|
| 1 | +""" |
| 2 | +Endpoint resolution for Contentstack services across all regions. |
| 3 | +
|
| 4 | +Reads a bundled regions.json (src/assets/regions.json) and resolves |
| 5 | +the correct base URL for any region + service combination. No runtime |
| 6 | +HTTP calls — the file is shipped with the package and updated via |
| 7 | +``python scripts/refresh_regions.py``. |
| 8 | +""" |
| 9 | + |
| 10 | +import json |
| 11 | +import os |
| 12 | +import re |
| 13 | +import urllib.request |
| 14 | +from typing import Dict, Optional, Union |
| 15 | + |
| 16 | +_REGIONS_URL = "https://artifacts.contentstack.com/regions.json" |
| 17 | +_ASSETS_DIR = os.path.join(os.path.dirname(__file__), "assets") |
| 18 | +_REGIONS_FILE = os.path.join(_ASSETS_DIR, "regions.json") |
| 19 | + |
| 20 | + |
| 21 | +class Endpoint: |
| 22 | + """ |
| 23 | + Resolve Contentstack service URLs for any region. |
| 24 | +
|
| 25 | + All public methods are static — no instantiation required. |
| 26 | +
|
| 27 | + Example:: |
| 28 | +
|
| 29 | + from contentstack_utils import Endpoint |
| 30 | +
|
| 31 | + # Full URL |
| 32 | + url = Endpoint.get_contentstack_endpoint("na", "contentDelivery") |
| 33 | + # → "https://cdn.contentstack.io" |
| 34 | +
|
| 35 | + # Host only (strip https://) — useful for SDK setHost() calls |
| 36 | + host = Endpoint.get_contentstack_endpoint("eu", "contentDelivery", omit_https=True) |
| 37 | + # → "eu-cdn.contentstack.com" |
| 38 | +
|
| 39 | + # All endpoints for a region |
| 40 | + all_endpoints = Endpoint.get_contentstack_endpoint("azure-na") |
| 41 | + # → {"contentDelivery": "...", "contentManagement": "...", ...} |
| 42 | + """ |
| 43 | + |
| 44 | + # Module-level cache — loaded once per Python process, shared across all calls. |
| 45 | + _regions_data: Optional[Dict] = None |
| 46 | + |
| 47 | + @staticmethod |
| 48 | + def get_contentstack_endpoint( |
| 49 | + region: str = "us", |
| 50 | + service: str = "", |
| 51 | + omit_https: bool = False, |
| 52 | + ) -> Union[str, Dict[str, str]]: |
| 53 | + """ |
| 54 | + Resolve a Contentstack service endpoint URL for the given region. |
| 55 | +
|
| 56 | + :param region: |
| 57 | + Region ID or any accepted alias (case-insensitive, ``-`` and ``_`` |
| 58 | + are interchangeable). Examples: ``'na'``, ``'us'``, ``'eu'``, |
| 59 | + ``'AWS-NA'``, ``'azure_eu'``, ``'gcp-na'``. |
| 60 | + Defaults to ``'us'`` (AWS North America). |
| 61 | + :param service: |
| 62 | + Optional service key. When provided, a single URL string is |
| 63 | + returned. When omitted, a dict of **all** service URLs is returned. |
| 64 | + Valid keys include: ``'contentDelivery'``, ``'contentManagement'``, |
| 65 | + ``'auth'``, ``'graphqlDelivery'``, ``'preview'``, ``'images'``, |
| 66 | + ``'assets'``, ``'automate'``, ``'launch'``, ``'developerHub'``, |
| 67 | + ``'brandKit'``, ``'genAI'``, ``'personalizeManagement'``, |
| 68 | + ``'personalizeEdge'``, ``'composableStudio'``, ``'assetManagement'``. |
| 69 | + :param omit_https: |
| 70 | + When ``True``, strips the ``https://`` (or ``http://``) scheme from |
| 71 | + every returned URL. Useful when passing the host to an SDK that |
| 72 | + constructs its own URLs (e.g. ``stack.set_host(host)``). |
| 73 | + :returns: |
| 74 | + - A ``str`` URL when *service* is specified. |
| 75 | + - A ``dict[str, str]`` mapping service keys → URLs when *service* |
| 76 | + is omitted. |
| 77 | + :raises ValueError: |
| 78 | + If *region* is an empty string. |
| 79 | + :raises LookupError: |
| 80 | + If *region* does not match any known region ID or alias, or if |
| 81 | + *service* is not present in the resolved region's endpoint map. |
| 82 | + :raises RuntimeError: |
| 83 | + If the bundled ``regions.json`` cannot be read or is malformed. |
| 84 | +
|
| 85 | + Examples:: |
| 86 | +
|
| 87 | + Endpoint.get_contentstack_endpoint("na", "contentDelivery") |
| 88 | + # → "https://cdn.contentstack.io" |
| 89 | +
|
| 90 | + Endpoint.get_contentstack_endpoint("eu", "contentDelivery", omit_https=True) |
| 91 | + # → "eu-cdn.contentstack.com" |
| 92 | +
|
| 93 | + Endpoint.get_contentstack_endpoint("azure-na") |
| 94 | + # → {"contentDelivery": "https://...", ...} |
| 95 | + """ |
| 96 | + if not region: |
| 97 | + raise ValueError("Empty region provided. Please put valid region.") |
| 98 | + |
| 99 | + data = Endpoint._load_regions() |
| 100 | + normalized = region.strip().lower() |
| 101 | + |
| 102 | + if not normalized: |
| 103 | + raise ValueError("Empty region provided. Please put valid region.") |
| 104 | + region_row = Endpoint._find_region_by_id_or_alias(data["regions"], normalized) |
| 105 | + |
| 106 | + if region_row is None: |
| 107 | + raise LookupError(f"Invalid region: {region}") |
| 108 | + |
| 109 | + if service: |
| 110 | + endpoints = region_row["endpoints"] |
| 111 | + if service not in endpoints: |
| 112 | + raise LookupError( |
| 113 | + f'Service "{service}" not found for region "{region_row["id"]}"' |
| 114 | + ) |
| 115 | + url = endpoints[service] |
| 116 | + return Endpoint._strip_https(url) if omit_https else url |
| 117 | + |
| 118 | + endpoints = dict(region_row["endpoints"]) |
| 119 | + return Endpoint._strip_https_from_map(endpoints) if omit_https else endpoints |
| 120 | + |
| 121 | + |
| 122 | + getContentstackEndpoint = get_contentstack_endpoint |
| 123 | + |
| 124 | + @staticmethod |
| 125 | + def reset_cache() -> None: |
| 126 | + """ |
| 127 | + Clear the in-memory region cache. |
| 128 | +
|
| 129 | + Intended for testing only — forces the next call to re-read |
| 130 | + ``regions.json`` from disk. |
| 131 | + """ |
| 132 | + Endpoint._regions_data = None |
| 133 | + |
| 134 | + # ------------------------------------------------------------------ |
| 135 | + # Internal helpers |
| 136 | + # ------------------------------------------------------------------ |
| 137 | + |
| 138 | + @staticmethod |
| 139 | + def _load_regions() -> Dict: |
| 140 | + """ |
| 141 | + Load and cache regions data. |
| 142 | +
|
| 143 | + Resolution order: |
| 144 | + 1. In-memory cache (zero I/O after the first call in a process) |
| 145 | + 2. Bundled ``contentstack_utils/assets/regions.json`` on disk |
| 146 | + 3. Live download from ``artifacts.contentstack.com`` (fallback when |
| 147 | + the file is absent — e.g. an editable install without assets) |
| 148 | + """ |
| 149 | + if Endpoint._regions_data is not None: |
| 150 | + return Endpoint._regions_data |
| 151 | + |
| 152 | + if not os.path.exists(_REGIONS_FILE): |
| 153 | + Endpoint._download_and_save(_REGIONS_FILE) |
| 154 | + |
| 155 | + if not os.path.exists(_REGIONS_FILE): |
| 156 | + raise RuntimeError( |
| 157 | + "contentstack_utils: regions.json not found and could not be downloaded. " |
| 158 | + "Run 'python scripts/refresh_regions.py' and ensure network access." |
| 159 | + ) |
| 160 | + |
| 161 | + try: |
| 162 | + with open(_REGIONS_FILE, "r", encoding="utf-8") as fh: |
| 163 | + raw = fh.read() |
| 164 | + except OSError as exc: |
| 165 | + raise RuntimeError( |
| 166 | + f"contentstack_utils: Could not read regions.json: {exc}" |
| 167 | + ) from exc |
| 168 | + |
| 169 | + try: |
| 170 | + decoded = json.loads(raw) |
| 171 | + except json.JSONDecodeError as exc: |
| 172 | + raise RuntimeError( |
| 173 | + "contentstack_utils: regions.json is corrupt. " |
| 174 | + "Run 'python scripts/refresh_regions.py' to re-download it." |
| 175 | + ) from exc |
| 176 | + |
| 177 | + if not isinstance(decoded, dict) or "regions" not in decoded: |
| 178 | + raise RuntimeError( |
| 179 | + "contentstack_utils: regions.json is corrupt. " |
| 180 | + "Run 'python scripts/refresh_regions.py' to re-download it." |
| 181 | + ) |
| 182 | + |
| 183 | + Endpoint._regions_data = decoded |
| 184 | + return Endpoint._regions_data |
| 185 | + |
| 186 | + @staticmethod |
| 187 | + def _download_and_save(dest: str) -> None: |
| 188 | + """ |
| 189 | + Fetch regions.json from the Contentstack CDN and write it to *dest*. |
| 190 | +
|
| 191 | + Silent on failure — the caller decides whether a missing file is fatal. |
| 192 | + """ |
| 193 | + os.makedirs(os.path.dirname(dest), exist_ok=True) |
| 194 | + try: |
| 195 | + with urllib.request.urlopen(_REGIONS_URL, timeout=30) as resp: |
| 196 | + data = resp.read().decode("utf-8") |
| 197 | + except Exception: |
| 198 | + return |
| 199 | + |
| 200 | + try: |
| 201 | + decoded = json.loads(data) |
| 202 | + except json.JSONDecodeError: |
| 203 | + return |
| 204 | + |
| 205 | + if isinstance(decoded, dict) and "regions" in decoded: |
| 206 | + with open(dest, "w", encoding="utf-8") as fh: |
| 207 | + fh.write(data) |
| 208 | + |
| 209 | + @staticmethod |
| 210 | + def _find_region_by_id_or_alias( |
| 211 | + regions: list, normalized_input: str |
| 212 | + ) -> Optional[Dict]: |
| 213 | + """ |
| 214 | + Find a region by its ``id`` field first, then by any alias. |
| 215 | +
|
| 216 | + Both passes are case-insensitive (caller must pass a lowercased string). |
| 217 | + Two-pass approach mirrors the PHP implementation: ID match wins over alias |
| 218 | + match, which avoids surprising behaviour when a future alias happens to |
| 219 | + collide with another region's canonical ID. |
| 220 | + """ |
| 221 | + # Pass 1 — exact id match |
| 222 | + for row in regions: |
| 223 | + if row["id"] == normalized_input: |
| 224 | + return row |
| 225 | + |
| 226 | + # Pass 2 — alias match |
| 227 | + for row in regions: |
| 228 | + for alias in row.get("alias", []): |
| 229 | + if alias.lower() == normalized_input: |
| 230 | + return row |
| 231 | + |
| 232 | + return None |
| 233 | + |
| 234 | + @staticmethod |
| 235 | + def _strip_https(url: str) -> str: |
| 236 | + """Strip ``https://`` or ``http://`` from the start of a URL.""" |
| 237 | + return re.sub(r"^https?://", "", url) |
| 238 | + |
| 239 | + @staticmethod |
| 240 | + def _strip_https_from_map(endpoints: Dict[str, str]) -> Dict[str, str]: |
| 241 | + """Return a new dict with the scheme stripped from every URL value.""" |
| 242 | + return {key: Endpoint._strip_https(url) for key, url in endpoints.items()} |
0 commit comments