Merge pull request #6 from expyriment/new_dif_procedure

New dif procedure

Author: lindemann09

dataintegrityfingerprint/__init__.py

@@ -1,7 +1,7 @@
 __author__ = 'Oliver Lindemann <oliver@expyriment.org>, ' \
              'Florian Krause <florian@expyriment.org>'
 
-__version__ = '0.5.4'
+__version__ = '0.6.0'
 
 from sys import version_info as _vi
 if _vi.major< 3:

dataintegrityfingerprint/cli.py

@@ -5,6 +5,7 @@
 from . import DataIntegrityFingerprint
 from . import __version__
 
+
 def run():
 
     def progress(count, total, status=''):
@@ -16,15 +17,19 @@ def progress(count, total, status=''):
         sys.stdout.flush()
 
     parser = argparse.ArgumentParser(
-        description="Create a Data Integrity Fingerprint (DIF). v{0}".format(
-        __version__),
-        epilog="(c) O. Lindemann & F. Krause")
+        description="""Data Integrity Fingerprint (DIF)
+Python Reference Implementation v{0}""".format(__version__),
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        epilog="""Authors:
+Oliver Lindemann <oliver@expyriment.org>
+Florian Krause <florian@expyriment.org""")
 
     parser.add_argument("PATH", nargs='?', default=None,
                         help="the path to the data folder or file")
-    parser.add_argument("-f", "--from-checksums-file", dest="fromchecksumsfile",
-                        action="store_true",
-                        help="Calculate dif from checksums file. PATH is a checksums file",
+    parser.add_argument("-f", "--from-checksums-file",
+                        dest="fromchecksumsfile", action="store_true",
+                        help="Calculate dif from checksums file. " +
+                             "PATH is a checksums file",
                         default=False)
     parser.add_argument("-a", "--algorithm", metavar="ALGORITHM",
                         type=str,
@@ -46,13 +51,14 @@ def progress(count, total, status=''):
                         action="store_true",
                         help="print available algorithms",
                         default=False)
-    parser.add_argument("-s", "--save-checksums-file", dest="savechecksumsfile",
-                        action="store_true",
+    parser.add_argument("-s", "--save-checksums-file",
+                        dest="savechecksumsfile", action="store_true",
                         help="save checksums to file",
                         default=False)
     parser.add_argument("-d", "--diff-checksums-file", metavar="CHECKSUMSFILE",
                         type=str,
-                        help="Calculate differences of checksums file to CHECKSUMSFILE")
+                        help="Calculate differences of checksums file to " +
+                             "CHECKSUMSFILE")
     parser.add_argument("-n", "--no-multi-processing", dest="nomultiprocess",
                         action="store_true",
                         help="switch of multi processing",
@@ -64,10 +70,10 @@ def progress(count, total, status=''):
     parser.add_argument("--non-crypthographic",
                         dest="noncrypto",
                         action="store_true",
-                        help="allow non crypthographic algorithms (Not suggested, please " +
-                             "read documentation carefully!) ",
+                        help="allow non crypthographic algorithms " +
+                             "(Not suggested, please read documentation " +
+                             "carefully!)",
                         default=False)
-
     args = vars(parser.parse_args())
 
     if args['listalgos']:
@@ -100,7 +106,7 @@ def progress(count, total, status=''):
         dif.generate(progress=progress)
         print("")
 
-    #output
+    # Output
     if args['savechecksumsfile']:
         outfile = os.path.split(
             dif.data)[-1] + ".{0}".format(dif._hash_algorithm)
@@ -124,15 +130,13 @@ def progress(count, total, status=''):
     elif args['checksums']:
         print(dif.checksums.strip())
     else:
-        print("Data Integrity Fingerprint (DIF) v{0}".format(__version__))
-
-        print("\nFolder: {0}".format(dif.data))
+        print("Data Integrity Fingerprint (DIF)".format(__version__))
+        print("")
+        print("Folder: {0}".format(dif.data))
         print("Files: {0} included".format(dif.count_files()))
         print("Algorithm: {0}".format(dif.hash_algorithm))
         print("DIF: {}".format(dif))
 
 
-
-
 if __name__ == "__main__":
     run()

dataintegrityfingerprint/dif.py

@@ -30,7 +30,6 @@ class DataIntegrityFingerprint:
 
     CRYPTOGRAPHIC_ALGORITHMS = OpenSSLHashAlgorithm.SUPPORTED_ALGORITHMS
     NON_CRYPTOGRAPHIC_ALGORITHMS = ZlibHashAlgorithm.SUPPORTED_ALGORITHMS
-    SEPARATOR = "_"
     CHECKSUM_FILENAME_SEPARATOR = "  "
 
     def __init__(self, data, from_checksums_file=False,
@@ -105,33 +104,24 @@ def file_hash_list(self):
     @property
     def checksums(self):
         rtn = ""
-        for h, fl in self.file_hash_list:
+        for h, fl in sorted(self.file_hash_list, key=lambda x: x[1]):
             rtn += u"{0}{1}{2}\n".format(h, self.CHECKSUM_FILENAME_SEPARATOR,
                                          fl)
         return rtn
 
     @property
-    def prefix(self):
-        return "".join(x for x in self.hash_algorithm.lower() if x.isalnum())
-
-    @property
-    def postfix(self):
+    def dif(self):
         if len(self.file_hash_list) < 1:
             return None
 
         hasher = new_hash_instance(self._hash_algorithm,
                                    self.allow_non_cryptographic_algorithms)
-        hasher.update(self.checksums.encode("utf-8"))
+        concat = "".join(["".join(x) for x in self.file_hash_list])
+        hasher.update(concat.encode("utf-8"))
         return hasher.checksum
 
-    @property
-    def dif(self):
-        if self.postfix is not None:
-            return self.prefix + self.SEPARATOR + self.postfix
-
     def count_files(self):
-        """Returns the number of files that are included in the dif
-        """
+        """Return the number of files that are included in the dif."""
 
         return len(self.get_files())
 

dataintegrityfingerprint/gui.py

@@ -17,6 +17,7 @@
 import tkinter.ttk as ttk
 from tkinter import filedialog, messagebox
 
+from . import __version__
 from .dif import DataIntegrityFingerprint as DIF
 from .dif import new_hash_instance
 
@@ -30,13 +31,12 @@ def __init__(self, master, *args, **kwargs):
         self.master = master
         self.master.title("Data Integrity Fingerprint (DIF)")
         self.about_text = """Data Integrity Fingerprint (DIF)
-
-Reference Python implementation
+Reference Python implementation v{0}
 
 Authors:
 Oliver Lindemann <oliver@expyriment.org>
 Florian Krause <florian@expyriment.org>
-"""
+""".format(__version__)
         self.dif = None
         self.create_widgets()
         self.dir_button.focus()
@@ -88,6 +88,7 @@ def create_widgets(self):
         self.algorithm_menu = tk.Menu(self.menubar)
         self.algorithm_var = tk.StringVar()
         self.algorithm_var.set("SHA-256")
+        self.algorithm_var.trace('w', self.set_dif_label)
         for algorithm in DIF.CRYPTOGRAPHIC_ALGORITHMS:
             self.algorithm_menu.add_radiobutton(label=algorithm,
                                                 value=algorithm,
@@ -172,8 +173,9 @@ def create_widgets(self):
         self.frame2 = ttk.Frame(self.master)
         self.frame2.grid(row=3, column=0, sticky="NSWE")
         self.frame2.grid_columnconfigure(1, weight=1)
-        self.dif_label = ttk.Label(self.frame2, text="DIF:")
+        self.dif_label = ttk.Label(self.frame2)
         self.dif_label.grid(row=0, column=0)
+        self.set_dif_label()
         self.dif_var = tk.StringVar()
         self.dif_var.set("")
         self.dif_entry = ttk.Entry(self.frame2, textvariable=self.dif_var,
@@ -208,6 +210,10 @@ def set_data_directory(self, data_dir=None, *args):
             self.statusbar["text"] = "Ready"
             self.unblock_gui(enable_save_checksums=False)
 
+    def set_dif_label(self, *args):
+        self.dif_label.config(text="DIF [{0}]:".format(
+            self.algorithm_var.get()))
+
     def block_gui(self):
         """Block GUI from user entry."""
 
@@ -266,6 +272,7 @@ def _progress(count, total, status=''):
         self.checksum_list.insert(1.0, self.dif.checksums.strip("\n"))
         self.checksum_list["state"] = tk.DISABLED
         self.dif_var.set(self.dif.dif)
+        self.set_dif_label()
         self.copy_button["state"] = tk.NORMAL
         self.copy_button.focus()
         self.statusbar["text"] = "Done"
@@ -304,6 +311,7 @@ def open_checksums(self, *args):
                 self.checksum_list.delete(1.0, tk.END)
                 self.checksum_list.insert(1.0, checksums)
                 self.checksum_list["state"] = tk.DISABLED
+                self.set_dif_label()
                 self.dif_var.set(master_hash)
                 self.copy_button["state"] = tk.NORMAL
                 self.copy_button.focus()
@@ -322,7 +330,7 @@ def save_checksums(self, *args):
         """Save checksums file."""
 
         if self.checksum_list.get(1.0, tk.END).strip("\n") != "":
-            algorithm = self.dif_var.get().split(DIF.SEPARATOR)[0]
+            algorithm = self.algorithm_var.get()
             extension = "".join(x for x in algorithm.lower() if x.isalnum())
             filename = filedialog.asksaveasfilename(
                 defaultextension=extension,