> ## Documentation Index
> Fetch the complete documentation index at: https://docs.moorcheh.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Jina Integration

> Use Jina jina-embeddings-v5-text-small with Moorcheh vector namespaces for multilingual semantic search.

## Jina + Moorcheh

This integration uses the **[Jina Embeddings API](https://jina.ai/embeddings)** with **`jina-embeddings-v5-text-small`** and **Moorcheh vector namespaces** to store and search vectors with ITS ranking.

The model produces **1024-dimensional** vectors by default (Matryoshka truncation is available via the API if you need a smaller size and match `vector_dimension` accordingly).

## Architecture

<CardGroup cols={2}>
  <Card title="Embedding generation" icon="brain">
    Call `POST https://api.jina.ai/v1/embeddings` with task `retrieval.passage` (index) or `retrieval.query` (search)
  </Card>

  <Card title="Vector storage" icon="database">
    Store vectors in Moorcheh vector namespaces
  </Card>

  <Card title="Semantic retrieval" icon="magnifying-glass">
    Query with the same model and `retrieval.query` for asymmetric retrieval
  </Card>

  <Card title="Authentication" icon="key">
    `Authorization: Bearer` your Jina API key
  </Card>
</CardGroup>

## Prerequisites

* `MOORCHEH_API_KEY` from the [Moorcheh Console](https://console.moorcheh.ai/)
* `JINA_API_KEY` from the [Jina API dashboard](https://jina.ai/api-dashboard/key-manager)
* Python 3.9+

```bash theme={null}
pip install moorcheh-sdk requests python-dotenv
```

### `.env` file

```bash theme={null}
MOORCHEH_API_KEY=your_moorcheh_key
JINA_API_KEY=your_jina_key
```

## Tasks

For asymmetric retrieval, use:

| Task                | When to use                   |
| ------------------- | ----------------------------- |
| `retrieval.passage` | Chunks or documents you index |
| `retrieval.query`   | User queries at search time   |

Other tasks (for example `text-matching`, `classification`) are supported by the API for different workloads; keep task and model consistent between index and query for retrieval.

## Vector dimensions

Default output for **`jina-embeddings-v5-text-small`** is **1024** dimensions. You can pass a **`dimensions`** field in the API request to truncate (Matryoshka); the Moorcheh namespace **`vector_dimension`** must match what you send at index and query time.

## End-to-end example

The following example loads keys from `.env` via `load_dotenv()`, embeds passages and a query through the Jina API, uploads vectors to Moorcheh, and runs similarity search.

```python theme={null}
import os
import textwrap
from typing import List

import requests
from dotenv import load_dotenv
from moorcheh_sdk import MoorchehClient

load_dotenv()

MOORCHEH_API_KEY = os.getenv("MOORCHEH_API_KEY", "").strip()
JINA_API_KEY = os.getenv("JINA_API_KEY", "").strip()
if not MOORCHEH_API_KEY or not JINA_API_KEY:
    raise SystemExit("Set MOORCHEH_API_KEY and JINA_API_KEY.")

JINA_EMBEDDINGS_URL = "https://api.jina.ai/v1/embeddings"
MODEL = "jina-embeddings-v5-text-small"
VECTOR_DIMENSION = 1024

NAMESPACE = "jina-v5-small-demo"
CHUNK_SIZE = 900
CHUNK_OVERLAP = 180


def jina_embed(texts: List[str], task: str) -> List[List[float]]:
    headers = {
        "Authorization": f"Bearer {JINA_API_KEY}",
        "Content-Type": "application/json",
    }
    payload = {
        "model": MODEL,
        "task": task,
        "input": texts,
        "normalized": True,
    }
    r = requests.post(JINA_EMBEDDINGS_URL, headers=headers, json=payload, timeout=120)
    r.raise_for_status()
    body = r.json()
    items = sorted(body["data"], key=lambda x: x["index"])
    return [item["embedding"] for item in items]


def to_float_vector(values: List[float]) -> List[float]:
    return [float(x) for x in values]


def chunk_text(text: str, chunk_size: int = CHUNK_SIZE, overlap: int = CHUNK_OVERLAP) -> List[str]:
    chunks: List[str] = []
    start = 0
    while start < len(text):
        end = min(start + chunk_size, len(text))
        chunks.append(text[start:end].strip())
        if end == len(text):
            break
        start = max(end - overlap, 0)
    return [c for c in chunks if c]


def extract_text(result: dict) -> str:
    if result.get("text"):
        return str(result["text"])
    metadata = result.get("metadata") or {}
    if isinstance(metadata, dict):
        return str(metadata.get("text") or metadata.get("raw_text") or metadata.get("content") or "")
    return ""


def clean_text(text: str) -> str:
    return " ".join(str(text).split())


def print_result(idx: int, result: dict) -> None:
    metadata = result.get("metadata") or {}
    text_value = clean_text(extract_text(result))
    wrapped = textwrap.fill(text_value, width=100)
    print(f"[{idx}] id={result.get('id')}")
    print(f"score={result.get('score')} label={result.get('label')}")
    print(f"section={metadata.get('section')} source_doc_id={metadata.get('source_doc_id')}")
    print("text:")
    print(wrapped if wrapped else "(no text returned)")
    print("-" * 120)


mc = MoorchehClient(api_key=MOORCHEH_API_KEY)

try:
    mc.namespaces.create(
        namespace_name=NAMESPACE,
        type="vector",
        vector_dimension=VECTOR_DIMENSION,
    )
except Exception:
    pass

source_documents = [
    {
        "id": "guide-vector-namespaces",
        "section": "vector-namespace-best-practices",
        "text": (
            "Moorcheh vector namespaces support bring-your-own-embedding workflows. "
            "With Jina jina-embeddings-v5-text-small, use task retrieval.passage for chunks and retrieval.query for queries. "
            "Match vector_dimension to the API output size (1024 by default)."
        ),
    },
    {
        "id": "guide-search-tuning",
        "section": "semantic-search-tuning",
        "text": (
            "Use coherent chunks with overlap. Tune similarity_search threshold and top_k for your domain."
        ),
    },
]

documents = []
for doc in source_documents:
    parts = chunk_text(doc["text"])
    for idx, chunk in enumerate(parts):
        documents.append(
            {
                "id": f"{doc['id']}-chunk-{idx}",
                "text": chunk,
                "source_doc_id": doc["id"],
                "section": doc["section"],
                "chunk_index": idx,
                "total_chunks": len(parts),
            }
        )

texts = [d["text"] for d in documents]
doc_embeddings = jina_embed(texts, task="retrieval.passage")

mc.vectors.upload(
    namespace_name=NAMESPACE,
    vectors=[
        {
            "id": documents[i]["id"],
            "vector": to_float_vector(doc_embeddings[i]),
            "text": documents[i]["text"],
            "source": "jina-embeddings-v5-text-small",
            "model": MODEL,
            "task": "retrieval.passage",
            "section": documents[i]["section"],
            "source_doc_id": documents[i]["source_doc_id"],
            "chunk_index": documents[i]["chunk_index"],
            "total_chunks": documents[i]["total_chunks"],
        }
        for i in range(len(documents))
    ],
)

query = "How do I use Jina retrieval tasks with Moorcheh vector namespaces?"
query_vecs = jina_embed([query], task="retrieval.query")
query_vec = to_float_vector(query_vecs[0])

results = mc.similarity_search.query(
    namespaces=[NAMESPACE],
    query=query_vec,
    top_k=5,
    kiosk_mode=True,
    threshold=0.15,
)

print(f"namespace={NAMESPACE} total_results={len(results.get('results', []))}")
print("=" * 120)
for idx, r in enumerate(results.get("results", []), start=1):
    print_result(idx, r)
```

## Runnable demo script

See `integrations/jina/jina_moorcheh_demo.py`.

## Important notes

<AccordionGroup>
  <Accordion title="Vector dimension must match">
    Default length is **1024** for `jina-embeddings-v5-text-small`. If you set `dimensions` in the Jina request, create the Moorcheh namespace with that same `vector_dimension`.
  </Accordion>

  <Accordion title="Use passage vs query tasks">
    Use **`retrieval.passage`** for stored chunks and **`retrieval.query`** for search queries so the retrieval adapter matches training.
  </Accordion>

  <Accordion title="Store text on each vector">
    Include `text` on each uploaded vector so search results can return the original chunk.
  </Accordion>

  <Accordion title="Keep model and task consistent">
    Use the same model name and task pairing at index and query time.
  </Accordion>
</AccordionGroup>

## Troubleshooting

* `401` / auth errors: Check `JINA_API_KEY` and `Authorization: Bearer` format.
* `Dimension mismatch`: Align Moorcheh `vector_dimension` with embedding length (default 1024).
* Low relevance: Adjust chunking, `threshold`, and `top_k`.

## Related docs

* [Create Namespace](/api-reference/namespaces/create)
* [Upload Vector Data](/api-reference/data/upload-vector)
* [Search Query](/api-reference/search/query)
* [Gemini integration](/integrations/gemini/overview)
* [Cohere integration](/integrations/cohere/overview)