1 files changed, 32 insertions, 9 deletions

diff --git a/lib/dtas/source/av_ff_common.rb b/lib/dtas/source/av_ff_common.rb
index 5299fdb..c600c48 100644
--- a/lib/dtas/source/av_ff_common.rb
+++ b/lib/dtas/source/av_ff_common.rb
@@ -7,10 +7,10 @@ require_relative '../replaygain'
 require_relative '../xs'
 require_relative 'file'
 
-# Common code for libav (avconv/avprobe) and ffmpeg (and ffprobe)
-# TODO: newer versions of both *probes support JSON, which will be easier to
-# parse.  However, the packaged libav version in Debian 7.0 does not
-# support JSON, so we have an ugly parser...
+# Common code for ffmpeg/ffprobe and the abandoned libav (avconv/avprobe).
+# TODO: newer versions of both *probes support JSON, which will be easier
+# to parse.  libav is abandoned, nowadays, and Debian only packages
+# ffmpeg+ffprobe nowadays.
 module DTAS::Source::AvFfCommon # :nodoc:
   include DTAS::Source::File
   include DTAS::XS
@@ -21,10 +21,23 @@ module DTAS::Source::AvFfCommon # :nodoc:
   attr_reader :format
   attr_reader :duration
 
+  CACHE_KEYS = [ :@duration, :@probe_harder, :@comments, :@astreams,
+                 :@format ].freeze
+
+  def mcache_lookup(infile)
+    (@mcache ||= DTAS::Mcache.new).lookup(infile) do |input, dst|
+      tmp = source_file_dup(infile, nil, nil)
+      tmp.av_ff_ok? or return nil
+      CACHE_KEYS.each { |k| dst[k] = tmp.instance_variable_get(k) }
+      dst
+    end
+  end
+
   def try(infile, offset = nil, trim = nil)
-    rv = source_file_dup(infile, offset, trim)
-    rv.av_ff_ok? or return
-    rv
+    ent = mcache_lookup(infile) or return
+    ret = source_file_dup(infile, offset, trim)
+    CACHE_KEYS.each { |k| ret.instance_variable_set(k, ent[k]) }
+    ret
   end
 
   def __parse_astream(cmd, stream)
@@ -104,13 +117,14 @@ module DTAS::Source::AvFfCommon # :nodoc:
       prev_cmd = cmd
     end while incomplete.compact[0]
 
+    enc = Encoding.default_external # typically Encoding::UTF_8
     # old avprobe
     s.scan(%r{^\[FORMAT\]\n(.*?)\n\[/FORMAT\]\n}m) do |_|
       f = $1.dup
       f =~ /^duration=([\d\.]+)\s*$/nm and @duration = $1.to_f
       # TODO: multi-line/multi-value/repeated tags
       f.gsub!(/^TAG:([^=]+)=(.*)$/ni) { |_|
-        @comments[$1.upcase] = -($2)
+        @comments[-DTAS.try_enc($1.upcase, enc)] = $2
       }
     end
 
@@ -118,13 +132,22 @@ module DTAS::Source::AvFfCommon # :nodoc:
     s.scan(%r{^\[format\.tags\]\n(.*?)\n\n}m) do |_|
       f = $1.dup
       f.gsub!(/^([^=]+)=(.*)$/ni) { |_|
-        @comments[$1.upcase] = -$2
+        @comments[-DTAS.try_enc($1.upcase, enc)] = $2
       }
     end
     s.scan(%r{^\[format\]\n(.*?)\n\n}m) do |_|
       f = $1.dup
       f =~ /^duration=([\d\.]+)\s*$/nm and @duration = $1.to_f
     end
+    comments.each do |k,v|
+      v.chomp!
+      comments[k] = -DTAS.try_enc(v, enc)
+    end
+
+    # ffprobe always uses "track", favor FLAC convention "TRACKNUMBER":
+    if @comments['TRACK'] && !@comments['TRACKNUMBER']
+      @comments['TRACKNUMBER'] = @comments.delete('TRACK')
+    end
 
     ! @astreams.compact.empty?
   end