read: Support arrays, character splitting

Enhance the `read` builtin to support creating an array with the --array
flag. With --array, only a single variable name is allowed and the
entire input is tokenized and placed into that variable as an array.

Also add custom behavior if IFS is empty or unset. In that event, split
the input on every character, instead of the previous behavior of doing
no splitting at all.

builtin.cpp

@@ -2301,6 +2301,7 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
     int exit_res=STATUS_BUILTIN_OK;
     const wchar_t *mode_name = READ_MODE_NAME;
     int shell = 0;
+    int array = 0;
 
     woptind=0;
 
@@ -2345,6 +2346,10 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
                 L"shell", no_argument, 0, 's'
             }
             ,
+            {
+                L"array", no_argument, 0, 'a'
+            }
+            ,
             {
                 L"help", no_argument, 0, 'h'
             }
@@ -2359,7 +2364,7 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
 
         int opt = wgetopt_long(argc,
                                argv,
-                               L"xglUup:c:hm:s",
+                               L"xglUup:c:hm:sa",
                                long_options,
                                &opt_index);
         if (opt == -1)
@@ -2414,6 +2419,10 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
                 shell = 1;
                 break;
 
+            case 'a':
+                array = 1;
+                break;
+
             case 'h':
                 builtin_print_help(parser, argv[0], stdout_buffer);
                 return STATUS_BUILTIN_OK;
@@ -2446,6 +2455,14 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
         return STATUS_BUILTIN_ERROR;
     }
 
+    if (array && woptind+1 != argc)
+    {
+        append_format(stderr_buffer, _(L"%ls: --array option requires a single variable name.\n"), argv[0]);
+        builtin_print_help(parser, argv[0], stderr_buffer);
+
+        return STATUS_BUILTIN_ERROR;
+    }
+
     /*
       Verify all variable names
     */
@@ -2580,18 +2597,64 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
         wchar_t *state;
 
         env_var_t ifs = env_get_string(L"IFS");
-        if (ifs.missing())
-            ifs = L"";
 
-        nxt = wcstok(buff, (i<argc-1)?ifs.c_str():L"", &state);
-
-        while (i<argc)
+        if (ifs.missing_or_empty())
         {
-            env_set(argv[i], nxt != 0 ? nxt: L"", place);
+            /* Every character is a separate token */
+            size_t bufflen = wcslen(buff);
+            if (array)
+            {
+                if (bufflen > 0)
+                {
+                    wcstring chars(bufflen+(bufflen-1), ARRAY_SEP);
+                    for (size_t j=0; j<bufflen; ++j)
+                    {
+                        chars[j*2] = buff[j];
+                    }
+                    env_set(argv[i], chars.c_str(), place);
+                }
+                else
+                {
+                    env_set(argv[i], NULL, place);
+                }
+            }
+            else
+            {
+                size_t j = 0;
+                for (; i+1 < argc; ++i)
+                {
+                    env_set(argv[i], j < bufflen ? (wchar_t[2]){buff[j], 0} : L"", place);
+                    if (j < bufflen) ++j;
+                }
+                if (i < argc) env_set(argv[i], &buff[j], place);
+            }
+        }
+        else if (array)
+        {
+            wcstring tokens;
+            tokens.reserve(wcslen(buff));
+            bool empty = true;
 
-            i++;
-            if (nxt != 0)
-                nxt = wcstok(0, (i<argc-1)?ifs.c_str():L"", &state);
+            for (nxt = wcstok(buff, ifs.c_str(), &state); nxt != 0; nxt = wcstok(0, ifs.c_str(), &state))
+            {
+                if (! tokens.empty()) tokens.push_back(ARRAY_SEP);
+                tokens.append(nxt);
+                empty = false;
+            }
+            env_set(argv[i], empty ? NULL : tokens.c_str(), place);
+        }
+        else
+        {
+            nxt = wcstok(buff, (i<argc-1)?ifs.c_str():L"", &state);
+
+            while (i<argc)
+            {
+                env_set(argv[i], nxt != 0 ? nxt: L"", place);
+
+                i++;
+                if (nxt != 0)
+                    nxt = wcstok(0, (i<argc-1)?ifs.c_str():L"", &state);
+            }
         }
     }
 

doc_src/read.txt

@@ -19,11 +19,17 @@ The following options are available:
 - <code>-u</code> or <code>--unexport</code> prevents the variables from being exported to child processes (default behaviour).
 - <code>-U</code> or <code>--universal</code> causes the specified shell variable to be made universal.
 - <code>-x</code> or <code>--export</code> exports the variables to child processes.
+- <code>-a</code> or <code>--array</code> stores the result as an array.
 
 \c read reads a single line of input from stdin, breaks it into tokens
 based on the <tt>IFS</tt> shell variable, and then assigns one
 token to each variable specified in <tt>VARIABLES</tt>. If there are more
 tokens than variables, the complete remainder is assigned to the last variable.
+As a special case, if \c IFS is set to the empty string, each character of the
+input is considered a separate token.
+
+If \c -a or \c --array is provided, only one variable name is allowed and the
+tokens are stored as an array in this variable.
 
 See the documentation for \c set for more details on the scoping rules for
 variables.

tests/read.in

@@ -0,0 +1,72 @@
+#
+# Test read builtin and IFS
+#
+
+count (echo one\ntwo)
+set -l IFS \t
+count (echo one\ntwo)
+set -l IFS
+count (echo one\ntwo)
+set -le IFS
+
+function print_vars --no-scope-shadowing
+	set -l space
+	set -l IFS \n # ensure our command substitution works right
+	for var in $argv
+		echo -n $space (count $$var) \'$$var\'
+		set space ''
+	end
+	echo
+end
+
+echo
+echo 'hello there' | read -l one two
+print_vars one two
+echo 'hello there' | read -l one
+print_vars one
+echo '' | read -l one
+print_vars one
+echo '' | read -l one two
+print_vars one two
+echo 'test' | read -l one two three
+print_vars one two three
+
+echo
+set -l IFS
+echo 'hello' | read -l one
+print_vars one
+echo 'hello' | read -l one two
+print_vars one two
+echo 'hello' | read -l one two three
+print_vars one two three
+echo '' | read -l one
+print_vars one
+echo 't' | read -l one two
+print_vars one two
+echo 't' | read -l one two three
+print_vars one two three
+echo ' t' | read -l one two
+print_vars one two
+set -le IFS
+
+echo
+echo 'hello there' | read -la ary
+print_vars ary
+echo 'hello' | read -la ary
+print_vars ary
+echo 'this is a bunch of words' | read -la ary
+print_vars ary
+echo '   one   two     three' | read -la ary
+print_vars ary
+echo '' | read -la ary
+print_vars ary
+
+echo
+set -l IFS
+echo 'hello' | read -la ary
+print_vars ary
+echo 'h' | read -la ary
+print_vars ary
+echo '' | read -la ary
+print_vars ary
+set -le IFS

tests/read.out

@@ -0,0 +1,27 @@
+2
+2
+1
+
+1 'hello' 1 'there'
+1 'hello there'
+1 ''
+1 '' 1 ''
+1 'test' 1 '' 1 ''
+
+1 'hello'
+1 'h' 1 'ello'
+1 'h' 1 'e' 1 'llo'
+1 ''
+1 't' 1 ''
+1 't' 1 '' 1 ''
+1 ' ' 1 't'
+
+2 'hello' 'there'
+1 'hello'
+6 'this' 'is' 'a' 'bunch' 'of' 'words'
+3 'one' 'two' 'three'
+0
+
+5 'h' 'e' 'l' 'l' 'o'
+1 'h'
+0

tests/read.status

@@ -0,0 +1 @@
+0

tests/top.out

@@ -1,5 +1,6 @@
 Testing high level script functionality
 File printf.in tested ok
+File read.in tested ok
 File test1.in tested ok
 File test2.in tested ok
 File test3.in tested ok