BugFix: allow run_in_windows_bash in MSYS/Cygwin

[SSE4]

related to #8476

running AutoToolsBuildEnvironment(self, win_bash=True) in MSYS2 results in an error:

ERROR: apr/1.7.0: Error in build() method, line 92
        autotools = self._configure_autotools()
while calling '_configure_autotools', line 76
        self._autotools.configure(args=conf_args, configure_dir=self._source_subfolder)
        ConanException: Command only for Windows operating system

the fix allows run_in_windows_bash to work in sub-systems as well.

Changelog: BugFix: Allow run_in_windows_bash in MSYS/Cygwin.
Docs: omit

• Refer to the issue that supports this Pull Request.
• If the issue has missing info, explain the purpose/use case/pain/need that covers this Pull Request.
• I've read the Contributing guide.
• I've followed the PEP8 style guides for Python code.
• I've opened another PR in the Conan docs repo to the develop branch, documenting this one.

^{Note: By default this PR will skip the slower tests and will use a limited set of python versions. Check here how to increase the testing level by writing some tags in the current PR body text.}

[memsharded]

Cygwin also returns "Windows" here, could be only Msys will be different

[SSE4]

didn't check Cygwin (may check tomorrow), msys returns a different value for sure

[memsharded]

The fix seems fine anyway, it was just an observation. Would you like to check something more, or can it be merged?

[SSE4]

for the completeness.
Cygwin:

sse4@DESKTOP-LAD63V0 ~
$ python -c "import platform; print(platform.system())"
CYGWIN_NT-10.0-18363

MSYS2:

sse4@DESKTOP-LAD63V0 MSYS ~
$ python -c "import platform; print(platform.system())"
MSYS_NT-10.0-18363

MinGW32 (MSYS2):

sse4@DESKTOP-LAD63V0 MINGW32 ~
$ python -c "import platform; print(platform.system())"
MINGW32_NT-10.0-18363

MinGW64 (MSYS2):

sse4@DESKTOP-LAD63V0 MINGW64 ~
$ python -c "import platform; print(platform.system())"
MINGW64_NT-10.0-18363

[memsharded]

Curiously, it seems it can change, I am not sure why:

Cygwin:

$ python -c "import platform; print(platform.system())"
Windows

In any case, doesn't seem relevant, if it can change in other computers, need to take it into account.

[SSE4]

this is probably not a python from a Cygwin itself. run which python to check.
if it's /usr/bin/python, then it's cygwin's one.
if it's something like /cygdrive/c/users..., then it's installed in the main Windows system.

[memsharded]

I think this can be merged, but I would like to note that run_in_windows_bash() is very complicated and obscure thing inside Conan codebase. I really want to improve this while we define the new helpers, mostly for Autotools, simplifying when possible, make it a bit more explicit and configurable, and being covered by real tests better.

[SSE4]

checking cygwin, it seems to be not enough:

apr/1.7.0: run_in_windows_bash: /usr/bin/bash --login -c cd "/cygdrive/home/sse4/.conan/data/apr/1.7.0/_/_/build/520c5865652102ac1331f89be33c82aad2520eb9" && /cygdrivesource_subfolder/configure '--with-installbuilddir=${prefix}/bin/build-1' '--enable-shared=no' '--enable-static=yes' '--prefix=/home/sse4/.conan/data/apr/1.7.0/_/_/package/520c5865652102ac1331f89be33c82aad2520eb9'
/bin/sh: /cygdrivesource_subfolder/configure: No such file or directory
apr/1.7.0:
apr/1.7.0: ERROR: Package '520c5865652102ac1331f89be33c82aad2520eb9' build failed
apr/1.7.0: WARN: Build folder /home/sse4/.conan/data/apr/1.7.0/_/_/build/520c5865652102ac1331f89be33c82aad2520eb9
ERROR: apr/1.7.0: Error in build() method, line 94
        autotools = self._configure_autotools()
while calling '_configure_autotools', line 78
        self._autotools.configure(args=conf_args, configure_dir=self._source_subfolder)
        ConanException: Error 127 while executing /cygdrivesource_subfolder/configure '--with-installbuilddir=${prefix}/bin/build-1' '--enable-shared=no' '--enable-static=yes' '--prefix=/home/sse4/.conan/data/apr/1.7.0/_/_/package/520c5865652102ac1331f89be33c82aad2520eb9'

I guess the error is in unix_path function which needs a simple fix:

diff --git a/conans/client/tools/win.py b/conans/client/tools/win.py
index 43db28b5a..bd69a3e47 100644
--- a/conans/client/tools/win.py
+++ b/conans/client/tools/win.py
@@ -620,7 +620,10 @@ def unix_path(path, path_flavor=None):
     if path_flavor in (MSYS, MSYS2):
         return path.lower()
     elif path_flavor == CYGWIN:
-        return '/cygdrive' + path.lower()
+        if os.path.isabs(path):
+            return '/cygdrive' + path.lower()
+        else:
+            return path.lower()
     elif path_flavor == WSL:
         return '/mnt' + path[0:2].lower() + path[2:]
     elif path_flavor == SFU:

which allows the build to start.
actually, it needs a more complete fix (e.g. for other sub-systems, plus it should translate only paths starting with drives like C:, D:, because paths like /home don't need to be adjusted).

[memsharded]

@SSE4 How would you like to proceed? Fixing those things in this PR?

[SSE4]

@SSE4 How would you like to proceed? Fixing those things in this PR?

nope, this deserves its own PR, and I want to write some tests for unix_path.