Merge pull request #3980 from ZyX-I/shell-unquote

shell: Unquote &shell* options before using them

2 files changed, 61 insertions, 12 deletions

diff --git a/src/nvim/os/shell.c b/src/nvim/os/shell.c
index 3813c45726..f5a1637c94 100644
--- a/src/nvim/os/shell.c
+++ b/src/nvim/os/shell.c
@@ -36,7 +36,6 @@ typedef struct {
 # include "os/shell.c.generated.h"
 #endif
 
-
 /// Builds the argument vector for running the user-configured 'shell' (p_sh)
 /// with an optional command prefixed by 'shellcmdflag' (p_shcf).
 ///
@@ -45,9 +44,10 @@ typedef struct {
 /// @return A newly allocated argument vector. It must be freed with
 ///         `shell_free_argv` when no longer needed.
 char **shell_build_argv(const char *cmd, const char *extra_args)
+  FUNC_ATTR_NONNULL_RET FUNC_ATTR_MALLOC
 {
-  size_t argc = tokenize(p_sh, NULL) + tokenize(p_shcf, NULL);
-  char **rv = xmalloc((unsigned)((argc + 4) * sizeof(char *)));
+  size_t argc = tokenize(p_sh, NULL) + (cmd ? tokenize(p_shcf, NULL) : 0);
+  char **rv = xmalloc((argc + 4) * sizeof(*rv));
 
   // Split 'shell'
   size_t i = tokenize(p_sh, rv);
@@ -338,24 +338,22 @@ static void out_data_cb(Stream *stream, RBuffer *buf, size_t count, void *data,
 /// @param argv The vector that will be filled with copies of the parsed
 ///        words. It can be NULL if the caller only needs to count words.
 /// @return The number of words parsed.
-static size_t tokenize(const char_u *str, char **argv)
+static size_t tokenize(const char_u *const str, char **const argv)
+  FUNC_ATTR_NONNULL_ARG(1)
 {
-  size_t argc = 0, len;
-  char_u *p = (char_u *) str;
+  size_t argc = 0;
+  const char *p = (const char *) str;
 
   while (*p != NUL) {
-    len = word_length(p);
+    const size_t len = word_length((const char_u *) p);
 
     if (argv != NULL) {
       // Fill the slot
-      argv[argc] = xmalloc(len + 1);
-      memcpy(argv[argc], p, len);
-      argv[argc][len] = NUL;
+      argv[argc] = vim_strnsave_unquoted(p, len);
     }
 
     argc++;
-    p += len;
-    p = skipwhite(p);
+    p = (const char *) skipwhite((char_u *) (p + len));
   }
 
   return argc;
@@ -377,6 +375,9 @@ static size_t word_length(const char_u *str)
     if (*p == '"') {
       // Found a quote character, switch the `inquote` flag
       inquote = !inquote;
+    } else if (*p == '\\' && inquote) {
+      p++;
+      length++;
     }
 
     p++;
diff --git a/src/nvim/strings.c b/src/nvim/strings.c
index 00dcf3cf46..fe91141375 100644
--- a/src/nvim/strings.c
+++ b/src/nvim/strings.c
@@ -119,6 +119,54 @@ char_u *vim_strsave_escaped_ext(const char_u *string, const char_u *esc_chars,
   return escaped_string;
 }
 
+/// Save a copy of an unquoted string
+///
+/// Turns string like `a\bc"def\"ghi\\\n"jkl` into `a\bcdef"ghi\\njkl`, for use
+/// in shell_build_argv: the only purpose of backslash is making next character
+/// be treated literally inside the double quotes, if this character is
+/// backslash or quote.
+///
+/// @param[in]  string  String to copy.
+/// @param[in]  length  Length of the string to copy.
+///
+/// @return [allocated] Copy of the string.
+char *vim_strnsave_unquoted(const char *const string, const size_t length)
+  FUNC_ATTR_MALLOC FUNC_ATTR_WARN_UNUSED_RESULT FUNC_ATTR_NONNULL_ALL
+  FUNC_ATTR_NONNULL_RET
+{
+#define ESCAPE_COND(p, inquote, string_end) \
+  (*p == '\\' && inquote && p + 1 < string_end && (p[1] == '\\' || p[1] == '"'))
+  size_t ret_length = 0;
+  bool inquote = false;
+  const char *const string_end = string + length;
+  for (const char *p = string; p < string_end; p++) {
+    if (*p == '"') {
+      inquote = !inquote;
+    } else if (ESCAPE_COND(p, inquote, string_end)) {
+      ret_length++;
+      p++;
+    } else {
+      ret_length++;
+    }
+  }
+
+  char *const ret = xmallocz(ret_length);
+  char *rp = ret;
+  inquote = false;
+  for (const char *p = string; p < string_end; p++) {
+    if (*p == '"') {
+      inquote = !inquote;
+    } else if (ESCAPE_COND(p, inquote, string_end)) {
+      *rp++ = *(++p);
+    } else {
+      *rp++ = *p;
+    }
+  }
+#undef ESCAPE_COND
+
+  return ret;
+}
+
 /*
  * Escape "string" for use as a shell argument with system().
  * This uses single quotes, except when we know we need to use double quotes