css-test/app/api/configs.py

"""ReconConfig API proxy / fake endpoints.

When RECON_API_BASE_URL is set in the environment, POST and PUT calls are
forwarded to the real API.  GET calls always use fake data in this workspace
so the editor has something to load without a real database.

Also provides a /api/configs/test-url endpoint that inspects file:// URLs and
returns metadata (no file contents are returned).
"""
from __future__ import annotations

import glob
import os
from datetime import datetime
from pathlib import Path
from typing import Any

import httpx
from fastapi import APIRouter, HTTPException, Request
from fastapi.responses import JSONResponse

from app.core.refdata import ReconConfigStatus, ReconPatterns
from app.service.fake_configs import get_fake_configs

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

_RECON_API_BASE = os.getenv("RECON_API_BASE_URL", "").rstrip("/")


# ── Helpers ────────────────────────────────────────────────────────────────

def _fake_config_full(ref: str, now: datetime) -> dict[str, Any] | None:
    """Build a full ReconConfigRequest-shaped dict from a FakeReconConfig."""
    fake = next((c for c in get_fake_configs(now) if c.reference == ref), None)
    if fake is None:
        return None
    return {
        "reference": fake.reference,
        "name": fake.name,
        "business_process": fake.business_process,
        "comment": "",
        "data_type": "Unknown",
        "pattern": ReconPatterns.ONE_TO_ONE.value,
        "status": fake.status,
        "frequency": fake.frequency,
        "start_datetime": fake.start_datetime.isoformat() if fake.start_datetime else None,
        "field_mapping": {},
        "sources": [
            {
                "name": "Source System",
                "url": "file:///data/input/{{as_at_date.strftime('%Y%m%d')}}.csv",
                "comment": "",
                "day_offset": 0,
                "system_schema": {},
                "obfuscate_fields": [],
                "pci_redact_fields": [],
                "field_widths": [],
                "profile_thresholds": [],
                "index_fields": [],
                "filter": "",
                "sql": "",
                "csv_spec": {"delimiter": ",", "header": True, "encoding": "utf_8", "trailer_rows": 0, "quoting": True},
                "xml_spec": None,
            }
        ],
        "destination": {
            "name": "Destination System",
            "url": "mssql://GROUPDW?db=BIODS_PROCESSING&table=EXAMPLE",
            "comment": "",
            "day_offset": 0,
            "system_schema": {},
            "obfuscate_fields": [],
            "pci_redact_fields": [],
            "field_widths": [],
            "profile_thresholds": [],
            "index_fields": [],
            "filter": "",
            "sql": "",
            "csv_spec": None,
            "xml_spec": None,
        },
    }


# ── List ────────────────────────────────────────────────────────────────────

@router.get("/")
async def list_configs():
    now = datetime.now()
    items = []
    for c in get_fake_configs(now):
        items.append({
            "reference": c.reference,
            "name": c.name,
            "status": c.status,
            "frequency": c.frequency,
            "business_process": c.business_process,
        })
    return items


# ── Get one ─────────────────────────────────────────────────────────────────

@router.get("/{reference}")
async def get_config(reference: str):
    now = datetime.now()

    if _RECON_API_BASE:
        async with httpx.AsyncClient(verify=False) as client:
            resp = await client.get(f"{_RECON_API_BASE}/api/configs/{reference}")
        if resp.status_code == 404:
            raise HTTPException(404, f"Config '{reference}' not found")
        return resp.json()

    data = _fake_config_full(reference, now)
    if data is None:
        raise HTTPException(404, f"Config '{reference}' not found")
    return data


# ── Create ──────────────────────────────────────────────────────────────────

@router.post("/")
async def create_config(request: Request):
    body = await request.json()

    if _RECON_API_BASE:
        async with httpx.AsyncClient(verify=False) as client:
            resp = await client.post(f"{_RECON_API_BASE}/api/configs/", json=body)
        return JSONResponse(content=resp.json(), status_code=resp.status_code)

    # Workspace stub: echo back with a fake revision
    return JSONResponse(
        content={**body, "revision_number": 1, "user": {"username": "workspace"}},
        status_code=201,
    )


# ── Update ──────────────────────────────────────────────────────────────────

@router.put("/{reference}")
async def update_config(reference: str, request: Request):
    body = await request.json()

    if _RECON_API_BASE:
        async with httpx.AsyncClient(verify=False) as client:
            resp = await client.put(f"{_RECON_API_BASE}/api/configs/{reference}", json=body)
        return JSONResponse(content=resp.json(), status_code=resp.status_code)

    return JSONResponse(
        content={**body, "revision_number": body.get("revision_number", 1) + 1, "user": {"username": "workspace"}},
        status_code=200,
    )


# ── Test URL ────────────────────────────────────────────────────────────────

@router.post("/test-url")
async def test_url(request: Request):
    """Return metadata for a URL pattern — no file contents ever returned."""
    body = await request.json()
    url: str = body.get("url", "")
    as_at_date: str = body.get("as_at_date", datetime.now().strftime("%Y%m%d"))

    if not url.startswith("file://"):
        return {"type": "non-file", "message": "Only file:// URLs can be tested here. Database connections are validated at run time."}

    # Resolve the path part (strip file://)
    raw_path = url[7:]

    # Replace simple template variables for preview purposes
    from datetime import date as date_type
    try:
        as_at = datetime.strptime(as_at_date, "%Y%m%d").date()
    except ValueError:
        as_at = date_type.today()

    resolved = (
        raw_path
        .replace("{{as_at_date.strftime('%Y%m%d')}}", as_at.strftime("%Y%m%d"))
        .replace("{{today.strftime('%Y%m%d')}}", date_type.today().strftime("%Y%m%d"))
        .replace("{{as_at_date.strftime('%Y-%m-%d')}}", as_at.strftime("%Y-%m-%d"))
        .replace("{{today.strftime('%Y-%m-%d')}}", date_type.today().strftime("%Y-%m-%d"))
    )

    # Treat as glob pattern for anything still containing {
    if "*" in resolved or "?" in resolved or "{" in resolved:
        matches = glob.glob(resolved)
        if not matches:
            return {"type": "file", "resolved": resolved, "found": False, "message": "No files matched the pattern."}
        file_infos = []
        for p in sorted(matches)[:10]:
            try:
                st = os.stat(p)
                file_infos.append({
                    "path": p,
                    "size_bytes": st.st_size,
                    "modified": datetime.fromtimestamp(st.st_mtime).isoformat(timespec="seconds"),
                })
            except OSError:
                file_infos.append({"path": p, "error": "Could not stat file"})
        return {"type": "file", "resolved": resolved, "found": True, "matches": len(matches), "files": file_infos}

    # Exact path
    p = Path(resolved)
    if not p.exists():
        return {"type": "file", "resolved": resolved, "found": False, "message": f"File not found: {resolved}"}
    try:
        st = p.stat()
        return {
            "type": "file",
            "resolved": resolved,
            "found": True,
            "matches": 1,
            "files": [{
                "path": str(p),
                "size_bytes": st.st_size,
                "modified": datetime.fromtimestamp(st.st_mtime).isoformat(timespec="seconds"),
            }],
        }
    except OSError as e:
        return {"type": "file", "resolved": resolved, "found": False, "message": str(e)}