Improve wordsplit.

* include/mailutils/wordsplit.h (mu_wordsplit) <ws_escape>: New member.
(MU_WRDSF_ESCAPE): New flag.
(mu_wordsplit_general_unquote_copy): New function.
* libmailutils/string/wordsplit.c: Remove empty nodes after whitespace
trimming.  If MU_WRDSF_ESCAPE is set, remove only backslashes appearing
in front of a character in ws_escape.
* libmailutils/tests/wordsplit.at: Test new features.

@@ -28,6 +28,7 @@ struct mu_wordsplit
   int ws_flags;
   int ws_flags;
   const char *ws_delim;
   const char *ws_delim;
   const char *ws_comment;
   const char *ws_comment;
+  const char *ws_escape;
   void (*ws_alloc_die) (struct mu_wordsplit *wsp);
   void (*ws_alloc_die) (struct mu_wordsplit *wsp);
   void (*ws_error) (const char *, ...)
   void (*ws_error) (const char *, ...)
 		    __attribute__ ((__format__ (__printf__, 1, 2)));
 		    __attribute__ ((__format__ (__printf__, 1, 2)));
@@ -110,6 +111,9 @@ struct mu_wordsplit
    stored in the element that follows its name. */
    stored in the element that follows its name. */
 #define MU_WRDSF_ENV_KV            0x4000000
 #define MU_WRDSF_ENV_KV            0x4000000
 
 
+/* ws_escape is set */
+#define MU_WRDSF_ESCAPE            0x8000000
+
 #define MU_WRDSF_DEFFLAGS	       \
 #define MU_WRDSF_DEFFLAGS	       \
   (MU_WRDSF_NOVAR | MU_WRDSF_NOCMD | \
   (MU_WRDSF_NOVAR | MU_WRDSF_NOCMD | \
    MU_WRDSF_QUOTE | MU_WRDSF_SQUEEZE_DELIMS | MU_WRDSF_CESCAPES)
    MU_WRDSF_QUOTE | MU_WRDSF_SQUEEZE_DELIMS | MU_WRDSF_CESCAPES)
@@ -132,6 +136,8 @@ int mu_wordsplit_c_unquote_char (int c);
 int mu_wordsplit_c_quote_char (int c);
 int mu_wordsplit_c_quote_char (int c);
 size_t mu_wordsplit_c_quoted_length (const char *str, int quote_hex,
 size_t mu_wordsplit_c_quoted_length (const char *str, int quote_hex,
 				     int *quote);
 				     int *quote);
+void mu_wordsplit_general_unquote_copy (char *dst, const char *src, size_t n,
+					const char *escapable);
 void mu_wordsplit_sh_unquote_copy (char *dst, const char *src, size_t n);
 void mu_wordsplit_sh_unquote_copy (char *dst, const char *src, size_t n);
 void mu_wordsplit_c_unquote_copy (char *dst, const char *src, size_t n);
 void mu_wordsplit_c_unquote_copy (char *dst, const char *src, size_t n);
 void mu_wordsplit_c_quote_copy (char *dst, const char *src, int quote_hex);
 void mu_wordsplit_c_quote_copy (char *dst, const char *src, int quote_hex);

@@ -500,7 +500,12 @@ wsnode_quoteremoval (struct mu_wordsplit *wsp)
 	      p->v.word = newstr;
 	      p->v.word = newstr;
 	      p->flags |= _WSNF_WORD;
 	      p->flags |= _WSNF_WORD;
 	    }
 	    }
-	  uqfn (p->v.word, str, slen);
+
+	  if (wsp->ws_flags & MU_WRDSF_ESCAPE)
+	    mu_wordsplit_general_unquote_copy (p->v.word, str, slen,
+					       wsp->ws_escape);
+	  else
+	    uqfn (p->v.word, str, slen);
 	}
 	}
     }
     }
   return 0;
   return 0;
@@ -906,32 +911,39 @@ node_expand_vars (struct mu_wordsplit *wsp, struct mu_wordsplit_node *node)
   return 0;
   return 0;
 }
 }
 
 
-static int
-mu_wordsplit_varexp (struct mu_wordsplit *wsp)
+/* Remove NULL lists */
+static void
+wsnode_nullelim (struct mu_wordsplit *wsp)
 {
 {
   struct mu_wordsplit_node *p;
   struct mu_wordsplit_node *p;
 
 
   for (p = wsp->ws_head; p;)
   for (p = wsp->ws_head; p;)
     {
     {
       struct mu_wordsplit_node *next = p->next;
       struct mu_wordsplit_node *next = p->next;
-      if (!(p->flags & _WSNF_NOEXPAND))
-	if (node_expand_vars (wsp, p))
-	  return 1;
+      if (p->flags & _WSNF_NULL)
+	{
+	  wsnode_remove (wsp, p);
+	  wsnode_free (p);
+	}
       p = next;
       p = next;
     }
     }
+}
+
+static int
+mu_wordsplit_varexp (struct mu_wordsplit *wsp)
+{
+  struct mu_wordsplit_node *p;
 
 
-  /* Remove NULL lists */
   for (p = wsp->ws_head; p;)
   for (p = wsp->ws_head; p;)
     {
     {
       struct mu_wordsplit_node *next = p->next;
       struct mu_wordsplit_node *next = p->next;
-      if (p->flags & _WSNF_NULL)
-	{
-	  wsnode_remove (wsp, p);
-	  wsnode_free (p);
-	}
+      if (!(p->flags & _WSNF_NOEXPAND))
+	if (node_expand_vars (wsp, p))
+	  return 1;
       p = next;
       p = next;
     }
     }
 
 
+  wsnode_nullelim (wsp);
   return 0;
   return 0;
 }
 }
 
 
@@ -959,7 +971,11 @@ mu_wordsplit_trimws (struct mu_wordsplit *wsp)
       for (n = p->v.segm.end; n > p->v.segm.beg && ISWS (wsp->ws_input[n-1]);
       for (n = p->v.segm.end; n > p->v.segm.beg && ISWS (wsp->ws_input[n-1]);
 	   n--);
 	   n--);
       p->v.segm.end = n;
       p->v.segm.end = n;
+      if (p->v.segm.beg == p->v.segm.end)
+	p->flags |= _WSNF_NULL;
     }
     }
+
+  wsnode_nullelim (wsp);
 }
 }
 
 
 static int
 static int
@@ -1217,6 +1233,21 @@ mu_wordsplit_c_quoted_length (const char *str, int quote_hex, int *quote)
 }
 }
 
 
 void
 void
+mu_wordsplit_general_unquote_copy (char *dst, const char *src, size_t n,
+				   const char *escapable)
+{
+  int i;
+
+  for (i = 0; i < n;)
+    {
+      if (src[i] == '\\' && i < n && strchr (escapable, src[i+1]))
+	i++;
+      *dst++ = src[i++];
+    }
+  *dst = 0;
+}
+
+void
 mu_wordsplit_sh_unquote_copy (char *dst, const char *src, size_t n)
 mu_wordsplit_sh_unquote_copy (char *dst, const char *src, size_t n)
 {
 {
   int i;
   int i;

@@ -313,6 +313,15 @@ TESTWSP([C escapes off],[],[-cescapes],
 3: newnline
 3: newnline
 ])
 ])
 
 
+TESTWSP([ws elimination],[],[delim ' ()' ws return_delims],
+[( list  items  )],
+[NF: 4
+0: (
+1: list
+2: items
+3: )
+])
+
 TESTWSP([empty quotes],[],[delim : ws return_delims],
 TESTWSP([empty quotes],[],[delim : ws return_delims],
 [t=""],
 [t=""],
 [NF: 1
 [NF: 1
@@ -338,4 +347,12 @@ TESTWSP([suppress ws trimming within quotes],[],
 4: "formatfield=In message %{text}, "
 4: "formatfield=In message %{text}, "
 ])
 ])
 
 
+TESTWSP([unescape],[],[-default novar nocmd quote escape '\"'],
+[\Seen "quote \"" "bs \\"],
+[NF: 3
+0: \\Seen
+1: "quote \""
+2: "bs \\"
+])
+
 m4_popdef([TESTWSP])
 m4_popdef([TESTWSP])