Adds metadata (captions and timestamps) to your exported Snapchat memories.
Snapchat allows you to save images/videos to their servers with their memories feature. When you export your memories from Snapchat from their accounts website, you get all the memories, but with no metadata and the captions are stored separately!
| Base Image | Overlay Image |
|---|---|
 |
 |
You may then go to an alternate project like ToTheMax's
Snapchat-All-Memories-Downloader,
which is fantastic! It uses a memories_history.json file to download all memories and add a timestamp, but this technique loses all the captions! They aren't included in memories_history.json :(
{
"Date": "2020-01-02 23:08:19 UTC",
"Media Type": "Image",
"Location": "Latitude, Longitude: 0.0, 0.0",
"Download Link": "https://app.snapchat.com/dmd/memories?..."
}So in short, memories_history.json has all the metadata we need, but we need
to combine it with the full memory export if we want captions. So this project
serves as a bridge between these two methods! We have all the metadata and we
have all the memory photos/videos and captions, we just need to combine them. The final result:
Note the caption is on the image, the file name and metadata has a timestamp! The file's creation date is also set to the correct date.
Note
This project and documentation is not very friendly to non-developers :( If you have questions, I'm happy to try and help if you make a new Github issue. I'm very open to any PRs that want to work on making this more user-friendly (see the Contribution section below!)
This project relies on two major libraries.
-
ffmpeg is "a complete, cross-platform solution to record, convert and stream audio and video." This project uses ffmpeg to overlay the captions onto videos. If you're on Linux, download ffmpeg from your package manager, if you're on MacOS, install
ffmpegfrom brew, and if you're on Windows, install it from winget withwinget install -e --id Gyan.FFmpeg(package here).- Note: Ffmpeg must be version 7.1 or higher. FFmpeg 7 is not supported, because the scale2ref filter is broken on that version but the scale filter didn't yet support relative heights. Ffmpeg 6 is not actively supported, but here's the last commit that supported it if that works for your environment. If you're stuck on Ffmpeg 7 (Fedora 41 users, this is for you!), consider using distrobox to spin up an arch linux container and run this script there. Sorry for the headache, hopefully Fedora 42 will bump to 7.1!
-
libvips is "a demand-driven, horizontally threaded image processing library. Compared to similar libraries, libvips runs quickly and uses little memory." This project uses it to overlay the captions onto images. If you're on Linux, download it from your package manager, if you're on macOS install it from brew, and if you're on Windows download a compiled version here and add the bin folder to your path.
Finally, this project requires Python to be installed on your computer. Once installed, you can clone this repo and run
python -m pip install -r requirements.txtto install the Python requirements for the project.
Note
This project and documentation is not very friendly to non-developers :( If you have questions, I'm happy to try and help if you make a new Github issue. I'm very open to any PRs that want to work on making this more user-friendly (see the Contribution section below!)
First, we need to get our data. Follow Snapchat's instructions to download your data. You'll need to do two separate exports:
-
Check "Include your Memories, Chat Media and Shared Stories" to download all your memories. (Note: this may take up a significant amount of space!) The status of the other checkmarks shouldn't matter, so feel free to export any additional data you'd like. Make sure the date range includes everything you want to export! If you choose to download multiple data packages, merge all the
memoriesfolders into onememoriesfolder. This is the only folder you need from the export for this script. -
Create a second export, without "Include your Memories, Chat Media and Shared Stories" checked, and with "Export JSON files" and "Memories and Other Media" checked. Use the same date range as your first export. This export should take much less time for Snapchat to send to you :) In this data export, make sure you have a
json/memories_history.json. This is the metadata for the memories, which is only generated when you don't export your memories.
Now, make a folder in this repo called input and put the memories folder
from the first export and the memories_history.json file from the second
export into it.
- Alternatively, you can provide the location of the
memories_history.jsonfile with the--memories-historyflag and the memories folder with thememoriesflag, for examplepython main.py --memories-history test/memories_history.json --memories-folder test/memories --output test_output.
Some other flags:
--image-onlywill only convert images and not videos!--video-onlywill only convert videos and not images!
With this folder prepared, you can now run python main.py to run the script!
It will create a new folder called output that will hold all memories with
timestamps and captions. (Note: this will create a new copy of each photo/video,
so make sure you have enough space!) Alternatively, use the --output flag to
provide a different location to dump the photos.
If you run into issues, try running the tests to see if they produce the
expected output. You can do this by running just test if you have
[just] installed, or by running python -m unittest tests/test.py.
The tests will have output files in tests/output, and there should be four.
There should be an image with no caption, an image with a caption, a video
with no caption, and a video with a caption. If there are issues with the
images, check your VIPS installation, and if there are issues with the videos,
check your ffmpeg installation.
If there are still issues, please run the script with the -v or --verbose
flag, which will output information on what the script is doing. If that
doesn't help you with the issue, open a github issue with that output! You
can also use the -vv flag to get logs from ffmpeg/vips.
Warning
If your captions are important to you, double check the final result against the photos in the Snapchat app! I found that the memories export did not include overlays/captions from some of my photos in 2017-2018. So check the final photos to make sure captions are there. If they aren't, you may need to download those impacted photos/videos directly from the Snapchat app. The majority of photos were successfully exported with all their captions.
Note: The timezone applied to the photos is set by your computer's local
timezone! I'm open to any PRs to improve this (see below), but for a quick
hack you can set the default timezone in SnapchatMemoriesCaptionAdder/adder.py
in the add_metadata signature.
- @n-katti for fixing issues on Windows and new Snapchat export format (issue #3)
- @Enricon27 for fixing a problem with the earlier fix :) (issue #3)
- @autumnesponda for adding location metadata
- @harrydayexe for adding FFMPEG 7.1 support (issue #7 and #14)
This achieved what I needed it to do, so I don't anticipate adding many more features. I'd be happy to accept PRs for some feaures I didn't implement:
- Making this much more user friendly. This is currently not accessible to
anyone who doesn't have knowledge of installing and using libraries.
- Bundling libraries with the app
- Creating a GUI
- Cleaning the project up :) Introducing python-ffmpeg's async definitely made this functional but messy.
- Snapchat uses UTC as the timezone for the timestamps. This script takes a guess that your computer's local time zone is the timezone you want the timestamps in. Allowing the user to specify a timezone, or automatically determine the timezone from the photo's location data, would be an improvement.
- I believe ffmpeg could also do photos, but it's less efficient than VIPS. Adding a fallback to ffmpeg would make installation easier for those who can't easily install VIPS.
MIT