backend/api/routes/report.py

"""Report generation: JSON preview and DOCX export from surveyor notes."""

from __future__ import annotations

from typing import Literal

from fastapi import APIRouter, Depends, HTTPException, Response, status
from pydantic import BaseModel, Field, model_validator

from backend.api.deps import get_current_tenant
from backend.core import pii_scrubber, report_assembler, section_mapper, template_discoverer
from backend.core.interference import resolve_interference_level
from backend.core.photo_store import draft_section_photo_paths
from backend.core.rag_store import TIER_REFERENCE, get_rag_store
from backend.models.schema import TemplateSchema

router = APIRouter(prefix="/api/report", tags=["report"])

_DOCX_MIME = "application/vnd.openxmlformats-officedocument.wordprocessingml.document"


class GenerateRequest(BaseModel):
    raw_notes: str = Field(
        default="",
        description="Surveyor field notes (any format). May be empty when using photos only.",
    )
    property_type: str = Field(
        min_length=1,
        description="Generic property descriptor, e.g. semi-detached house.",
    )
    tenure: str = Field(min_length=1, description="e.g. freehold or leasehold.")
    title: str = Field(default="Survey Report", max_length=200)
    include_footer: bool = Field(default=False)
    report_draft_id: str | None = Field(
        default=None,
        description=(
            "Draft id from POST /api/report/drafts. When set, selected section photos "
            "(max 2 per section) are analyzed and mapped into the relevant sections."
        ),
    )
    survey_level: int = Field(
        default=3,
        ge=1,
        le=3,
        description=(
            "RICS survey product tier (1/2/3). When interference_level is omitted, "
            "Level 1→minimum, Level 2→medium, Level 3→maximum composition."
        ),
    )
    interference_level: Literal["minimum", "medium", "maximum"] | None = Field(
        default=None,
        description=(
            "Report composition depth. When omitted, derived from survey_level. "
            "minimum: baseline preservation with in-place fact updates. "
            "medium: technical in-place edit on past report baseline. "
            "maximum: full long-form narrative edit retaining baseline scaffolding."
        ),
    )

    @model_validator(mode="after")
    def notes_or_draft(self) -> GenerateRequest:
        if not self.raw_notes.strip() and not self.report_draft_id:
            raise ValueError("raw_notes or report_draft_id is required")
        return self


def _require_generation_ready(
    tenant_id: str,
    interference_level: Literal["minimum", "medium", "maximum"],
) -> TemplateSchema:
    schema = template_discoverer.ensure_canonical_schema(tenant_id)
    store = get_rag_store()
    if store.count(tenant_id, TIER_REFERENCE) == 0:
        raise HTTPException(
            status_code=status.HTTP_400_BAD_REQUEST,
            detail=(
                "No past reports have been ingested. Upload reference reports "
                f"for {interference_level}-AI mapping."
            ),
        )
    return schema


@router.post("/preview")
def preview(
    req: GenerateRequest,
    tenant_id: str = Depends(get_current_tenant),
) -> dict:
    il = resolve_interference_level(req.interference_level, req.survey_level)
    schema = _require_generation_ready(tenant_id, il)
    result = section_mapper.generate_report(
        tenant_id,
        req.raw_notes,
        property_type=req.property_type,
        tenure=req.tenure,
        interference_level=il,
        survey_level=req.survey_level,
        report_draft_id=req.report_draft_id,
    )
    return report_assembler.to_preview(result, schema)


@router.post("/generate")
def generate(
    req: GenerateRequest,
    tenant_id: str = Depends(get_current_tenant),
) -> Response:
    il = resolve_interference_level(req.interference_level, req.survey_level)
    schema = _require_generation_ready(tenant_id, il)
    result = section_mapper.generate_report(
        tenant_id,
        req.raw_notes,
        property_type=req.property_type,
        tenure=req.tenure,
        interference_level=il,
        survey_level=req.survey_level,
        report_draft_id=req.report_draft_id,
    )
    section_photos = draft_section_photo_paths(
        tenant_id,
        req.report_draft_id,
        schema.section_ids(),
    )
    try:
        data = report_assembler.to_docx(
            result,
            schema,
            title=req.title,
            include_footer=req.include_footer,
            section_photo_paths=section_photos,
        )
    except pii_scrubber.PiiDetectedError:
        for s in result.sections:
            s.text = pii_scrubber.scrub(s.text).text
        result.unassigned_text = pii_scrubber.scrub(result.unassigned_text).text
        data = report_assembler.to_docx(
            result,
            schema,
            title=req.title,
            include_footer=req.include_footer,
            section_photo_paths=section_photos,
        )
    return Response(
        content=data,
        media_type=_DOCX_MIME,
        headers={"Content-Disposition": 'attachment; filename="survey_report_draft.docx"'},
    )