how to format a filenaming if/else option within a postprocessor?

[Jeddostotle7]

After figuring out that the problem I was having with gallery-dl downloading text replies in spite of my image-filter was due to the "content" postprocessor ignoring the twitter extractor options for some reason,, I decided to nix the predefined "content" postprocessor I was using and just make a nearly-identical bespoke one within the twitter extractor so that I can fit in "filter": "quote_id or retweet_id or (user is author)" to bring it in line with the main extractor's image-filter.

Currently, it looks like this:

"postprocessors": [
    {
        "name": "metadata",

        "event": "post",
        "filter": "quote_id or retweet_id or (user is author)",
        "filename": {
            "locals().get('retweet_id')": "{retweet_id}.txt",
            ""                          : "{tweet_id}.txt"
        },
        "mode" : "custom",
        "format": "{content}\n"
    }
]

However, I'm having a problem: the postprocessor is erroring due to my filenaming if/else. For context, it's the same if/else filenaming option I have on the twitter exctractor itself (with .txt instead of .{extension}, of course), and I added it because I realized that this whole time gallery-dl's default filenaming command for twitter using tweet_id meant that for retweets it would use the retweet's id rather than the actual "real" id of the tweet iself, and that was a problem for me, now having a bunch of mis-named files.

The if/else works fine in the twitter extractor itself obviously, but with the version in the postprocessor, when I try to download a twitter url the postprocessor errors out and gives me the error postprocessor: 'metadata' initialization failed: TypeError: unhashable type: 'dict'. I'm not sure what change to make to the if condition to get it to work, if I can't call on the metadata tags like that.

[allendema]

metadata.filename expects a string not a dict.
https://github.com/mikf/gallery-dl/blob/master/docs/configuration.rst#metadatafilename

Create two PPs:

1: Filter retweet_id in locals() and use a string as filename.
2: FIlter retweet_id not in locals()and use a string as filename.

[Jeddostotle7]

Oh right, that would make sense (plus I should've looked back harder at the other postprocessors under my other extractors in my config cuz I've used similar "separate postprocessors that do the same thing with slight changes with a different filter for each" previously), but in trying to implement it, I've run into a new problem.

First, I configured it like so:

"postprocessors": [
    {
        "name": "metadata",

        "event": "post",
        "filter": ["quote_id or retweet_id or (user is author)", "retweet_id in locals()"],
        "filename": "{retweet_id}.txt",
        "mode" : "custom",
        "format": "{content}\n"
    },
    {
        "name": "metadata",

        "event": "post",
        "filter": ["quote_id or retweet_id or (user is author)", "retweet_id not in locals()"],
        "filename": "{tweet_id}.txt",
        "mode" : "custom",
        "format": "{content}\n"
    }
],

But then I got an error (which will be described further down), and so thinking maybe perhaps a postprocessor's filter can't handle multiple strings the way an image-filter can, and they were somewhat redundant anyway, I simplified to this:

"postprocessors": [
    {
        "name": "metadata",

        "event": "post",
        "filter": ["retweet_id in locals()"],
        "filename": "{retweet_id}.txt",
        "mode" : "custom",
        "format": "{content}\n"
    },
    {
        "name": "metadata",

        "event": "post",
        "filter": ["quote_id or (user is author)"],
        "filename": "{tweet_id}.txt",
        "mode" : "custom",
        "format": "{content}\n"
    }
],

But then, like the unsimplified version before, I got the same error: postprocessor: 'metadata' initialization failed: TypeError: compile() arg 1 must be a string, bytes or AST object, twice, before downloading the main media itself.

In case it helps, when I ran it --verbose, the full error it gave me (on both unsimplified and simplified versions) was:

postprocessor: 'metadata' initialization failed:  TypeError: compile() arg 1 must be a string, bytes or AST object
postprocessor:
Traceback (most recent call last):
  File "C:\Python311\Lib\site-packages\gallery_dl\job.py", line 521, in initialize
    pp_obj = pp_cls(self, pp_dict)
             ^^^^^^^^^^^^^^^^^^^^^
  File "C:\Python311\Lib\site-packages\gallery_dl\postprocessor\metadata.py", line 83, in __init__
    job.register_hooks({event: self.run for event in events}, options)
  File "C:\Python311\Lib\site-packages\gallery_dl\job.py", line 539, in register_hooks
    condition = util.compile_expression(expr)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Python311\Lib\site-packages\gallery_dl\util.py", line 624, in compile_expression
    code_object = compile(expr, name, "eval")
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: compile() arg 1 must be a string, bytes or AST object

(twice, again)

I'm not sure exactly what part is breaking it, since a modified form of the version I put in my original post (with the filename simply set to "{tweet_id}.txt") was working fine when testing it to make sure the filter worked while waiting for help on filename discriminating, and the simplified versions aren't really all that different from it, especially the second of the two which is nearly identical.

EDIT: aaaaah, never mind, finally realized the error was definitely basically telling me to not use a list of strings in the filter, and that even my simplified versions still didn't work because despite being single strings they were still in brackets. It works fine now. Thanks!