[facebook] Photo/Set extractor only captures per-image caption, not the post's own text

[FMicro-soft]

Description

When downloading photos from a Facebook post via the set/photo extractor (URLs like facebook.com/media/set/?set=... or facebook.com/photo/?fbid=...&set=...), the caption` metadata field only captures the per-image caption that can optionally be added to an individual photo inside a multi-photo post — it does not capture the post's own main text (the message written when the post itself was created), which is a separate piece of content on Facebook.

This means:

• If the author wrote a per-image caption for a photo, caption is populated.
• If the author did not write any per-image captions (common — this is an optional, secondary feature), caption is empty for every photo in that post, even though the post itself has a full text message attached.

Example (public posts, no login required to view)

Post A — has a long post message, and some (not all) photos in it have an additional per-image caption:
https://www.facebook.com/MCDucBao/posts/pfbid02hj9e4Coo1pBjGpeHoaFeAemT1wiG4VzHsuwQQXGp4FHiAXVE7mXqSeWce2UVLKoNl
→ caption is correctly extracted for the photos that have a per-image caption, but empty for the others in the same post — even though the post text itself exists and applies to all of them.

Post B — has a full post message, but no photo in it has a per-image caption:
https://www.facebook.com/MCDucBao/posts/pfbid034bCSxzumBpjEw4BUMq44uqx1zS3P74VnztdJv3Jopaa3MxU1EgHDLsT8qxTmKExcl
→ caption is empty for every photo, even though the post's own text is visible and non-empty when viewing the post directly.

Expected behavior

It would be very useful to also expose the post's own message text as a separate metadata field (e.g. post_text or similar), independent of the per-image caption, so users can save/use the actual post content regardless of whether individual photos have their own captions.

Environment

• gallery-dl version: 1.32.8 (Windows executable, up to date via -U)
• Extractor: facebook (set/photo)

[FMicro-soft]

Additional finding: mention/tag info is discarded during caption extraction

While looking into this, I found that the raw photo page HTML actually does contain data about mentions/tags inside a photo caption (e.g. when someone is tagged as a clickable link inside the caption text), but the current extraction logic discards it. Specifically, in FacebookExtractor.parse_photo_page():

"caption": self.decode_all(text.extr(
    photo_page,
    '"message":{"delight_ranges"',
    '"},"message_preferred_body"'
).rsplit('],"text":"', 1)[-1]),

This grabs the whole message block — which starts with delight_ranges (an array describing formatting/entity ranges, presumably including tagged users/pages) — but then only keeps the final plain "text" value via .rsplit('],"text":"', 1)[-1], throwing away the delight_ranges array
itself.

As a result, a caption like:

2 ca sĩ chính của chương trình: MC Đức Bảo và BTV Nguyen Diep Chi hứa hẹn...

(where "Nguyen Diep Chi" is a clickable tag/link to that person's profile) comes out as fully plain text, with no way to tell which part of the string was a tag.
It would be great if the extracted caption could optionally preserve this, e.g. by parsing delight_ranges and wrapping tagged substrings with a marker like @Name, so users can distinguish tagged mentions from plain text in the saved caption.

[mikf]

f1bf5477fe [facebook] extract post_text metadata (cb#309 gh#9679)

---

This grabs the whole message block — which starts with delight_ranges (an array describing formatting/entity ranges, presumably including tagged users/pages) — but then only keeps the final plain "text" value via .rsplit('],"text":"', 1)[-1], throwing away the delight_ranges array
itself.

I really don't want to write code to parse yet another completely optional metadata field, especially since extracting data from FB is already a PITA and this would make it even worse.