coreutils/coreutils-i18n-un-expand-BOM.patch

From 7a7c776a4e228d180e74614fd8c8afcad5d4bdf7 Mon Sep 17 00:00:00 2001
From: Jakub Martisko <jamartis@redhat.com>
Date: Thu, 7 Jul 2016 12:53:26 +0200
Subject: [PATCH] coreutils-i18n-un-expand-BOM.patch

---
 src/expand-common.c  | 114 +++++++++++++++++++++++++++++++++++++++++++++++++++
 src/expand-common.h  |  12 ++++++
 src/expand.c         |  45 +++++++++++++++++++-
 src/unexpand.c       |  43 ++++++++++++++++++-
 tests/expand/mb.sh   |  71 ++++++++++++++++++++++++++++++++
 tests/unexpand/mb.sh |  59 ++++++++++++++++++++++++++
 6 files changed, 342 insertions(+), 2 deletions(-)

diff --git a/src/expand-common.c b/src/expand-common.c
index 4657e46..97cbb09 100644
--- a/src/expand-common.c
+++ b/src/expand-common.c
@@ -18,6 +18,7 @@
 
 #include <stdio.h>
 #include <sys/types.h>
+#include <mbfile.h>
 #include "system.h"
 #include "die.h"
 #include "error.h"
@@ -105,6 +106,119 @@ set_extend_size (uintmax_t tabval)
   return ok;
 }
 
+extern int
+set_utf_locale (void)
+{
+      /*try using some predefined locale */
+      const char* predef_locales[] = {"C.UTF8","en_US.UTF8","en_GB.UTF8"};
+
+      const int predef_locales_count=3;
+      for (int i=0;i<predef_locales_count;i++)
+        {
+          if (setlocale(LC_ALL,predef_locales[i])!=NULL)
+          {
+            break;
+          }
+          else if (i==predef_locales_count-1)
+          {
+            return 1;
+            error (EXIT_FAILURE, errno, _("cannot set UTF-8 locale"));
+          }
+        }
+        return 0;
+}
+
+extern bool
+check_utf_locale(void)
+{
+  char* locale = setlocale (LC_CTYPE , NULL);
+  if (locale == NULL)
+  {
+    return false;
+  }
+  else if (strcasestr(locale, "utf8") == NULL && strcasestr(locale, "utf-8") == NULL)
+  {
+    return false;
+  }
+  return true;
+}
+
+extern bool
+check_bom(FILE* fp, mb_file_t *mbf)
+{
+  int c;
+
+
+  c=fgetc(fp);
+
+  /*test BOM header of the first file */
+  mbf->bufcount=0;
+  if (c == 0xEF)
+  {
+    c=fgetc(fp);
+  }
+  else
+  {
+    if (c != EOF)
+    {
+      ungetc(c,fp);
+    }
+    return false;
+  }
+
+  if (c == 0xBB)
+  {
+    c=fgetc(fp);
+  }
+  else
+  {
+    if ( c!= EOF )
+    {
+      mbf->buf[0]=(unsigned char) 0xEF;
+      mbf->bufcount=1;
+      ungetc(c,fp);
+      return false;
+    }
+    else
+    {
+      ungetc(0xEF,fp);
+      return false;
+    }
+  }
+  if (c == 0xBF)
+  {
+    mbf->bufcount=0;
+    return true;
+  }
+  else
+  {
+    if (c != EOF)
+    {
+      mbf->buf[0]=(unsigned char) 0xEF;
+      mbf->buf[1]=(unsigned char) 0xBB;
+      mbf->bufcount=2;
+      ungetc(c,fp);
+      return false;
+    }
+    else
+    {
+      mbf->buf[0]=(unsigned char) 0xEF;
+      mbf->bufcount=1;
+      ungetc(0xBB,fp);
+      return false;
+    }
+  }
+  return false;
+}
+
+extern void
+print_bom(void)
+{
+  putc (0xEF, stdout);
+  putc (0xBB, stdout);
+  putc (0xBF, stdout);
+}
+
 /* Add the comma or blank separated list of tab stops STOPS
    to the list of tab stops.  */
 extern void
diff --git a/src/expand-common.h b/src/expand-common.h
index 8cb2079..763bfda 100644
--- a/src/expand-common.h
+++ b/src/expand-common.h
@@ -34,6 +34,18 @@ extern size_t max_column_width;
 /* The desired exit status.  */
 extern int exit_status;
 
+extern int
+set_utf_locale (void);
+
+extern bool
+check_utf_locale(void);
+
+extern bool
+check_bom(FILE* fp, mb_file_t *mbf);
+
+extern void
+print_bom(void);
+
 /* Add tab stop TABVAL to the end of 'tab_list'.  */
 extern void
 add_tab_stop (uintmax_t tabval);
diff --git a/src/expand.c b/src/expand.c
index 310b349..4136824 100644
--- a/src/expand.c
+++ b/src/expand.c
@@ -105,11 +105,33 @@ expand (void)
   FILE *fp = next_file (NULL);
   mb_file_t mbf;
   mbf_char_t c;
+  /* True if the starting locale is utf8.  */
+  bool using_utf_locale;
+
+  /* True if the first file contains BOM header.  */
+  bool found_bom;
+  using_utf_locale=check_utf_locale();
 
   if (!fp)
     return;
-
   mbf_init (mbf, fp);
+  found_bom=check_bom(fp,&mbf);
+
+  if (using_utf_locale == false && found_bom == true)
+  {
+    /*try using some predefined locale */
+
+    if (set_utf_locale () != 0)
+    {
+      error (EXIT_FAILURE, errno, _("cannot set UTF-8 locale"));
+    }
+  }
+
+
+  if (found_bom == true)
+  {
+    print_bom();
+  }
 
   while (true)
     {
@@ -134,6 +156,27 @@ expand (void)
             if ((mb_iseof (c)) && (fp = next_file (fp)))
               {
                 mbf_init (mbf, fp);
+                if (fp!=NULL)
+                {
+                  if (check_bom(fp,&mbf)==true)
+                  {
+                    /*Not the first file - check BOM header*/
+                    if (using_utf_locale==false && found_bom==false)
+                    {
+                      /*BOM header in subsequent file but not in the first one. */
+                      error (EXIT_FAILURE, errno, _("combination of files with and without BOM header"));
+                    }
+                  }
+                  else
+                  {
+                    if(using_utf_locale==false && found_bom==true)
+                    {
+                      /*First file conatined BOM header - locale was switched to UTF
+                      /*all subsequent files should contain BOM. */
+                      error (EXIT_FAILURE, errno, _("combination of files with and without BOM header"));
+                    }
+                  }
+                }
                 continue;
               }
             else
diff --git a/src/unexpand.c b/src/unexpand.c
index 863a90a..5681b58 100644
--- a/src/unexpand.c
+++ b/src/unexpand.c
@@ -116,16 +116,36 @@ unexpand (void)
      include characters other than spaces, so the blanks must be
      stored, not merely counted.  */
   mbf_char_t *pending_blank;
+  /* True if the starting locale is utf8.  */
+  bool using_utf_locale;
+
+  /* True if the first file contains BOM header.  */
+  bool found_bom;
+  using_utf_locale=check_utf_locale();
 
   if (!fp)
     return;
+  mbf_init (mbf, fp);
+  found_bom=check_bom(fp,&mbf);
+
+  if (using_utf_locale == false && found_bom == true)
+  {
+    /*try using some predefined locale */
 
+    if (set_utf_locale () != 0)
+    {
+      error (EXIT_FAILURE, errno, _("cannot set UTF-8 locale"));
+    }
+  }
   /* The worst case is a non-blank character, then one blank, then a
      tab stop, then MAX_COLUMN_WIDTH - 1 blanks, then a non-blank; so
      allocate MAX_COLUMN_WIDTH bytes to store the blanks.  */
   pending_blank = xmalloc (max_column_width * sizeof (mbf_char_t));
 
-  mbf_init (mbf, fp);
+  if (found_bom == true)
+  {
+    print_bom();
+  }
 
   while (true)
     {
@@ -169,6 +189,27 @@ unexpand (void)
             if ((mb_iseof (c)) && (fp = next_file (fp)))
               {
                 mbf_init (mbf, fp);
+                if (fp!=NULL)
+                {
+                  if (check_bom(fp,&mbf)==true)
+                  {
+                    /*Not the first file - check BOM header*/
+                    if (using_utf_locale==false && found_bom==false)
+                    {
+                      /*BOM header in subsequent file but not in the first one. */
+                      error (EXIT_FAILURE, errno, _("combination of files with and without BOM header"));
+                    }
+                  }
+                  else
+                  {
+                    if(using_utf_locale==false && found_bom==true)
+                    {
+                      /*First file conatined BOM header - locale was switched to UTF
+                      /*all subsequent files should contain BOM. */
+                      error (EXIT_FAILURE, errno, _("combination of files with and without BOM header"));
+                    }
+                  }
+                }
                 continue;
               }
             else
diff --git a/tests/expand/mb.sh b/tests/expand/mb.sh
index 031be7a..1621c84 100755
--- a/tests/expand/mb.sh
+++ b/tests/expand/mb.sh
@@ -109,4 +109,75 @@ env printf '12345678
 expand < in > out || fail=1
 compare exp out > /dev/null 2>&1 || fail=1
 
+
+
+#BOM header test 1
+printf "\xEF\xBB\xBF" > in; cat <<\EOF >> in || framework_failure_
+1234567812345678123456781
+.       .       .       .
+a	b	c	d
+.       .       .       .
+ä	ö	ü	ß
+.       .       .       .
+EOF
+env printf '   äöü\t.    öüä.   \tä xx\n' >> in || framework_failure_
+
+printf "\xEF\xBB\xBF" > exp; cat <<\EOF >> exp || framework_failure_
+1234567812345678123456781
+.       .       .       .
+a       b       c       d
+.       .       .       .
+ä       ö       ü       ß
+.       .       .       .
+   äöü  .    öüä.       ä xx
+EOF
+
+
+expand < in > out || fail=1
+compare exp out > /dev/null 2>&1 || fail=1
+
+LANG=C expand < in > out || fail=1
+compare exp out > /dev/null 2>&1 || fail=1
+
+LC_ALL=C expand < in > out || fail=1
+compare exp out > /dev/null 2>&1 || fail=1
+
+
+printf '\xEF\xBB\xBF' > in1; cat <<\EOF >> in1 || framework_failure_
+1234567812345678123456781
+.       .       .       .
+a	b	c	d
+.       .       .       .
+ä	ö	ü	ß
+.       .       .       .
+EOF
+env printf '   äöü\t.    öüä.   \tä xx\n' >> in1 || framework_failure_
+
+
+printf '\xEF\xBB\xBF' > exp; cat <<\EOF >> exp || framework_failure_
+1234567812345678123456781
+.       .       .       .
+a       b       c       d
+.       .       .       .
+ä       ö       ü       ß
+.       .       .       .
+   äöü  .    öüä.       ä xx
+1234567812345678123456781
+.       .       .       .
+a       b       c       d
+.       .       .       .
+ä       ö       ü       ß
+.       .       .       .
+   äöü  .    öüä.       ä xx
+EOF
+
+expand in1 in1 > out || fail=1
+compare exp out > /dev/null 2>&1 || fail=1
+
+LANG=C expand in1 in1  > out || fail=1
+compare exp out > /dev/null 2>&1 || fail=1
+
+LC_ALL=C expand in1 in1 > out || fail=1
+compare exp out > /dev/null 2>&1 || fail=1
+
 exit $fail
diff --git a/tests/unexpand/mb.sh b/tests/unexpand/mb.sh
index 8d75652..9d4ee3e 100755
--- a/tests/unexpand/mb.sh
+++ b/tests/unexpand/mb.sh
@@ -111,3 +111,62 @@ env printf '12345678
 
 unexpand -a < in > out || fail=1
 compare exp out > /dev/null 2>&1 || fail=1
+
+#BOM header test 1
+printf "\xEF\xBB\xBF" > in; cat <<\EOF >> in || framework_failure_
+1234567812345678123456781
+.       .       .       .
+a       b       c       d
+.       .       .       .
+ä       ö       ü       ß
+.       .       .       .
+   äöü  .    öüä.       ä xx
+EOF
+env printf '   äöü\t.    öüä.   \tä xx\n' >> in || framework_failure_
+
+printf "\xEF\xBB\xBF" > exp; cat <<\EOF >> exp || framework_failure_
+1234567812345678123456781
+.	.	.	.
+a	b	c	d
+.	.	.	.
+ä	ö	ü	ß
+.	.	.	.
+   äöü	.    öüä.	ä xx
+EOF
+
+unexpand < in > out || fail=1
+compare exp out > /dev/null 2>&1 || fail=1
+
+LANG=C unexpand < in > out || fail=1
+compare exp out > /dev/null 2>&1 || fail=1
+
+LC_ALL=C unexpand < in > out || fail=1
+compare exp out > /dev/null 2>&1 || fail=1
+
+
+printf "\xEF\xBB\xBF" > exp; cat <<\EOF >> exp || framework_failure_
+1234567812345678123456781
+.	.	.	.
+a	b	c	d
+.	.	.	.
+ä	ö	ü	ß
+.	.	.	.
+   äöü	.    öüä.	ä xx
+1234567812345678123456781
+.	.	.	.
+a	b	c	d
+.	.	.	.
+ä	ö	ü	ß
+.	.	.	.
+   äöü	.    öüä.	ä xx
+EOF
+
+
+unexpand in in > out || fail=1
+compare exp out > /dev/null 2>&1 || fail=1
+
+LANG=C unexpand in in > out || fail=1
+compare exp out > /dev/null 2>&1 || fail=1
+
+LC_ALL=C unexpand in in > out || fail=1
+compare exp out > /dev/null 2>&1 || fail=1
-- 
2.9.3