Fix hash when Program.source_files contains a subdirectory (#339)

bin/program.py

@@ -405,6 +405,10 @@ def build(self, bar):
         for f in self.source_files:
             ensure_symlink(self.tmpdir / f.name, f)
             self.input_files.append(self.tmpdir / f.name)
+            if not f.is_file():
+                self.ok = False
+                self.bar.error(f'{str(f)} is not a file')
+                return False
             hashes.append(hash_file(f))
         self.hash = combine_hashes(hashes)
 

bin/util.py

@@ -13,6 +13,7 @@
 import hashlib
 import tempfile
 import yaml as yamllib
+import errno
 
 from enum import Enum
 from pathlib import Path
@@ -1000,15 +1001,36 @@ def inc_label(label):
     return 'A' + label
 
 
+def combine_hashes(values):
+    values.sort()
+    hasher = hashlib.sha256(usedforsecurity=False)
+    for item in values:
+        hasher.update(item.encode())
+    return hasher.hexdigest()
+
+
+def combine_hashes_dict(d):
+    hasher = hashlib.sha256(usedforsecurity=False)
+    for key in d:
+        hasher.update(key.encode())
+        if d[key] is not None:
+            hasher.update(d[key].encode())
+    return hasher.hexdigest()
+
+
 def hash_string(string):
     sha = hashlib.sha256(usedforsecurity=False)
     sha.update(string.encode())
     return sha.hexdigest()
 
 
 def hash_file(file, buffer_size=65536):
-    assert file.is_file(), f"File {file} does not exist"
+    if not file.is_file():
+        raise FileNotFoundError(errno.ENOENT, os.strerror(errno.ENOENT), str(file))
     sha = hashlib.sha256(usedforsecurity=False)
+    name = file.name.encode('utf-8')
+    sha.update(len(name).to_bytes(8))
+    sha.update(name)
 
     with open(file, 'rb') as f:
         while True:
@@ -1020,18 +1042,10 @@ def hash_file(file, buffer_size=65536):
     return sha.hexdigest()
 
 
-def combine_hashes(list):
-    list.sort()
-    hasher = hashlib.sha256(usedforsecurity=False)
-    for item in list:
-        hasher.update(item.encode())
-    return hasher.hexdigest()
-
-
-def combine_hashes_dict(d):
-    hasher = hashlib.sha256(usedforsecurity=False)
-    for key in d:
-        hasher.update(key.encode())
-        if d[key] is not None:
-            hasher.update(d[key].encode())
-    return hasher.hexdigest()
+def hash_file_or_dir(file_or_dir, buffer_size=65536):
+    if file_or_dir.is_dir():
+        return combine_hashes(
+            [hash_string(file_or_dir.name)] + [hash_file_or_dir(f) for f in file_or_dir.iterdir()]
+        )
+    else:
+        return hash_file(file_or_dir)