)
     assert diag.messages == []
 
 
-def test_looser_declared_range_warns_and_suggests_tightening(diag):
+def test_looser_declared_range_warns_and_suggests_tightening(diag: RecordingDiagnostics) -> None:
     group = make_group('g', [make_testcase('a')], range_='0 100', accept_score=10, reject_score=0)
     assert _check_score_range(group, False, diag) == (0, 10)
     assert diag.messages == [
@@ -173,7 +174,7 @@ def test_looser_declared_range_warns_and_suggests_tightening(diag):
     ]
 
 
-def test_no_declared_range_warns_and_suggests_one(diag):
+def test_no_declared_range_warns_and_suggests_one(diag: RecordingDiagnostics) -> None:
     group = make_group('g', [make_testcase('a')], accept_score=10, reject_score=0)  # range left at the default
     assert _check_score_range(group, False, diag) == (0, 10)
     assert diag.messages == [
@@ -187,7 +188,7 @@ def test_no_declared_range_warns_and_suggests_one(diag):
     ]
 
 
-def test_narrower_declared_range_is_trusted_without_warning(diag):
+def test_narrower_declared_range_is_trusted_without_warning(diag: RecordingDiagnostics) -> None:
     # A "bad guarantee": the group's own children can clearly reach 100, but the author declared a
     # narrower range. We don't warn -- that's a promise checked elsewhere (checks.submissions) --
     # but we do trust it for the returned (propagated) value.
@@ -196,7 +197,7 @@ def test_narrower_declared_range_is_trusted_without_warning(diag):
     assert diag.messages == []
 
 
-def test_narrower_declared_range_propagates_to_parent(diag):
+def test_narrower_declared_range_propagates_to_parent(diag: RecordingDiagnostics) -> None:
     # The parent's aggregate must reflect the child's effective (trusted) range, not its raw
     # aggregate -- so a narrow declaration deep in the tree is reflected in ancestors' results too.
     bad_child = make_group('g.secret', [make_testcase('a')], range_='0 10', accept_score=100, reject_score=0)
@@ -214,7 +215,7 @@ def test_narrower_declared_range_propagates_to_parent(diag):
     ]
 
 
-def test_disjoint_declared_range_warns_distinctly(diag):
+def test_disjoint_declared_range_warns_distinctly(diag: RecordingDiagnostics) -> None:
     group = make_group('g', [make_testcase('a')], range_='0 10', accept_score=100, reject_score=50)
     assert _check_score_range(group, False, diag) == (0, 10)
     assert diag.messages == [
@@ -225,7 +226,7 @@ def test_disjoint_declared_range_warns_distinctly(diag):
     ]
 
 
-def test_root_with_unbounded_aggregate_and_no_declared_range_still_warns(diag):
+def test_root_with_unbounded_aggregate_and_no_declared_range_still_warns(diag: RecordingDiagnostics) -> None:
     group = make_group('g', [make_testcase('a')], grading='custom', is_root=True)
     assert _check_score_range(group, False, diag) == (-INF, INF)
     assert diag.messages == [
@@ -240,13 +241,13 @@ def test_root_with_unbounded_aggregate_and_no_declared_range_still_warns(diag):
     ]
 
 
-def test_non_root_with_unbounded_aggregate_and_no_declared_range_is_silent(diag):
+def test_non_root_with_unbounded_aggregate_and_no_declared_range_is_silent(diag: RecordingDiagnostics) -> None:
     group = make_group('g', [make_testcase('a')], grading='custom', is_root=False)
     assert _check_score_range(group, False, diag) == (-INF, INF)
     assert diag.messages == []
 
 
-def test_root_with_negative_minimum_and_no_other_issue_warns(diag):
+def test_root_with_negative_minimum_and_no_other_issue_warns(diag: RecordingDiagnostics) -> None:
     group = make_group('g', [make_testcase('a')], accept_score=10, reject_score=-5, range_='-5 10', is_root=True)
     assert _check_score_range(group, False, diag) == (-5, 10)
     assert diag.messages == [
@@ -260,14 +261,14 @@ def test_root_with_negative_minimum_and_no_other_issue_warns(diag):
     ]
 
 
-def test_negative_minimum_below_root_is_not_flagged(diag):
+def test_negative_minimum_below_root_is_not_flagged(diag: RecordingDiagnostics) -> None:
     # negative scores for non-root test groups is a bit weird, but IMHO not weird enough to warn about
     group = make_group('g', [make_testcase('a')], accept_score=10, reject_score=-5, range_='-5 10', is_root=False)
     assert _check_score_range(group, False, diag) == (-5, 10)
     assert diag.messages == []
 
 
-def test_negative_minimum_at_root_is_suppressed_by_other_warnings(diag):
+def test_negative_minimum_at_root_is_suppressed_by_other_warnings(diag: RecordingDiagnostics) -> None:
     # Warning about a negative range is low priority. If we can instead suggest to narrow the range,
     # that's typically a better warning.
     group = make_group('g', [make_testcase('a')], accept_score=10, reject_score=0, range_='-10 200', is_root=True)
@@ -276,14 +277,14 @@ def test_negative_minimum_at_root_is_suppressed_by_other_warnings(diag):
     assert 'looser than the computed range' in diag.messages[0][1]
 
 
-def test_invalid_range_format_errors_and_falls_back_to_aggregate(diag):
+def test_invalid_range_format_errors_and_falls_back_to_aggregate(diag: RecordingDiagnostics) -> None:
     group = make_group('g', [make_testcase('a')], range_='not a range', accept_score=10, reject_score=0)
     assert _check_score_range(group, False, diag) == (0, 10)
     assert diag.errors == 1
     assert "Invalid format 'not a range'" in diag.messages[0][1]
 
 
-def test_min_greater_than_max_errors_and_falls_back_to_aggregate(diag):
+def test_min_greater_than_max_errors_and_falls_back_to_aggregate(diag: RecordingDiagnostics) -> None:
     group = make_group('g', [make_testcase('a')], range_='10 0', accept_score=10, reject_score=0)
     assert _check_score_range(group, False, diag) == (0, 10)
     assert diag.errors == 1
diff --git a/tests/test_verify_hello.py b/tests/test_verify_hello.py
index e0c63768..976c7d4a 100644
--- a/tests/test_verify_hello.py
+++ b/tests/test_verify_hello.py
@@ -25,7 +25,7 @@ def test_load_hello(diag: Diagnostics) -> None:
     assert problem.output_validators.uses_default(problem.format_version, problem.metadata), 'Hello uses the default validator'
 
 
-def test_load_twice(diag: Diagnostics):
+def test_load_twice(diag: Diagnostics) -> None:
     probdir = example_directory('hello')
 
     model.load_problem(probdir, diag)
diff --git a/tests/test_xss.py b/tests/test_xss.py
index 821a91b8..5a6ce556 100644
--- a/tests/test_xss.py
+++ b/tests/test_xss.py
@@ -1,11 +1,13 @@
 import os
 import tempfile
+from pathlib import Path
 
 from problemtools import problem2html, problem2pdf
+from problemtools.diagnostics import Diagnostics
 from tests.conftest import datadir
 
 
-def render(problem_path, diag):
+def render(problem_path: Path, diag: Diagnostics) -> str:
     with tempfile.TemporaryDirectory() as temp_dir:
         args, _unknown = problem2html.get_parser().parse_known_args(
             ['--problem', str(problem_path.resolve()), '--dest-dir', str(temp_dir)]
@@ -16,26 +18,26 @@ def render(problem_path, diag):
             return html
 
 
-def renderpdf(problem_path):
+def renderpdf(problem_path: Path) -> None:
     with tempfile.TemporaryDirectory() as temp_dir:
         outpath = os.path.join(temp_dir, 'out.pdf')
         args, _unknown = problem2pdf.get_parser().parse_known_args(['--problem', str(problem_path.resolve()), '--o', outpath])
         problem2pdf.convert(args)
 
 
-def test_no_xss_statement(diag):
+def test_no_xss_statement(diag: Diagnostics) -> None:
     problem_path = datadir() / 'problems' / 'statementxss'
     html = render(problem_path, diag)
     assert 'alert' not in html
 
 
-def test_no_xss_problemname(diag):
+def test_no_xss_problemname(diag: Diagnostics) -> None:
     problem_path = datadir() / 'problems' / 'problemnamexss'
     html = render(problem_path, diag)
     assert '<script>' not in html
 
 
-def test_no_xss_sample(diag):
+def test_no_xss_sample(diag: Diagnostics) -> None:
     problem_path = datadir() / 'problems' / 'samplexss'
     html = render(problem_path, diag)
     assert '<script>' not in html

From af3868df0f14262aab28fc43c9e230b6ba38c636 Mon Sep 17 00:00:00 2001
From: Gunnar Kreitz <gkreitz@kattis.com>
Date: Wed, 2 Sep 2026 14:27:45 +0200
Subject: [PATCH 6/6] Add mypy coverage of tests

---
 .github/workflows/python-app.yml | 3 +++
 run_tests.sh                     | 1 +
 2 files changed, 4 insertions(+)

diff --git a/.github/workflows/python-app.yml b/.github/workflows/python-app.yml
index 8b6a5d89..0c25180a 100644
--- a/.github/workflows/python-app.yml
+++ b/.github/workflows/python-app.yml
@@ -43,6 +43,9 @@ jobs:
     - name: Run mypy
       run: |
         venv/bin/mypy --non-interactive --config-file mypy.ini -p problemtools
+    - name: Run mypy on tests
+      run: |
+        venv/bin/mypy --non-interactive --config-file mypy.ini --exclude '^tests/data/' tests/
 
   packages: # Use a separate job to test debian packaging to speed things up (no need to test this for every python version above)
     runs-on: ubuntu-latest
diff --git a/run_tests.sh b/run_tests.sh
index 2cc673b8..dd29bd3c 100755
--- a/run_tests.sh
+++ b/run_tests.sh
@@ -10,3 +10,4 @@ venv/bin/ruff check .
 venv/bin/ruff format --check .
 venv/bin/pytest
 venv/bin/mypy --non-interactive --config-file mypy.ini -p problemtools
+venv/bin/mypy --non-interactive --config-file mypy.ini --exclude '^tests/data/' tests/