From 0cfd9a7f26488567b9a3e5ec192099a8b80ad9df Mon Sep 17 00:00:00 2001

From: Lumir Balhar <lbalhar@redhat.com>

Date: Tue, 19 Jan 2021 07:55:37 +0100

Subject: [PATCH] [PATCH] bpo-37064: Add -k and -a options to pathfix.py tool

 (GH-16387)

* bpo-37064: Add option -k to Tools/scripts/pathfix.py (GH-15548)

Add flag -k to pathscript.py script: preserve shebang flags.

(cherry picked from commit 50254ac4c179cb412e90682098c97db786143929)

* bpo-37064: Add option -a to pathfix.py tool (GH-15717)

Add option -a to Tools/Scripts/pathfix.py script: add flags.

(cherry picked from commit 1dc1acbd73f05f14c974b7ce1041787d7abef31e)

---

 Lib/test/test_tools/test_pathfix.py | 104 ++++++++++++++++++++++++++++

 Tools/scripts/pathfix.py            |  64 +++++++++++++++--

 2 files changed, 163 insertions(+), 5 deletions(-)

 create mode 100644 Lib/test/test_tools/test_pathfix.py

diff --git a/Lib/test/test_tools/test_pathfix.py b/Lib/test/test_tools/test_pathfix.py

new file mode 100644

index 0000000..1f0585e

--- /dev/null

+++ b/Lib/test/test_tools/test_pathfix.py

@@ -0,0 +1,104 @@

+import os

+import subprocess

+import sys

+import unittest

+from test import support

+from test.test_tools import import_tool, scriptsdir

+

+

+class TestPathfixFunctional(unittest.TestCase):

+    script = os.path.join(scriptsdir, 'pathfix.py')

+

+    def setUp(self):

+        self.temp_file = support.TESTFN

+        self.addCleanup(support.unlink, support.TESTFN)

+

+    def pathfix(self, shebang, pathfix_flags, exitcode=0, stdout='', stderr=''):

+        with open(self.temp_file, 'w', encoding='utf8') as f:

+            f.write(f'{shebang}\n' + 'print("Hello world")\n')

+

+        proc = subprocess.run(

+            [sys.executable, self.script,

+             *pathfix_flags, '-n', self.temp_file],

+            universal_newlines=True, stdout=subprocess.PIPE,

+            stderr=subprocess.PIPE)

+

+        if stdout == '' and proc.returncode == 0:

+            stdout = f'{self.temp_file}: updating\n'

+        self.assertEqual(proc.returncode, exitcode, proc)

+        self.assertEqual(proc.stdout, stdout, proc)

+        self.assertEqual(proc.stderr, stderr, proc)

+

+        with open(self.temp_file, '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_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()

diff --git a/Tools/scripts/pathfix.py b/Tools/scripts/pathfix.py

index c5bf984..2dfa6e8 100755

--- a/Tools/scripts/pathfix.py

+++ b/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 @@ rep = sys.stdout.write

 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')

@@ -70,10 +88,14 @@ def main():

             if fix(arg): bad = 1

     sys.exit(bad)

+

 ispythonprog = re.compile(r'^[a-zA-Z0-9_]+\.py$')

+

+

 def ispython(name):

     return bool(ispythonprog.match(name))

+

 def recursedown(dirname):

     dbg('recursedown(%r)\n' % (dirname,))

     bad = 0

@@ -96,6 +118,7 @@ def recursedown(dirname):

         if recursedown(fullname): bad = 1

     return bad

+

 def fix(filename):

 ##  dbg('fix(%r)\n' % (filename,))

     try:

@@ -166,12 +189,43 @@ 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()

-- 

2.29.2