--only-print-filenames actually downloads some live photos instead of printing their names

[Jackabomb]

Overview

The file size de-duplication interacts with --only-print-filenames and live photos in a surprising way.
If a live photo needs to be deduplicated in this manner, then the video portion of the LP will be downloaded from the server, even though --only-print-filenames was given.

Steps to Reproduce

Using version 1.30.0 on Debian 13 (Trixie) via the downloadable standalone executable.

1. Put two live photos into the same album (Live, in my example). They must have different sizes but the same name and date such that they will need deduplication when downloaded. I don't know exactly how this comes about. For me, the live photo is named lp_image which is unusual in itself.
2. mkdir -p albums/Live
3. Run icloudpd -d albums/Live -a Live to download both of the live photos (2 JPGs and 2 MOVs)
4. Delete the Live Photo (both its JPG and MOV) that has a size in the name.
5. Run icloudpd --only-print-filenames -a Live -d albums/Live

Expected Behavior

The missing live photo's JPG and MOV filepaths should be printed to standard output, with their deduplicated size-appended names.
Neither file should be downloaded to Live/albums.

Actual Behavior

The JPG file name is printed as expected, but the MOV file name is not.
The MOV file is actually downloaded to the target directory. --only-print-filenames suppresses the logging output, so it will not be obvious this is happening just from watching the command output.

Context

Downloading and then selectively deleting files probably seems a weird thing to do. It's just there to simulate the case where deduplication is needed during the --only-print-filenames pass. Examples might be a partial previous download, or two downloads during the same day that the photos were taken (or received via text), so on.

I ran across this while using --only-print-filenames to generate lists of files to pass to a script. The script itself isn't relevant to this bug, but for the curious, I'm using it to set up my albums as a "tree of symlinks" as discussed in #125 (comment)

As weird as what I'm doing might be, I think it's fairly reasonable that an option called --only-print-filenames would not in fact download any files.

[Jackabomb]

My understanding could be way off here. I'm new to this project's code. But I think the problem comes from this line:

icloud_photos_downloader/src/icloudpd/base.py

Lines 1134 to 1136 in 8290e6e

	if only_print_filenames and not lp_file_exists:
	print(lp_download_path)
	else:

only_print_filenames is only evaluated once, before the deduplication renaming logic kicks in. Then, once the new name is generated, the download commences without reference to only_print_filenames:

icloud_photos_downloader/src/icloudpd/base.py

Lines 1149 to 1152 in 8290e6e

	if not lp_file_exists:
	truncated_path = truncate_middle(lp_download_path, 96)
	logger.debug("Downloading %s...", truncated_path)
	download_result = download.download_media(

Compare and contrast this logic with the logic for the still photo itself, earlier, on lines 1068-1111 of the same file. Here the renaming/deduplicating happens before the check of only_print_filename and before any actual actions get kicked off.

[AndreyNikiforov]

your expectation that --only-print-filename should not download anything is correct.

[AndreyNikiforov]

fixed in v1.32.2