Refactor/cleanup argument parsing functions

src/os/shell.c

@@ -9,108 +9,105 @@
 #include "option_defs.h"
 #include "charset.h"
 
+static int tokenize(char_u *str, char **argv);
+static int word_length(char_u *command);
 
-void shell_skip_word(char_u **cmd)
+// Returns the argument vector for running a shell, with an optional command
+// and extra shell option.
+char ** shell_build_argv(char_u *cmd, char_u *extra_shell_opt)
 {
-  char_u *p = *cmd;
+  int i;
-  bool inquote = false;
+  char **rv;
+  int argc = tokenize(p_sh, NULL) + tokenize(p_shcf, NULL);
 
-  // Move `p` to the end of shell word by advancing the pointer it while it's
+  rv = (char **)alloc((unsigned)((argc + 4) * sizeof(char *)));
-  // inside a quote or it's a non-whitespace character
+
-  while (*p && (inquote || (*p != ' ' && *p != TAB))) {
+  if (rv == NULL) {
-    if (*p == '"')
+    // out of memory
-      // Found a quote character, switch the `inquote` flag
+    return NULL;
-      inquote = !inquote;
-    ++p;
   }
   
-  *cmd = p;
+  // Split 'shell'
-}
+  i = tokenize(p_sh, rv);
 
-int shell_count_argc(char_u **ptr)
+  if (extra_shell_opt != NULL) {
-{
+    // Push a copy of `extra_shell_opt`
-  int rv = 0;
+    rv[i++] = strdup((char *)extra_shell_opt);
-  char_u *p = *ptr;
-
-  while (true) {
-    rv++;
-    shell_skip_word(&p);
-    if (*p == NUL)
-      break;
-    // Move to the next word
-    p = skipwhite(p);
   }
 
-  // Account for multiple args in p_shcf('shellcmdflag' option)
+  if (cmd != NULL) {
-  p = p_shcf;
+    // Split 'shellcmdflag'
-  while (true) {
+    i += tokenize(p_shcf, rv + i);
-    // Same as above, but doesn't need to take quotes into consideration
+    rv[i++] = strdup((char *)cmd);
-    p = skiptowhite(p);
-    if (*p == NUL)
-      break;
-    rv++;
-    p = skipwhite(p);
   }
 
-  *ptr = p;
+  rv[i] = NULL;
 
   return rv;
 }
 
-char ** shell_build_argv(int argc, char_u *cmd,
+void shell_free_argv(char **argv)
-    char_u *extra_shell_arg, char_u **ptr, char_u **p_shcf_copy_ptr)
 {
-  char **argv;
+  char **p = argv;
-  char_u *p_shcf_copy = *p_shcf_copy_ptr;
-  char_u *p = *ptr;
-  // Allocate argv memory
-  argv = (char **)alloc((unsigned)((argc + 4) * sizeof(char *)));
-  if (argv == NULL) // out of memory
-    return NULL;
 
-  // Build argv[]
+  if (p == NULL) {
-  argc = 0;
+    // Nothing was allocated, return
-  while (true) {
+    return;
-    argv[argc] = (char *)p;
-    ++argc;
-    shell_skip_word(&p);
-    if (*p == NUL)
-      break;
-    // Terminate the word
-    *p++ = NUL;
-    p = skipwhite(p);
   }
-  if (cmd != NULL) {
-    char_u  *s;
 
-    if (extra_shell_arg != NULL)
+  while (*p != NULL) {
-      argv[argc++] = (char *)extra_shell_arg;
+    // Free each argument 
+    free(*p);
+    p++;
+  }
 
-    // Break 'shellcmdflag' into white separated parts.  This doesn't
-    // handle quoted strings, they are very unlikely to appear.
-    p_shcf_copy = alloc((unsigned)STRLEN(p_shcf) + 1);
-    if (p_shcf_copy == NULL) {
-      // out of memory 
   free(argv);
-      return NULL;
+}
+
+// Walks through a string and returns the number of shell tokens it contains.
+// If a non-null `argv` parameter is passed, it will be filled with copies
+// of the tokens.
+static int tokenize(char_u *str, char **argv)
+{
+  int argc = 0, len;
+  char_u *p = str;
+
+  while (*p != NUL) {
+    len = word_length(p);
+
+    if (argv != NULL) {
+      // Fill the slot
+      argv[argc] = malloc(len + 1);
+      memcpy(argv[argc], p, len);
+      argv[argc][len] = NUL;
     }
 
-    s = p_shcf_copy;
+    argc++;
-    p = p_shcf;
+    p += len;
-    while (*p != NUL) {
-      argv[argc++] = (char *)s;
-      while (*p && *p != ' ' && *p != TAB)
-        *s++ = *p++;
-      *s++ = NUL;
     p = skipwhite(p);
   }
 
-    argv[argc++] = (char *)cmd;
+  return argc;
+}
+
+// Returns the length of the shell token in `str`
+static int word_length(char_u *str)
+{
+  char_u *p = str;
+  bool inquote = false;
+  int length = 0;
+
+  // Move `p` to the end of shell word by advancing the pointer while it's
+  // inside a quote or it's a non-whitespace character
+  while (*p && (inquote || (*p != ' ' && *p != TAB))) {
+    if (*p == '"') {
+      // Found a quote character, switch the `inquote` flag
+      inquote = !inquote;
     }
 
-  argv[argc] = NULL;
+    p++;
-  *ptr = p;
+    length++;
-  *p_shcf_copy_ptr = p_shcf_copy;
+  }
 
-  return argv;
+  return length;
 }

src/os/shell.h

@@ -5,10 +5,8 @@
 
 #include "types.h"
 
-void shell_skip_word(char_u **ptr);
+char ** shell_build_argv(char_u *cmd, char_u *extra_shell_arg);
-int shell_count_argc(char_u **ptr);
+void shell_free_argv(char **argv);
-char ** shell_build_argv(int argc, char_u *cmd,
-    char_u *extra_shell_arg, char_u **ptr, char_u **p_shcf_copy_ptr);
 
 #endif  // NEOVIM_OS_SHELL_H
 

src/os_unix.c

@@ -1688,7 +1688,6 @@ int mch_call_shell(char_u *cmd, int options, char_u *extra_shell_arg)
 # define EXEC_FAILED 122    /* Exit code when shell didn't execute.  Don't use
                                127, some shells use that already */
 
-  char_u      *newcmd = NULL;
   pid_t pid;
   pid_t wpid = 0;
   pid_t wait_pid = 0;
@@ -1699,8 +1698,6 @@ int mch_call_shell(char_u *cmd, int options, char_u *extra_shell_arg)
 # endif
   int retval = -1;
   char        **argv = NULL;
-  int argc;
-  char_u      *p_shcf_copy = NULL;
   int i;
   char_u      *p;
   int pty_master_fd = -1;                   /* for pty's */
@@ -1710,19 +1707,11 @@ int mch_call_shell(char_u *cmd, int options, char_u *extra_shell_arg)
   char envbuf[50];
   int did_settmode = FALSE;             /* settmode(TMODE_RAW) called */
 
-  newcmd = vim_strsave(p_sh);
-  if (newcmd == NULL)           /* out of memory */
-    goto error;
-
   out_flush();
   if (options & SHELL_COOKED)
     settmode(TMODE_COOK);               /* set to normal mode */
 
-  // Count the number of arguments for the shell
+  argv = shell_build_argv(cmd, extra_shell_arg);
-  p = newcmd;
-  argc = shell_count_argc(&p);
-  p = newcmd;
-  argv = shell_build_argv(argc, cmd, extra_shell_arg, &p, &p_shcf_copy);
 
   if (argv == NULL) {
     goto error;
@@ -2304,15 +2293,13 @@ finished:
         MSG_PUTS(_("\nCommand terminated\n"));
     }
   }
-  vim_free(argv);
+  shell_free_argv(argv);
-  vim_free(p_shcf_copy);
 
 error:
   if (!did_settmode)
     if (tmode == TMODE_RAW)
       settmode(TMODE_RAW);              /* set to raw mode */
   resettitle();
-  vim_free(newcmd);
 
   return retval;
 }