fix: extract transcript from comment voice replies
This commit is contained in:
ThetaDev
parent
1d1ae17ffc
commit
30f60c30f9
4 changed files with 3825 additions and 1 deletions
|
|
@ -624,6 +624,7 @@ pub(crate) struct CommentViewModelWrap {
|
|||
pub(crate) struct CommentViewModel {
|
||||
pub comment_id: String,
|
||||
pub comment_key: String,
|
||||
pub comment_surface_key: String,
|
||||
pub toolbar_state_key: String,
|
||||
}
|
||||
|
||||
|
|
@ -695,6 +696,7 @@ pub(crate) struct AuthorCommentBadgeRenderer {
|
|||
#[serde(rename_all = "camelCase")]
|
||||
pub(crate) enum Payload {
|
||||
CommentEntityPayload(CommentEntityPayload),
|
||||
CommentSurfaceEntityPayload(CommentSurfaceEntityPayload),
|
||||
#[serde(rename_all = "camelCase")]
|
||||
EngagementToolbarStateEntityPayload {
|
||||
heart_state: HeartState,
|
||||
|
|
@ -716,6 +718,13 @@ pub(crate) struct CommentEntityPayload {
|
|||
pub avatar: ImageView,
|
||||
}
|
||||
|
||||
#[serde_as]
|
||||
#[derive(Debug, Deserialize)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub(crate) struct CommentSurfaceEntityPayload {
|
||||
pub voice_reply_container_view_model: Option<VoiceReplyContainer>,
|
||||
}
|
||||
|
||||
#[serde_as]
|
||||
#[derive(Debug, Deserialize)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
|
|
@ -772,3 +781,17 @@ pub(crate) struct ContinuationButton {
|
|||
pub(crate) struct ContinuationButtonRenderer {
|
||||
pub command: ContinuationEndpoint,
|
||||
}
|
||||
|
||||
#[derive(Debug, Deserialize)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub(crate) struct VoiceReplyContainer {
|
||||
pub voice_reply_container_view_model: VoiceReplyContainer2,
|
||||
}
|
||||
|
||||
#[serde_as]
|
||||
#[derive(Debug, Deserialize)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub(crate) struct VoiceReplyContainer2 {
|
||||
#[serde_as(as = "AttributedText")]
|
||||
pub transcript_text: TextComponents,
|
||||
}
|
||||
|
|
|
|||
|
|
@ -0,0 +1,356 @@
|
|||
---
|
||||
source: src/client/video_details.rs
|
||||
expression: map_res.c
|
||||
---
|
||||
Paginator(
|
||||
count: None,
|
||||
items: [
|
||||
Comment(
|
||||
id: "Ugw4wdQ25m4lN301Rol4AaABAg.AC1r0Q9vExEAC1rR-OMlS-",
|
||||
text: RichText([
|
||||
Text(
|
||||
text: "Transcribed voice reply:",
|
||||
style: Style(
|
||||
bold: true,
|
||||
),
|
||||
),
|
||||
Text(
|
||||
text: " \"oh you\'re right I made a mistake whoops\"",
|
||||
),
|
||||
]),
|
||||
author: Some(ChannelTag(
|
||||
id: "UCQSpnDG3YsFNf5-qHocF-WQ",
|
||||
name: "@ThioJoe",
|
||||
avatar: [
|
||||
Thumbnail(
|
||||
url: "https://yt3.ggpht.com/ytc/AIdro_nlxlwCvy3lBTSpqObahldOpRO3LpeOZFmFCJ7vW2i2ZNQ=s88-c-k-c0x00ffffff-no-rj",
|
||||
width: 88,
|
||||
height: 88,
|
||||
),
|
||||
],
|
||||
verification: verified,
|
||||
subscriber_count: None,
|
||||
)),
|
||||
publish_date: "[date]",
|
||||
publish_date_txt: "3 days ago",
|
||||
like_count: Some(2100),
|
||||
reply_count: 0,
|
||||
replies: Paginator(
|
||||
count: Some(0),
|
||||
items: [],
|
||||
ctoken: None,
|
||||
endpoint: browse,
|
||||
),
|
||||
by_owner: true,
|
||||
pinned: false,
|
||||
hearted: false,
|
||||
),
|
||||
Comment(
|
||||
id: "Ugw4wdQ25m4lN301Rol4AaABAg.AC1r0Q9vExEAC1rXfHrZv_",
|
||||
text: RichText([
|
||||
Text(
|
||||
text: "that was a K like the old windows which represents Kilobytes not Megabytes",
|
||||
),
|
||||
]),
|
||||
author: Some(ChannelTag(
|
||||
id: "UCWSUHd0FSYNF6cjkMCC8Y9w",
|
||||
name: "@Pls2M",
|
||||
avatar: [
|
||||
Thumbnail(
|
||||
url: "https://yt3.ggpht.com/A493Oz1QwDx8Fa93HfAieUdYe-HFKlvzQimUpCySjCbOAPcgHQQgQ3PM-MNjxdmdk9ogpOTZ_Y0=s88-c-k-c0x00ffffff-no-rj",
|
||||
width: 88,
|
||||
height: 88,
|
||||
),
|
||||
],
|
||||
verification: none,
|
||||
subscriber_count: None,
|
||||
)),
|
||||
publish_date: "[date]",
|
||||
publish_date_txt: "3 days ago",
|
||||
like_count: Some(41),
|
||||
reply_count: 0,
|
||||
replies: Paginator(
|
||||
count: Some(0),
|
||||
items: [],
|
||||
ctoken: None,
|
||||
endpoint: browse,
|
||||
),
|
||||
by_owner: false,
|
||||
pinned: false,
|
||||
hearted: false,
|
||||
),
|
||||
Comment(
|
||||
id: "Ugw4wdQ25m4lN301Rol4AaABAg.AC1r0Q9vExEAC1rZObfQ9y",
|
||||
text: RichText([
|
||||
Text(
|
||||
text: "so yes",
|
||||
),
|
||||
]),
|
||||
author: Some(ChannelTag(
|
||||
id: "UCWSUHd0FSYNF6cjkMCC8Y9w",
|
||||
name: "@Pls2M",
|
||||
avatar: [
|
||||
Thumbnail(
|
||||
url: "https://yt3.ggpht.com/A493Oz1QwDx8Fa93HfAieUdYe-HFKlvzQimUpCySjCbOAPcgHQQgQ3PM-MNjxdmdk9ogpOTZ_Y0=s88-c-k-c0x00ffffff-no-rj",
|
||||
width: 88,
|
||||
height: 88,
|
||||
),
|
||||
],
|
||||
verification: none,
|
||||
subscriber_count: None,
|
||||
)),
|
||||
publish_date: "[date]",
|
||||
publish_date_txt: "3 days ago",
|
||||
like_count: Some(18),
|
||||
reply_count: 0,
|
||||
replies: Paginator(
|
||||
count: Some(0),
|
||||
items: [],
|
||||
ctoken: None,
|
||||
endpoint: browse,
|
||||
),
|
||||
by_owner: false,
|
||||
pinned: false,
|
||||
hearted: false,
|
||||
),
|
||||
Comment(
|
||||
id: "Ugw4wdQ25m4lN301Rol4AaABAg.AC1r0Q9vExEAC1rnkWZoH7",
|
||||
text: RichText([
|
||||
YouTube(
|
||||
text: "@ThioJoe",
|
||||
target: Channel(
|
||||
id: "UCQSpnDG3YsFNf5-qHocF-WQ",
|
||||
),
|
||||
),
|
||||
Text(
|
||||
text: " wow voice note reply. since when was that a thing",
|
||||
),
|
||||
]),
|
||||
author: Some(ChannelTag(
|
||||
id: "UCSNxfsFLRzIZ-99aIGZ_dIg",
|
||||
name: "@N1r4",
|
||||
avatar: [
|
||||
Thumbnail(
|
||||
url: "https://yt3.ggpht.com/gpEcInEquxyAqzFxVmv6m83DSI3OHXXbQr1HgTZaySg-sbpdj6nOp-W-NYdmq8jDa2-Pf8nk134=s88-c-k-c0x00ffffff-no-rj",
|
||||
width: 88,
|
||||
height: 88,
|
||||
),
|
||||
],
|
||||
verification: none,
|
||||
subscriber_count: None,
|
||||
)),
|
||||
publish_date: "[date]",
|
||||
publish_date_txt: "3 days ago",
|
||||
like_count: Some(569),
|
||||
reply_count: 0,
|
||||
replies: Paginator(
|
||||
count: Some(0),
|
||||
items: [],
|
||||
ctoken: None,
|
||||
endpoint: browse,
|
||||
),
|
||||
by_owner: false,
|
||||
pinned: false,
|
||||
hearted: false,
|
||||
),
|
||||
Comment(
|
||||
id: "Ugw4wdQ25m4lN301Rol4AaABAg.AC1r0Q9vExEAC1sTNcHMYN",
|
||||
text: RichText([
|
||||
YouTube(
|
||||
text: "@N1r4",
|
||||
target: Channel(
|
||||
id: "UCSNxfsFLRzIZ-99aIGZ_dIg",
|
||||
),
|
||||
),
|
||||
Text(
|
||||
text: " is it supported for you? for me it has a transcription and says voice reply is not supported",
|
||||
),
|
||||
]),
|
||||
author: Some(ChannelTag(
|
||||
id: "UCCCvp1-zqEQetYuZduIdd7A",
|
||||
name: "@xapplezapple3453",
|
||||
avatar: [
|
||||
Thumbnail(
|
||||
url: "https://yt3.ggpht.com/ytc/AIdro_lfpjOSUn38QauvP05T14pFJ2vGX_wDkuh0E9M6_qOO=s88-c-k-c0x00ffffff-no-rj",
|
||||
width: 88,
|
||||
height: 88,
|
||||
),
|
||||
],
|
||||
verification: none,
|
||||
subscriber_count: None,
|
||||
)),
|
||||
publish_date: "[date]",
|
||||
publish_date_txt: "3 days ago",
|
||||
like_count: Some(141),
|
||||
reply_count: 0,
|
||||
replies: Paginator(
|
||||
count: Some(0),
|
||||
items: [],
|
||||
ctoken: None,
|
||||
endpoint: browse,
|
||||
),
|
||||
by_owner: false,
|
||||
pinned: false,
|
||||
hearted: false,
|
||||
),
|
||||
Comment(
|
||||
id: "Ugw4wdQ25m4lN301Rol4AaABAg.AC1r0Q9vExEAC1tu_pIsg4",
|
||||
text: RichText([
|
||||
Text(
|
||||
text: "\u{200b}",
|
||||
),
|
||||
YouTube(
|
||||
text: "@ThioJoe",
|
||||
target: Channel(
|
||||
id: "UCQSpnDG3YsFNf5-qHocF-WQ",
|
||||
),
|
||||
),
|
||||
Text(
|
||||
text: " since when can you reply with audio\nnever seen this",
|
||||
),
|
||||
]),
|
||||
author: Some(ChannelTag(
|
||||
id: "UCTR5xEDXSi92_GQPONLU_vw",
|
||||
name: "@Moshiur_Rahman",
|
||||
avatar: [
|
||||
Thumbnail(
|
||||
url: "https://yt3.ggpht.com/NLBULvx76jIqiPmRaJD0OWYXs644vyEahh6S4TsMRRSG_PsM3B0rYMsctP8jNMHM7iWTBRz2Sw=s88-c-k-c0x00ffffff-no-rj",
|
||||
width: 88,
|
||||
height: 88,
|
||||
),
|
||||
],
|
||||
verification: none,
|
||||
subscriber_count: None,
|
||||
)),
|
||||
publish_date: "[date]",
|
||||
publish_date_txt: "3 days ago (edited)",
|
||||
like_count: Some(106),
|
||||
reply_count: 0,
|
||||
replies: Paginator(
|
||||
count: Some(0),
|
||||
items: [],
|
||||
ctoken: None,
|
||||
endpoint: browse,
|
||||
),
|
||||
by_owner: false,
|
||||
pinned: false,
|
||||
hearted: false,
|
||||
),
|
||||
Comment(
|
||||
id: "Ugw4wdQ25m4lN301Rol4AaABAg.AC1r0Q9vExEAC1virGMTo7",
|
||||
text: RichText([
|
||||
YouTube(
|
||||
text: "@xapplezapple3453",
|
||||
target: Channel(
|
||||
id: "UCCCvp1-zqEQetYuZduIdd7A",
|
||||
),
|
||||
),
|
||||
Text(
|
||||
text: " I believe you have to view it in the YouTube iOS or Android app, not desktop",
|
||||
),
|
||||
]),
|
||||
author: Some(ChannelTag(
|
||||
id: "UCQSpnDG3YsFNf5-qHocF-WQ",
|
||||
name: "@ThioJoe",
|
||||
avatar: [
|
||||
Thumbnail(
|
||||
url: "https://yt3.ggpht.com/ytc/AIdro_nlxlwCvy3lBTSpqObahldOpRO3LpeOZFmFCJ7vW2i2ZNQ=s88-c-k-c0x00ffffff-no-rj",
|
||||
width: 88,
|
||||
height: 88,
|
||||
),
|
||||
],
|
||||
verification: verified,
|
||||
subscriber_count: None,
|
||||
)),
|
||||
publish_date: "[date]",
|
||||
publish_date_txt: "3 days ago (edited)",
|
||||
like_count: Some(236),
|
||||
reply_count: 0,
|
||||
replies: Paginator(
|
||||
count: Some(0),
|
||||
items: [],
|
||||
ctoken: None,
|
||||
endpoint: browse,
|
||||
),
|
||||
by_owner: true,
|
||||
pinned: false,
|
||||
hearted: false,
|
||||
),
|
||||
Comment(
|
||||
id: "Ugw4wdQ25m4lN301Rol4AaABAg.AC1r0Q9vExEAC27uYKqxUy",
|
||||
text: RichText([
|
||||
YouTube(
|
||||
text: "@ThioJoe",
|
||||
target: Channel(
|
||||
id: "UCQSpnDG3YsFNf5-qHocF-WQ",
|
||||
),
|
||||
),
|
||||
Text(
|
||||
text: " Insane new YouTube feature spotted",
|
||||
),
|
||||
]),
|
||||
author: Some(ChannelTag(
|
||||
id: "UC8TLX4SSKMpXl3yuZeQbLDQ",
|
||||
name: "@RippleXRPcorp-2024",
|
||||
avatar: [
|
||||
Thumbnail(
|
||||
url: "https://yt3.ggpht.com/2FTo9EkfZUvYfpxs-kzbjmKl8SG0pesUxVkUMQAc4yejxPt0qbCU2Bwway6gWR3syeXDmUN7dw=s88-c-k-c0x00ffffff-no-rj",
|
||||
width: 88,
|
||||
height: 88,
|
||||
),
|
||||
],
|
||||
verification: none,
|
||||
subscriber_count: None,
|
||||
)),
|
||||
publish_date: "[date]",
|
||||
publish_date_txt: "3 days ago",
|
||||
like_count: Some(136),
|
||||
reply_count: 0,
|
||||
replies: Paginator(
|
||||
count: Some(0),
|
||||
items: [],
|
||||
ctoken: None,
|
||||
endpoint: browse,
|
||||
),
|
||||
by_owner: false,
|
||||
pinned: false,
|
||||
hearted: false,
|
||||
),
|
||||
Comment(
|
||||
id: "Ugw4wdQ25m4lN301Rol4AaABAg.AC1r0Q9vExEAC28KAyUNVu",
|
||||
text: RichText([
|
||||
Text(
|
||||
text: "Voice Replies?? Seems a lil unnecessary imo but i dont mind",
|
||||
),
|
||||
]),
|
||||
author: Some(ChannelTag(
|
||||
id: "UC3ODAuviniQtvaq47mTZyVA",
|
||||
name: "@RealZerenaFan",
|
||||
avatar: [
|
||||
Thumbnail(
|
||||
url: "https://yt3.ggpht.com/Lzs-ycGJPuNL2Ag7oB-z0FGWup5yfv76Y2AnPxdjrDbREAmX44029LhobUkwTkZ5MDudvSVJgg=s88-c-k-c0x00ffffff-no-rj",
|
||||
width: 88,
|
||||
height: 88,
|
||||
),
|
||||
],
|
||||
verification: none,
|
||||
subscriber_count: None,
|
||||
)),
|
||||
publish_date: "[date]",
|
||||
publish_date_txt: "3 days ago",
|
||||
like_count: Some(84),
|
||||
reply_count: 0,
|
||||
replies: Paginator(
|
||||
count: Some(0),
|
||||
items: [],
|
||||
ctoken: None,
|
||||
endpoint: browse,
|
||||
),
|
||||
by_owner: false,
|
||||
pinned: false,
|
||||
hearted: false,
|
||||
),
|
||||
],
|
||||
ctoken: Some("Eg0SC04yWElmX1RVbU1ZGAYy0wEKTGdldF9jb21tZW50X3dpdGhfcmVwbGllc19zdHJlYW0tLUNnZ0lnQVFWRjdmUk9CSUZDS0FnR0FFaURRb0xDUGZYOTdvR0VQaWVfRDQaUBIaVWd3NHdkUTI1bTRsTjMwMVJvbDRBYUFCQWciAggAKhhVQ1FTcG5ERzNZc0ZOZjUtcUhvY0YtV1EyC04yWElmX1RVbU1ZQAFIMoIBAggBKAlCL2NvbW1lbnQtcmVwbGllcy1pdGVtLVVndzR3ZFEyNW00bE4zMDFSb2w0QWFBQkFn"),
|
||||
endpoint: browse,
|
||||
)
|
||||
|
|
@ -593,6 +593,14 @@ fn map_comment_vm(
|
|||
} else {
|
||||
false
|
||||
};
|
||||
let voice_reply = if let Some(Payload::CommentSurfaceEntityPayload(sf)) =
|
||||
mutations.remove(&vm.comment_surface_key)
|
||||
{
|
||||
sf.voice_reply_container_view_model
|
||||
.map(|vr| vr.voice_reply_container_view_model.transcript_text)
|
||||
} else {
|
||||
None
|
||||
};
|
||||
|
||||
let mut parse_num = |s: &str| -> Option<u32> {
|
||||
if s.is_empty() || s == " " {
|
||||
|
|
@ -606,7 +614,10 @@ fn map_comment_vm(
|
|||
|
||||
Some(Comment {
|
||||
id: vm.comment_id,
|
||||
text: ce.properties.content.into(),
|
||||
text: voice_reply
|
||||
.filter(|_| ce.properties.content.is_empty())
|
||||
.unwrap_or(ce.properties.content)
|
||||
.into(),
|
||||
by_owner: ce.author.as_ref().map(|a| a.is_creator).unwrap_or_default(),
|
||||
author: ce.author.map(|a| ChannelTag {
|
||||
id: a.channel_id,
|
||||
|
|
@ -697,6 +708,7 @@ mod tests {
|
|||
#[case::latest("latest")]
|
||||
#[case::frameworkupd("20240401_frameworkupd")]
|
||||
#[case::frameworkupd_reply("20240401_frameworkupd_reply")]
|
||||
#[case::voice_reply("20241218_voice_reply")]
|
||||
fn map_comments(#[case] name: &str) {
|
||||
let json_path = path!(*TESTFILES / "video_details" / format!("comments_{name}.json"));
|
||||
let json_file = File::open(json_path).unwrap();
|
||||
|
|
|
|||
3433
testfiles/video_details/comments_20241218_voice_reply.json
Normal file
3433
testfiles/video_details/comments_20241218_voice_reply.json
Normal file
File diff suppressed because it is too large
Load diff
Reference in a new issue