Location byte position

ext/rbs_extension/legacy_location.c

@@ -193,16 +193,16 @@ VALUE rbs_new_location(VALUE buffer, rbs_range_t rg) {
     rbs_loc *loc;
     VALUE obj = TypedData_Make_Struct(RBS_Location, rbs_loc, &location_type, loc);
 
-    rbs_loc_init(loc, buffer, (rbs_loc_range) { rg.start.char_pos, rg.end.char_pos });
+    rbs_loc_init(loc, buffer, (rbs_loc_range) { rg.start.byte_pos, rg.end.byte_pos });
 
     return obj;
 }
 
-VALUE rbs_new_location2(VALUE buffer, int start_char, int end_char) {
+VALUE rbs_new_location2(VALUE buffer, int start_byte, int end_byte) {
     rbs_loc *loc;
     VALUE obj = TypedData_Make_Struct(RBS_Location, rbs_loc, &location_type, loc);
 
-    rbs_loc_init(loc, buffer, (rbs_loc_range) { .start = start_char, .end = end_char });
+    rbs_loc_init(loc, buffer, (rbs_loc_range) { start_byte, end_byte });
 
     return obj;
 }

ext/rbs_extension/legacy_location.h

@@ -23,7 +23,7 @@ SUPPRESS_RUBY_HEADER_DIAGNOSTICS_END
 extern VALUE RBS_Location;
 
 /**
- * Range of character index for `rbs_loc` locations.
+ * Range of byte index for `rbs_loc` locations.
  */
 typedef struct {
     int start;
@@ -57,7 +57,7 @@ typedef struct {
  * */
 VALUE rbs_new_location(VALUE buffer, rbs_range_t rg);
 
-VALUE rbs_new_location2(VALUE buffer, int start_char, int end_char);
+VALUE rbs_new_location2(VALUE buffer, int start_byte, int end_byte);
 
 /**
  * Return rbs_loc associated with the RBS::Location object.

lib/rbs/ast/ruby/comment_block.rb

@@ -25,9 +25,9 @@ def initialize(source_buffer, comments)
 
             offsets << tuple
 
-            start_char = comment.location.start_character_offset + tuple[1]
-            end_char = comment.location.end_character_offset
-            ranges << (start_char ... end_char)
+            start_byte = comment.location.start_offset + tuple[1]
+            end_byte = comment.location.end_offset
+            ranges << (start_byte ... end_byte)
           end
 
           @comment_buffer = source_buffer.sub_buffer(lines: ranges)
@@ -53,7 +53,7 @@ def end_line
 
         def line_starts
           offsets.map do |comment, prefix_size|
-            comment.location.start_character_offset + prefix_size
+            comment.location.start_offset + prefix_size
           end
         end
 
@@ -168,7 +168,7 @@ def yield_annotation(start_line, end_line, current_line, variables, &block)
 
         def text(comment_index)
           range = comment_buffer.ranges[comment_index]
-          comment_buffer.content[range] or raise
+          comment_buffer.content.byteslice(range) or raise
         end
 
         def line_location(start_line, end_line)

lib/rbs/ast/ruby/declarations.rb

@@ -44,8 +44,8 @@ def location
               if type_annotation
                 Location.new(
                   type_name_location.buffer,
-                  type_name_location.start_pos,
-                  type_annotation.location.end_pos
+                  type_name_location.start_byte,
+                  type_annotation.location.end_byte
                 )
               else
                 type_name_location

lib/rbs/ast/ruby/helpers/location_helper.rb

@@ -6,7 +6,7 @@ module Ruby
       module Helpers
         module LocationHelper
           def rbs_location(location)
-            Location.new(buffer, location.start_character_offset, location.end_character_offset)
+            Location.new(buffer, location.start_offset, location.end_offset)
           end
         end
       end

lib/rbs/buffer.rb

@@ -20,7 +20,7 @@ def initialize(name: nil, content:, parent: nil)
     end
 
     def lines
-      ranges.map { self.content[_1] || raise } #$ String
+      ranges.map { content.byteslice(_1) || raise } #$ String
     end
 
     def line_count
@@ -31,7 +31,6 @@ def ranges
       @ranges ||= begin
         if content.empty?
           ranges = [0...0] #: Array[Range[Integer]]
-          lines = [""]
         else
           lines = content.lines
           lines << "" if content.end_with?("\n")
@@ -40,9 +39,9 @@ def ranges
           offset = 0
 
           lines.each do |line|
-            size0 = line.size
+            size0 = line.bytesize
             line = line.chomp
-            range = offset...(offset+line.size)
+            range = offset...(offset+line.bytesize)
             ranges << range
 
             offset += size0
@@ -89,9 +88,9 @@ def inspect
 
     def rbs_location(location, loc2=nil)
       if loc2
-        Location.new(self.top_buffer, location.start_character_offset, loc2.end_character_offset)
+        Location.new(self.top_buffer, location.start_offset, loc2.end_offset)
       else
-        Location.new(self.top_buffer, location.start_character_offset, location.end_character_offset)
+        Location.new(self.top_buffer, location.start_offset, location.end_offset)
       end
     end
 
@@ -100,7 +99,7 @@ def sub_buffer(lines:)
       lines.each_with_index do |range, index|
         start_pos = range.begin
         end_pos = range.end
-        slice = content[start_pos...end_pos] or raise
+        slice = content.byteslice(start_pos, end_pos - start_pos) or raise
         if slice.include?("\n")
           raise "Line #{index + 1} cannot contain newline character."
         end

lib/rbs/errors.rb

@@ -34,8 +34,8 @@ def detailed_message(highlight: false, **)
       # Support only one line
       return msg unless location.start_line == location.end_line
 
-      indent = " " * location.start_column
-      marker = "^" * ([location.end_column - location.start_column, 1].max or raise)
+      indent = " " * location.start_byte_column
+      marker = "^" * ([location.end_byte_column - location.start_byte_column, 1].max or raise)
 
       io = StringIO.new
       io.puts msg

lib/rbs/location_aux.rb

@@ -10,7 +10,7 @@ def inspect
       else
         source.each_line.first&.chomp&.inspect
       end
-      "#<#{self.class}:#{self.__id__} buffer=#{buffer.name}, start=#{start_line}:#{start_column}, pos=#{start_pos}...#{end_pos}, children=#{(rks + ops).join(",")} source=#{src}>"
+      "#<#{self.class}:#{self.__id__} buffer=#{buffer.name}, start=#{start_line}:#{start_byte_column}, pos=#{start_byte}...#{end_byte}, children=#{(rks + ops).join(",")} source=#{src}>"
     end
 
     def self.new(buffer_ = nil, start_pos_ = nil, end_pos_ = nil, buffer: nil, start_pos: nil, end_pos: nil)
@@ -28,14 +28,36 @@ def self.new(buffer_ = nil, start_pos_ = nil, end_pos_ = nil, buffer: nil, start
 
     WithChildren = self
 
-    def start_pos
+    def start_byte
       buffer.absolute_position(_start_pos) || raise
     end
 
-    def end_pos
+    def end_byte
       buffer.absolute_position(_end_pos) || raise
     end
 
+    def start_char
+      top = buffer.top_buffer
+      top.content.byteslice(0, start_byte)&.length || 0
+    end
+
+    def end_char
+      top = buffer.top_buffer
+      top.content.byteslice(0, end_byte)&.length || 0
+    end
+
+    # @deprecated Use `start_char` or `start_byte` instead.
+    def start_pos
+      RBS.print_warning { "`RBS::Location#start_pos` is deprecated. Use `start_char` or `start_byte` instead." }
+      start_char
+    end
+
+    # @deprecated Use `end_char` or `end_byte` instead.
+    def end_pos
+      RBS.print_warning { "`RBS::Location#end_pos` is deprecated. Use `end_char` or `end_byte` instead." }
+      end_char
+    end
+
     def name
       buffer.name
     end
@@ -44,54 +66,80 @@ def start_line
       start_loc[0]
     end
 
-    def start_column
+    def start_byte_column
       start_loc[1]
     end
 
+    def start_char_column
+      top = buffer.top_buffer
+      line, byte_col = top.pos_to_loc(start_byte)
+      line_range = top.ranges[line - 1] or return 0
+      top.content.byteslice(line_range.begin, byte_col)&.length || 0
+    end
+
+    def end_byte_column
+      end_loc[1]
+    end
+
+    def end_char_column
+      top = buffer.top_buffer
+      line, byte_col = top.pos_to_loc(end_byte)
+      line_range = top.ranges[line - 1] or return 0
+      top.content.byteslice(line_range.begin, byte_col)&.length || 0
+    end
+
+    # @deprecated Use `start_byte_column` or `start_char_column` instead.
+    def start_column
+      RBS.print_warning { "`RBS::Location#start_column` is deprecated. Use `start_byte_column` or `start_char_column` instead." }
+      start_byte_column
+    end
+
     def end_line
       end_loc[0]
     end
 
+    # @deprecated Use `end_byte_column` or `end_char_column` instead.
     def end_column
-      end_loc[1]
+      RBS.print_warning { "`RBS::Location#end_column` is deprecated. Use `end_byte_column` or `end_char_column` instead." }
+      end_byte_column
     end
 
     def start_loc
-      @start_loc ||= buffer.top_buffer.pos_to_loc(start_pos)
+      @start_loc ||= buffer.top_buffer.pos_to_loc(start_byte)
     end
 
     def end_loc
-      @end_loc ||= buffer.top_buffer.pos_to_loc(end_pos)
+      @end_loc ||= buffer.top_buffer.pos_to_loc(end_byte)
     end
 
     def range
-      @range ||= start_pos...end_pos
+      @range ||= start_byte...end_byte
     end
 
     def source
-      @source ||= (buffer.top_buffer.content[range] || raise)
+      @source ||= (buffer.top_buffer.content.byteslice(start_byte, end_byte - start_byte) || raise)
     end
 
     def to_s
-      "#{name || "-"}:#{start_line}:#{start_column}...#{end_line}:#{end_column}"
+      "#{name || "-"}:#{start_line}:#{start_byte_column}...#{end_line}:#{end_byte_column}"
     end
 
     def ==(other)
       other.is_a?(Location) &&
         other.buffer == buffer &&
-        other.start_pos == start_pos &&
-        other.end_pos == end_pos
+        other.start_byte == start_byte &&
+        other.end_byte == end_byte
     end
 
     def to_json(state = nil)
       {
         start: {
           line: start_line,
-          column: start_column
+          column: start_byte_column
         },
         end: {
           line: end_line,
-          column: end_column
+          column: end_byte_column
         },
         buffer: {
           name: name&.to_s

lib/rbs/locator.rb

@@ -10,8 +10,16 @@ def initialize(buffer:, dirs:, decls:)
       @decls = decls
     end
 
-    def find(line:, column:)
-      pos = buffer.loc_to_pos([line, column])
+    # Find RBS elements at the given position.
+    #
+    # Returns a list of components, inner component comes first.
+    #
+    # @param line [Integer] Line number (1-origin)
+    # @param byte_column [Integer] Byte offset within the line (0-origin)
+    # @param char_column [Integer] Character offset within the line (0-origin). Converted to byte offset internally.
+    # @param column [Integer] Deprecated. Treated as char column for backwards compatibility.
+    def find(line:, column: nil, byte_column: nil, char_column: nil)
+      pos = resolve_position(line: line, column: column, byte_column: byte_column, char_column: char_column)
 
       dirs.each do |dir|
         array = [] #: Array[component]
@@ -26,8 +34,14 @@ def find(line:, column:)
       []
     end
 
-    def find2(line:, column:)
-      path = find(line: line, column: column)
+    # Find RBS elements at the given position, returning the inner most symbol and outer components.
+    #
+    # @param line [Integer] Line number (1-origin)
+    # @param byte_column [Integer] Byte offset within the line (0-origin)
+    # @param char_column [Integer] Character offset within the line (0-origin). Converted to byte offset internally.
+    # @param column [Integer] Deprecated. Treated as char column for backwards compatibility.
+    def find2(line:, column: nil, byte_column: nil, char_column: nil)
+      path = find(line: line, column: column, byte_column: byte_column, char_column: char_column)
 
       return if path.empty?
 
@@ -238,10 +252,33 @@ def find_in_loc(pos, location:, array:)
 
     def test_loc(pos, location:)
       if location
-        location.start_pos <= pos && pos <= location.end_pos
+        location.start_byte <= pos && pos <= location.end_byte
       else
         false
       end
     end
+
+    private
+
+    # Converts char column to byte column within a line.
+    def char_column_to_byte_column(line, char_column) #: Integer
+      line_range = buffer.ranges[line - 1] or return 0
+      line_content = buffer.content.byteslice(line_range) or return 0
+      prefix = line_content.encode(Encoding::UTF_8).chars.first(char_column).join
+      prefix.bytesize
+    end
+
+    def resolve_position(line:, column:, byte_column:, char_column:) #: Integer
+      if byte_column
+        buffer.loc_to_pos([line, byte_column])
+      elsif char_column
+        buffer.loc_to_pos([line, char_column_to_byte_column(line, char_column)])
+      elsif column
+        RBS.print_warning { "`column` keyword argument of RBS::Locator#find is deprecated. Use `char_column` or `byte_column` instead." }
+        buffer.loc_to_pos([line, char_column_to_byte_column(line, column)])
+      else
+        raise ArgumentError, "One of `byte_column`, `char_column`, or `column` must be specified"
+      end
+    end
   end
 end

lib/rbs/parser_aux.rb

@@ -23,7 +23,7 @@ def self.parse_signature(source)
       resolved = magic_comment(buf)
       start_pos =
         if resolved
-          (resolved.location || raise).end_pos
+          (resolved.location || raise).end_byte
         else
           0
         end