wildcard matcher

src/base/base_strings.c

@@ -325,6 +325,76 @@ str8_match(String8 a, String8 b, StringMatchFlags flags)
   return result;
 }
 
+internal B32
+str8_char_match(U8 a, U8 b, StringMatchFlags flags)
+{
+  U8 at = a;
+  U8 bt = b;
+  if (flags & StringMatchFlag_CaseInsensitive) {
+    at = upper_from_char(at);
+    bt = upper_from_char(bt);
+  }
+  if (flags & StringMatchFlag_SlashInsensitive) {
+    if (at == '\\') { at = '/'; }
+    if (bt == '\\') { bt = '/'; }
+  }
+  return (at == bt);
+}
+
+internal B32
+str8_match_wildcard(String8 string, String8 pattern, StringMatchFlags flags)
+{
+  B32 matched = 0;
+
+  U64 pattern_cursor = 0;
+  U64 string_cursor  = 0;
+
+  U64 pattern_start = max_U64;
+  U64 string_start  = 0;
+
+  for (;;) {
+    if (pattern_cursor == pattern.size) {
+      if (string_cursor == string.size || (flags & StringMatchFlag_RightSideSloppy)) {
+        matched = 1;
+        break;
+      }
+    }
+
+    if (string_cursor == string.size) {
+      while (pattern_cursor < pattern.size && pattern.str[pattern_cursor] == '*') {
+        pattern_cursor += 1;
+      }
+      matched = (pattern_cursor == pattern.size);
+      break;
+    }
+
+    if (pattern_cursor < pattern.size && pattern.str[pattern_cursor] == '*') {
+      pattern_start = pattern_cursor;
+      string_start = string_cursor;
+      pattern_cursor += 1;
+      continue;
+    }
+
+
+    if (pattern_cursor < pattern.size && (pattern.str[pattern_cursor] == '?' || str8_char_match(string.str[string_cursor], pattern.str[pattern_cursor], flags))) {
+      string_cursor  += 1;
+      pattern_cursor += 1;
+      continue;
+    }
+
+    if (pattern_start != max_U64) {
+      pattern_cursor = pattern_start + 1;
+      string_start += 1;
+      string_cursor = string_start;
+      continue;
+    }
+
+    break;
+  }
+
+  return matched;
+}
+
 internal U64
 str8_find_needle(String8 string, U64 start_pos, String8 needle, StringMatchFlags flags)
 {

src/base/base_strings.h

@@ -197,6 +197,7 @@ internal String8 backslashed_from_str8(Arena *arena, String8 string);
 #define str8_match_lit(a_lit, b, flags)   str8_match(str8_lit(a_lit), (b), (flags))
 #define str8_match_cstr(a_cstr, b, flags) str8_match(str8_cstring(a_cstr), (b), (flags))
 internal B32 str8_match(String8 a, String8 b, StringMatchFlags flags);
+internal B32 str8_match_wildcard(String8 string, String8 pattern, StringMatchFlags flags);
 internal U64 str8_find_needle(String8 string, U64 start_pos, String8 needle, StringMatchFlags flags);
 internal U64 str8_find_needle_reverse(String8 string, U64 start_pos, String8 needle, StringMatchFlags flags);
 internal B32 str8_is_before(String8 a, String8 b);

src/torture/torture_base.c

@@ -104,4 +104,76 @@ TEST(count_digits)
   T_Ok(count_digits_u64(999999, 10) == 6);
 }
 
+TEST(match_wildcard)
+{
+  // empty strings
+  T_Ok(str8_match_wildcard(str8_lit(""),  str8_lit(""),   0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit(""),  str8_lit("*"),  0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit(""),  str8_lit("**"), 0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit(""),  str8_lit("?"),  0) == 0);
+  T_Ok(str8_match_wildcard(str8_lit(""),  str8_lit("*?"), 0) == 0);
+  T_Ok(str8_match_wildcard(str8_lit(""),  str8_lit("?*"), 0) == 0);
+  T_Ok(str8_match_wildcard(str8_lit("a"), str8_lit(""),   0) == 0);
+
+  // exact
+  T_Ok(str8_match_wildcard(str8_lit("a"), str8_lit("a"), 0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("a"), str8_lit("A"), 0) == 0);
+
+  // ?
+  T_Ok(str8_match_wildcard(str8_lit("a"),  str8_lit("?"),    0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit(""),   str8_lit("?"),    0) == 0);
+  T_Ok(str8_match_wildcard(str8_lit("ab"), str8_lit("?"),    0) == 0);
+  T_Ok(str8_match_wildcard(str8_lit("ab"), str8_lit("a?"),   0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("ab"), str8_lit("?b"),   0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("ab"), str8_lit("??"),   0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("abc"), str8_lit("a?c"), 0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("ab"), str8_lit("???"),  0) == 0);
+
+  // *
+  T_Ok(str8_match_wildcard(str8_lit(""),    str8_lit("*"),      0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("a"),   str8_lit("*"),      0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("abc"), str8_lit("*"),      0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("abc"), str8_lit("a*"),     0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("abc"), str8_lit("*c"),     0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("abc"), str8_lit("*b*"),    0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("abc"), str8_lit("a*c"),    0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("abc"), str8_lit("b*"),     0) == 0);
+  T_Ok(str8_match_wildcard(str8_lit("abc"), str8_lit("**"),     0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("abc"), str8_lit("a**c"),   0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("abc"), str8_lit("a*b*c"),  0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("abc"), str8_lit("*a*d*"),  0) == 0);
+
+  T_Ok(str8_match_wildcard(str8_lit("abcd"),              str8_lit("a*d"),        0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("abefcdgiescdfimde"), str8_lit("ab*cd?i*de"), 0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("mississippi"),       str8_lit("m*iss*ppi"),  0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("abc"),               str8_lit("*b"),         0) == 0);
+  T_Ok(str8_match_wildcard(str8_lit("a"),                 str8_lit("aa"),         0) == 0);
+  T_Ok(str8_match_wildcard(str8_lit("aa"),                str8_lit("a"),          0) == 0);
+
+  // case insensitive
+  T_Ok(str8_match_wildcard(str8_lit("a"),      str8_lit("A"),      StringMatchFlag_CaseInsensitive) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("FooBar"), str8_lit("foobar"), StringMatchFlag_CaseInsensitive) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("Foobar"), str8_lit("foo*"),   StringMatchFlag_CaseInsensitive) == 1);
+
+  // right side sloppy
+  T_Ok(str8_match_wildcard(str8_lit("abc"), str8_lit("ab"), StringMatchFlag_RightSideSloppy) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("abc"), str8_lit(""),   StringMatchFlag_RightSideSloppy) == 1);
+  T_Ok(str8_match_wildcard(str8_lit(""),    str8_lit("a"),  StringMatchFlag_RightSideSloppy) == 0);
+
+  // slash insensitive
+  T_Ok(str8_match_wildcard(str8_lit("a/b"),   str8_lit("a\\b"),    0)                                == 0);
+  T_Ok(str8_match_wildcard(str8_lit("a/b"),   str8_lit("a\\b"),    StringMatchFlag_SlashInsensitive) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("a/b/c"), str8_lit("a\\*\\c"), StringMatchFlag_SlashInsensitive) == 1);
+
+  // combined
+  T_Ok(str8_match_wildcard(str8_lit("Ab\\Cde"), str8_lit("ab/*e"), StringMatchFlag_CaseInsensitive|StringMatchFlag_SlashInsensitive) == 1);
+
+  T_Ok(str8_match_wildcard(str8_lit("abc"), str8_lit("*?*"), 0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("a"),   str8_lit("*?*"), 0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit(""),    str8_lit("*?*"), 0) == 0);
+  T_Ok(str8_match_wildcard(str8_lit("abc"), str8_lit("?*?"), 0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("ab"),  str8_lit("?*?"), 0) == 1);
+  T_Ok(str8_match_wildcard(str8_lit("a"),   str8_lit("?*?"), 0) == 0);
+}
+
 #undef T_Group