hush: optional support for ${var:N:M} bashism

function                                             old     new   delta
expand_vars_to_list                                 1999    2183    +184
handle_dollar                                        682     623     -59

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

3 files changed, 120 insertions, 39 deletions

diff --git a/shell/hush.c b/shell/hush.c
index 06e9af21a..0333395f8 100644
--- a/shell/hush.c
+++ b/shell/hush.c
@@ -46,7 +46,6 @@
  *      brace expansion: one/{two,three,four}
  *      reserved words: function select
  *      advanced test: [[ ]]
- *      substrings: ${var:1:5}
  *      process substitution: <(list) and >(list)
  *      =~: regex operator
  *      let EXPR [EXPR...]
@@ -2586,7 +2585,7 @@ static NOINLINE int expand_vars_to_list(o_string *output, int n, char *arg, char
                                 var++;
                         } else {
                                 /* maybe handle parameter expansion */
-                                exp_saveptr = var + strcspn(var, ":-=+?%#");
+                                exp_saveptr = var + strcspn(var, "%#:-=+?");
                                 exp_save = *exp_saveptr;
                                 if (exp_save) {
                                         exp_word = exp_saveptr;
@@ -2636,9 +2635,38 @@ static NOINLINE int expand_vars_to_list(o_string *output, int n, char *arg, char
                                                 else if (loc) /* % or %% and match was found */
                                                         *loc = '\0';
                                         }
-                                } else { /* one of :-=+? */
-//TODO: check validity of exp_op. Currently it can be anything.
-//TODO: handle ${VAR:N[:M]} here. N, M can be expressions similar to $((EXPR)): 2+2, 2+var etc
+                                } else if (!strchr("%#:-=+?"+3, exp_op)) {
+#if ENABLE_HUSH_BASH_COMPAT
+        /* exp_op is ':' and next char isn't a subst operator.
+         * Assuming it's ${var:[N][:M]} bashism.
+         * TODO: N, M can be expressions similar to $((EXPR)): 2+2, 2+var etc
+         */
+                                        char *end;
+                                        unsigned len = INT_MAX;
+                                        unsigned beg = 0;
+                                        end = --exp_word;
+                                        if (*exp_word != ':') /* not ${var::...} */
+                                                beg = bb_strtou(exp_word, &end, 0);
+                                                //bb_error_msg("beg:'%s'=%u end:'%s'", exp_word, beg, end);
+                                        if (*end == ':') {
+                                                len = bb_strtou(end + 1, &end, 0);
+                                                //bb_error_msg("len:%u end:'%s'", len, end);
+                                        }
+                                        if (*end == '\0'
+                                         && end != exp_word /* not "${var:}" */
+                                        ) {
+                                                //bb_error_msg("from val:'%s'", val);
+                                                if (!val || beg >= strlen(val))
+                                                        val = "";
+                                                else
+                                                        val = dyn_val = xstrndup(val + beg, len);
+                                                //bb_error_msg("val:'%s'", val);
+                                        } else
+#endif
+                                        {
+                                                die_if_script("malformed ${%s...}", var);
+                                        }
+                                } else { /* one of "-=+?" */
         /* Standard-mandated substitution ops:
          * ${var?word} - indicate error if unset
          *      If var is unset, word (or a message indicating it is unset
@@ -2693,7 +2721,7 @@ static NOINLINE int expand_vars_to_list(o_string *output, int n, char *arg, char
                                 }
 
                                 *exp_saveptr = exp_save;
-                        }
+                        } /* if (exp_op) */
 
                         arg[0] = first_ch;
 #if ENABLE_HUSH_TICK
@@ -5926,14 +5954,15 @@ static int handle_dollar(o_string *as_string,
                 goto make_one_char_var;
         case '{': {
                 bool first_char, all_digits;
-                int expansion;
+                bool in_expansion_param;
 
                 ch = i_getch(input);
                 nommu_addchr(as_string, ch);
                 o_addchr(dest, SPECIAL_VAR_SYMBOL);
 
-                /* TODO: maybe someone will try to escape the '}' */
-                expansion = 0;
+// TODO: need to handle "a=ab}; echo ${a%\}}"
+// and "a=abc; c=c; echo ${a%${c}}"
+                in_expansion_param = false;
                 first_char = true;
                 all_digits = false;
                 while (1) {
@@ -5957,45 +5986,22 @@ static int handle_dollar(o_string *as_string,
                                         goto char_ok;
                         }
 
-                        if (expansion < 2
-                         && (  (all_digits && !isdigit(ch))
-                            || (!all_digits && !isalnum(ch) && ch != '_')
+                        if (!in_expansion_param
+                         && (  (all_digits && !isdigit(ch)) /* met non-digit: 123w */
+                            || (!all_digits && !isalnum(ch) && ch != '_') /* met non-name char: abc% */
                             )
                         ) {
                                 /* handle parameter expansions
                                  * http://www.opengroup.org/onlinepubs/009695399/utilities/xcu_chap02.html#tag_02_06_02
                                  */
-                                if (first_char)
-                                        goto case_default;
-                                switch (ch) {
-                                case ':': /* null modifier */
-                                        if (expansion == 0) {
-                                                debug_printf_parse(": null modifier\n");
-                                                ++expansion;
-                                                break;
-                                        }
-                                        goto case_default;
-                                case '#': /* remove prefix */
-                                case '%': /* remove suffix */
-                                        if (expansion == 0) {
-                                                debug_printf_parse(": remove suffix/prefix\n");
-                                                expansion = 2;
-                                                break;
-                                        }
-                                        goto case_default;
-                                case '-': /* default value */
-                                case '=': /* assign default */
-                                case '+': /* alternative */
-                                case '?': /* error indicate */
-                                        debug_printf_parse(": parameter expansion\n");
-                                        expansion = 2;
-                                        break;
-                                default:
-                                case_default:
+                                if (first_char /* bad (empty var name): "${%..." */
+                                 || !strchr("%#:-=+?", ch) /* bad: "${var<bad_char>..." */
+                                ) {
                                         syntax_error_unterm_str("${name}");
                                         debug_printf_parse("handle_dollar return 1: unterminated ${name}\n");
                                         return 1;
                                 }
+                                in_expansion_param = true;
                         }
  char_ok:
                         debug_printf_parse(": '%c'\n", ch);
diff --git a/shell/hush_test/hush-vars/param_expand_bash_substring.right b/shell/hush_test/hush-vars/param_expand_bash_substring.right
new file mode 100644
index 000000000..9cd465938
--- /dev/null
+++ b/shell/hush_test/hush-vars/param_expand_bash_substring.right
@@ -0,0 +1,29 @@
+hush: syntax error: unterminated ${name}
+hush: syntax error: unterminated ${name}
+hush: syntax error: unterminated ${name}
+hush: syntax error: unterminated ${name}
+0123456789
+1    =||
+1:1  =||
+1:1:2=||
+1::2 =||
+1    =|0123|
+1:1  =|123|
+1:1:2=|12|
+1::2 =|01|
+f    =||
+f:1  =||
+f:1:2=||
+f::2 =||
+f    =||
+f:1  =||
+f:1:2=||
+f::2 =||
+f    =|a|
+f:1  =||
+f:1:2=||
+f::2 =|a|
+f    =|0123456789|
+f:1  =|123456789|
+f:1:2=|12|
+f::2 =|01|
diff --git a/shell/hush_test/hush-vars/param_expand_bash_substring.tests b/shell/hush_test/hush-vars/param_expand_bash_substring.tests
new file mode 100755
index 000000000..6a1765559
--- /dev/null
+++ b/shell/hush_test/hush-vars/param_expand_bash_substring.tests
@@ -0,0 +1,46 @@
+# do all of these in subshells since it's supposed to error out
+
+export var=0123456789
+
+# first try some invalid patterns
+"$THIS_SH" -c 'echo ${:}'
+"$THIS_SH" -c 'echo ${::}'
+"$THIS_SH" -c 'echo ${:1}'
+"$THIS_SH" -c 'echo ${::1}'
+
+#this also is not valid in bash, but we accept it:
+"$THIS_SH" -c 'echo ${var:}'
+
+# then some funky ones
+# UNFIXED BUG: this should work: "$THIS_SH" -c 'echo ${?:0}'
+
+# now some valid ones
+"$THIS_SH" -c 'set --; echo "1    =|${1}|"'
+"$THIS_SH" -c 'set --; echo "1:1  =|${1:1}|"'
+"$THIS_SH" -c 'set --; echo "1:1:2=|${1:1:2}|"'
+"$THIS_SH" -c 'set --; echo "1::2 =|${1::2}|"'
+
+"$THIS_SH" -c 'set -- 0123; echo "1    =|${1}|"'
+"$THIS_SH" -c 'set -- 0123; echo "1:1  =|${1:1}|"'
+"$THIS_SH" -c 'set -- 0123; echo "1:1:2=|${1:1:2}|"'
+"$THIS_SH" -c 'set -- 0123; echo "1::2 =|${1::2}|"'
+
+"$THIS_SH" -c 'unset f; echo "f    =|$f|"'
+"$THIS_SH" -c 'unset f; echo "f:1  =|${f:1}|"'
+"$THIS_SH" -c 'unset f; echo "f:1:2=|${f:1:2}|"'
+"$THIS_SH" -c 'unset f; echo "f::2 =|${f::2}|"'
+
+"$THIS_SH" -c 'f=; echo "f    =|$f|"'
+"$THIS_SH" -c 'f=; echo "f:1  =|${f:1}|"'
+"$THIS_SH" -c 'f=; echo "f:1:2=|${f:1:2}|"'
+"$THIS_SH" -c 'f=; echo "f::2 =|${f::2}|"'
+
+"$THIS_SH" -c 'f=a; echo "f    =|$f|"'
+"$THIS_SH" -c 'f=a; echo "f:1  =|${f:1}|"'
+"$THIS_SH" -c 'f=a; echo "f:1:2=|${f:1:2}|"'
+"$THIS_SH" -c 'f=a; echo "f::2 =|${f::2}|"'
+
+"$THIS_SH" -c 'f=0123456789; echo "f    =|$f|"'
+"$THIS_SH" -c 'f=0123456789; echo "f:1  =|${f:1}|"'
+"$THIS_SH" -c 'f=0123456789; echo "f:1:2=|${f:1:2}|"'
+"$THIS_SH" -c 'f=0123456789; echo "f::2 =|${f::2}|"'