Repository navigation
feat: Rework outfit recommendation slot rules to treat full body items better - #264
Merged
Merged
Conversation
…ith garment vocabulary, ensuring consistency and avoiding manual updates. Added tests for vocabulary-based role types and prompt rendering.
Contributor
|
Thanks for your first pull request to Wardrowbe! Please make sure you've read the CONTRIBUTING guide, especially the PR title format section. A maintainer will review this soon. |
Reviewer's GuideThe PR updates recommendation and pairing prompts to treat separates and one-piece outfits as equal silhouettes, while rendering every garment list from the validated ITEM_ROLE taxonomy to prevent prompt drift. It adds focused coverage for role derivation and prompt rendering; database-backed recommendation tests were not run because they require Postgres. Sequence diagram for taxonomy-driven outfit prompt loadingsequenceDiagram
participant Service as Recommendation_or_Pairing_Service
participant Loader as load_prompt
participant Renderer as render_outfit_prompt
participant Vocabulary as ITEM_ROLE_Vocabulary
Service->>Loader: load_prompt(template)
Loader-->>Service: Prompt_template
Service->>Renderer: render_outfit_prompt(Prompt_template)
loop Each role in ROLES
Renderer->>Vocabulary: types_for_role(role)
Vocabulary-->>Renderer: Role_garment_types
Renderer->>Renderer: Replace_ROLE_TYPES_token
end
Renderer-->>Service: Rendered_outfit_prompt
Flow diagram for equal-weight outfit silhouettesflowchart LR
Prompt[Outfit recommendation or pairing prompt] --> Silhouettes{Equal-weight silhouettes}
Silhouettes --> Separates[Separates: top and bottom roles]
Silhouettes --> OnePiece[One-piece: full-body garment fills top and bottom roles]
Separates --> Legwear[Optional legwear]
OnePiece --> Legwear
File-Level Changes
Assessment against linked issues
Possibly linked issues
Tips and commandsInteracting with Sourcery
Customizing Your ExperienceAccess your dashboard to:
Getting Help
|
geekygirlsarah
marked this pull request as ready for review
October 13, 2026 06:59
![sourcery-ai[bot]](https://avatars.githubusercontent.com/in/48477?s=60&v=4){kind=small}
There was a problem hiding this comment.
Hey - I've found 2 issues
Prompt for AI Agents
Please address the comments from this code review:
## Individual Comments
### Comment 1
<location path="backend/app/services/pairing_service.py" line_range="21" />
<code_context>
logger = logging.getLogger(__name__)
-PAIRING_PROMPT_TEMPLATE = load_prompt("item_pairing")
+PAIRING_PROMPT_TEMPLATE = render_outfit_prompt(load_prompt("item_pairing"))
# A pairing is an internally-generated row, or an externally-authored one carrying the
</code_context>
<issue_to_address>
**Pairings omit their source item**
When the source item is a top or bottom and the model selects a full-body garment, `generate_pairings` adds the source to `valid_ids`, but `deduplicate_by_body_slot` processes the multi-slot garment first and removes the source, so the service can persist and return a pairing without its required source item.
Ensure deduplication preserves the source item when resolving body-slot conflicts.
</issue_to_address>
### Comment 2
<location path="backend/tests/test_garment_vocabulary.py" line_range="79-87" />
<code_context>
assert prompt_types == set(DEFAULT_WASH_INTERVALS)
+def test_types_for_role_is_derived_from_the_vocabulary():
+ assert types_for_role("base_top") == [
+ "shirt", "t-shirt", "top", "sweater", "polo", "blouse", "tank-top",
+ ] # fmt: skip
+ assert types_for_role("full_body") == ["dress", "jumpsuit"]
+ assert set(types_for_role("bottom")) == {"pants", "jeans", "shorts", "skirt"}
+ assert set(types_for_role("footwear")) == {"shoes", "sneakers", "boots", "sandals"}
+ assert types_for_role("legwear") == ["tights"]
+ assert all(ITEM_ROLE[garment] == "base_top" for garment in types_for_role("base_top"))
+
+
</code_context>
<issue_to_address>
**New garment roles go unlisted**
When a garment is added to `ITEM_ROLE` for one of the tested roles while `types_for_role` keeps returning its old hard-coded list, `test_types_for_role_is_derived_from_the_vocabulary` checks fixed expected lists at lines 80–86, so an implementation that returns those hard-coded lists still passes after a new garment is added to a role. The assertion at line 87 only checks the listed base-top garments’ roles, not that every taxonomy entry is returned.
Compare each role’s result against the corresponding entries derived from `ITEM_ROLE`, so the test fails when a taxonomy addition is omitted.
</issue_to_address>Sourcery assessment
Approval pending. 2 findings to address first.
Blocking findings: backend/app/services/pairing_service.py:21, backend/tests/test_garment_vocabulary.py:87
test_types_for_role_is_derived_from_the_vocabulary pinned fixed per-role lists, so a hard-coded types_for_role would still pass after a garment is added to a role. Compare each role against the entries derived from ITEM_ROLE and assert the per-role lists cover the taxonomy.
Owner
|
Thanks a bunch for your help @geekygirlsarah! 🙏🏽 |
TheBlackSide243
added a commit
to TheBlackSide243/wardrowbe
that referenced
this pull request
Oct 18, 2026
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
…ith garment vocabulary, ensuring consistency and avoiding manual updates. Added tests for vocabulary-based role types and prompt rendering.
Description
Reworks the outfit silhouettes in both AI outfit prompts so separates and one-piece looks are equal peers, and derives the garment type lists from the
ITEM_ROLEtaxonomy so the prompts can't drift from what the tagger validates.Previously
recommendation.txtframed the formula as separates-first withOR pick ONE dressas a less-preferred fallback, anditem_pairing.txtdid the same. The slot enumeration was also a hand-written subset that omitted several realbase_toptypes (e.g.top,polo,tank-top) and never namedjumpsuit. For a wardrobe heavy in dresses, this wording biased the model toward separating.Changes:
backend/app/utils/garment_vocabulary.py: addtypes_for_role(role)andrender_outfit_prompt(template), which fills<<ROLE_TYPES>>tokens fromITEM_ROLE(mirrors the existingrender_tagging_promptpattern).backend/app/prompts/recommendation.txt: replace theORcontinuation with two equal-weight silhouette bullets (separates / one-piece) plus an optional legwear bullet; generalize the proportion guidance to cover one-piece waists.backend/app/prompts/item_pairing.txt: same silhouette treatment, and state that a full-body garment fills both top and bottom slots.backend/app/services/recommendation_service.py,backend/app/services/pairing_service.py: render the prompts at load time.backend/tests/test_garment_vocabulary.py: tests asserting the role lists come from the vocabulary and that both prompts render and contain every role's types.Related Issue
Resolves #213
Type of Change
Checklist
Testing
Test Environment
Tests Performed
The DB-backed tests in
test_recommendation_service.pywere not run (they require Postgres viaTEST_DATABASE_URL); only the prompt/template tests in that file were exercised.Screenshots (if applicable)
N/A — prompt-only change, no UI.
Additional Notes
garment_vocabulary.jsonis untouched, sotest_type_lists_agreecontinues to hold.leggingsis apantssubtype inclothing_analysis.txt, not aTYPE, so it is not a member of the derived bottom list (pants, jeans, shorts, skirt). It remains identifiable through subtype tagging..txtfiles stay readable while the lists stay derived; happy to switch if reviewers prefer statically literal prompts.recommendation_service.pyreadsoutfit_data.get("layers", {}), but no prompt asks the model to emit alayerskey, so that field is always empty.Summary by Sourcery
Improve outfit recommendations and pairings by making one-piece garments first-class options and keeping prompt garment types aligned with the validated taxonomy.
New Features:
Bug Fixes:
Enhancements:
Tests: