|
| 1 | +from __future__ import annotations |
| 2 | + |
| 3 | +import base64 |
| 4 | +import json |
| 5 | +import logging |
| 6 | +import re |
| 7 | +import time |
| 8 | +from datetime import datetime, timedelta, timezone |
| 9 | +from os import makedirs |
| 10 | +from pathlib import Path |
| 11 | +from typing import Any |
| 12 | +from urllib.parse import urlparse |
| 13 | + |
| 14 | +import requests |
| 15 | +from requests.auth import HTTPBasicAuth |
| 16 | +from tqdm import tqdm |
| 17 | + |
| 18 | +from earthdaily.platform._resolver_base import AssetResolver, DomainPatternResolver |
| 19 | + |
| 20 | +logger = logging.getLogger(__name__) |
| 21 | + |
| 22 | +_EARTHDATA_DOMAINS = [ |
| 23 | + "data.lpdaac.earthdatacloud.nasa.gov", |
| 24 | + "data.laadsdaac.earthdatacloud.nasa.gov", |
| 25 | + "cmr.earthdata.nasa.gov", |
| 26 | + "ladsweb.modaps.eosdis.nasa.gov", |
| 27 | + "data.nsidc.earthdatacloud.nasa.gov", |
| 28 | +] |
| 29 | + |
| 30 | +_S3_HTTPS_PATTERN = re.compile(r"(s3-|s3\.)?(.*)\.amazonaws\.com") |
| 31 | + |
| 32 | + |
| 33 | +class EarthDataResolver(DomainPatternResolver): |
| 34 | + """Resolver for NASA EarthData assets. |
| 35 | +
|
| 36 | + Accepts a JWT token directly. The token is validated for expiry before |
| 37 | + each request; callers are responsible for providing a non-expired token. |
| 38 | + """ |
| 39 | + |
| 40 | + needs_credentials = True |
| 41 | + |
| 42 | + domain_pattern = re.compile("|".join(re.escape(d) for d in _EARTHDATA_DOMAINS)) |
| 43 | + |
| 44 | + def __init__(self, token: str) -> None: |
| 45 | + self._raw_token = token |
| 46 | + |
| 47 | + @classmethod |
| 48 | + def from_config(cls, config: Any) -> EarthDataResolver | None: |
| 49 | + token = getattr(config, "earthdata_token", "") or "" |
| 50 | + return cls(token=token) if token else None |
| 51 | + |
| 52 | + @staticmethod |
| 53 | + def _add_base64_padding(data: str) -> str: |
| 54 | + return data + "=" * (-len(data) % 4) |
| 55 | + |
| 56 | + @classmethod |
| 57 | + def _validate_jwt(cls, jwt_token: str) -> dict: |
| 58 | + """Decode and validate a JWT token, raising on expiry or bad format.""" |
| 59 | + parts = jwt_token.split(".") |
| 60 | + if len(parts) != 3: |
| 61 | + raise ValueError("Invalid EarthData JWT: expected three dot-separated parts.") |
| 62 | + |
| 63 | + try: |
| 64 | + decoded = base64.urlsafe_b64decode(cls._add_base64_padding(parts[1])).decode("utf-8") |
| 65 | + except Exception as exc: |
| 66 | + raise ValueError("Invalid EarthData JWT: payload is not valid Base64.") from exc |
| 67 | + |
| 68 | + try: |
| 69 | + payload = json.loads(decoded) |
| 70 | + except json.JSONDecodeError as exc: |
| 71 | + raise ValueError("Invalid EarthData JWT: payload is not valid JSON.") from exc |
| 72 | + |
| 73 | + exp = payload.get("exp") |
| 74 | + if not exp: |
| 75 | + raise ValueError("Invalid EarthData JWT: missing 'exp' claim.") |
| 76 | + if time.time() > exp: |
| 77 | + raise ValueError(f"EarthData JWT expired on {time.strftime('%Y-%m-%d %H:%M:%S', time.gmtime(exp))}.") |
| 78 | + return payload |
| 79 | + |
| 80 | + def get_download_url(self, url: str) -> str: |
| 81 | + return url |
| 82 | + |
| 83 | + def get_headers(self, url: str) -> dict[str, str]: |
| 84 | + self._validate_jwt(self._raw_token) |
| 85 | + return {"Authorization": f"Bearer {self._raw_token}"} |
| 86 | + |
| 87 | + |
| 88 | +class EUMETSATResolver(DomainPatternResolver): |
| 89 | + """Resolver for EUMETSAT assets. |
| 90 | +
|
| 91 | + Performs OAuth2 ``client_credentials`` token exchange against |
| 92 | + ``https://api.eumetsat.int/token`` and caches the bearer token with a |
| 93 | + 60-second safety buffer before expiry. |
| 94 | + """ |
| 95 | + |
| 96 | + needs_credentials = True |
| 97 | + |
| 98 | + domain_pattern = re.compile(r"api\.eumetsat\.int$") |
| 99 | + |
| 100 | + _TOKEN_URL = "https://api.eumetsat.int/token" |
| 101 | + |
| 102 | + def __init__(self, client_id: str, client_secret: str) -> None: |
| 103 | + self._client_id = client_id |
| 104 | + self._client_secret = client_secret |
| 105 | + self._bearer_token: str | None = None |
| 106 | + self._token_expires: datetime | None = None |
| 107 | + |
| 108 | + @classmethod |
| 109 | + def from_config(cls, config: Any) -> EUMETSATResolver | None: |
| 110 | + client_id = getattr(config, "eumetsat_client_id", "") or "" |
| 111 | + client_secret = getattr(config, "eumetsat_client_secret", "") or "" |
| 112 | + if client_id and client_secret: |
| 113 | + return cls(client_id=client_id, client_secret=client_secret) |
| 114 | + return None |
| 115 | + |
| 116 | + def _refresh_token_if_needed(self) -> None: |
| 117 | + if self._bearer_token and self._token_expires and self._token_expires > datetime.now(timezone.utc): |
| 118 | + return |
| 119 | + |
| 120 | + response = requests.post( |
| 121 | + self._TOKEN_URL, |
| 122 | + auth=HTTPBasicAuth(self._client_id, self._client_secret), |
| 123 | + timeout=15, |
| 124 | + data={"grant_type": "client_credentials"}, |
| 125 | + ) |
| 126 | + response.raise_for_status() |
| 127 | + body = response.json() |
| 128 | + self._bearer_token = body["access_token"] |
| 129 | + self._token_expires = datetime.now(timezone.utc) + timedelta(seconds=int(body["expires_in"]) - 60) |
| 130 | + |
| 131 | + def get_download_url(self, url: str) -> str: |
| 132 | + return url |
| 133 | + |
| 134 | + def get_headers(self, url: str) -> dict[str, str]: |
| 135 | + self._refresh_token_if_needed() |
| 136 | + return {"Authorization": f"Bearer {self._bearer_token}"} |
| 137 | + |
| 138 | + |
| 139 | +class S3Resolver(AssetResolver): |
| 140 | + """Resolver for S3 assets (``s3://`` and S3 HTTPS URLs). |
| 141 | +
|
| 142 | + Uses a ``boto3`` S3 client for ``s3://`` URLs and falls back to plain HTTP |
| 143 | + for S3 HTTPS URLs. |
| 144 | +
|
| 145 | + When constructed without an explicit *s3_client*, a default client is |
| 146 | + lazily created from the environment's AWS credential chain on first |
| 147 | + download. This requires ``boto3`` to be installed |
| 148 | + (``pip install 'earthdaily[download]'``). |
| 149 | +
|
| 150 | + Parameters |
| 151 | + ---------- |
| 152 | + s3_client : |
| 153 | + A ``boto3`` S3 client (``boto3.client("s3")``). If ``None``, one is |
| 154 | + created lazily using default AWS credentials. |
| 155 | + requester_pays : |
| 156 | + Whether to send ``RequestPayer: requester`` for ``s3://`` downloads. |
| 157 | + """ |
| 158 | + |
| 159 | + needs_credentials = False |
| 160 | + |
| 161 | + def __init__( |
| 162 | + self, |
| 163 | + s3_client: Any = None, |
| 164 | + requester_pays: bool = True, |
| 165 | + ) -> None: |
| 166 | + self._s3_client = s3_client |
| 167 | + self._requester_pays = requester_pays |
| 168 | + |
| 169 | + def _get_s3_client(self) -> Any: |
| 170 | + """Return the S3 client, lazily creating a default one if needed.""" |
| 171 | + if self._s3_client is None: |
| 172 | + try: |
| 173 | + import boto3 |
| 174 | + except ImportError: |
| 175 | + raise ImportError( |
| 176 | + "boto3 is required for S3 downloads. Install it with: pip install 'earthdaily[download]'" |
| 177 | + ) |
| 178 | + self._s3_client = boto3.client("s3") |
| 179 | + return self._s3_client |
| 180 | + |
| 181 | + @staticmethod |
| 182 | + def _extract_s3_href(asset_dict: dict) -> str: |
| 183 | + """Prefer ``alternate.s3.href`` over the top-level ``href``.""" |
| 184 | + alt = asset_dict.get("alternate", {}).get("s3", {}).get("href", "") |
| 185 | + return alt if alt else asset_dict.get("href", "") |
| 186 | + |
| 187 | + def can_handle(self, url: str) -> bool: |
| 188 | + parsed = urlparse(url, allow_fragments=False) |
| 189 | + if parsed.scheme == "s3": |
| 190 | + return not parsed.path.endswith("/") |
| 191 | + if parsed.scheme == "https" and _S3_HTTPS_PATTERN.match(parsed.netloc): |
| 192 | + return not parsed.path.endswith("/") |
| 193 | + return False |
| 194 | + |
| 195 | + def get_download_url(self, url: str) -> str: |
| 196 | + return url |
| 197 | + |
| 198 | + def get_headers(self, url: str) -> dict[str, str]: |
| 199 | + return {} |
| 200 | + |
| 201 | + def download( |
| 202 | + self, url: str, output_dir: Path, quiet: bool = False, asset_metadata: dict | None = None |
| 203 | + ) -> Path | None: |
| 204 | + """Download an S3 asset directly via boto3. |
| 205 | +
|
| 206 | + For ``s3://`` URLs this uses ``s3_client.download_file``. For HTTPS |
| 207 | + S3 URLs the method returns ``None`` to fall back to the standard HTTP |
| 208 | + download path. |
| 209 | +
|
| 210 | + When *asset_metadata* contains a ``file:local_path`` field the file is |
| 211 | + saved at ``output_dir / <file:local_path>`` instead of the default |
| 212 | + flat filename derived from the S3 key. |
| 213 | + """ |
| 214 | + parsed = urlparse(url, allow_fragments=False) |
| 215 | + |
| 216 | + if parsed.scheme != "s3": |
| 217 | + return None |
| 218 | + |
| 219 | + bucket = parsed.netloc |
| 220 | + key = parsed.path.lstrip("/") |
| 221 | + |
| 222 | + local_path = (asset_metadata or {}).get("file:local_path") |
| 223 | + if local_path: |
| 224 | + dest = output_dir / local_path |
| 225 | + else: |
| 226 | + dest = output_dir / key |
| 227 | + |
| 228 | + makedirs(dest.parent, exist_ok=True) |
| 229 | + |
| 230 | + extra_args: dict | None = None |
| 231 | + if self._requester_pays: |
| 232 | + extra_args = {"RequestPayer": "requester"} |
| 233 | + |
| 234 | + s3 = self._get_s3_client() |
| 235 | + logger.info("Downloading %s to %s", url, dest) |
| 236 | + |
| 237 | + if quiet: |
| 238 | + s3.download_file(bucket, key, str(dest), ExtraArgs=extra_args) |
| 239 | + else: |
| 240 | + head = s3.head_object(Bucket=bucket, Key=key, **(extra_args or {})) |
| 241 | + total = head["ContentLength"] |
| 242 | + with tqdm(total=total, unit="B", unit_scale=True, desc=dest.name) as pbar: |
| 243 | + s3.download_file(bucket, key, str(dest), ExtraArgs=extra_args, Callback=pbar.update) |
| 244 | + |
| 245 | + return dest |
![internal-python-client-app[bot]](https://avatars.githubusercontent.com/u/50711777?v=4&size=40){kind=avatar}
0 commit comments