|
| 1 | +"""High-level Mixpeek client — ergonomic wrapper around the generated SDK.""" |
| 2 | + |
| 3 | +from __future__ import annotations |
| 4 | + |
| 5 | +import json |
| 6 | +import os |
| 7 | +from typing import Any, Dict, List, Optional |
| 8 | + |
| 9 | +import urllib3 |
| 10 | + |
| 11 | + |
| 12 | +class Mixpeek: |
| 13 | + """One-liner client for the Mixpeek API. |
| 14 | +
|
| 15 | + Usage:: |
| 16 | +
|
| 17 | + from mixpeek import Mixpeek |
| 18 | +
|
| 19 | + mp = Mixpeek("sk_xxx", namespace="ns_xxx") |
| 20 | + results = mp.search("red car", collection="products") |
| 21 | + """ |
| 22 | + |
| 23 | + DEFAULT_BASE_URL = "https://api.mixpeek.com/v1" |
| 24 | + |
| 25 | + def __init__( |
| 26 | + self, |
| 27 | + api_key: Optional[str] = None, |
| 28 | + *, |
| 29 | + namespace: Optional[str] = None, |
| 30 | + base_url: Optional[str] = None, |
| 31 | + timeout: float = 30.0, |
| 32 | + ) -> None: |
| 33 | + self.api_key = api_key or os.environ.get("MIXPEEK_API_KEY", "") |
| 34 | + if not self.api_key: |
| 35 | + raise ValueError( |
| 36 | + "An API key is required. Pass api_key= or set MIXPEEK_API_KEY." |
| 37 | + ) |
| 38 | + self.namespace = namespace or os.environ.get("MIXPEEK_NAMESPACE") |
| 39 | + self.base_url = (base_url or self.DEFAULT_BASE_URL).rstrip("/") |
| 40 | + self.timeout = timeout |
| 41 | + self._http = urllib3.PoolManager( |
| 42 | + timeout=urllib3.Timeout(connect=5.0, read=timeout), |
| 43 | + retries=urllib3.Retry(total=2, backoff_factor=0.3), |
| 44 | + ) |
| 45 | + |
| 46 | + # Resource managers (lazy-style but immediate for discoverability) |
| 47 | + from mixpeek._client.resources import ( |
| 48 | + Buckets, |
| 49 | + Collections, |
| 50 | + Documents, |
| 51 | + Namespaces, |
| 52 | + Retrievers, |
| 53 | + ) |
| 54 | + |
| 55 | + self.namespaces = Namespaces(self) |
| 56 | + self.buckets = Buckets(self) |
| 57 | + self.collections = Collections(self) |
| 58 | + self.retrievers = Retrievers(self) |
| 59 | + self.documents = Documents(self) |
| 60 | + |
| 61 | + # ---- Convenience shortcuts ------------------------------------------------ |
| 62 | + |
| 63 | + def search( |
| 64 | + self, |
| 65 | + query: str, |
| 66 | + *, |
| 67 | + collection: Optional[str] = None, |
| 68 | + filters: Optional[Dict[str, Any]] = None, |
| 69 | + limit: int = 10, |
| 70 | + namespace: Optional[str] = None, |
| 71 | + ) -> Dict[str, Any]: |
| 72 | + """Zero-setup semantic search using an adhoc retriever. |
| 73 | +
|
| 74 | + Constructs a ``feature_search`` stage on the fly and executes it |
| 75 | + via ``POST /v1/retrievers/execute`` (adhoc mode). |
| 76 | + """ |
| 77 | + ns = namespace or self.namespace |
| 78 | + stages: List[Dict[str, Any]] = [ |
| 79 | + { |
| 80 | + "type": "feature_search", |
| 81 | + "feature_extractor": {"type": "text"}, |
| 82 | + "query": query, |
| 83 | + "collection_ids": [collection] if collection else [], |
| 84 | + "limit": limit, |
| 85 | + } |
| 86 | + ] |
| 87 | + if filters: |
| 88 | + stages.insert( |
| 89 | + 0, |
| 90 | + { |
| 91 | + "type": "attribute_filter", |
| 92 | + "conditions": filters, |
| 93 | + }, |
| 94 | + ) |
| 95 | + body: Dict[str, Any] = { |
| 96 | + "stages": stages, |
| 97 | + "inputs": {"query": query}, |
| 98 | + } |
| 99 | + return self._request( |
| 100 | + "POST", "/retrievers/execute", body=body, namespace=ns |
| 101 | + ) |
| 102 | + |
| 103 | + def index( |
| 104 | + self, |
| 105 | + source: str, |
| 106 | + *, |
| 107 | + collection: Optional[str] = None, |
| 108 | + metadata: Optional[Dict[str, Any]] = None, |
| 109 | + namespace: Optional[str] = None, |
| 110 | + ) -> Dict[str, Any]: |
| 111 | + """Upload a file URL to a bucket and optionally trigger a collection. |
| 112 | +
|
| 113 | + ``source`` can be an S3/GCS URI or any public URL. |
| 114 | + If ``collection`` is provided the collection is triggered after upload. |
| 115 | + """ |
| 116 | + ns = namespace or self.namespace |
| 117 | + |
| 118 | + # Find or pick the first bucket in the namespace |
| 119 | + buckets = self._request("POST", "/buckets/list", namespace=ns) |
| 120 | + if not buckets: |
| 121 | + raise ValueError("No buckets found in the target namespace.") |
| 122 | + bucket_id = buckets[0]["bucket_id"] if isinstance(buckets, list) else buckets["results"][0]["bucket_id"] |
| 123 | + |
| 124 | + upload_body: Dict[str, Any] = {"blob": {"url": source}} |
| 125 | + if metadata: |
| 126 | + upload_body["metadata"] = metadata |
| 127 | + |
| 128 | + result = self._request( |
| 129 | + "POST", f"/buckets/{bucket_id}/objects", body=upload_body, namespace=ns |
| 130 | + ) |
| 131 | + |
| 132 | + if collection: |
| 133 | + self._request( |
| 134 | + "POST", f"/collections/{collection}/trigger", namespace=ns |
| 135 | + ) |
| 136 | + |
| 137 | + return result |
| 138 | + |
| 139 | + # ---- HTTP helper ---------------------------------------------------------- |
| 140 | + |
| 141 | + def _request( |
| 142 | + self, |
| 143 | + method: str, |
| 144 | + path: str, |
| 145 | + *, |
| 146 | + body: Optional[Any] = None, |
| 147 | + namespace: Optional[str] = None, |
| 148 | + ) -> Any: |
| 149 | + url = f"{self.base_url}{path}" |
| 150 | + headers = { |
| 151 | + "Authorization": f"Bearer {self.api_key}", |
| 152 | + "Content-Type": "application/json", |
| 153 | + "Accept": "application/json", |
| 154 | + } |
| 155 | + ns = namespace or self.namespace |
| 156 | + if ns: |
| 157 | + headers["X-Namespace"] = ns |
| 158 | + |
| 159 | + encoded_body = json.dumps(body).encode("utf-8") if body is not None else None |
| 160 | + |
| 161 | + resp = self._http.request( |
| 162 | + method, |
| 163 | + url, |
| 164 | + body=encoded_body, |
| 165 | + headers=headers, |
| 166 | + ) |
| 167 | + |
| 168 | + if resp.status >= 400: |
| 169 | + try: |
| 170 | + detail = json.loads(resp.data.decode("utf-8")) |
| 171 | + except Exception: |
| 172 | + detail = resp.data.decode("utf-8", errors="replace") |
| 173 | + raise MixpeekAPIError(resp.status, detail) |
| 174 | + |
| 175 | + if not resp.data: |
| 176 | + return None |
| 177 | + |
| 178 | + return json.loads(resp.data.decode("utf-8")) |
| 179 | + |
| 180 | + |
| 181 | +class MixpeekAPIError(Exception): |
| 182 | + """Raised when the Mixpeek API returns an error response.""" |
| 183 | + |
| 184 | + def __init__(self, status: int, detail: Any) -> None: |
| 185 | + self.status = status |
| 186 | + self.detail = detail |
| 187 | + super().__init__(f"Mixpeek API error {status}: {detail}") |
0 commit comments