EmbeddedMetadata

Plugins Scrapers
, ,
1
:placard: Summary A scraper that pulls metadata from the files itself.
:link: Repository https://github.com/stashapp/CommunityScrapers/tree/master/scrapers/EmbeddedMetadata
:information_source: Source URL https://stashapp.github.io/CommunityScrapers/stable/index.yml
:open_book: Install How to install a scraper?

A scraper that pulls metadata from the files itself.

Right now, the following sources are supported:

  • JPEG EXIF data
  • JPEG IPTC data
  • JPEG XMP data

Installation

This scraper requires certain python dependencies, it should automatically install them.

Internally, the prefered source is pyexiv2. If, however, this module fails to load (for example, it requires a failry recent GLIBC that can be a problem on very long support cycle Linux Distributions), it will fall back to a wrapper around exiftool. But this will require exiftool to be available in the PATH, so you might want to install it manually in these scenarios.

Configuration

This scraper supports an optional config file config.py, to enable some maybe unwanted features.

If your Stash installation uses Authentication, make sure the API key is configured in the config.ini in py_common:

Some other things that can be enabled

Setting Description
skip_ensure_requirements = False enable skip dependency install / check, makes scraper much faster, but should not be enabled untile scrapper has successfully run once
details_date_fields = False enable to add all seen date field values to details (since the scraper has to potentially pick from one of several conflicting fields and might be wrong)
details_title_fields = False enable to add all senn title field values to details (since the scraper has to potentially pick from one of several conflicting fields and might be wrong)
details_author_fields = False enable to add all seen studio / photographer field values to details (since the scraper has to potentially pick from one of several conflicting fields and might be wrong)
details_camera_infos = False enable to add camera make, model and lens make and model as one line to details
details_upprocessed_fields = False enable to add fields to explicitly process and not ignored to details
details_upprocessed_fields_ignored = [] additional fields to not add to details (default list is already quite extensive and should ignore most boring stuff like camera settings)
details_upprocessed_fields_unignored = [] fields to add to details that are on the default ignore list (default list is already quite extensive and should ignore most boring stuff like camera settings)

Metadata fields

exiv2 has a very comprehensive documentation on EXIF, IPTC and XMP tags, which can be found under Exiv2 - Image metadata library and tools

2

I was having an issue with the scraper but I think I fixed it:

I was able to import title and details, but not tags (which are the only three fields I’ve tried importing so far). Turned out the issue was the script trying to reference value_parts under the else: statement (line 228), but the default value was only set under the “if.”

I just repeated the declaration under else:.

		else:
			value_parts = []
			value_parts.extend(value.split(','))

It may not be the most elegant, and Notepad++ gave me trouble by using spaces instead of tabs or indentation, but I don’t actual know anything more than RegEx so it’s the most I can do lol.

Also, for anyone curious who got intimated by the Exiv2 documentation, you can use a dumber of different tags that will all by imported to the same field. The mapping I’m using is
ImageTitle => Title
DateTimeOriginal => Date
(The format is “YYYY:mm:dd HH:MM:SS” and it requires all of those elements or Exiftool won’t update the tag)
ImageDescription => Details
XPKeywords => Tags
(These just need to be a comma seperated list)

The script indicates that XPAuthor and Artist can be used to set “Studio” but I haven’t actually tried it yet. I’m not sure if the script distinguishes between “Studio” and “Photographer.”

1 Like
4

Could you maybe create an issue at the CommunityScrapers repo? I’ll see if I can look into it over the weekend. This sucker is kanda hard to test though due to the wildly different code paths depending on available tooling and the source image.

Maybe you could even create a draft pull request with your changes? THat’d be helpful too.

(Sorry about the deleted post, I’m an idiot)

5

Done and done. Thanks for making this!