Merge branch 'master' of github.com:tlvince/scripts

README.txt

@@ -1,2 +1,6 @@
-Various multi-purpose Python scripts that are too small to warrant their own
+Various multi-purpose scripts that are too small to warrant their own
 repository. See individual files for usage descriptions.
+
+This repository contains a branch per language.
+
+This is the Python branch.

chromium-update.py

@@ -45,7 +45,7 @@ def download(remote, local):
 def verifyDownload(fileSize, local):
     """Make some attempts to check the file was downloaded correctly."""
     try:
-        assert os.stat(local).st_size == total
+        assert os.stat(local).st_size == fileSize
     except AssertionError:
         logging.warning("Downloaded file size does not match expect.")
         logging.info("The file maybe corrupt or incomplete.")

cmd.py

@@ -4,23 +4,13 @@
 """Setup cmd with some sane defaults."""
 
 import os
-import ctypes
 import subprocess
 import getpass
-import sys
 import logging
 
 from urllib.request import getproxies_registry
 from urllib.parse import urlparse
 
-def getMyDocs():
-    """Return the path to My Documents.
-    http://stackoverflow.com/questions/3927259/how-do-you-get-the-exact-path-to-my-documents/3927493#3927493"""
-    dll = ctypes.windll.shell32
-    buf = ctypes.create_unicode_buffer(300)
-    dll.SHGetSpecialFolderPathW(None, buf, 0x0005, False)
-    return buf.value
-
 def setProxies(user, password, host, port):
     """Set proxy variables for the currently running shell."""
     schemes = ["http", "https"]
@@ -34,12 +24,6 @@ def setProxies(user, password, host, port):
             "{0}://{1}:{2}@{3}:{4}".format(base, user, password, host, port)
         )
 
-def postHook():
-    """Perform some platform-specific post operations."""
-    if sys.platform == "win32":
-        cmd = ["cmd.exe", "/k", "cd", getMyDocs()]
-        subprocess.call(cmd)
-
 def getProxyHost():
     """Return the proxy host and port from Windows registry."""
     try:
@@ -60,7 +44,9 @@ def main():
         port = input("Port: ")
 
     setProxies(un, pw, host, port)
-    postHook()
+
+    # Start a child shell to inherit the new environment variables
+    subprocess.call(["cmd.exe", "/k", "title", "cmd"])
 
 if __name__ == "__main__":
     main()

sanitise.py

@@ -6,6 +6,7 @@
 import argparse
 import string
 import unicodedata
+import re
 
 def removeAccents(str):
     """Remove any form of UTF-8 accents.
@@ -15,19 +16,33 @@ def removeAccents(str):
     nkfd_form = unicodedata.normalize('NFKD', str)
     return "".join([c for c in nkfd_form if not unicodedata.combining(c)])
 
-def sanitise(str):
-    """Perform substitutions and return the string."""
+def regexSanitise(str):
+    """Perform detailed sanitising substitutions using regex."""
 
-    str = removeAccents(str)
+    # List of (pattern, replacement) tuples
+    regex = [
+        ("&", "and"),           # Replace ampersand with a safe string
+        ("( |_)", "-"),         # See: http://webmasters.stackexchange.com/q/374
+        ("(\.|-){2,}", "\\1"),  # Flatten a series of two or more dots or dashes
+        ("^-", ""),             # Remove a leading dash
+        ("(-$|\.$)", ""),       # Remove a trailing dash or dot
+    ]
+
+    for handler in regex:
+        pattern, replacement = handler
+        str = re.sub(pattern, replacement, str)
 
-    str = str.replace("&", "and")
-    str = str.replace("_", "-")
-    str = str.replace(" ", "-")
+    return str
 
-    valid = string.ascii_letters + string.digits + "-."
-    str = "".join([chr for chr in str if chr in valid])
+def sanitise(str):
+    """Perform substitutions and return the string."""
+    str = str.lower()
+    str = removeAccents(str)
+    str = regexSanitise(str)
 
-    return str.lower()
+    # Permit only letters, digits, dash (seperator) and dot (file extension)
+    valid = string.ascii_lowercase + string.digits + "-."
+    return "".join([chr for chr in str if chr in valid])
 
 def parseArguments():
     """Parse the command-line arguments."""

test/test-sanitise.py

@@ -12,11 +12,20 @@ def testAll():
         "éèêàùçÇ": "eeeaucc",
         "CRaZyCASE": "crazycase",
         "Bill & Ted": "bill-and-ted",
-        "file.mp3": "file.mp3",
+        "file.MP3": "file.mp3",
+        ".dotfile.txt": ".dotfile.txt",
+        "---heLLO": "hello",
+        "infix----dashes": "infix-dashes",
+        "---hello....txt.": "hello.txt",
     }
 
+    i = 1
     for case in cases:
-        assert sanitise(case) == cases[case]
+        expected = cases[case]
+        actual = sanitise(case)
+        print("({0}) {1}: {2} -> {3}".format(i, case, expected, actual))
+        assert expected == actual
+        i += 1
 
 def main():
     """Start execution of test-sanitise."""