`Sequence.contains()` : Errors with sequences that contain unrelated numbers

[robin-vagner]

Hello,

I think I came across a parsing bug while using pyseq==0.9.0 in one of our tools.
Here is a minimal script to reproduce it, tested using multiple versions of python (from 3.9 to 3.11):

import pyseq


if __name__ == "__main__":
    images = [
        "s001_0030_1.jpg",
        "s001_0030_2.jpg",
        "s001_0090_1.jpg",
        "s001_0090_2.jpg",
    ]
    sequences = pyseq.get_sequences(images)

    # This prints "Sequences found:  [<pyseq.Sequence "s001_0030_1-2.jpg">, <pyseq.Sequence "s001_0090_1-2.jpg">]"
    # which is the expected result.
    print("Sequences found: ", sequences)

    # This prints the following :
    # s001_0030_2.jpg is in s001_0090_1-90.jpg  <- This one is NOT expected
    # s001_0090_1.jpg is in s001_0090_1-2.jpg
    # s001_0090_2.jpg is in s001_0090_1-2.jpg
    for image in images:
        if sequences[1].contains(image):
            print(f"{image} is in {sequences[1]}")

As stated in the script's comments, here is what I have noticed to be "wrong":

• s001_0030_2.jpg should not be included in s001_0090_1-2.jpg
• after calling contains() on the sequence, its __str__() method returns s001_0090_1-90.jpg instead of s001_0090_1-2.jpg

This is not a blocking issue on our part, I will tell our users to place their files in different folders as a workaround.
I'm at your disposal if you need more details.

Thanks in advance :)

[rsgalloway]

hey @robin-vagner ! thanks for the report, this is an interesting one. I was able to reproduce this issue on my end as well.

I think there is some confusion about what contains() and includes() is meant to do. as implemented, they both just call is_sibling which checks if a file/image should be part of the sequence or not (not whether it is in the sequence already). there may be a bug here where if a sequence has >=2 items in it already, then it's not validating membership correctly on items after that. I'll investigate that.

in the meantime, I was able to tweak your test slightly to give expected results:

    for image in images:
        if image in sequences[1]:
            print(f"{image} is in {sequences[1]}")
        else:
            print(f"{image} is NOT in {sequences[1]}")

which outputs:

Sequences found:  [<pyseq.Sequence "s001_0030_1-2.jpg">, <pyseq.Sequence "s001_0090_1-2.jpg">]
s001_0030_1.jpg is NOT in s001_0090_1-2.jpg
s001_0030_2.jpg is NOT in s001_0090_1-2.jpg
s001_0090_1.jpg is in s001_0090_1-2.jpg
s001_0090_2.jpg is in s001_0090_1-2.jpg

does that help?

[robin-vagner]

Hi!

I tried out using in and it works even in my more complex use-case.

Thanks for you help :)

[robin-vagner]

Just in case, I want to point out that there is still the formatting bug that occurs only when calling contains(). Here is an example :

import pyseq


if __name__ == "__main__":
    images = [
        "s001_0030_1.jpg",
        "s001_0030_2.jpg",
        "s001_0090_1.jpg",
        "s001_0090_2.jpg",
    ]
    sequences = pyseq.get_sequences(images)

    sequences = [
        sequence
        for sequence in sequences
        if any(image in sequence for image in images)
    ]
    single_images = [
        image
        for image in images
        # Using `image in sequence` instead of `sequence.contains(image)` here makes the formatting work as expected
        if not any(sequence.contains(image) for sequence in sequences)
    ]

    # This prints the following :
    # Sequence:  s001_0030_1-30.jpg   <- The formatting bug is here
    # Images:  [<pyseq.Item "s001_0030_1.jpg">, <pyseq.Item "s001_0030_2.jpg">]
    # Sequence:  s001_0090_1-2.jpg
    # Images:  [<pyseq.Item "s001_0090_1.jpg">, <pyseq.Item "s001_0090_2.jpg">]
    for sequence in sequences:
        print("Sequence: ", sequence)
        print("Images: ", [s for s in sequence])

This is no longer a problem for me since I only use in now, but I just wanted to make sure you knew.

Thanks again!

[rsgalloway]

hi @robin-vagner ! glad it worked out and this is less urgent now. I'll keep this issue open to address the parsing error. thanks again for using pyseq and for the report!