Merge pull request #18653 from dundargoc/refactor/move-reverse-text

refactor: move reverse_text to strings.c as it's a string operation

src/nvim/search.c

@@ -204,31 +204,6 @@ char_u *get_search_pat(void)
   return mr_pattern;
 }
 
-/*
- * Reverse text into allocated memory.
- * Returns the allocated string.
- *
- * TODO(philix): move reverse_text() to strings.c
- */
-char_u *reverse_text(char_u *s) FUNC_ATTR_NONNULL_RET
-{
-  /*
-   * Reverse the pattern.
-   */
-  size_t len = STRLEN(s);
-  char_u *rev = xmalloc(len + 1);
-  size_t rev_i = len;
-  for (size_t s_i = 0; s_i < len; s_i++) {
-    const int mb_len = utfc_ptr2len((char *)s + s_i);
-    rev_i -= mb_len;
-    memmove(rev + rev_i, s + s_i, mb_len);
-    s_i += mb_len - 1;
-  }
-  rev[len] = NUL;
-
-  return rev;
-}
-
 void save_re_pat(int idx, char_u *pat, int magic)
 {
   if (spats[idx].pat != pat) {

src/nvim/strings.c

@@ -1507,3 +1507,24 @@ int kv_do_printf(StringBuilder *str, const char *fmt, ...)
   str->size += (size_t)printed;
   return printed;
 }
+
+/// Reverse text into allocated memory.
+///
+/// @return  the allocated string.
+char_u *reverse_text(char_u *s)
+  FUNC_ATTR_NONNULL_RET
+{
+  // Reverse the pattern.
+  size_t len = STRLEN(s);
+  char_u *rev = xmalloc(len + 1);
+  size_t rev_i = len;
+  for (size_t s_i = 0; s_i < len; s_i++) {
+    const int mb_len = utfc_ptr2len((char *)s + s_i);
+    rev_i -= (size_t)mb_len;
+    memmove(rev + rev_i, s + s_i, (size_t)mb_len);
+    s_i += (size_t)mb_len - 1;
+  }
+  rev[len] = NUL;
+
+  return rev;
+}

test/unit/search_spec.lua

@@ -5,6 +5,8 @@ local to_cstr = helpers.to_cstr
 local eq      = helpers.eq
 
 local search = helpers.cimport("./src/nvim/search.h")
+local globals = helpers.cimport('./src/nvim/globals.h')
+local ffi = helpers.ffi
 
 itp('pat_has_uppercase', function()
   -- works on empty string
@@ -31,3 +33,25 @@ itp('pat_has_uppercase', function()
   eq(false, search.pat_has_uppercase(to_cstr("aa\\%Ab")))
   eq(true, search.pat_has_uppercase(to_cstr("aab\\%AU")))
 end)
+
+describe('search_regcomp', function()
+  local search_regcomp = function(pat, pat_save, pat_use, options )
+    local regmatch = ffi.new("regmmatch_T")
+    local fail = search.search_regcomp(to_cstr(pat), pat_save, pat_use, options, regmatch)
+    return fail, regmatch
+  end
+
+  local get_search_pat = function()
+    return helpers.internalize(search.get_search_pat())
+  end
+
+  itp("accepts regexp pattern with invalid utf", function()
+    --crafted to call reverse_text with invalid utf
+    globals.curwin.w_onebuf_opt.wo_rl = 1
+    globals.curwin.w_onebuf_opt.wo_rlc = to_cstr('s')
+    globals.cmdmod.keeppatterns = 1
+    local fail = search_regcomp("a\192", 0,0,0)
+    eq(1, fail)
+    eq("\192a", get_search_pat())
+  end)
+end)

test/unit/strings_spec.lua

@@ -150,3 +150,38 @@ describe('strcase_save()' , function()
     eq("a", strcase_save("\xc1\x81", false))
   end)
 end)
+
+describe("reverse_text", function()
+  local reverse_text = function(str)
+    return helpers.internalize(strings.reverse_text(to_cstr(str)))
+  end
+
+  itp("handles empty string", function()
+    eq("", reverse_text(""))
+  end)
+
+  itp("handles simple cases", function()
+    eq("a", reverse_text("a"))
+    eq("ba", reverse_text("ab"))
+  end)
+
+  itp("handles multibyte characters", function()
+    eq("bα", reverse_text("αb"))
+    eq("Yötön yö", reverse_text("öy nötöY"))
+  end)
+
+  itp("handles combining chars", function()
+    local utf8_COMBINING_RING_ABOVE = "\204\138"
+    local utf8_COMBINING_RING_BELOW = "\204\165"
+    eq("bba" .. utf8_COMBINING_RING_ABOVE .. utf8_COMBINING_RING_BELOW .. "aa",
+       reverse_text("aaa" .. utf8_COMBINING_RING_ABOVE .. utf8_COMBINING_RING_BELOW .. "bb"))
+  end)
+
+  itp("treats invalid utf as separate characters", function()
+    eq("\192ba", reverse_text("ab\192"))
+  end)
+
+  itp("treats an incomplete utf continuation sequence as valid", function()
+    eq("\194ba", reverse_text("ab\194"))
+  end)
+end)