00335: Backport pathfix change

Tools/scripts/pathfix.py backports
Add -k and -a command line options to preserve and add shebang flags
In upstream since 3.8: https://bugs.python.org/issue37064
Assume all .py files are Python scripts when working recursively:
In upstream since 3.8: https://bugs.python.org/issue38347

Co-authored-by: Victor Stinner <vstinner@redhat.com>

Author: 2 people

Lib/test/test_tools/test_pathfix.py

@@ -0,0 +1,129 @@
+import os
+import subprocess
+import sys
+import unittest
+from test import support
+from test.test_tools import import_tool, scriptsdir, skip_if_missing
+
+
+# need Tools/script/ directory: skip if run on Python installed on the system
+skip_if_missing()
+
+
+class TestPathfixFunctional(unittest.TestCase):
+    script = os.path.join(scriptsdir, 'pathfix.py')
+
+    def setUp(self):
+        self.addCleanup(support.unlink, support.TESTFN)
+
+    def pathfix(self, shebang, pathfix_flags, exitcode=0, stdout='', stderr='',
+                directory=''):
+        if directory:
+            # bpo-38347: Test filename should contain lowercase, uppercase,
+            # "-", "_" and digits.
+            filename = os.path.join(directory, 'script-A_1.py')
+            pathfix_arg = directory
+        else:
+            filename = support.TESTFN
+            pathfix_arg = filename
+
+        with open(filename, 'w', encoding='utf8') as f:
+            f.write(f'{shebang}\n' + 'print("Hello world")\n')
+
+        proc = subprocess.run(
+            [sys.executable, self.script,
+             *pathfix_flags, '-n', pathfix_arg],
+            capture_output=True, text=1)
+
+        if stdout == '' and proc.returncode == 0:
+            stdout = f'{filename}: updating\n'
+        self.assertEqual(proc.returncode, exitcode, proc)
+        self.assertEqual(proc.stdout, stdout, proc)
+        self.assertEqual(proc.stderr, stderr, proc)
+
+        with open(filename, 'r', encoding='utf8') as f:
+            output = f.read()
+
+        lines = output.split('\n')
+        self.assertEqual(lines[1:], ['print("Hello world")', ''])
+        new_shebang = lines[0]
+
+        if proc.returncode != 0:
+            self.assertEqual(shebang, new_shebang)
+
+        return new_shebang
+
+    def test_recursive(self):
+        tmpdir = support.TESTFN + '.d'
+        self.addCleanup(support.rmtree, tmpdir)
+        os.mkdir(tmpdir)
+        expected_stderr = f"recursedown('{os.path.basename(tmpdir)}')\n"
+        self.assertEqual(
+            self.pathfix(
+                '#! /usr/bin/env python',
+                ['-i', '/usr/bin/python3'],
+                directory=tmpdir,
+                stderr=expected_stderr),
+            '#! /usr/bin/python3')
+
+    def test_pathfix(self):
+        self.assertEqual(
+            self.pathfix(
+                '#! /usr/bin/env python',
+                ['-i', '/usr/bin/python3']),
+            '#! /usr/bin/python3')
+        self.assertEqual(
+            self.pathfix(
+                '#! /usr/bin/env python -R',
+                ['-i', '/usr/bin/python3']),
+            '#! /usr/bin/python3')
+
+    def test_pathfix_keeping_flags(self):
+        self.assertEqual(
+            self.pathfix(
+                '#! /usr/bin/env python -R',
+                ['-i', '/usr/bin/python3', '-k']),
+            '#! /usr/bin/python3 -R')
+        self.assertEqual(
+            self.pathfix(
+                '#! /usr/bin/env python',
+                ['-i', '/usr/bin/python3', '-k']),
+            '#! /usr/bin/python3')
+
+    def test_pathfix_adding_flag(self):
+        self.assertEqual(
+            self.pathfix(
+                '#! /usr/bin/env python',
+                ['-i', '/usr/bin/python3', '-a', 's']),
+            '#! /usr/bin/python3 -s')
+        self.assertEqual(
+            self.pathfix(
+                '#! /usr/bin/env python -S',
+                ['-i', '/usr/bin/python3', '-a', 's']),
+            '#! /usr/bin/python3 -s')
+        self.assertEqual(
+            self.pathfix(
+                '#! /usr/bin/env python -V',
+                ['-i', '/usr/bin/python3', '-a', 'v', '-k']),
+            '#! /usr/bin/python3 -vV')
+        self.assertEqual(
+            self.pathfix(
+                '#! /usr/bin/env python',
+                ['-i', '/usr/bin/python3', '-a', 'Rs']),
+            '#! /usr/bin/python3 -Rs')
+        self.assertEqual(
+            self.pathfix(
+                '#! /usr/bin/env python -W default',
+                ['-i', '/usr/bin/python3', '-a', 's', '-k']),
+            '#! /usr/bin/python3 -sW default')
+
+    def test_pathfix_adding_errors(self):
+        self.pathfix(
+            '#! /usr/bin/env python -E',
+            ['-i', '/usr/bin/python3', '-a', 'W default', '-k'],
+            exitcode=2,
+            stderr="-a option doesn't support whitespaces")
+
+
+if __name__ == '__main__':
+    unittest.main()

Misc/NEWS.d/next/Tools-Demos/2019-05-27-15-26-12.bpo-37064.k_SPW2.rst

@@ -0,0 +1,2 @@
+Add option -k to pathscript.py script: preserve shebang flags.
+Add option -a to pathscript.py script: add flags.

Tools/scripts/pathfix.py

@@ -1,6 +1,6 @@
 #!/usr/bin/env python3
 
-# Change the #! line occurring in Python scripts.  The new interpreter
+# Change the #! line (shebang) occurring in Python scripts.  The new interpreter
 # pathname must be given with a -i option.
 #
 # Command line arguments are files or directories to be processed.
@@ -10,7 +10,13 @@
 # arguments).
 # The original file is kept as a back-up (with a "~" attached to its name),
 # -n flag can be used to disable this.
-#
+
+# Sometimes you may find shebangs with flags such as `#! /usr/bin/env python -si`.
+# Normally, pathfix overwrites the entire line, including the flags.
+# To change interpreter and keep flags from the original shebang line, use -k.
+# If you want to keep flags and add to them one single literal flag, use option -a.
+
+
 # Undoubtedly you can do this using find and sed or perl, but this is
 # a nice example of Python code that recurses down a directory tree
 # and uses regular expressions.  Also note several subtleties like
@@ -33,16 +39,21 @@
 new_interpreter = None
 preserve_timestamps = False
 create_backup = True
+keep_flags = False
+add_flags = b''
 
 
 def main():
     global new_interpreter
     global preserve_timestamps
     global create_backup
-    usage = ('usage: %s -i /interpreter -p -n file-or-directory ...\n' %
+    global keep_flags
+    global add_flags
+
+    usage = ('usage: %s -i /interpreter -p -n -k -a file-or-directory ...\n' %
              sys.argv[0])
     try:
-        opts, args = getopt.getopt(sys.argv[1:], 'i:pn')
+        opts, args = getopt.getopt(sys.argv[1:], 'i:a:kpn')
     except getopt.error as msg:
         err(str(msg) + '\n')
         err(usage)
@@ -54,6 +65,13 @@ def main():
             preserve_timestamps = True
         if o == '-n':
             create_backup = False
+        if o == '-k':
+            keep_flags = True
+        if o == '-a':
+            add_flags = a.encode()
+            if b' ' in add_flags:
+                err("-a option doesn't support whitespaces")
+                sys.exit(2)
     if not new_interpreter or not new_interpreter.startswith(b'/') or \
            not args:
         err('-i option or file-or-directory missing\n')
@@ -96,6 +114,7 @@ def recursedown(dirname):
         if recursedown(fullname): bad = 1
     return bad
 
+
 def fix(filename):
 ##  dbg('fix(%r)\n' % (filename,))
     try:
@@ -166,12 +185,43 @@ def fix(filename):
     # Return success
     return 0
 
+
+def parse_shebang(shebangline):
+    shebangline = shebangline.rstrip(b'\n')
+    start = shebangline.find(b' -')
+    if start == -1:
+        return b''
+    return shebangline[start:]
+
+
+def populate_flags(shebangline):
+    old_flags = b''
+    if keep_flags:
+        old_flags = parse_shebang(shebangline)
+        if old_flags:
+            old_flags = old_flags[2:]
+    if not (old_flags or add_flags):
+        return b''
+    # On Linux, the entire string following the interpreter name
+    # is passed as a single argument to the interpreter.
+    # e.g. "#! /usr/bin/python3 -W Error -s" runs "/usr/bin/python3 "-W Error -s"
+    # so shebang should have single '-' where flags are given and
+    # flag might need argument for that reasons adding new flags is
+    # between '-' and original flags
+    # e.g. #! /usr/bin/python3 -sW Error
+    return b' -' + add_flags + old_flags
+
+
 def fixline(line):
     if not line.startswith(b'#!'):
         return line
+
     if b"python" not in line:
         return line
-    return b'#! ' + new_interpreter + b'\n'
+
+    flags = populate_flags(line)
+    return b'#! ' + new_interpreter + flags + b'\n'
+
 
 if __name__ == '__main__':
     main()