[lit] Always quote arguments containing '[' on windows

[mstorsjo]

Summary

This avoids breaking clang-tidy/infrastructure/validate-check-names.cpp
if 'not' is evaluated as a lit internal tool (making TestRunner
invoke 'grep' directly in that test, instead of invoking 'not', which
then invokes 'grep').

The quoting of arguments is still brittle if the executable is an
MSYS based tool though, as MSYS based tools incorrectly unescape
backslashes in quoted arguments (contrary to regular win32 argument
parsing rules), see D99406 and
https://github.com/msys2/msys2-runtime/issues/36 for more examples
of the issues.

[mstorsjo]

Updated to include a testcase for the cases it fixes, and a clear example of an argument that this particular workaround still doesn't handle.

[rnk]

lgtm

llvm/test/Other/lit-quoting.txt

18–19 ↗

(On Diff #335464)

RUN lines that don't work as expected are often written as RUNX: to disable them. Maybe that's an old convention.

llvm/utils/lit/lit/TestRunner.py

def quote_windows_command(seq):
 
     Some core util distributions (MSys) don't tokenize command line arguments
     the same way that MSVC CRT does. Lit rolls its own quoting logic similar to
     the stock CPython logic to paper over these quoting and tokenization rule
     differences.
 
     We use the same algorithm from MSDN as CPython
     (http://msdn.microsoft.com/en-us/library/17w5ykft.aspx), but we treat more
-    characters as needing quoting, such as double quotes themselves.
+    characters as needing quoting, such as double quotes themselves, and square
+    brackets.
+
+    For MSys based tools, this is very brittle though, because quoting an
+    argument makes the MSys based tool unescape backslashes where it shouldn't
+    (e.g. "a\b\\c\\\\d" becomes "a\b\c\\d" where it should stay as it was,
+    according to regular win32 command line parsing rules).
     """
     result = []
     needquote = False
     for arg in seq:
         bs_buf = []
 
         # Add a space to separate this argument from the others
         if result:
             result.append(' ')
 
         # This logic differs from upstream list2cmdline.
-        needquote = (" " in arg) or ("\t" in arg) or ("\"" in arg) or not arg
+        needquote = (" " in arg) or ("\t" in arg) or ("\"" in arg) or ("[" in arg) or not arg
         if needquote:
             result.append('"')
 
         for c in arg:
             if c == '\\':
                 # Don't know if we need to double yet.
                 bs_buf.append(c)
             elif c == '"':