Chat Archive and Captioning Upload

NotQuiteApex commented 3 years ago

Closes #32

This pull request will add the ability to download chat logs from Twitch VODs alongside the VOD download itself, as well as the ability to reformat and export said logs into useful closed captioning formats like SAMI and RealText.

NotQuiteApex commented 3 years ago

PR TODO:

[x] Chat downloading
- [x] Initial query
- [x] Pagination
[x] Chat file archiving
- [x] Parsing downloaded JSON
- [x] Reformatting dates as second offsets from downloaded VOD starting form 0
- [x] Writing each line of the chat log file as something like offset=0;color=#FFFFFF;user="...";msg="...".
  - note: possibly format the message with proper escape codes? or find certain escape codes in ascii to use as separators. look into allowed characters in IRC standard? SOLUTION: null character separators, no quotes anywhere.
  - also note: this is not the file that gets uploaded to YouTube. This acts a base for the exported closed captioning file to work from.
- [x] Preamble of all the unique chatters with unique identifiers so certain info isnt repeated as often in the file (like display name and chat color)
[x] Chat exporting
- [x] Slicing chat from specific offsets of seconds from 0
- [x] Combining lists of chat messages of video slices with updated offsets
- [x] Taking archived chat logs and transcribing them into different formats
  - [x] SAMI
  - [x] RealText
[x] Chat uploading
- [x] New Google API Scopes
- [x] Additional fields for stage files to include exported captions such as a flag and path
- [x] service pulling captions list and inserting new media file upload for the captions file (with appropriate flags)
- [x] Rename chat caption track appropriately

NotQuiteApex commented 3 years ago

SAMI Format Document: https://www.w3.org/WAI/sami.html Notes: uses milliseconds for timestamps, doesn't seem to have a stop parameter for SYNC (because of this, may need multilines to stack?). very html/xml-like but has oddities Related stackoverflow question: https://stackoverflow.com/questions/55201241/need-help-to-format-youtube-sami-smi-caption Example:

<SAMI>
    <BODY>
        <SYNC Start=200><P Class=ENUSCC><font color="#ff0000"><b>NotQuiteApex</b></font><font color="#ffffff">: faek PogChamp :y</font>
        <SYNC Start=10000><P Class=ENUSCC><font color="#f0f000"><b>NotQuiteApex</b></font><font color="#ffffff">: faek PogChamp 2!!!!!!!</font>
    </BODY>
</SAMI>

RealText Format Document: http://sites.mpc.com.br/ric/realtext/RealText_tutor.html Notes: uses seconds in quotes (???), text can overlap itself (useful maybe?) Example:

<window type="generic"
 wordwrap="true" />

<font face="arial">

<!-- overlapping chat -->
<time begin="0" end="5"/><font color="#ff0000"><b>NotQuiteApex</b></font><font color="#ffffff">: faek PogChamp :y</font>
<time begin="2" end="10"/><font color="#f0f000"><b>NotQuiteApex</b></font><font color="#ffffff">: faek PogChamp 2!!!!!!!</font>

</font>
</window>

Definitely gonna need some formatting rules and config flags.

NotQuiteApex commented 3 years ago

Also, YouTube let's subtitle tracks be named. VodBot should absolutely name the chat upload appropriately automatically.

EDIT: setting the values of snippet.videoId (the video ID at the end of a youtube url), language ("en-US", should be changeable in config, see here for stuff about bcp-47 names), and name (call it "Twitch Chat" ?) will perform this action.

NotQuiteApex commented 3 years ago

I would like to add support for TTML as well before calling this finished, since that can align text properly.

NotQuiteApex commented 3 years ago

A basic version of TTML is working but colors are broken and it is not adjusting the text to the left as it is supposed to, so I'm leaving it behind for now.

FriendTeamInc / VodBot

Chat Archive and Captioning Upload #33