Embeddings

POST https://hooshvana.ir/api/v1/embeddings

curl https://hooshvana.ir/api/v1/embeddings \
  -H "Authorization: Bearer $HOOSHVANA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/text-embedding-3-small",
    "input": "هزینه‌ی ارسال سفارش به شهرستان چقدر است؟"
  }'

{
  "object": "list",
  "model": "openai/text-embedding-3-small",
  "data": [
    { "object": "embedding", "index": 0, "embedding": [0.0123, -0.0456, 0.0789] }
  ],
  "usage": { "prompt_tokens": 14, "total_tokens": 14 }
}

import os
from openai import OpenAI

client = OpenAI(base_url="https://hooshvana.ir/api/v1", api_key=os.environ["HOOSHVANA_API_KEY"])

def embed_all(texts, batch_size=100):
    vectors = []
    for start in range(0, len(texts), batch_size):
        batch = texts[start:start + batch_size]
        resp = client.embeddings.create(
            model="openai/text-embedding-3-small",
            input=batch,
        )
        # Sort by index to be safe, then keep only the vectors
        vectors.extend(item.embedding for item in sorted(resp.data, key=lambda d: d.index))
    return vectors

docs = [
    "ارسال به تهران یک روزه انجام می‌شود.",
    "ارسال به شهرستان‌ها دو تا چهار روز کاری زمان می‌برد.",
    "امکان مرجوعی کالا تا هفت روز وجود دارد.",
]
doc_vectors = embed_all(docs)

resp = client.embeddings.create(
    model="openai/text-embedding-3-large",
    input="جستجوی معنایی",
    dimensions=1024,
)
print(len(resp.data[0].embedding))  # 1024

import math

def cosine(a, b):
    dot = sum(x * y for x, y in zip(a, b))
    norm_a = math.sqrt(sum(x * x for x in a))
    norm_b = math.sqrt(sum(y * y for y in b))
    return dot / (norm_a * norm_b)

query = "چند روز طول می‌کشد سفارش به اصفهان برسد؟"
q_vec = client.embeddings.create(
    model="openai/text-embedding-3-small", input=query
).data[0].embedding

ranked = sorted(
    zip(docs, doc_vectors),
    key=lambda pair: cosine(q_vec, pair[1]),
    reverse=True,
)
for text, vec in ranked:
    print(round(cosine(q_vec, vec), 3), text)

import OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://hooshvana.ir/api/v1", apiKey: process.env.HOOSHVANA_API_KEY });

const { data } = await client.embeddings.create({
  model: "openai/text-embedding-3-small",
  input: ["خرید اقساطی", "پرداخت قسطی"],
});

const [a, b] = data.map((d) => d.embedding);
const dot = a.reduce((sum, x, i) => sum + x * b[i], 0);
console.log("similarity:", dot.toFixed(3));

import re

ARABIC_TO_PERSIAN = str.maketrans({
    "ي": "ی", "ى": "ی", "ك": "ک", "ة": "ه",
    "٠": "0", "١": "1", "٢": "2", "٣": "3", "٤": "4",
    "٥": "5", "٦": "6", "٧": "7", "٨": "8", "٩": "9",
    "۰": "0", "۱": "1", "۲": "2", "۳": "3", "۴": "4",
    "۵": "5", "۶": "6", "۷": "7", "۸": "8", "۹": "9",
})

def normalize_fa(text: str) -> str:
    text = text.translate(ARABIC_TO_PERSIAN)
    text = re.sub(r"[\u064B-\u0652]", "", text)  # remove Arabic diacritics
    text = re.sub(r"[ \t]+", " ", text)
    return text.strip()