pymake

commit bcab7973f2109cacb013447226ecba899d920ae4
parent 71e96d09f2ea76e14d1759006ccfe19211a3ddce
Author: gracefu <81774659+gracefuu@users.noreply.github.com>
Date:   Sun, 20 Apr 2025 07:09:34 +0800

Make the two tar sketches more similar

Diffstat:
M
tar-sketch/a.txt
 | 
2
+-
M
tar-sketch/tar.py
 | 
43
++++++++++++++++++++++++++-----------------
M
tar-sketch/tar2.py
 | 
1
-

3 files changed, 27 insertions(+), 19 deletions(-)
diff --git a/tar-sketch/a.txt b/tar-sketch/a.txt
@@ -1 +1 @@
-Sun Apr 20 07:04:58 AM +08 2025
+Sun Apr 20 07:08:59 AM +08 2025
diff --git a/tar-sketch/tar.py b/tar-sketch/tar.py
@@ -3,6 +3,8 @@ import os
 import pickle
 import subprocess
 
+from io import BufferedReader
+
 db: dict = dict()
 
 try:
@@ -14,20 +16,25 @@ except BaseException as e:
 file_modtime = lambda f: os.stat(f.fileno()).st_mtime_ns
 
 
-def file_hash(filename):
-    with open(filename, "rb") as f:
-        modtime = file_modtime(f)
-        if modtime == db.get(f"file:{filename}:modtime", None):
-            print("skipping hash", db[f"file:{filename}:hash"])
-            return db[f"file:{filename}:hash"]
-        db[f"file:{filename}:modtime"] = modtime
-        h = hashlib.sha256()
-        for chunk in f:
-            h.update(chunk)
-        d = h.hexdigest()
-        db[f"file:{filename}:hash"] = d
-        print("hash", filename, d)
-        return d
+def _file_hash(f: BufferedReader):
+    modtime = file_modtime(f)
+    if modtime == db.get(f"file:{f.name}:modtime", None):
+        return db[f"file:{f.name}:hash"]
+    db[f"file:{f.name}:modtime"] = modtime
+    h = hashlib.sha256()
+    for chunk in f:
+        h.update(chunk)
+    d = h.hexdigest()
+    db[f"file:{f.name}:hash"] = d
+    print("hash", f.name, d)
+    return d
+
+
+def file_hash(f: BufferedReader | bytes | str):
+    if isinstance(f, bytes) or isinstance(f, str):
+        with open(f, "rb") as f:
+            return _file_hash(f)
+    return _file_hash(f)
 
 
 def tar(manifest=b"manifest", output=b"archive.tar.gz"):
@@ -45,10 +52,12 @@ def tar(manifest=b"manifest", output=b"archive.tar.gz"):
         file_hashes[manifest] = file_hash(manifest)
         for fname in manifest_lines:
             file_hashes[fname] = file_hash(fname)
-        db["tar:file_hashes"] = file_hashes
 
-    print("tar", "cvzf", output, *manifest_lines)
-    subprocess.run([b"tar", b"cvzf", output, *manifest_lines])
+        print("tar", "cvzf", output, *manifest_lines)
+        subprocess.run([b"tar", b"cvzf", output, *manifest_lines])
+
+        file_hashes[output] = file_hash(output)
+        db["tar:file_hashes"] = file_hashes
 
 
 tar()
diff --git a/tar-sketch/tar2.py b/tar-sketch/tar2.py
@@ -2,7 +2,6 @@ import contextvars
 import functools
 import inspect
 import pickle
-import types
 from typing import Any
 
 db: dict = dict()