Repository navigation
Releases
Releases · Kyrstears/TagExtract
TagExtract v1.2.0
Added
- Rating filter: Flag or exclude posts by rating (safe / sensitive / questionable / explicit), with cross-site normalization. Danbooru's letter ratings are disambiguated (
g/smap to safe and sensitive on Danbooru;smaps to safe everywhere else). Flag mode shows an inline warning on post pages; exclude mode drops tags from disallowed posts and listings. Gelbooru listings do not expose ratings, so posts on those listings are never excluded. - NoobAI prefix preset:
masterpiece, best quality, newest, absurdres, highres, safe, verified against the official Laxhar Lab model cards. Also available as a per-site override. - Tag category order: User-defined output order of the five tag categories (character, copyright, artist, meta, general), configurable on the options page.
- .txt download: Save the current tag output as a
.txtfile via the selection panel's toolbar. File names are derived from the site and post ID. - Replacement rules: Up to 100 literal find → replace rules applied to tags before copying (for example,
1girl→1girl, solo), with a total size cap. - Per-tag selection: Collapsible checkbox panel on post pages for copying a subset of tags, with select all / none, copy selected, and .txt buttons.
- Listing multiselect for Gelbooru and Moebooru: The merged copy now works on all supported listing pages, not just Danbooru and e621. Moebooru tags and ratings are read from the preview image title; Gelbooru tags are read from
span.thumbpreviews. - Tag count cap warning: Listing pages warn when the merged tag set exceeds 75 tags, since many models degrade beyond that point.
Changed
- The options page gains sections for the new settings: a tag order picker (positions swap automatically to stay a valid permutation), a replacement rule editor with a live count, and a rating filter with mode and allowed-rating checkboxes.
- The content script is rebuilt around
sanitizeSettings, which fully validates the new setting shapes, including prototype-pollution guards. Family-aware rating normalization now lives inshared.js.
v1.1.2
Extracted shared constants and helpers (DEFAULTS, ALLOWED_HOSTS, LIMITS, prefix presets, validation sets, clampWeight) into a new shared.js module loaded by both the content script and the options page. The duplicated copies had already drifted (LIMITS.pattern was missing in options.js); this is now a single source of truth. No user-facing behavior change.
v1.1.1
[1.1.1] - 2026-09-08
Added
- MIT
LICENSEfile, so the license badge and topics are backed by an actual license grant.
Changed
- README badges now use dynamic shields: release version follows GitHub releases, license reads the repo license. Added dependencies (none) and supported sites (16) badges.
v1.1.0
[1.1.0] - 2026-09-08
Added
- Support for five additional image boards:
aibooru.online,xbooru.com,hypnohub.net,tbib.org, andlolibooru.moe, bringing the extension to 32 URL patterns. - The new sites require additional host permissions and also appear in the per-site overrides dropdown on the options page.
TagExtract 1.0.0
TagExtract copies tags, metadata, and copyright from supported image boards in one click, ready for your prompt. This is the initial release.
Supported Sites
- Danbooru:
danbooru.donmai.us - Gelbooru:
gelbooru.com,safebooru.org,rule34.xxx,rule34.us,rule34hentai.net - e621:
e621.net,e926.org - Moebooru:
yande.re,konachan.com,konachan.net
Features
- One-click Copy Tags and Copy All buttons on post pages, with a floating panel fallback that does not cover the image.
- 5 output formats: space, comma, SD prompt, JSON, and custom templates with
{tags},{copyright},{artist},{character},{meta}placeholders. - SD weights: per-category wrapping as
(tag:1.10), clamped to 0.1-2.0, with parentheses escaped as\( \)for A1111 and ComfyUI. - Prefix presets: SD 1.5, Pony, Illustrious, and Anima, plus custom presets.
- Anima mode: lowercase tags with spaces, optional
@artistprefixing. - Blacklist: glob (
*) patterns, up to 200 entries, with the excluded count reported on copy. - Tag normalization: convert between spaces and underscores, optionally strip
character (series)qualifiers. - Multi-post merge: select thumbnails on listing pages and copy a deduplicated tag set.
- Live tag count badge on the Copy Tags button.
- Keyboard shortcuts:
Ctrl+Shift+1Copy Tags,Ctrl+Shift+2Copy All (remappable atchrome://extensions/shortcuts). - Auto-copy on post load, once per URL, with SPA navigation support.
- History: last 10 copies in the popup, with re-copy and clear.
- Per-site overrides for format, prefix preset, and tag style.
- Optional
Source: <url>line appended to the output. - Theme-aware UI for light and dark modes, without flash on load.
Installation
- Download
TagExtract-v1.0.0.zipbelow and unzip it. - Go to
chrome://extensionsand enable Developer mode. - Click Load unpacked and select the unzipped folder.
Permissions
storage (settings and history), activeTab, and clipboardWrite. No network, login, or download permissions.