Primarily rely on date instead of id when determining message freshness

Workaround for vysheng/tg#1077, variation of PR #15

dumpers/lib/dumper_interface.rb

@@ -7,9 +7,9 @@
 # * Implement one or more of the functions listed below (at least dump_msg)
 
 # Note:
-#   Dumpers are a low-level construct and as of v2.0.0 dumpers they are no
-#   longer used for custom output formats. Instead, custom formatters have been
-#   introduced for this purpose.
+#   Dumpers are a low-level construct and as of v2.0.0 they are no longer used
+#   for implementing custom output formats. Instead, custom formatters have been
+#   introduced for this purpose (see /formatters/lib/formatter_base.rb).
 
 class DumperInterface
 
@@ -30,7 +30,16 @@ def start_dialog(dialog, progress)
   # This default makes sense in simple cases, override for advanced custom logic
   def msg_fresh?(msg, progress)
     # msg: Hash, progress: DumpProgress
-    !progress.last_id || msg['id'] > progress.last_id
+
+    return true unless progress.newest_date
+
+    if msg['date'] && msg['date'] > progress.newest_date
+      return true
+    elsif msg['date'] == progress.newest_date && progress.newest_id
+      return true if msg['id'] && msg['id'] > progress.newest_id
+    end
+
+    false
   end
 
   # Will be called for each message to dump (from newest to oldest)

lib/dump_progress.rb

@@ -2,25 +2,29 @@
 
 class DumpProgress
 
-  attr_reader :last_id
-  attr_reader :last_date
+  attr_reader :newest_id
+  attr_reader :newest_date
   attr_reader :dumper_state
   attr_writer :dumper_state
 
-  def initialize(last_id = nil, last_date = nil, dumper_state = {})
-    @last_id = last_id
-    @last_date = last_date
+  def initialize(newest_id = nil, newest_date = nil, dumper_state = {})
+    @newest_id = newest_id
+    @newest_date = newest_date
     @dumper_state = dumper_state
   end
 
   def self.from_hash(hash)
-    self.new(hash['last_id'], hash['last_date'], hash['dumper_state'])
+    self.new(
+      hash['newest_id'],
+      hash['newest_date'] || hash['last_date'], # last_date is v2.0.x compat
+      hash['dumper_state']
+    )
   end
 
   def to_hash
     {
-      :last_id => @last_id,
-      :last_date => @last_date,
+      :newest_id => @newest_id,
+      :newest_date => @newest_date,
       :dumper_state => @dumper_state
     }
   end
@@ -29,12 +33,11 @@ def to_json(*a)
     to_hash.to_json(*a)
   end
 
-  def bump_id(id)
-    @last_id = id if !@last_id || id > @last_id
-  end
-
-  def bump_date(date)
-    @last_date = date if !@last_date || date > @last_date
+  def update(msg)
+    if !@newest_date || (msg['date'] && msg['date'] >= @newest_date)
+      @newest_date = msg['date'] || @newest_date
+      @newest_id = msg['id'] || @newest_id
+    end
   end
 
 end

telegram-history-dump.rb

@@ -66,23 +66,22 @@ def dump_dialog(dialog)
     raise 'Expected array' unless msg_chunk.is_a?(Array)
     msg_chunk.reverse_each do |msg|
       dump_msg = true
-      if msg['id']
-        cur_progress.bump_id(msg['id'])
-      else
+      unless msg['id']
         $log.warn('Dropping message without id: %s' % msg)
         dump_msg = false
       end
-      if msg['date']
-        cur_progress.bump_date(msg['date'])
-      else
-        $log.warn('Message without date: %s' % msg)
+      unless msg['date']
+        $log.warn('Dropping message without date: %s' % msg)
+        dump_msg = false
       end
 
+      cur_progress.update(msg)
+
       if msg['text'] && filter_regex && filter_regex =~ msg['text']
         dump_msg = false
       end
 
-      unless $dumper.msg_fresh?(msg, old_progress)
+      unless msg['date'] && $dumper.msg_fresh?(msg, old_progress)
         if keep_dumping
           $log.info('Reached end of new messages since last backup')
         end