|
| 1 | +from __future__ import annotations |
| 2 | + |
| 3 | +import os |
| 4 | +from datetime import datetime, timezone |
| 5 | +from typing import Literal |
| 6 | +from urllib.parse import urlparse |
| 7 | +from uuid import uuid4 |
| 8 | + |
| 9 | +from fastapi import Depends, FastAPI, HTTPException, Query, status |
| 10 | +from pydantic import BaseModel, Field |
| 11 | +from sqlalchemy import Column, DateTime, String, create_engine, select, text |
| 12 | +from sqlalchemy.orm import Session, declarative_base, sessionmaker |
| 13 | + |
| 14 | + |
| 15 | +DEFAULT_DATABASE_URL = "sqlite:///./runs.db" |
| 16 | +RunStatus = Literal["submitted", "running", "succeeded", "failed", "unknown"] |
| 17 | +ArtifactName = Literal["report", "timeline", "trace", "dag", "multiqc"] |
| 18 | + |
| 19 | +Base = declarative_base() |
| 20 | + |
| 21 | + |
| 22 | +def utc_now() -> datetime: |
| 23 | + return datetime.now(timezone.utc) |
| 24 | + |
| 25 | + |
| 26 | +class RunModel(Base): |
| 27 | + __tablename__ = "runs" |
| 28 | + |
| 29 | + id = Column(String, primary_key=True) |
| 30 | + name = Column(String, nullable=False) |
| 31 | + status = Column(String, nullable=False, default="submitted") |
| 32 | + s3_prefix = Column(String, nullable=True) |
| 33 | + nextflow_report = Column(String, nullable=False, default="pipeline_info/report.html") |
| 34 | + timeline = Column(String, nullable=False, default="pipeline_info/timeline.html") |
| 35 | + trace = Column(String, nullable=False, default="pipeline_info/trace.txt") |
| 36 | + dag = Column(String, nullable=False, default="pipeline_info/dag.dot") |
| 37 | + multiqc_report = Column(String, nullable=False, default="multiqc/multiqc_report.html") |
| 38 | + created_at = Column(DateTime(timezone=True), nullable=False, default=utc_now) |
| 39 | + updated_at = Column(DateTime(timezone=True), nullable=False, default=utc_now, onupdate=utc_now) |
| 40 | + |
| 41 | + |
| 42 | +class RunCreate(BaseModel): |
| 43 | + run_id: str | None = Field(default=None, description="Stable external run id. Defaults to a UUID.") |
| 44 | + name: str = Field(min_length=1) |
| 45 | + status: RunStatus = "submitted" |
| 46 | + s3_prefix: str | None = Field(default=None, description="S3 prefix containing published Nextflow results.") |
| 47 | + nextflow_report: str = "pipeline_info/report.html" |
| 48 | + timeline: str = "pipeline_info/timeline.html" |
| 49 | + trace: str = "pipeline_info/trace.txt" |
| 50 | + dag: str = "pipeline_info/dag.dot" |
| 51 | + multiqc_report: str = "multiqc/multiqc_report.html" |
| 52 | + |
| 53 | + |
| 54 | +class RunUpdate(BaseModel): |
| 55 | + name: str | None = Field(default=None, min_length=1) |
| 56 | + status: RunStatus | None = None |
| 57 | + s3_prefix: str | None = None |
| 58 | + nextflow_report: str | None = None |
| 59 | + timeline: str | None = None |
| 60 | + trace: str | None = None |
| 61 | + dag: str | None = None |
| 62 | + multiqc_report: str | None = None |
| 63 | + |
| 64 | + |
| 65 | +class RunOut(BaseModel): |
| 66 | + id: str |
| 67 | + name: str |
| 68 | + status: str |
| 69 | + s3_prefix: str | None |
| 70 | + nextflow_report: str |
| 71 | + timeline: str |
| 72 | + trace: str |
| 73 | + dag: str |
| 74 | + multiqc_report: str |
| 75 | + created_at: datetime |
| 76 | + updated_at: datetime |
| 77 | + |
| 78 | + model_config = {"from_attributes": True} |
| 79 | + |
| 80 | + |
| 81 | +def model_data(model: BaseModel, *, exclude_unset: bool = False) -> dict: |
| 82 | + if hasattr(model, "model_dump"): |
| 83 | + return model.model_dump(exclude_unset=exclude_unset) |
| 84 | + return model.dict(exclude_unset=exclude_unset) |
| 85 | + |
| 86 | + |
| 87 | +def engine_options(database_url: str) -> dict: |
| 88 | + if database_url.startswith("sqlite"): |
| 89 | + return {"connect_args": {"check_same_thread": False}} |
| 90 | + return {"pool_pre_ping": True} |
| 91 | + |
| 92 | + |
| 93 | +def parse_s3_uri(uri: str) -> tuple[str, str]: |
| 94 | + parsed = urlparse(uri) |
| 95 | + if parsed.scheme != "s3" or not parsed.netloc or not parsed.path.strip("/"): |
| 96 | + raise HTTPException(status_code=400, detail=f"Invalid S3 URI: {uri}") |
| 97 | + return parsed.netloc, parsed.path.lstrip("/") |
| 98 | + |
| 99 | + |
| 100 | +def resolve_artifact_uri(run: RunModel, artifact: ArtifactName) -> str: |
| 101 | + attr = "nextflow_report" if artifact == "report" else "multiqc_report" if artifact == "multiqc" else artifact |
| 102 | + value = getattr(run, attr) |
| 103 | + if value.startswith("s3://"): |
| 104 | + return value |
| 105 | + if not run.s3_prefix: |
| 106 | + raise HTTPException(status_code=404, detail="Run has no s3_prefix for relative artefact paths") |
| 107 | + return f"{run.s3_prefix.rstrip('/')}/{value.lstrip('/')}" |
| 108 | + |
| 109 | + |
| 110 | +def get_s3_client(): |
| 111 | + import boto3 |
| 112 | + |
| 113 | + return boto3.client("s3", region_name=os.getenv("AWS_REGION")) |
| 114 | + |
| 115 | + |
| 116 | +def create_app(database_url: str | None = None) -> FastAPI: |
| 117 | + db_url = database_url or os.getenv("DATABASE_URL", DEFAULT_DATABASE_URL) |
| 118 | + engine = create_engine(db_url, **engine_options(db_url)) |
| 119 | + session_local = sessionmaker(bind=engine, autoflush=False, autocommit=False) |
| 120 | + Base.metadata.create_all(bind=engine) |
| 121 | + |
| 122 | + app = FastAPI( |
| 123 | + title="RNA-seq Report Portal", |
| 124 | + version="0.1.0", |
| 125 | + description="Registers Nextflow RNA-seq runs and signs S3 report artefact URLs.", |
| 126 | + ) |
| 127 | + |
| 128 | + def get_db(): |
| 129 | + db = session_local() |
| 130 | + try: |
| 131 | + yield db |
| 132 | + finally: |
| 133 | + db.close() |
| 134 | + |
| 135 | + @app.get("/health") |
| 136 | + def health(db: Session = Depends(get_db)) -> dict: |
| 137 | + db.execute(text("SELECT 1")) |
| 138 | + return {"status": "ok"} |
| 139 | + |
| 140 | + @app.post("/runs", response_model=RunOut, status_code=status.HTTP_201_CREATED) |
| 141 | + def create_run(payload: RunCreate, db: Session = Depends(get_db)) -> RunModel: |
| 142 | + run_id = payload.run_id or str(uuid4()) |
| 143 | + if db.get(RunModel, run_id): |
| 144 | + raise HTTPException(status_code=409, detail=f"Run already exists: {run_id}") |
| 145 | + data = model_data(payload) |
| 146 | + data.pop("run_id", None) |
| 147 | + run = RunModel(id=run_id, **data) |
| 148 | + db.add(run) |
| 149 | + db.commit() |
| 150 | + db.refresh(run) |
| 151 | + return run |
| 152 | + |
| 153 | + @app.get("/runs", response_model=list[RunOut]) |
| 154 | + def list_runs(status_filter: RunStatus | None = Query(default=None, alias="status"), db: Session = Depends(get_db)): |
| 155 | + stmt = select(RunModel).order_by(RunModel.created_at.desc()) |
| 156 | + if status_filter: |
| 157 | + stmt = stmt.where(RunModel.status == status_filter) |
| 158 | + return list(db.scalars(stmt)) |
| 159 | + |
| 160 | + @app.get("/runs/{run_id}", response_model=RunOut) |
| 161 | + def get_run(run_id: str, db: Session = Depends(get_db)) -> RunModel: |
| 162 | + run = db.get(RunModel, run_id) |
| 163 | + if not run: |
| 164 | + raise HTTPException(status_code=404, detail=f"Unknown run: {run_id}") |
| 165 | + return run |
| 166 | + |
| 167 | + @app.patch("/runs/{run_id}", response_model=RunOut) |
| 168 | + def update_run(run_id: str, payload: RunUpdate, db: Session = Depends(get_db)) -> RunModel: |
| 169 | + run = db.get(RunModel, run_id) |
| 170 | + if not run: |
| 171 | + raise HTTPException(status_code=404, detail=f"Unknown run: {run_id}") |
| 172 | + for key, value in model_data(payload, exclude_unset=True).items(): |
| 173 | + setattr(run, key, value) |
| 174 | + run.updated_at = utc_now() |
| 175 | + db.add(run) |
| 176 | + db.commit() |
| 177 | + db.refresh(run) |
| 178 | + return run |
| 179 | + |
| 180 | + @app.get("/runs/{run_id}/artifacts/{artifact}/presign") |
| 181 | + def presign_artifact( |
| 182 | + run_id: str, |
| 183 | + artifact: ArtifactName, |
| 184 | + expires: int = Query(default=3600, ge=60, le=604800), |
| 185 | + db: Session = Depends(get_db), |
| 186 | + s3_client=Depends(get_s3_client), |
| 187 | + ) -> dict: |
| 188 | + run = db.get(RunModel, run_id) |
| 189 | + if not run: |
| 190 | + raise HTTPException(status_code=404, detail=f"Unknown run: {run_id}") |
| 191 | + s3_uri = resolve_artifact_uri(run, artifact) |
| 192 | + bucket, key = parse_s3_uri(s3_uri) |
| 193 | + url = s3_client.generate_presigned_url( |
| 194 | + "get_object", |
| 195 | + Params={"Bucket": bucket, "Key": key}, |
| 196 | + ExpiresIn=expires, |
| 197 | + ) |
| 198 | + return {"artifact": artifact, "s3_uri": s3_uri, "url": url, "expires_in": expires} |
| 199 | + |
| 200 | + return app |
| 201 | + |
| 202 | + |
| 203 | +app = create_app() |
0 commit comments