intermediate checkin

cmp_by_hash.py

@@ -0,0 +1,70 @@
+#!python3
+# Created 18 Sep 2019
+# @author: andreas
+
+""" Module: Main Template  """
+
+from argparse import ArgumentParser
+import logging
+import sys
+import os
+import hashlib
+from pathlib import Path
+import datetime as dt
+import pickle
+import shutil
+
+LOG = logging.getLogger()
+PROXY_DIR = "."
+
+   
+def cli(argv=None):
+    # command line interface
+    if argv is None:
+        argv = sys.argv
+    LOG.info("%s %s", os.path.basename(argv[0]), " ".join(argv[1:]))
+    parser = ArgumentParser(description="Module Template")
+    parser.add_argument("--src")
+    parser.add_argument("--dest")
+    args = parser.parse_args(argv[1:])
+    argd = vars(args)
+    # arguments
+    for k, v in argd.items(): 
+        print(k, v)
+    # feature
+    srcd = build_hashdict(Path(args.src))
+    destd = build_hashdict(Path(args.dest))
+    srch = set(srcd.keys())
+    desth = set(destd.keys())
+    xor_hashes = srch ^ desth
+    for h in xor_hashes:
+        if h in srcd:
+            print(srcd[h])
+    for h in xor_hashes:
+        if h in destd:
+            print(destd[h])
+    LOG.info("done")
+    return 0
+
+
+def build_hashdict(base):
+    files_by_hash = dict()
+    for fp in base.glob("**/*.jpg"):
+        h = file_hash(fp)
+        files_by_hash[h] = fp
+    return files_by_hash    
+  
+
+def file_hash(filepath):
+    h = hashlib.sha256()
+    with filepath.open("rb") as fh:
+        h.update(fh.read())
+    return h.digest()
+
+
+if __name__ == "__main__":
+    logging.Formatter.default_time_format = '%H:%M:%S'
+    logging.Formatter.default_msec_format = '%s.%03d'
+    logging.basicConfig(level=logging.DEBUG,
+                        format='%(asctime)s [%(process)i] %(levelname).4s %(module)s.%(funcName)s: %(message)s')
+    sys.exit(cli())