add Sublime-like fuzzy search to file browser

Author: cdacamar

src/base/base_string.c

@@ -1699,6 +1699,85 @@ fuzzy_match_find(Arena *arena, String8 needle, String8 haystack)
   scratch_end(scratch);
   return result;
 }
+
+internal ScoredFuzzyMatchRangeList
+scored_fuzzy_match_find(Arena *arena, String8 needle, String8 haystack)
+{
+  Temp scratch = scratch_begin(0, 0);
+  // We're going to implement a very simple scoring mechanism similar to that described in
+  // https://www.forrestthewoods.com/blog/reverse_engineering_sublime_texts_fuzzy_match/.
+#define scored_fuzzy_match_unmatched -1
+#define scored_fuzzy_match_consecutive 5
+#define scored_fuzzy_match_unmatched_leading -3
+  ScoredFuzzyMatchRangeList invalid = {0};
+  ScoredFuzzyMatchRangeList result = {0};
+  // Simplify to a single needle which has common delimiters removed.
+  String8List needles = str8_split(scratch.arena, needle, (U8*)" ", 1, 0);
+  needle = str8_list_join(scratch.arena, &needles, 0);
+  if (needle.size == 0)
+  {
+    scratch_end(scratch);
+    return invalid;
+  }
+  String8 tmp_str = str8(needle.str, 1);
+  U64 find_pos = 0;
+  find_pos = str8_find_needle(haystack, find_pos, tmp_str, StringMatchFlag_CaseInsensitive);
+  if (find_pos >= haystack.size)
+  {
+    scratch_end(scratch);
+    return invalid;
+  }
+  // Leading character penalty.
+  // Only go to a max of 3 based on the article.
+  result.score += Min(find_pos, 3) * scored_fuzzy_match_unmatched_leading;
+  // We also want to deduct for additional unmatched characters between start and find_pos.
+  if (find_pos > 3)
+  {
+    result.score += (find_pos - 3) * scored_fuzzy_match_unmatched;
+  }
+  Rng1U64 range = r1u64(find_pos, find_pos + 1);
+  FuzzyMatchRangeNode *n = push_array(arena, FuzzyMatchRangeNode, 1);
+  n->range = range;
+  SLLQueuePush(result.list.first, result.list.last, n);
+  result.list.count += 1;
+  // Match the rest.
+  U64 prev_found = find_pos;
+  U64 search_start = 0;
+  find_pos += 1;
+  for (U64 idx = 1; idx < needle.size; ++idx)
+  {
+    tmp_str = str8(needle.str + idx, 1);
+    search_start = find_pos;
+    find_pos = str8_find_needle(haystack, find_pos, tmp_str, StringMatchFlag_CaseInsensitive);
+    if (find_pos >= haystack.size)
+    {
+      scratch_end(scratch);
+      return invalid;
+    }
+    // Compute consecutive bonus.
+    if (prev_found + 1 == find_pos)
+    {
+      result.score += scored_fuzzy_match_consecutive;
+      // We can reuse the existing node and simply extend it.
+      result.list.last->range.max = find_pos + 1;
+    }
+    else
+    {
+      result.score += (find_pos - search_start) * scored_fuzzy_match_unmatched;
+      Rng1U64 range = r1u64(find_pos, find_pos + 1);
+      FuzzyMatchRangeNode *n = push_array(arena, FuzzyMatchRangeNode, 1);
+      n->range = range;
+      SLLQueuePush(result.list.first, result.list.last, n);
+      result.list.count += 1;
+    }
+    prev_found = find_pos;
+    find_pos += 1;
+  }
+  // Compute final unmatched characters.
+  result.score += (haystack.size - find_pos) * scored_fuzzy_match_unmatched;
+  scratch_end(scratch);
+  return result;
+}
 ////////////////////////////////
 //~ NOTE(allen): Serialization Helpers
 

src/base/base_string.h

@@ -148,6 +148,13 @@ struct FuzzyMatchRangeList
   U64 total_dim;
 };
 
+typedef struct ScoredFuzzyMatchRangeList ScoredFuzzyMatchRangeList;
+struct ScoredFuzzyMatchRangeList
+{
+  FuzzyMatchRangeList list;
+  S32 score;
+};
+
 ////////////////////////////////
 //~ rjf: Character Classification & Conversion Functions
 
@@ -342,6 +349,7 @@ internal Vec4F32 rgba_from_hex_string_4f32(String8 hex_string);
 //~ rjf: String Fuzzy Matching
 
 internal FuzzyMatchRangeList fuzzy_match_find(Arena *arena, String8 needle, String8 haystack);
+internal ScoredFuzzyMatchRangeList scored_fuzzy_match_find(Arena *arena, String8 needles, String8 haystack);
 
 ////////////////////////////////
 //~ NOTE(allen): Serialization Helpers

src/df/gfx/df_views.c

@@ -26,16 +26,15 @@ df_qsort_compare_file_info__default(DF_FileInfo *a, DF_FileInfo *b)
 internal int
 df_qsort_compare_file_info__default_filtered(DF_FileInfo *a, DF_FileInfo *b)
 {
-  int result = 0;
-  if(a->filename.size < b->filename.size)
+  if (a->match_ranges.score > b->match_ranges.score)
   {
-    result = -1;
+    return -1;
   }
-  else if(a->filename.size > b->filename.size)
+  if (a->match_ranges.score < b->match_ranges.score)
   {
-    result = +1;
+    return 1;
   }
-  return result;
+  return 0;
 }
 
 internal int
@@ -2242,8 +2241,8 @@ DF_VIEW_UI_FUNCTION_DEF(FileSystem)
       OS_FileIter *it = os_file_iter_begin(scratch.arena, path_query.path, 0);
       for(OS_FileInfo info = {0}; os_file_iter_next(scratch.arena, it, &info);)
       {
-        FuzzyMatchRangeList match_ranges = fuzzy_match_find(fs->cached_files_arena, path_query.search, info.name);
-        B32 fits_search = (path_query.search.size == 0 || match_ranges.count == match_ranges.needle_part_count);
+        ScoredFuzzyMatchRangeList match_ranges = scored_fuzzy_match_find(fs->cached_files_arena, path_query.search, info.name);
+        B32 fits_search = (path_query.search.size == 0 || match_ranges.list.count != 0);
         B32 fits_dir_only = !!(info.props.flags & FilePropertyFlag_IsFolder) || !dir_selection;
         if(fits_search && fits_dir_only)
         {
@@ -2542,7 +2541,7 @@ DF_VIEW_UI_FUNCTION_DEF(FileSystem)
           UI_PrefWidth(ui_pct(1, 0))
           {
             UI_Box *box = ui_build_box_from_stringf(UI_BoxFlag_DrawText, "%S##%p", file->filename, view);
-            df_box_equip_fuzzy_match_range_list_vis(box, file->match_ranges);
+            df_box_equip_fuzzy_match_range_list_vis(box, file->match_ranges.list);
           }
         }
 

src/df/gfx/df_views.h

@@ -22,7 +22,7 @@ struct DF_FileInfo
 {
   String8 filename;
   FileProperties props;
-  FuzzyMatchRangeList match_ranges;
+  ScoredFuzzyMatchRangeList match_ranges;
 };
 
 typedef struct DF_FileInfoNode DF_FileInfoNode;