Module: Kernai::Parser
- Defined in:
- lib/kernai/parser.rb
Constant Summary collapse
- BLOCK_PATTERN =
Canonical: <block type=“TYPE” name=“NAME”>content</block>
%r{<block\s+type="([^"]+)"(?:\s+name="([^"]*)")?\s*>(.*?)</block>}m- SHORTHAND_TYPES =
Shorthand: <TYPE name=“NAME”>content</TYPE> (e.g. <final>answer</final>)
Block::TYPES.map(&:to_s).join('|')
- SHORTHAND_PATTERN =
%r{<(#{SHORTHAND_TYPES})(?:\s+name="([^"]*)")?\s*>(.*?)</\1>}m
Class Method Summary collapse
-
.parse(text) ⇒ Object
Scans the response text once and weaves blocks and text segments back in source order.
Class Method Details
.parse(text) ⇒ Object
Scans the response text once and weaves blocks and text segments back in source order. The method is linear but touches a lot of locals (matches, positions, segments) which drives AbcSize up without adding real complexity. rubocop:disable Metrics/AbcSize
18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 |
# File 'lib/kernai/parser.rb', line 18 def parse(text) blocks = [] text_segments = [] # Find all matches from both patterns with their positions matches = [] text.scan(BLOCK_PATTERN) do m = Regexp.last_match matches << { pos: m.begin(0), end_pos: m.end(0), type: m[1], name: m[2], content: m[3] } end text.scan(SHORTHAND_PATTERN) do m = Regexp.last_match # Skip if this region overlaps with an already-found canonical block next if matches.any? { |existing| m.begin(0) >= existing[:pos] && m.begin(0) < existing[:end_pos] } matches << { pos: m.begin(0), end_pos: m.end(0), type: m[1], name: m[2], content: m[3] } end matches.sort_by! { |m| m[:pos] } last_end = 0 matches.each do |m| if m[:pos] > last_end segment = text[last_end...m[:pos]] text_segments << segment unless segment.strip.empty? end blocks << Block.new(type: m[:type].to_sym, content: m[:content], name: m[:name]) last_end = m[:end_pos] end if last_end < text.length segment = text[last_end..] text_segments << segment unless segment.strip.empty? end { blocks: blocks, text_segments: text_segments } end |