padelnomics/web/src/padelnomics/seo/_bing.py

"""Bing Webmaster Tools sync via REST API.

Uses an API key for auth. Fetches query stats and page stats.
"""

from datetime import UTC, datetime, timedelta
from urllib.parse import urlparse

import httpx

from ..core import config, execute, utcnow, utcnow_iso

_TIMEOUT_SECONDS = 30


def _normalize_url(full_url: str) -> str:
    """Strip a full URL to just the path."""
    parsed = urlparse(full_url)
    return parsed.path or "/"


async def sync_bing(days_back: int = 3, timeout_seconds: int = _TIMEOUT_SECONDS) -> int:
    """Sync Bing Webmaster query stats into seo_search_metrics. Returns rows synced."""
    assert 1 <= days_back <= 90, "days_back must be 1-90"
    assert 1 <= timeout_seconds <= 120, "timeout_seconds must be 1-120"

    if not config.BING_WEBMASTER_API_KEY or not config.BING_SITE_URL:
        return 0  # Bing not configured — skip silently

    started_at = utcnow()

    try:
        rows_synced = 0
        async with httpx.AsyncClient(timeout=timeout_seconds) as client:
            # Fetch query stats for the date range
            response = await client.get(
                "https://ssl.bing.com/webmaster/api.svc/json/GetQueryStats",
                params={
                    "apikey": config.BING_WEBMASTER_API_KEY,
                    "siteUrl": config.BING_SITE_URL,
                },
            )
            response.raise_for_status()
            data = response.json()

            # Bing returns {"d": [{"Query": ..., "Date": ..., ...}, ...]}
            entries = data.get("d", []) if isinstance(data, dict) else data
            if not isinstance(entries, list):
                entries = []

            cutoff = utcnow() - timedelta(days=days_back)

            for entry in entries:
                # Bing date format: "/Date(1708905600000)/" (ms since epoch)
                date_str = entry.get("Date", "")
                if "/Date(" in date_str:
                    ms = int(date_str.split("(")[1].split(")")[0])
                    entry_date = datetime.fromtimestamp(ms / 1000, tz=UTC)
                else:
                    continue

                if entry_date < cutoff:
                    continue

                metric_date = entry_date.strftime("%Y-%m-%d")
                query = entry.get("Query", "")

                await execute(
                    """INSERT OR REPLACE INTO seo_search_metrics
                       (source, metric_date, page_url, query, country, device,
                        clicks, impressions, ctr, position_avg)
                       VALUES ('bing', ?, '/', ?, NULL, NULL, ?, ?, ?, ?)""",
                    (
                        metric_date, query,
                        entry.get("Clicks", 0),
                        entry.get("Impressions", 0),
                        entry.get("AvgCTR", 0.0),
                        entry.get("AvgClickPosition", 0.0),
                    ),
                )
                rows_synced += 1

            # Also fetch page-level stats
            page_response = await client.get(
                "https://ssl.bing.com/webmaster/api.svc/json/GetPageStats",
                params={
                    "apikey": config.BING_WEBMASTER_API_KEY,
                    "siteUrl": config.BING_SITE_URL,
                },
            )
            page_response.raise_for_status()
            page_data = page_response.json()

            page_entries = page_data.get("d", []) if isinstance(page_data, dict) else page_data
            if not isinstance(page_entries, list):
                page_entries = []

            for entry in page_entries:
                date_str = entry.get("Date", "")
                if "/Date(" in date_str:
                    ms = int(date_str.split("(")[1].split(")")[0])
                    entry_date = datetime.fromtimestamp(ms / 1000, tz=UTC)
                else:
                    continue

                if entry_date < cutoff:
                    continue

                metric_date = entry_date.strftime("%Y-%m-%d")
                page_url = _normalize_url(entry.get("Url", "/"))

                await execute(
                    """INSERT OR REPLACE INTO seo_search_metrics
                       (source, metric_date, page_url, query, country, device,
                        clicks, impressions, ctr, position_avg)
                       VALUES ('bing', ?, ?, '', NULL, NULL, ?, ?, NULL, NULL)""",
                    (
                        metric_date, page_url,
                        entry.get("Clicks", 0),
                        entry.get("Impressions", 0),
                    ),
                )
                rows_synced += 1

        duration_ms = int((utcnow() - started_at).total_seconds() * 1000)
        await execute(
            """INSERT INTO seo_sync_log
               (source, status, rows_synced, started_at, completed_at, duration_ms)
               VALUES ('bing', 'success', ?, ?, ?, ?)""",
            (rows_synced, started_at.strftime("%Y-%m-%dT%H:%M:%S"), utcnow_iso(), duration_ms),
        )
        return rows_synced

    except Exception as exc:
        duration_ms = int((utcnow() - started_at).total_seconds() * 1000)
        await execute(
            """INSERT INTO seo_sync_log
               (source, status, rows_synced, error, started_at, completed_at, duration_ms)
               VALUES ('bing', 'failed', 0, ?, ?, ?, ?)""",
            (str(exc), started_at.strftime("%Y-%m-%dT%H:%M:%S"), utcnow_iso(), duration_ms),
        )
        raise