legion-data 1.6.11 → 1.6.12

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c0e428fc5d4a92e8fe8b3c24f2277f5842e13b0c9814f712c82cc7cc41a43cfc
-  data.tar.gz: b9d1dd28db840e02b4308d2b29b20a1ff89b7932c07952f243b64cad6b45d5c1
+  metadata.gz: 17ce25b6562386687fec0e9647a28005e54f430a2186ffd841c345b8e5c98a55
+  data.tar.gz: 6227a1417584976e995d4868d4a8ef6d5084a9e7cad49963f7155cb9a7e99b46
 SHA512:
-  metadata.gz: df6941f0d373dbb1d18aadd1b75ee092ec58d8da8fb82b273e6b847f8cf420e999212d0323171bfdf965fe4b105e0e4f13bbe62d3f29ac77f63c35477a4f78c7
-  data.tar.gz: d64f91ecafd301f33153e5bc77988014f11ceb2a3cc0eacf0201bcb19250adc7f71888004643d45ab967cf230ca82fb1986b166b389f1b08a4c098890fed98c1
+  metadata.gz: b066c7ff24ba343941e507616899e77eeb3804a6b83ee0b3954c5646e3ea097daf5402785ef27dc7bba01d5c8742fe7fb23b15d5134a5db058dec6999b5b4b26
+  data.tar.gz: f27b514fec0fb8c2d9d8157f23a69f0c7f81d0eade28ec8f824ef44132d48d40db0c5ff1e97da9e7e8a647fd97314be5d4b3fe3edad8187ebfe42a541d73a281

data/CHANGELOG.md

@@ -1,5 +1,15 @@
 # Legion::Data Changelog
 
+## [1.6.12] - 2026-03-28
+
+### Added
+- VTT (WebVTT) extract handler for meeting transcript parsing (`Handlers::Vtt`)
+  - Parses speaker tags (`<v SpeakerName>`), timestamps, and WEBVTT header
+  - `preserve_speakers: true` (default) prefixes each line with speaker name
+  - Accepts inline VTT string content or a file path
+  - Returns `{ text:, metadata: { bytes:, speakers:, line_count: } }`
+- `.vtt` extension registered in `TypeDetector::EXTENSION_MAP` (maps to `:vtt`)
+
 ## [1.6.11] - 2026-03-28
 
 ### Added

data/lib/legion/data/extract/handlers/vtt.rb

@@ -0,0 +1,65 @@
+# frozen_string_literal: true
+
+module Legion
+  module Data
+    module Extract
+      module Handlers
+        class Vtt < Base
+          TIMESTAMP_PATTERN = /^\d{2}:\d{2}:\d{2}\.\d{3} --> \d{2}:\d{2}:\d{2}\.\d{3}/
+          SPEAKER_TAG_PATTERN = /^<v ([^>]+)>(.*)$/
+
+          def self.type = :vtt
+          def self.extensions = %w[.vtt]
+          def self.gem_name = nil
+
+          def self.extract(source, preserve_speakers: true)
+            content = if source.respond_to?(:read)
+                        source.read
+                      elsif source.is_a?(String) && source.include?("\n")
+                        source
+                      else
+                        File.read(source.to_s)
+                      end
+            lines = parse_vtt(content, preserve_speakers: preserve_speakers)
+            text = lines.join("\n")
+            speakers = extract_speakers(content)
+            {
+              text:     text,
+              metadata: {
+                bytes:      content.bytesize,
+                speakers:   speakers,
+                line_count: lines.size
+              }
+            }
+          rescue StandardError => e
+            { text: nil, error: e.message }
+          end
+
+          def self.parse_vtt(content, preserve_speakers: true)
+            lines = []
+            content.each_line do |raw|
+              line = raw.strip
+              next if line.empty?
+              next if line == 'WEBVTT'
+              next if TIMESTAMP_PATTERN.match?(line)
+
+              if (match = SPEAKER_TAG_PATTERN.match(line))
+                speaker = match[1].strip
+                text = match[2].strip
+                lines << (preserve_speakers ? "#{speaker}: #{text}" : text)
+              else
+                lines << line
+              end
+            end
+            lines
+          end
+
+          def self.extract_speakers(content)
+            content.scan(SPEAKER_TAG_PATTERN).map { |m| m[0].strip }.uniq
+          end
+          private_class_method :parse_vtt, :extract_speakers
+        end
+      end
+    end
+  end
+end

data/lib/legion/data/extract/type_detector.rb

@@ -17,7 +17,8 @@ module Legion
           '.json'     => :json,
           '.jsonl'    => :jsonl,
           '.html'     => :html,
-          '.htm'      => :html
+          '.htm'      => :html,
+          '.vtt'      => :vtt
         }.freeze
 
         module_function

data/lib/legion/data/version.rb

@@ -2,6 +2,6 @@
 
 module Legion
   module Data
-    VERSION = '1.6.11'
+    VERSION = '1.6.12'
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: legion-data
 version: !ruby/object:Gem::Version
-  version: 1.6.11
+  version: 1.6.12
 platform: ruby
 authors:
 - Esity
@@ -123,6 +123,7 @@ files:
 - lib/legion/data/extract/handlers/pdf.rb
 - lib/legion/data/extract/handlers/pptx.rb
 - lib/legion/data/extract/handlers/text.rb
+- lib/legion/data/extract/handlers/vtt.rb
 - lib/legion/data/extract/handlers/xlsx.rb
 - lib/legion/data/extract/type_detector.rb
 - lib/legion/data/helper.rb