improve checker caching: file hash instead of mtime

1 files changed, 9 insertions, 6 deletions

diff --git a/check_checkers.py b/check_checkers.py
index e46bb12..4b78509 100755
--- a/check_checkers.py
+++ b/check_checkers.py
@@ -1,4 +1,5 @@
 #!/usr/bin/env python3
+import hashlib
 import html
 import json
 import os
@@ -153,7 +154,7 @@ class PyrightStrict(Pyright):
 
 class Puzzle(TypedDict):
     checker_results: Dict[str, List[Error]]
-    last_modified: int
+    puzzle_hash: str
 
 
 def run_checkers(checkers: List[Checker], puzzle: str, typeshed_path: str):
@@ -225,13 +226,16 @@ def run(
 
     for puzzle in puzzles:
         print(puzzle)
-        last_modified = int(os.stat(puzzle).st_mtime)
-        if puzzle in cache and last_modified == cache[puzzle]['last_modified']:
+        with open(puzzle, 'rb') as f:
+            code = f.read()
+
+        puzzle_hash = hashlib.sha256(code).hexdigest()
+        if puzzle in cache and puzzle_hash == cache[puzzle]['puzzle_hash']:
             checker_results = cache[puzzle]['checker_results']
         else:
             checker_results = run_checkers(checkers, puzzle, default_typeshed)
             cache[puzzle] = {
-                'last_modified': last_modified,
+                'puzzle_hash': puzzle_hash,
                 'checker_results': checker_results,
             }
 
@@ -241,8 +245,7 @@ def run(
             .replace('_', '-')
         )
         out.write('<tr><td id="{0}"><a href="#{0}">{0}</a>'.format(anchor))
-        with open(puzzle) as f:
-            code = f.read()
+        code = code.decode('utf-8')
         out.write(highlight(code, python_lexer, html_formatter))
         error_ok = '# error' in code or '# maybe error' in code
         no_error_ok = '# error' not in code