bash: Fix CR handling

Fixes #1839

`0005-bash-4.3-msys2-fix-lineendings.patch` adds CRLF support.
However, `0001-bash-4.4-cygwin.patch` already added `igncr` option to
Bash to support CRLF.

I confirmed that the Cygwin version of Bash has also the same issue
with #1839 when the `igncr` option is set.

After debugging, I found that there is an issue in
`rewind_input_string()` in `parser.y` that it doesn't take the CR into
account.

This PR adds the following changes:
* Modify `rewind_input_string()` to take the CR into account.
  (It might be better to apply a similar change to the Cygwin version of
  Bash.)
* Remove all the changes from `y.tab.c`. This file should be
  automatically generated from `parser.y`.
* Set `LC_ALL=C.UTF-8` when running `make check` for running on
  non-English locales.
* Add two tests: `run-ps1lf` and `run-crlf`.

Note: This patch contains a line with CRLF. So, .gitattributes is also
updated to keep the CRLF.

bash/.gitattributes

@@ -0,0 +1 @@
+*-bash-*-msys2-fix-lineendings.patch	-text

bash/0005-bash-4.3-msys2-fix-lineendings.patch

@@ -81,7 +81,7 @@ index c1135ec..b388af6 100644
      *niflp = invfl;
    if  (vlp)
 diff --git a/parse.y b/parse.y
-index 0ae3458..c6e9520 100644
+index 8fd24a1c..232a33dc 100644
 --- a/parse.y
 +++ b/parse.y
 @@ -1459,7 +1459,13 @@ yy_input_name ()
@@ -99,6 +99,23 @@ index 0ae3458..c6e9520 100644
  }
 
  /* Call this to unget C.  That is, to make C the next character
+@@ -1684,7 +1690,15 @@ rewind_input_string ()
+      into account, e.g., $(...\n) */
+   xchars = shell_input_line_len - shell_input_line_index;
+   if (bash_input.location.string[-1] == '\n')
+-    xchars++;
++    {
++      xchars++;
++#ifdef __MSYS__
++      {
++	if (bash_input.location.string[-2] == '\r')
++	  xchars++;
++      }
++#endif
++    }
+
+   /* XXX - how to reflect bash_input.location.string back to string passed to
+      parse_and_execute or xparse_dolparen? xparse_dolparen needs to know how
 diff --git a/shell.c b/shell.c
 index ee9d445..8f25726 100644
 --- a/shell.c
@@ -156,6 +173,63 @@ index 2001b4e..3ba2029 100644
  	    }
  	  else
  	    break;
+diff --git a/tests/crlf.right b/tests/crlf.right
+new file mode 100644
+index 0000000..d7fd195
+--- /dev/null
++++ b/tests/crlf.right
+@@ -0,0 +1,8 @@
++Line with LF
++Line with CR
++Line with CRLF
++Line with
++ LF
++Line with CR
++Line with
++ CRLF
+diff --git a/tests/crlf.tests b/tests/crlf.tests
+new file mode 100644
+index 0000000..b2c4da7
+--- /dev/null
++++ b/tests/crlf.tests
+@@ -0,0 +1,6 @@
++echo $(echo -e "Line with\n LF")
++echo $(echo -e "Line with\r CR")
++echo $(echo -e "Line with\r\n CRLF")
++echo "$(echo -e "Line with\n LF")"
++echo "$(echo -e "Line with\r CR")"
++echo "$(echo -e "Line with\r\n CRLF")"
+diff --git a/tests/ps1lf.right b/tests/ps1lf.right
+new file mode 100644
+index 0000000..ee83131
+--- /dev/null
++++ b/tests/ps1lf.right
+@@ -0,0 +1,2 @@
++foo
++$ exit
+diff --git a/tests/ps1lf.tests b/tests/ps1lf.tests
+new file mode 100644
+index 0000000..01a2265
+--- /dev/null
++++ b/tests/ps1lf.tests
+@@ -0,0 +1 @@
++PS1='$(echo foo)\n\$ '
+diff --git a/tests/run-crlf b/tests/run-crlf
+new file mode 100644
+index 0000000..3f6037c
+--- /dev/null
++++ b/tests/run-crlf
+@@ -0,0 +1,2 @@
++${THIS_SH} ./crlf.tests > ${BASH_TSTOUT}
++diff ${BASH_TSTOUT} crlf.right && rm -f ${BASH_TSTOUT}
+diff --git a/tests/run-ps1lf b/tests/run-ps1lf
+new file mode 100644
+index 0000000..1617108
+--- /dev/null
++++ b/tests/run-ps1lf
+@@ -0,0 +1,2 @@
++${THIS_SH} --rcfile ./ps1lf.tests -i < /dev/null 2>&1 | tr -d '\r' > ${BASH_TSTOUT}
++diff ${BASH_TSTOUT} ps1lf.right && rm -f ${BASH_TSTOUT}
 diff --git a/variables.c b/variables.c
 index 028667c..a10594d 100644
 --- a/variables.c
@@ -181,33 +255,3 @@ index 028667c..a10594d 100644
    if (shell_variables == 0)
      create_variable_tables ();
 
-diff --git a/y.tab.c b/y.tab.c
-index 32b4c7c..ac70820 100644
---- a/y.tab.c
-+++ b/y.tab.c
-@@ -3770,7 +3770,13 @@ yy_input_name ()
- static int
- yy_getc ()
- {
--  return (*(bash_input.getter)) ();
-+#ifdef __MSYS__
-+  int c;
-+  while ((c = (*(bash_input.getter)) ()) == '\r');
-+  return c;
-+#else
-+   return (*(bash_input.getter)) ();
-+#endif
- }
-
- /* Call this to unget C.  That is, to make C the next character
-@@ -4746,6 +4752,10 @@ shell_getc (remove_quoted_newline)
- 	  else
- 	    RESIZE_MALLOCED_BUFFER (shell_input_line, i, 2, shell_input_line_size, 256);
-
-+#ifdef __MSYS__
-+	  if (c == '\r')
-+	    continue;
-+#endif
- 	  if (c == EOF)
- 	    {
- 	      if (bash_input.type == st_stream)

bash/PKGBUILD

@@ -6,7 +6,7 @@ pkgname=('bash' 'bash-devel')
 _basever=5.2
 _patchlevel=037 #prepare for some patches
 pkgver=${_basever}.${_patchlevel}
-pkgrel=2
+pkgrel=3
 pkgdesc="The GNU Bourne Again shell"
 arch=('i686' 'x86_64')
 license=('GPL')
@@ -82,7 +82,7 @@ build() {
 
 check() {
   cd ${srcdir}/${pkgname}-$_basever
-  make check
+  LC_ALL=C.UTF-8 make check
 }
 
 package_bash() {
@@ -129,7 +129,7 @@ sha256sums=('a139c166df7ff4471c5e0733051642ee5556c1cc8a4a78f145583c5c81ab32fb'
             'SKIP'
             '948b8b5401dcb4e5eb577cfa6543e740e2e3bd0690939d8e77d078d75d110097'
             '16584e119db9418030912171f89aecae319858ecd357d3e56c95eba83667dae7'
-            'c55c24110fbe90a2000411239e6399c1baed2843a61220b4e8a7a036f4a7436a'
+            'b598a3dcfab16eb2bd0cee4228a7f0041fc6f10892e5ffc9da1ae1c2e1d7570e'
             '500c75c64593a70276585345a55c807226c0cc220d08b7cccece2ab005b3bcea'
             'cbae1aa81d56eba4e916bdaf2b2983731d6e2537dd8d606a3b378e49bcb81e79'
             'f42f2fee923bc2209f406a1892772121c467f44533bedfe00a176139da5d310a'