amagi-captions 1.3.2

data/lib/captions/formats/vtt.rb

@@ -0,0 +1,122 @@
+module Captions
+  class VTT < Base
+
+    # Header used for all VTT files
+    VTT_HEADER = "WEBVTT"
+
+    # VTT file comments/style section
+    VTT_METADATA = /^NOTE|^STYLE/
+
+    # Auto Keyword used in Alignment
+    AUTO_KEYWORD = "auto"
+
+    # Alignment Data
+    ALIGNMENT_VALUES = {
+      "middle" => "middle",
+      "center" => "middle",
+      "left"   => "left",
+      "right"  => "right",
+      "start"  => "start",
+      "end"    => "end",
+    }
+
+    # Parse VTT file and update CueList
+    def parse
+      base_parser do
+        count = 1
+        cue_count = 0
+        meta_data_section = false
+        cue = nil
+        raise InvalidSubtitle, "Invalid VTT Signature" unless validate_header(@file.gets)
+        while(line = @file.gets) do
+          line = line.strip
+          if line.empty?
+            meta_data_section = false
+          elsif is_meta_data?(line)
+            meta_data_section = true
+          elsif is_time?(line)
+            @cue_list.append(cue) if cue
+            cue_count += 1
+            cue = Cue.new
+            cue.number = cue_count
+            line = line.split
+            cue.set_time(line[0], line[2])
+            set_properties(cue, line[3..-1])
+          elsif !meta_data_section and is_text?(line)
+            cue.add_text(line)
+          end
+        end
+        @cue_list.append(cue) if cue
+      end
+    end
+
+    # Export CueList to VTT file
+    def dump(file)
+      base_dump(file) do |file|
+        file.write(VTT_HEADER)
+        @cue_list.each do |cue|
+          file.write("\n\n")
+          file.write(msec_to_timecode(cue.start_time))
+          file.write(" --> ")
+          file.write(msec_to_timecode(cue.end_time))
+          file.write("\n")
+          file.write(cue.text)
+        end
+      end
+    end
+
+    # Check whether its a VTT_HEADER or not
+    def validate_header(line)
+      !!line.strip.match(/^#{VTT_HEADER}/)
+    end
+
+    # Check whether its a meta-data or not
+    def is_meta_data?(text)
+      !!text.match(VTT_METADATA)
+    end
+
+    # Timecode format used in VTT file
+    def is_time?(text)
+      !!text.match(/^(\d{2}:)?\d{2}:\d{2}.\d{3}.*(\d{2}:)?\d{2}:\d{2}.\d{3}/)
+    end
+
+    # Check whether if its subtilte text or not
+    def is_text?(text)
+      !text.empty? and text.is_a?(String) and text != VTT_HEADER
+    end
+
+    def set_properties(cue, properties)
+      properties.each do |prop|
+        prop, value = prop.split(":")
+        value.gsub!("%","")
+        case prop
+        when "align"
+          cue.alignment = get_alignment(value)
+        when "line"
+          value = value.split(",")[0]
+          cue.position = get_line(value)
+        end
+      end
+    end
+
+    def get_alignment(value)
+      raise InvalidSubtitle, "Invalid VTT Alignment Property" unless ALIGNMENT_VALUES[value]
+      return ALIGNMENT_VALUES[value]
+    end
+
+    def get_line(value)
+      raise InvalidSubtitle, "VTT Line property should be a valid number" if !is_integer?(value) and value != AUTO_KEYWORD
+      return value.to_i
+    end
+
+    def get_position(value)
+      raise InvalidSubtitle, "VTT Position should be a valid number" if !is_integer?(value)
+      raise InvalidSubtitle, "VTT Position should be a number between 0 to 100" if (value.to_i < 0) or (value.to_i > 100)
+      return value.to_i
+    end
+
+    def is_integer?(val)
+      val.to_i.to_s == val
+    end
+  end
+end

data/lib/captions/list.rb

@@ -0,0 +1,53 @@
+module Captions
+  class CueList
+    include Enumerable
+
+    # Creates a new CueList
+    def initialize(frame_rate, list=[])
+      @fps = frame_rate
+      @list = list
+    end
+
+    # Returns frame-rate of the list
+    def fps
+      @fps
+    end
+
+    # Returns the parsed subtitles
+    def entries
+      @list
+    end
+
+    # Hide all cues when inspecting CueList
+    # Show only necessary info rather than printing everything
+    def inspect
+      "#<Captions::CueList:#{object_id} @fps=#{fps} @cues=#{@list.count}>"
+    end
+
+    # Changes the frame rate of CueList
+    # This also changes frame rate in already parsed
+    # subtitles
+    def frame_rate=(rate)
+      @list.each { |c| c.change_frame_rate(@fps, rate) }
+      @fps = rate
+    end
+
+    # Inserts the subtitle into the CueList
+    # Subtitle is serialized before its inserted
+    def append(cue)
+      cue.serialize(@fps)
+      @list << cue
+    end
+
+    # Iterate through CueList
+    def each
+      return to_enum(:each) unless block_given?
+      @list.each { |c| yield(c) }
+    end
+
+    # Array based enumerables for cuelist
+    def [](index)
+      @list[index]
+    end
+  end
+end

data/lib/captions/util.rb

@@ -0,0 +1,87 @@
+module Captions
+  module Util
+
+    # TC should be HH:MM:SS:FF (frames) or HH:MM:SS.MSC (milliseconds).
+    # FF(2 digits) and MSC(3 digits) are optional.
+    TIMECODE_REGEX = /^-?([01]\d|2[0-3]):[0-5]\d:[0-5]\d(:\d{2}|\.\d{3})?$/
+
+    # Currently considering frame rate as 25
+    # Converts time-code in HH:MM:SS.MSEC (or) HH:MM:SS:FF (or) MM:SS.MSEC
+    # to milliseconds.
+    def convert_to_msec(tc, ms_per_frame=40)
+      msec = 0
+      negative_multiplier = 1
+      if tc[0] == '-'
+        tc = tc[1..-1]  # remove -ve sign
+        negative_multiplier = -1
+      end
+      tc_split = tc.split('.')
+      time_split = tc_split[0].split(':')
+
+      # To handle MM:SS.MSEC format
+      if time_split.length == 2
+        time_split.unshift('00')
+      end
+
+      if tc_split[1]  # msec component exists
+        msec = tc_split[1].ljust(3, '0').to_i  # pad with trailing 0s to make it 3 digit
+      elsif time_split.length == 4  # FF (frame) component exists
+        msec = time_split[-1].to_i * ms_per_frame.to_f
+        time_split.pop  # so that below code can work from last index
+      end
+
+      min = 60
+      hour = min * 60
+      # Get HH:MM:SS in seconds
+      sec = time_split[-1].to_i
+      sec += time_split[-2].to_i * min
+      sec += time_split[-3].to_i * hour
+
+      msec += sec * 1000
+
+      return (negative_multiplier * msec.round) # to be consistent with tc_to_frames which also rounds
+    end
+
+    # Converts milliseconds calculated in one frame-rate to another frame-rate
+    def convert_frame_rate(msec, old_fps, new_fps)
+      old_ms_per_frame = (1000.0 / old_fps)
+      new_ms_per_frame = (1000.0 / new_fps)
+      frames = (msec / old_ms_per_frame).round  # Number of frames in old fps
+      sec = frames / old_fps
+      frames = frames % old_fps
+      new_frames = sec * new_fps
+      new_frames += frames # Number of frames in new fps
+      return (new_frames * new_ms_per_frame).round  # MSEC in new fps
+    end
+
+    # Converts milliseconds to timecode format
+    # Currently returns HH:MM:SS.MSEC
+    # Supports upto 60 hours
+    def msec_to_timecode(milliseconds)
+      seconds = milliseconds / 1000
+      msec = milliseconds % 1000
+      secs = seconds % 60
+
+      seconds = seconds / 60
+      mins = seconds % 60
+
+      seconds = seconds / 60
+      hours = seconds % 60
+
+      format("%02d:%02d:%02d.%03d",hours, mins, secs ,msec)
+    end
+
+    # Parses time-code and converts it to milliseconds.
+    # If time cannot be converted to milliseconds,
+    # it throws InvalidInput Error
+    def sanitize(time, frame_rate)
+      if time.is_a?(String)
+        if TIMECODE_REGEX.match(time)
+          time = convert_to_msec(time, frame_rate)
+        end
+      end
+      raise InvalidInput, 'Input should be in Milliseconds or Timecode' unless time.is_a? (Fixnum)
+      return time
+    end
+  end
+end

data/lib/captions/version.rb

@@ -0,0 +1,3 @@
+module Captions
+  VERSION = "1.3.2"
+end

data/lib/captions.rb

@@ -0,0 +1,10 @@
+require 'captions/version'
+require 'captions/error'
+require 'captions/util'
+require 'captions/base'
+require 'captions/list'
+require 'captions/cue'
+require 'captions/formats/srt'
+require 'captions/formats/vtt'
+
+require 'captions/export'