feat: Added scored dynamic related-techniques query returning up to 4 r…

- "backend/schemas.py"
- "backend/routers/techniques.py"
- "backend/tests/test_public_api.py"

GSD-Task: S02/T01
This commit is contained in:
jlightner 2026-03-31 06:13:59 +00:00
parent 6a793c2c9a
commit c25db471f7
3 changed files with 303 additions and 0 deletions

View file

@ -29,6 +29,87 @@ logger = logging.getLogger("chrysopedia.techniques")
router = APIRouter(prefix="/techniques", tags=["techniques"]) router = APIRouter(prefix="/techniques", tags=["techniques"])
async def _find_dynamic_related(
db: AsyncSession,
page: TechniquePage,
exclude_slugs: set[str],
limit: int,
) -> list[RelatedLinkItem]:
"""Score and return dynamically related technique pages.
Scoring:
- Same creator + same topic_category: +3
- Same creator, different category: +2
- Same topic_category, different creator: +2
- Each overlapping topic_tag: +1
"""
exclude_ids = {page.id}
# Base: all other technique pages, eagerly load creator for name
stmt = (
select(TechniquePage)
.options(selectinload(TechniquePage.creator))
.where(TechniquePage.id != page.id)
)
if exclude_slugs:
stmt = stmt.where(TechniquePage.slug.notin_(exclude_slugs))
result = await db.execute(stmt)
candidates = result.scalars().all()
if not candidates:
return []
current_tags = set(page.topic_tags) if page.topic_tags else set()
scored: list[tuple[int, str, TechniquePage]] = []
for cand in candidates:
score = 0
reasons: list[str] = []
same_creator = cand.creator_id == page.creator_id
same_category = cand.topic_category == page.topic_category
if same_creator and same_category:
score += 3
reasons.append("Same creator, same topic")
elif same_creator:
score += 2
reasons.append("Same creator")
elif same_category:
score += 2
reasons.append(f"Also about {page.topic_category}")
# Tag overlap scoring
if current_tags:
cand_tags = set(cand.topic_tags) if cand.topic_tags else set()
shared = current_tags & cand_tags
if shared:
score += len(shared)
reasons.append(f"Shared tags: {', '.join(sorted(shared))}")
if score > 0:
scored.append((score, "; ".join(reasons), cand))
# Sort descending by score, then by title for determinism
scored.sort(key=lambda x: (-x[0], x[2].title))
results: list[RelatedLinkItem] = []
for score, reason, cand in scored[:limit]:
creator_name = cand.creator.name if cand.creator else ""
results.append(
RelatedLinkItem(
target_title=cand.title,
target_slug=cand.slug,
relationship="dynamic",
creator_name=creator_name,
topic_category=cand.topic_category,
reason=reason,
)
)
return results
@router.get("", response_model=PaginatedResponse) @router.get("", response_model=PaginatedResponse)
async def list_techniques( async def list_techniques(
category: Annotated[str | None, Query()] = None, category: Annotated[str | None, Query()] = None,
@ -165,6 +246,23 @@ async def get_technique(
) )
) )
# Supplement with dynamic related techniques (up to 4 total)
curated_slugs = {link.target_slug for link in related_links}
max_related = 4
if len(related_links) < max_related:
remaining = max_related - len(related_links)
try:
dynamic_links = await _find_dynamic_related(
db, page, curated_slugs, remaining
)
related_links.extend(dynamic_links)
except Exception:
logger.warning(
"Dynamic related query failed for %s, continuing with curated only",
slug,
exc_info=True,
)
base = TechniquePageRead.model_validate(page) base = TechniquePageRead.model_validate(page)
# Count versions for this page # Count versions for this page

View file

@ -247,6 +247,9 @@ class RelatedLinkItem(BaseModel):
target_title: str = "" target_title: str = ""
target_slug: str = "" target_slug: str = ""
relationship: str = "" relationship: str = ""
creator_name: str = ""
topic_category: str = ""
reason: str = ""
class CreatorInfo(BaseModel): class CreatorInfo(BaseModel):

View file

@ -594,3 +594,205 @@ async def test_technique_detail_includes_version_count(client, db_engine):
resp2 = await client.get(f"{TECHNIQUES_URL}/{seed['tp1_slug']}") resp2 = await client.get(f"{TECHNIQUES_URL}/{seed['tp1_slug']}")
assert resp2.status_code == 200 assert resp2.status_code == 200
assert resp2.json()["version_count"] == 1 assert resp2.json()["version_count"] == 1
# ── Dynamic Related Techniques Tests ─────────────────────────────────────────
async def _seed_related_data(db_engine) -> dict:
"""Seed 2 creators and 5 technique pages with overlapping tags/categories.
Returns a dict of slugs and metadata for related-technique assertions.
"""
session_factory = async_sessionmaker(
db_engine, class_=AsyncSession, expire_on_commit=False
)
async with session_factory() as session:
creator_a = Creator(
name="Creator A",
slug="creator-a",
genres=["Ambient"],
folder_name="CreatorA",
)
creator_b = Creator(
name="Creator B",
slug="creator-b",
genres=["Techno"],
folder_name="CreatorB",
)
session.add_all([creator_a, creator_b])
await session.flush()
# tp1: creator_a, "Sound design", tags: [reverb, delay]
tp1 = TechniquePage(
creator_id=creator_a.id,
title="Reverb Chains",
slug="reverb-chains",
topic_category="Sound design",
topic_tags=["reverb", "delay"],
summary="Chaining reverbs for depth",
)
# tp2: creator_a, "Sound design", tags: [reverb, modulation]
# Same creator + same category = score 3, shared tag 'reverb' = +1 → 4
tp2 = TechniquePage(
creator_id=creator_a.id,
title="Advanced Reverb Modulation",
slug="advanced-reverb-modulation",
topic_category="Sound design",
topic_tags=["reverb", "modulation"],
summary="Modulating reverb tails",
)
# tp3: creator_a, "Mixing", tags: [delay, sidechain]
# Same creator, different category = score 2, shared tag 'delay' = +1 → 3
tp3 = TechniquePage(
creator_id=creator_a.id,
title="Delay Mixing Tricks",
slug="delay-mixing-tricks",
topic_category="Mixing",
topic_tags=["delay", "sidechain"],
summary="Using delay in mix context",
)
# tp4: creator_b, "Sound design", tags: [reverb]
# Different creator, same category = score 2, shared tag 'reverb' = +1 → 3
tp4 = TechniquePage(
creator_id=creator_b.id,
title="Plate Reverb Techniques",
slug="plate-reverb-techniques",
topic_category="Sound design",
topic_tags=["reverb"],
summary="Plate reverb for vocals",
)
# tp5: creator_b, "Mastering", tags: [limiting]
# Different creator, different category, no shared tags = score 0
tp5 = TechniquePage(
creator_id=creator_b.id,
title="Mastering Limiter Setup",
slug="mastering-limiter-setup",
topic_category="Mastering",
topic_tags=["limiting"],
summary="Setting up a mastering limiter",
)
session.add_all([tp1, tp2, tp3, tp4, tp5])
await session.commit()
return {
"tp1_slug": tp1.slug,
"tp2_slug": tp2.slug,
"tp3_slug": tp3.slug,
"tp4_slug": tp4.slug,
"tp5_slug": tp5.slug,
}
@pytest.mark.asyncio
async def test_dynamic_related_techniques(client, db_engine):
"""Dynamic related links are scored and ranked correctly."""
seed = await _seed_related_data(db_engine)
resp = await client.get(f"{TECHNIQUES_URL}/{seed['tp1_slug']}")
assert resp.status_code == 200
data = resp.json()
related = data["related_links"]
# tp1 should match tp2 (score 4), tp3 (score 3), tp4 (score 3), NOT tp5 (score 0)
related_slugs = [r["target_slug"] for r in related]
assert seed["tp5_slug"] not in related_slugs
assert len(related) <= 4
# tp2 should be first (highest score: same creator + same category + shared tag)
assert related[0]["target_slug"] == seed["tp2_slug"]
# All results should have enriched fields populated
for r in related:
assert r["creator_name"] != ""
assert r["topic_category"] != ""
assert r["reason"] != ""
@pytest.mark.asyncio
async def test_dynamic_related_excludes_self(client, db_engine):
"""The technique itself never appears in its own related_links."""
seed = await _seed_related_data(db_engine)
resp = await client.get(f"{TECHNIQUES_URL}/{seed['tp1_slug']}")
assert resp.status_code == 200
related_slugs = {r["target_slug"] for r in resp.json()["related_links"]}
assert seed["tp1_slug"] not in related_slugs
@pytest.mark.asyncio
async def test_dynamic_related_no_peers(client, db_engine):
"""A technique with no matching peers returns empty related_links."""
session_factory = async_sessionmaker(
db_engine, class_=AsyncSession, expire_on_commit=False
)
async with session_factory() as session:
creator = Creator(
name="Solo Artist",
slug="solo-artist",
genres=["Noise"],
folder_name="SoloArtist",
)
session.add(creator)
await session.flush()
tp = TechniquePage(
creator_id=creator.id,
title="Unique Technique",
slug="unique-technique",
topic_category="Experimental",
topic_tags=["unique-tag-xyz"],
summary="Completely unique",
)
session.add(tp)
await session.commit()
resp = await client.get(f"{TECHNIQUES_URL}/unique-technique")
assert resp.status_code == 200
assert resp.json()["related_links"] == []
@pytest.mark.asyncio
async def test_dynamic_related_null_tags(client, db_engine):
"""Technique with NULL topic_tags still scores on creator_id/topic_category."""
session_factory = async_sessionmaker(
db_engine, class_=AsyncSession, expire_on_commit=False
)
async with session_factory() as session:
creator = Creator(
name="Tag-Free Creator",
slug="tag-free-creator",
genres=["Pop"],
folder_name="TagFreeCreator",
)
session.add(creator)
await session.flush()
tp_main = TechniquePage(
creator_id=creator.id,
title="No Tags Main",
slug="no-tags-main",
topic_category="Production",
topic_tags=None,
summary="Main page with no tags",
)
tp_peer = TechniquePage(
creator_id=creator.id,
title="No Tags Peer",
slug="no-tags-peer",
topic_category="Production",
topic_tags=None,
summary="Peer page, same creator and category",
)
session.add_all([tp_main, tp_peer])
await session.commit()
resp = await client.get(f"{TECHNIQUES_URL}/no-tags-main")
assert resp.status_code == 200
related = resp.json()["related_links"]
assert len(related) == 1
assert related[0]["target_slug"] == "no-tags-peer"
assert related[0]["reason"] == "Same creator, same topic"