When using wordexp, only try to use the expand command

when something actually got expanded.  Skip it otherwise.
 -Erik

3 files changed, 42 insertions, 36 deletions

diff --git a/lash.c b/lash.c
index 8f19e526a..d321c929b 100644
--- a/lash.c
+++ b/lash.c
@@ -941,20 +941,22 @@ static int expand_arguments(char *command)
                 return FALSE;
         }
         
-        /* Convert from char** (one word per string) to a simple char*,
-         * but don't overflow command which is BUFSIZ in length */
-        *command = '\0';
-        while (i < expand_result.we_wordc && total_length < BUFSIZ) {
-                length=strlen(expand_result.we_wordv[i])+1;
-                if (BUFSIZ-total_length-length <= 0) {
-                        error_msg(out_of_space);
-                        return FALSE;
+        if (expand_result.we_wordc > 0) {
+                /* Convert from char** (one word per string) to a simple char*,
+                 * but don't overflow command which is BUFSIZ in length */
+                *command = '\0';
+                while (i < expand_result.we_wordc && total_length < BUFSIZ) {
+                        length=strlen(expand_result.we_wordv[i])+1;
+                        if (BUFSIZ-total_length-length <= 0) {
+                                error_msg(out_of_space);
+                                return FALSE;
+                        }
+                        strcat(command+total_length, expand_result.we_wordv[i++]);
+                        strcat(command+total_length, " ");
+                        total_length+=length;
                 }
-                strcat(command+total_length, expand_result.we_wordv[i++]);
-                strcat(command+total_length, " ");
-                total_length+=length;
+                wordfree (&expand_result);
         }
-        wordfree (&expand_result);
 #else
 
         /* Ok.  They don't have a recent glibc and they don't have uClibc.  Chances
diff --git a/sh.c b/sh.c
index 8f19e526a..d321c929b 100644
--- a/sh.c
+++ b/sh.c
@@ -941,20 +941,22 @@ static int expand_arguments(char *command)
                 return FALSE;
         }
         
-        /* Convert from char** (one word per string) to a simple char*,
-         * but don't overflow command which is BUFSIZ in length */
-        *command = '\0';
-        while (i < expand_result.we_wordc && total_length < BUFSIZ) {
-                length=strlen(expand_result.we_wordv[i])+1;
-                if (BUFSIZ-total_length-length <= 0) {
-                        error_msg(out_of_space);
-                        return FALSE;
+        if (expand_result.we_wordc > 0) {
+                /* Convert from char** (one word per string) to a simple char*,
+                 * but don't overflow command which is BUFSIZ in length */
+                *command = '\0';
+                while (i < expand_result.we_wordc && total_length < BUFSIZ) {
+                        length=strlen(expand_result.we_wordv[i])+1;
+                        if (BUFSIZ-total_length-length <= 0) {
+                                error_msg(out_of_space);
+                                return FALSE;
+                        }
+                        strcat(command+total_length, expand_result.we_wordv[i++]);
+                        strcat(command+total_length, " ");
+                        total_length+=length;
                 }
-                strcat(command+total_length, expand_result.we_wordv[i++]);
-                strcat(command+total_length, " ");
-                total_length+=length;
+                wordfree (&expand_result);
         }
-        wordfree (&expand_result);
 #else
 
         /* Ok.  They don't have a recent glibc and they don't have uClibc.  Chances
diff --git a/shell/lash.c b/shell/lash.c
index 8f19e526a..d321c929b 100644
--- a/shell/lash.c
+++ b/shell/lash.c
@@ -941,20 +941,22 @@ static int expand_arguments(char *command)
                 return FALSE;
         }
         
-        /* Convert from char** (one word per string) to a simple char*,
-         * but don't overflow command which is BUFSIZ in length */
-        *command = '\0';
-        while (i < expand_result.we_wordc && total_length < BUFSIZ) {
-                length=strlen(expand_result.we_wordv[i])+1;
-                if (BUFSIZ-total_length-length <= 0) {
-                        error_msg(out_of_space);
-                        return FALSE;
+        if (expand_result.we_wordc > 0) {
+                /* Convert from char** (one word per string) to a simple char*,
+                 * but don't overflow command which is BUFSIZ in length */
+                *command = '\0';
+                while (i < expand_result.we_wordc && total_length < BUFSIZ) {
+                        length=strlen(expand_result.we_wordv[i])+1;
+                        if (BUFSIZ-total_length-length <= 0) {
+                                error_msg(out_of_space);
+                                return FALSE;
+                        }
+                        strcat(command+total_length, expand_result.we_wordv[i++]);
+                        strcat(command+total_length, " ");
+                        total_length+=length;
                 }
-                strcat(command+total_length, expand_result.we_wordv[i++]);
-                strcat(command+total_length, " ");
-                total_length+=length;
+                wordfree (&expand_result);
         }
-        wordfree (&expand_result);
 #else
 
         /* Ok.  They don't have a recent glibc and they don't have uClibc.  Chances