Can't set retweets to download to a separate folder

[ExeArco]

According to the following issues on github this feature should be available.
##1421
##1334

I am currently on version 1.17.2.

config.set(("extractor", "twitter"), "directory", ["twitter","{author[name]}","archive"])
config.set(("extractor", "twitter"), "retweets", ("directory", ["Twitter", "{user_likes}", "Likes"]))

I have tried a variety of things to set for "retweets" but nothing changes unless I set it to False.
Then it doesn't download retweets, as should be expected.
When I feed gallery-dl a url that is a users twitter page I want to download the users tweets in a base folder then their retweets in a separate folder within their folder.
For Example:

targetUser
targetUser/retweets

That is what I want.

[ExeArco]

I "fixed" this by downloading to a directory further in with retweets enabled, with a script that follows the downloaded images out if they match the target username.
Not an elegant solution, but I'm writing this here for anyone wondering the same thing.

[Scripter17]

I thought it might've been possible to do this with "{'.' if retweet_id==0 else 'Retweets'}" but the function that handles this doesn't actually use fstrings, probably to avoid arbitrary code execution problems

Sorry to bug you, @mikf, but you got any ideas?

[rautamiekka]

I thought it might've been possible to do this with "{'.' if retweet_id==0 else 'Retweets'}" but [https://github.com/mikf/gallery-dl/blob/d09bc5bd3462b75a784c8406c549e1c1858f9852/gallery_dl/util.py#L599) doesn't actually use fstrings, probably to avoid arbitrary code execution problems

Sorry to bug you, @mikf, but you got any ideas?

I suspect it's more cuz f-strings are executed in runtime (can't be stored, then executed later) while str.format() can use templates saved into a string like the {author[name]}.

Plus it's faster and way more powerful than the old printf-style percentage format still used by youtube-dl for compatibility, something that couldn't do something like {author[name]}. While the f-strings are even faster and more powerful, which really should be used when templates aren't needed.

[mikf]

"{retweet_id:?//L0/Retweets/}" seems to work.
It produces an empty string when retweet_id is 0 and Retweets otherwise.

f-strings are Python 3.6+ only and I kind of want to keep Python 3.4 compatibility for gallery-dl v1.x.

[Scripter17]

"{retweet_id:?//L0/Retweets/}" seems to work.

Care to explain how on earth that works and how you managed to come up with it? I am very confused

This feels like it should be documented somewhere

[mikf]

?// returns an empty string when its input (retweet_id) evaluates to False (e.g. is 0) and otherwise returns its input as a string. It also stops any further processing.

L0/Retweets/ returns its input if its len() is <= 0 and Retweets otherwise.

So if retweet_id == 0 -> empty string through ?//, otherwise Retweets through L0/…/

This feels like it should be documented somewhere

I know, and it kind of is here.
The whole string formatting system needs to be redone with a proper parser and all that at some point, and I've been delaying writing any docs until that is done ...

[Hrxn]

config.set(("extractor", "twitter"), "retweets", ("directory", ["Twitter", "{user_likes}", "Likes"]))

I think this cannot possibly work, by the way.
Because you can only set the "retweets" option for Twitter to a boolean value (or to the "original" special value).

When I feed gallery-dl a url that is a users twitter page I want to download the users tweets in a base folder then their retweets in a separate folder within their folder.
For Example:

targetUser
targetUser/retweets

That is what I want.

The good news: This should be possible now.

Look here: https://github.com/mikf/gallery-dl/blob/master/docs/configuration.rst#extractordirectory

The "directory" option can - analogous to the "filename" option - be set to an object containing Python expression mappings.
So it should be possible to do something like this (in the Twitter section of your config file):

"directory": {
    "retweet_id != 0"    : ["Twitter", "{user[name]}", "Retweets"],
    ""                   : ["Twitter", "{user[name]}"]
}

@ExeArco Could you please try this?

[ExeArco]

config.set(
	("extractor", "twitter"), 
	"directory", [
			{
				"retweet_id != 0" : ["{category}", folder,"archive", "retweets"],
				""	: ["{category}", folder,"archive"]
			}
		]
	)

This is what I attempted to use, however it gives me the following error:
[twitter][error] DirectoryFormatError: Applying directory format string failed (TypeError: expected str, got dict)

And just to confirm, I am on 1.18.0.

[Scripter17]

It was implemented one commit after 1.18.0
You'll want to overwrite C:/Python39/Lib/site-packages/gallery_dl/util.py with this version of util.py

[ExeArco]

Alright I updated to 1.18.1-dev and I get this error instead
[twitter][error] DirectoryFormatError: Applying directory format string failed (TypeError: unhashable type: 'dict')

[mikf]

Your directory value is a list with a dict as element. It should just be a dict:

config.set(
    ("extractor", "twitter"), 
    "directory", {
        "retweet_id != 0" : ["{category}", folder, "archive", "retweets"],
        "" : ["{category}", folder, "archive"],
    },
)

[ExeArco]

Alright well I fixed that but now it doesn't seem to be separating them at all, it seems that it doesn't ever trigger the retweet != 0 condition.
Here are all the config.sets related to twitter before I run that downloadJob.

config.set(("extractor", "twitter"), "filename", "{user[name]}_{tweet_id}_{date}_{num}.{extension}")
config.set(("extractor", "twitter"), "quoted", True)
config.set(("extractor", "twitter"), "text-tweets", True)
config.set(("extractor", "twitter"), "retweets", "original") 

config.set(
	("extractor","twitter"),
	'postprocessors', [
            {
                "name": "metadata",
                "event": "post",
                "filename": "{user[name]}_{tweet_id}_{date}_{num}.data.json"
            }
        ]
    )

config.set(
	("extractor", "twitter"), 
	"directory", {
				"retweet_id != 0" : ["{category}", folder,"archive", "retweets"],
				""	: ["{category}", folder,"archive"]
	},
)

[mikf]

It is not really possible to differentiate between Tweets and Retweets when setting retweets to "original". It replaces each Retweet entry with its original Tweet, which has a retweeted_status_id_str/retweet_id value of 0.
Since you are using Python, you could change the condition before each new Twitter URL to check if the author['name'] field matches the expected username and determine if it's a Retweet like that. Or gallery-dl could update the retweet_id value in such cases, which shouldn't break anything else.

[ExeArco]

I'm using this to archive the user profile themselves, not passing through individual retweet/tweet/status links, so I can't check each URL itself.
It would be a much more elegant solution if I could just do it all in one go.

[Hrxn]

What's the specific reason for not using extractor.twitter.retweets = true?

[ExeArco]

If I set retweets = true it sets the username of the retweet to the target of my download, not the original poster of what is being retweeted.
EX:
I want to archive posts done by A.
If I set retweets to true, my sub folder with all the retweets will be full of content from B,C,D,E,F,G however they will all have the file name set as if A actually tweeted them.

[Scripter17]

You can use retweets: true and replace the {user[name]} in filename with {author[name]}

[Hrxn]

Yes, that is exactly what I am using here.

[ExeArco]

If I do as Scripter17 has suggested it works however it now doesn't sort out quoted retweets., and looking through the metadata json that I have, it appears there is not anything I can use to sort that out, unless I am missing anything.

[mikf]

quoted retweets

Do you mean regular quote tweets or quotes of a retweet?
I don't think I've ever seen the latter, but maybe that exists as well.

it appears there is not anything I can use to sort that out

Quoted tweets have a non-zero quote_id, so it'd be something like

"directory": {
    "retweet_id": ["{category}", folder, "archive", "retweets"],
    "quote_id"  : ["{category}", folder, "archive", "quotes"],
    ""	        : ["{category}", folder, "archive"],
}

---

I've also updated the behavior of "retweets"; "original" to have a non-zero value for retweet_id (414bdc9)

[ExeArco]

I'll be honest I'm not sure exactly what this is called either, but take this example.
https://twitter.com/BarackObama
This is the link I send to download, later down Obama quotes a NetflixFilm account(https://twitter.com/BarackObama/status/1408818108224131074), and I would like that put in retweets.
Looking through the metadata, it appears quote_id, reply_id, and retweet_id are all 0 so I am not sure how to separate this one out.

[ExeArco]

I am still looking for a fix to this, does anyone have one?
I have just updated to the latest version and there still seems to be no way to properly separate those tweets out.

[Hrxn]

[..] it appears quote_id, reply_id, and retweet_id are all 0 so I am not sure how to separate this one out.

Not sure, but if this is what gets returned by Twitter..

PS D:\> gallery-dl --ignore-config -K 'https://twitter.com/BarackObama/status/1408818108224131074' --option '"quoted"=true' | sls "tweet|quote|reply" -NoEmphasis -Context 0,1

> quote_count
    439
> quote_id
    0
> reply_count
    268
> reply_id
    0
> retweet_count
    792
> retweet_id
    0
>   tweet
> tweet_id
    1408510014847959053
> quote_count
    439
> quote_id
    0
> reply_count
    268
> reply_id
    0
> retweet_count
    792
> retweet_id
    0
>   tweet
> tweet_id
    1408510014847959053

PS D:\>

This tweet seems a bit strange.. If I don't use -o '"quoted"=true' gallery-dl skips this tweet with default settings ("skipping quoted tweet")..

So, something is a bit off here, I guess. I blame Twitter.

[ExeArco]

I mean it is a quoted tweet right?
So then why is quote_id 0?
I really would like to separate these out through gallery-dl

[mikf]

As it turns out, quote_id is the ID of the quoted Tweet and is only nonzero for Tweets that contain a quote, not for the Tweet being quoted. It is the exact opposite of how retweets behave and I have no idea why I implemented it like that back when.

Instead of checking for quote_id, you could compare user and author and classify it as a quote when both are different.

"directory": {
    "retweet_id"    : ["{category}", folder, "archive", "retweets"],
    "user != author": ["{category}", folder, "archive", "quotes"],
    ""	            : ["{category}", folder, "archive"],
}

[afterdelight]

use "locals().get('quote_by')"