@ -7,7 +7,7 @@ require "../helpers/serialized_yt_data"
private ITEM_CONTAINER_EXTRACTOR = { private ITEM_CONTAINER_EXTRACTOR = {
Extractors :: YouTubeTabs ,
Extractors :: YouTubeTabs ,
Extractors :: SearchResults ,
Extractors :: SearchResults ,
Extractors :: Continuation ,
Extractors :: Continuation Content ,
} }
private ITEM_PARSERS = { private ITEM_PARSERS = {
@ -18,8 +18,11 @@ private ITEM_PARSERS = {
Parsers :: CategoryRendererParser ,
Parsers :: CategoryRendererParser ,
Parsers :: RichItemRendererParser ,
Parsers :: RichItemRendererParser ,
Parsers :: ReelItemRendererParser ,
Parsers :: ReelItemRendererParser ,
Parsers :: ContinuationItemRendererParser ,
} }
private alias InitialData = Hash ( String , JSON :: Any )
record AuthorFallback , name : String , id : String record AuthorFallback , name : String , id : String
# Namespace for logic relating to parsing InnerTube data into various datastructs. # Namespace for logic relating to parsing InnerTube data into various datastructs.
@ -345,14 +348,9 @@ private module Parsers
content_container = item_contents [ " contents " ]
content_container = item_contents [ " contents " ]
end
end
raw_contents = content_container [ " items " ]? . try & . as_a
content_container [ " items " ]? . try & . as_a . each do | item |
if ! raw_contents . nil?
result = parse_item ( item , author_fallback . name , author_fallback . id )
raw_contents . each do | item |
contents << result if result . is_a? ( SearchItem )
result = extract_item ( item )
if ! result . nil?
contents << result
end
end
end
end
Category . new ( {
Category . new ( {
@ -384,7 +382,9 @@ private module Parsers
end
end
private def self . parse ( item_contents , author_fallback )
private def self . parse ( item_contents , author_fallback )
return VideoRendererParser . process ( item_contents , author_fallback )
child = VideoRendererParser . process ( item_contents , author_fallback )
child || = ReelItemRendererParser . process ( item_contents , author_fallback )
return child
end
end
def self . parser_name
def self . parser_name
@ -408,9 +408,19 @@ private module Parsers
private def self . parse ( item_contents , author_fallback )
private def self . parse ( item_contents , author_fallback )
video_id = item_contents [ " videoId " ] . as_s
video_id = item_contents [ " videoId " ] . as_s
video_details_container = item_contents . dig (
reel_player_overlay = item_contents . dig (
" navigationEndpoint " , " reelWatchEndpoint " ,
" navigationEndpoint " , " reelWatchEndpoint " ,
" overlay " , " reelPlayerOverlayRenderer " ,
" overlay " , " reelPlayerOverlayRenderer "
)
# Sometimes, the "reelPlayerOverlayRenderer" object is missing the
# important part of the response. We use this exception to tell
# the calling function to fetch the content again.
if ! reel_player_overlay . as_h . has_key? ( " reelPlayerHeaderSupportedRenderers " )
raise RetryOnceException . new
end
video_details_container = reel_player_overlay . dig (
" reelPlayerHeaderSupportedRenderers " ,
" reelPlayerHeaderSupportedRenderers " ,
" reelPlayerHeaderRenderer "
" reelPlayerHeaderRenderer "
)
)
@ -436,9 +446,9 @@ private module Parsers
# View count
# View count
view_count_text = video_details_container . dig? ( " viewCountText " , " simpleText " )
# View count used to be in the reelWatchEndpoint, but that changed?
view_count_text ||= video_details_container
view_count_text = item_contents . dig? ( " viewCountText " , " simpleText " )
. dig? ( " viewCountText " , " accessibility" , " accessibilityData " , " label " )
view_count_text || = video_details_container . dig? ( " viewCountText " , " simpleText " )
view_count = view_count_text . try & . as_s . gsub ( / \ D+ / , " " ) . to_i64? || 0 _i64
view_count = view_count_text . try & . as_s . gsub ( / \ D+ / , " " ) . to_i64? || 0 _i64
@ -450,8 +460,8 @@ private module Parsers
regex_match = / - (?<min> \ d+ minutes? )?(?<sec> \ d+ seconds?)+ - / . match ( a11y_data )
regex_match = / - (?<min> \ d+ minutes? )?(?<sec> \ d+ seconds?)+ - / . match ( a11y_data )
minutes = regex_match . try & . [ " min " ] . to_i ( strict : false ) || 0
minutes = regex_match . try & . [ " min " ] ?. try & . to_i ( strict : false ) || 0
seconds = regex_match . try & . [ " sec " ] . to_i ( strict : false ) || 0
seconds = regex_match . try & . [ " sec " ] ?. try & . to_i ( strict : false ) || 0
duration = ( minutes * 60 + seconds )
duration = ( minutes * 60 + seconds )
@ -475,6 +485,35 @@ private module Parsers
return {{ @type . name }}
return {{ @type . name }}
end
end
end
end
# Parses an InnerTube continuationItemRenderer into a Continuation.
# Returns nil when the given object isn't a continuationItemRenderer.
#
# continuationItemRenderer contains various metadata ued to load more
# content (i.e when the user scrolls down). The interesting bit is the
# protobuf object known as the "continutation token". Previously, those
# were generated from sratch, but recent (as of 11/2022) Youtube changes
# are forcing us to extract them from replies.
#
module ContinuationItemRendererParser
def self . process ( item : JSON :: Any , author_fallback : AuthorFallback )
if item_contents = item [ " continuationItemRenderer " ]?
return self . parse ( item_contents )
end
end
private def self . parse ( item_contents )
token = item_contents
. dig? ( " continuationEndpoint " , " continuationCommand " , " token " )
. try & . as_s
return Continuation . new ( token ) if token
end
def self . parser_name
return {{ @type . name }}
end
end
end end
# The following are the extractors for extracting an array of items from # The following are the extractors for extracting an array of items from
@ -510,7 +549,7 @@ private module Extractors
# }]
# }]
#
#
module YouTubeTabs
module YouTubeTabs
def self . process ( initial_data : Hash ( String , JSON :: Any ) )
def self . process ( initial_data : InitialData )
if target = initial_data [ " twoColumnBrowseResultsRenderer " ]?
if target = initial_data [ " twoColumnBrowseResultsRenderer " ]?
self . extract ( target )
self . extract ( target )
end
end
@ -575,7 +614,7 @@ private module Extractors
# }
# }
#
#
module SearchResults
module SearchResults
def self . process ( initial_data : Hash ( String , JSON :: Any ) )
def self . process ( initial_data : InitialData )
if target = initial_data [ " twoColumnSearchResultsRenderer " ]?
if target = initial_data [ " twoColumnSearchResultsRenderer " ]?
self . extract ( target )
self . extract ( target )
end
end
@ -608,8 +647,8 @@ private module Extractors
# The way they are structured is too varied to be accurately written down here.
# The way they are structured is too varied to be accurately written down here.
# However, they all eventually lead to an array of parsable items after traversing
# However, they all eventually lead to an array of parsable items after traversing
# through the JSON structure.
# through the JSON structure.
module Continuation
module Continuation Content
def self . process ( initial_data : Hash ( String , JSON :: Any ) )
def self . process ( initial_data : InitialData )
if target = initial_data [ " continuationContents " ]?
if target = initial_data [ " continuationContents " ]?
self . extract ( target )
self . extract ( target )
elsif target = initial_data [ " appendContinuationItemsAction " ]?
elsif target = initial_data [ " appendContinuationItemsAction " ]?
@ -691,8 +730,7 @@ end
# Parses an item from Youtube's JSON response into a more usable structure. # Parses an item from Youtube's JSON response into a more usable structure.
# The end result can either be a SearchVideo, SearchPlaylist or SearchChannel. # The end result can either be a SearchVideo, SearchPlaylist or SearchChannel.
def extract_item ( item : JSON :: Any , author_fallback : String ? = " " , def parse_item ( item : JSON :: Any , author_fallback : String ? = " " , author_id_fallback : String ? = " " )
author_id_fallback : String ? = " " )
# We "allow" nil values but secretly use empty strings instead. This is to save us the
# We "allow" nil values but secretly use empty strings instead. This is to save us the
# hassle of modifying every author_fallback and author_id_fallback arg usage
# hassle of modifying every author_fallback and author_id_fallback arg usage
# which is more often than not nil.
# which is more often than not nil.
@ -702,24 +740,23 @@ def extract_item(item : JSON::Any, author_fallback : String? = "",
# Each parser automatically validates the data given to see if the data is
# Each parser automatically validates the data given to see if the data is
# applicable to itself. If not nil is returned and the next parser is attempted.
# applicable to itself. If not nil is returned and the next parser is attempted.
ITEM_PARSERS . each do | parser |
ITEM_PARSERS . each do | parser |
LOGGER . trace ( " extract _item: Attempting to parse item using \" #{ parser . parser_name } \" (cycling...) " )
LOGGER . trace ( " parse _item: Attempting to parse item using \" #{ parser . parser_name } \" (cycling...) " )
if result = parser . process ( item , author_fallback )
if result = parser . process ( item , author_fallback )
LOGGER . debug ( " extract_item: Successfully parsed via #{ parser . parser_name } " )
LOGGER . debug ( " parse_item: Successfully parsed via #{ parser . parser_name } " )
return result
return result
else
else
LOGGER . trace ( " extract _item: Parser \" #{ parser . parser_name } \" does not apply. Cycling to the next one... " )
LOGGER . trace ( " parse _item: Parser \" #{ parser . parser_name } \" does not apply. Cycling to the next one... " )
end
end
end
end
end end
# Parses multiple items from YouTube's initial JSON response into a more usable structure. # Parses multiple items from YouTube's initial JSON response into a more usable structure.
# The end result is an array of SearchItem. # The end result is an array of SearchItem.
def extract_items ( initial_data : Hash ( String , JSON :: Any ) , author_fallback : String ? = nil , #
author_id_fallback : String ? = nil ) : Array ( SearchItem ) # This function yields the container so that items can be parsed separately.
items = [ ] of SearchItem #
def extract_items ( initial_data : InitialData , & block )
if unpackaged_data = initial_data [ " contents " ]? . try & . as_h
if unpackaged_data = initial_data [ " contents " ]? . try & . as_h
elsif unpackaged_data = initial_data [ " response " ]? . try & . as_h
elsif unpackaged_data = initial_data [ " response " ]? . try & . as_h
elsif unpackaged_data = initial_data . dig? ( " onResponseReceivedActions " , 0 ) . try & . as_h
elsif unpackaged_data = initial_data . dig? ( " onResponseReceivedActions " , 0 ) . try & . as_h
@ -727,24 +764,37 @@ def extract_items(initial_data : Hash(String, JSON::Any), author_fallback : Stri
unpackaged_data = initial_data
unpackaged_data = initial_data
end
end
# This is identical to the parser cycling of extract _item().
# This is identical to the parser cycling of parse _item().
ITEM_CONTAINER_EXTRACTOR . each do | extractor |
ITEM_CONTAINER_EXTRACTOR . each do | extractor |
LOGGER . trace ( " extract_items: Attempting to extract item container using \" #{ extractor . extractor_name } \" (cycling...) " )
LOGGER . trace ( " extract_items: Attempting to extract item container using \" #{ extractor . extractor_name } \" (cycling...) " )
if container = extractor . process ( unpackaged_data )
if container = extractor . process ( unpackaged_data )
LOGGER . debug ( " extract_items: Successfully unpacked container with \" #{ extractor . extractor_name } \" " )
LOGGER . debug ( " extract_items: Successfully unpacked container with \" #{ extractor . extractor_name } \" " )
# Extract items in container
# Extract items in container
container . each do | item |
container . each { | item | yield item }
if parsed_result = extract_item ( item , author_fallback , author_id_fallback )
else
items << parsed_result
LOGGER . trace ( " extract_items: Extractor \" #{ extractor . extractor_name } \" does not apply. Cycling to the next one... " )
end
end
end
end end
break
# Wrapper using the block function above
else
def extract_items (
LOGGER . trace ( " extract_items: Extractor \" #{ extractor . extractor_name } \" does not apply. Cycling to the next one... " )
initial_data : InitialData ,
author_fallback : String ? = nil ,
author_id_fallback : String ? = nil
) : { Array ( SearchItem ) , String ?}
items = [ ] of SearchItem
continuation = nil
extract_items ( initial_data ) do | item |
parsed = parse_item ( item , author_fallback , author_id_fallback )
case parsed
when . is_a? ( Continuation ) then continuation = parsed . token
when . is_a? ( SearchItem ) then items << parsed
end
end
end
end
return items
return items , continuation
end end
Samantaz Fox
2 years ago