From 4b081fe983f24d594e6ae084e6fcc9df675fa85f Mon Sep 17 00:00:00 2001
From: John Kupchanko <john.kupchanko@qdrant.com>
Date: Wed, 5 Aug 2026 16:03:24 -0700
Subject: [PATCH 1/5] Add mxbai + hybrid search to the backend (no reranker)

Semantic and keyword now run on the 1024d mxbai model via Cloud inference, and a
hybrid mode fuses dense + keyword with RRF. Keeps the existing neural= query
param working (neural=false -> keyword, neural=true -> hybrid) so the current
frontend needs no change; explicit mode= is also supported. Bumps qdrant-client
for Cloud inference. Frontend untouched.
---
 Dockerfile                     |  4 +-
 qdrant_demo/config.py          | 17 +++++++--
 qdrant_demo/neural_searcher.py | 70 ++++++++++++++++++++++++++--------
 qdrant_demo/service.py         | 42 ++++++++++++++++----
 qdrant_demo/sparse.py          | 26 +++++++++++++
 5 files changed, 132 insertions(+), 27 deletions(-)
 create mode 100644 qdrant_demo/sparse.py

diff --git a/Dockerfile b/Dockerfile
index aeade05..b5d1ea5 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -38,7 +38,9 @@ COPY ./poetry.lock /app
 COPY --from=build-step /app/dist /app/static
 
 RUN poetry install --no-interaction --no-ansi --no-root --without dev
-RUN python -c 'from fastembed.embedding import DefaultEmbedding; DefaultEmbedding("sentence-transformers/all-MiniLM-L6-v2")'
+# Bump the client past the lock so Cloud inference (mxbai) and the hybrid query
+# API work. The query is embedded server-side, so no local model download needed.
+RUN pip install -U "qdrant-client==1.18.0"
 
 # Finally copy the application source code and install root
 COPY qdrant_demo /app/qdrant_demo
diff --git a/qdrant_demo/config.py b/qdrant_demo/config.py
index 5b6567c..63cdc67 100644
--- a/qdrant_demo/config.py
+++ b/qdrant_demo/config.py
@@ -8,7 +8,18 @@
 QDRANT_URL = os.environ.get("QDRANT_URL", "http://localhost:6333/")
 QDRANT_API_KEY = os.environ.get("QDRANT_API_KEY", "")
 
-COLLECTION_NAME = os.environ.get("COLLECTION_NAME", "text-demo")
-EMBEDDINGS_MODEL = os.environ.get("EMBEDDINGS_MODEL", "sentence-transformers/all-MiniLM-L6-v2")
+COLLECTION_NAME = os.environ.get("COLLECTION_NAME", "startups_hybrid")
+EMBEDDINGS_MODEL = os.environ.get("EMBEDDINGS_MODEL", "mixedbread-ai/mxbai-embed-large-v1")
 
-TEXT_FIELD_NAME = "document"
+TEXT_FIELD_NAME = os.environ.get("TEXT_FIELD_NAME", "description")
+
+# Named vectors on the hybrid collection. Leave DENSE_VECTOR_NAME empty for a
+# collection with a single unnamed vector.
+DENSE_VECTOR_NAME = os.environ.get("DENSE_VECTOR_NAME", "dense")
+SPARSE_VECTOR_NAME = os.environ.get("SPARSE_VECTOR_NAME", "sparse")
+
+# Embed the query with Qdrant Cloud server-side inference instead of downloading
+# the model locally. Parse leniently: some hosts keep the surrounding quotes.
+CLOUD_INFERENCE = os.environ.get("CLOUD_INFERENCE", "0").strip().strip('"').strip("'").lower() in ("1", "true", "yes")
+RESULT_LIMIT = int(os.environ.get("RESULT_LIMIT", "20"))
+HYBRID_PREFETCH = int(os.environ.get("HYBRID_PREFETCH", "40"))
diff --git a/qdrant_demo/neural_searcher.py b/qdrant_demo/neural_searcher.py
index cc8c945..5060db2 100644
--- a/qdrant_demo/neural_searcher.py
+++ b/qdrant_demo/neural_searcher.py
@@ -1,27 +1,65 @@
+import os
 import time
-from typing import List
 
 from qdrant_client import QdrantClient, models
 
-from qdrant_demo.config import QDRANT_URL, QDRANT_API_KEY, EMBEDDINGS_MODEL
+from qdrant_demo.config import (
+    QDRANT_URL, QDRANT_API_KEY, EMBEDDINGS_MODEL,
+    DENSE_VECTOR_NAME, SPARSE_VECTOR_NAME, RESULT_LIMIT, HYBRID_PREFETCH,
+    CLOUD_INFERENCE,
+)
+from qdrant_demo.sparse import to_sparse
 
 
 class NeuralSearcher:
+    """Dense (semantic) and hybrid (dense + keyword, fused with RRF) search over a
+    collection with a named dense vector and a sparse keyword vector."""
 
     def __init__(self, collection_name: str):
         self.collection_name = collection_name
-        self.qdrant_client = QdrantClient(url=QDRANT_URL, api_key=QDRANT_API_KEY, prefer_grpc=True)
-
-    def search(self, text: str, filter_: dict = None) -> List[dict]:
-        start_time = time.time()
-        hits = self.qdrant_client.query_points(
-            collection_name=self.collection_name,
-            query=models.Document(
-                text=text,
-                model=EMBEDDINGS_MODEL,
-            ),
-            query_filter=models.Filter(**filter_) if filter_ else None,
-            limit=5
+        # Generous timeout: the first Cloud-inference call after idle has to load
+        # the model server-side, which can take longer than the default timeout.
+        timeout = int(os.environ.get("QDRANT_TIMEOUT", "60"))
+        self.qdrant_client = QdrantClient(
+            url=QDRANT_URL, api_key=QDRANT_API_KEY,
+            cloud_inference=CLOUD_INFERENCE, timeout=timeout,
         )
-        print(f"Search took {time.time() - start_time} seconds")
-        return [hit.payload for hit in hits.points]
+
+    def _dense(self, text: str):
+        return models.Document(text=text, model=EMBEDDINGS_MODEL)
+
+    def search(self, text: str, hybrid: bool = False) -> dict:
+        using = DENSE_VECTOR_NAME or None
+        t0 = time.perf_counter()
+
+        if hybrid:
+            idx, val = to_sparse(text)
+            hits = self.qdrant_client.query_points(
+                collection_name=self.collection_name,
+                prefetch=[
+                    models.Prefetch(query=self._dense(text), using=using, limit=HYBRID_PREFETCH),
+                    models.Prefetch(
+                        query=models.SparseVector(indices=idx, values=val),
+                        using=SPARSE_VECTOR_NAME, limit=HYBRID_PREFETCH,
+                    ),
+                ],
+                query=models.FusionQuery(fusion=models.Fusion.RRF),
+                limit=RESULT_LIMIT,
+            ).points
+        else:
+            hits = self.qdrant_client.query_points(
+                collection_name=self.collection_name,
+                query=self._dense(text),
+                using=using,
+                limit=RESULT_LIMIT,
+            ).points
+
+        latency_ms = round((time.perf_counter() - t0) * 1000)
+        results = [{**hit.payload, "score": hit.score} for hit in hits]
+        stats = {
+            "mode": "hybrid" if hybrid else "semantic",
+            "embedding_model": EMBEDDINGS_MODEL,
+            "latency_ms": latency_ms,
+            "results": len(results),
+        }
+        return {"results": results, "stats": stats}
diff --git a/qdrant_demo/service.py b/qdrant_demo/service.py
index 3248bd0..a1e4fad 100644
--- a/qdrant_demo/service.py
+++ b/qdrant_demo/service.py
@@ -1,9 +1,12 @@
 import os
+from typing import Optional
 
-from fastapi import FastAPI
+from fastapi import FastAPI, HTTPException
 from fastapi.staticfiles import StaticFiles
 
-from qdrant_demo.config import COLLECTION_NAME, STATIC_DIR
+from qdrant_demo.config import (
+    COLLECTION_NAME, STATIC_DIR, RESULT_LIMIT, CLOUD_INFERENCE, EMBEDDINGS_MODEL,
+)
 from qdrant_demo.neural_searcher import NeuralSearcher
 from qdrant_demo.text_searcher import TextSearcher
 
@@ -24,11 +27,36 @@
 
 
 @app.get("/api/search")
-async def read_item(q: str, neural: bool = True):
-    return {
-        "result": neural_searcher.search(text=q)
-        if neural else text_searcher.search(query=q)
-    }
+async def read_item(q: str, mode: Optional[str] = None, neural: Optional[bool] = None):
+    """mode = semantic (dense) | keyword (full-text) | hybrid (dense + keyword).
+
+    Back-compat: the older frontend passes `neural` (bool). Map it so that
+    neural=false is keyword and neural=true is hybrid (dense + keyword), which is
+    the stronger default. Explicit `mode` always wins."""
+    if mode is None:
+        mode = "keyword" if neural is False else "hybrid"
+    if not q.strip():
+        return {"result": [], "stats": {"mode": mode}}
+    try:
+        if mode == "keyword":
+            return {"result": text_searcher.search(query=q, top=RESULT_LIMIT), "stats": {"mode": "keyword"}}
+        out = neural_searcher.search(text=q, hybrid=(mode == "hybrid"))
+        return {"result": out["results"], "stats": out["stats"]}
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"{type(e).__name__}: {str(e)[:300]}")
+
+
+@app.get("/api/stats")
+async def stats():
+    """Live collection size, so the frontend can show off the scale."""
+    try:
+        count = neural_searcher.qdrant_client.count(COLLECTION_NAME).count
+        return {
+            "count": count, "collection": COLLECTION_NAME,
+            "cloud_inference": CLOUD_INFERENCE, "model": EMBEDDINGS_MODEL,
+        }
+    except Exception as e:
+        return {"count": None, "error": f"{type(e).__name__}: {str(e)[:120]}"}
 
 
 # Mount the static files directory once the search endpoint is defined
diff --git a/qdrant_demo/sparse.py b/qdrant_demo/sparse.py
new file mode 100644
index 0000000..d24200a
--- /dev/null
+++ b/qdrant_demo/sparse.py
@@ -0,0 +1,26 @@
+"""Keyword sparse encoder for hybrid search. Tokenize, term-frequency, stable
+hash to a u32 index. Qdrant applies IDF at query time (sparse index modifier=IDF),
+so index-time values carry the term-frequency component only. The same function
+runs for both documents and queries so tokenization stays consistent."""
+import re
+import zlib
+import math
+from collections import Counter
+
+_STOP = set(
+    "the a an and or of to in for on with is are was were be been being by at from as "
+    "it its this that these those i you he she we they your our their his her".split()
+)
+
+
+def tokenize(text):
+    return [w for w in re.findall(r"[a-z0-9]+", (text or "").lower()) if len(w) > 1 and w not in _STOP]
+
+
+def to_sparse(text):
+    tf = Counter(tokenize(text))
+    indices, values = [], []
+    for tok, c in tf.items():
+        indices.append(zlib.crc32(tok.encode("utf-8")) & 0x7FFFFFFF)
+        values.append(1.0 + math.log(c))
+    return indices, values

From 9f4db074f3c807dfcf599cc5ed58a1af2a1f94cf Mon Sep 17 00:00:00 2001
From: John Kupchanko <john.kupchanko@qdrant.com>
Date: Wed, 5 Aug 2026 21:28:47 -0700
Subject: [PATCH 2/5] Address review feedback on the hybrid search backend

init_collection_startups now builds the collection the search path expects:
a named dense (mxbai) vector, a sparse keyword vector with IDF, and the text
index, all from the demo data, so the collection is reproducible instead of
assumed.

neural_searcher prepends the mxbai query prompt and falls back to dense-only
when a query has no keyword tokens, and it reports the mode that actually ran.

service maps the legacy neural flag back to semantic, presents the payload
under the keys the frontend reads, and returns a generic error to the client
while logging the detail on the server.

config defaults cloud inference on and parses the flag even when the host
keeps the surrounding quotes.

sparse keeps single-character and symbol tokens like c and c++ and uses the
full 32-bit index range.

Dockerfile drops the unneeded client bump.
---
 Dockerfile                              |   3 -
 qdrant_demo/config.py                   |   7 +-
 qdrant_demo/init_collection_startups.py | 133 +++++++++++-------------
 qdrant_demo/neural_searcher.py          |  39 ++++---
 qdrant_demo/service.py                  |  40 +++++--
 qdrant_demo/sparse.py                   |  14 ++-
 6 files changed, 135 insertions(+), 101 deletions(-)

diff --git a/Dockerfile b/Dockerfile
index b5d1ea5..1dd7882 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -38,9 +38,6 @@ COPY ./poetry.lock /app
 COPY --from=build-step /app/dist /app/static
 
 RUN poetry install --no-interaction --no-ansi --no-root --without dev
-# Bump the client past the lock so Cloud inference (mxbai) and the hybrid query
-# API work. The query is embedded server-side, so no local model download needed.
-RUN pip install -U "qdrant-client==1.18.0"
 
 # Finally copy the application source code and install root
 COPY qdrant_demo /app/qdrant_demo
diff --git a/qdrant_demo/config.py b/qdrant_demo/config.py
index 63cdc67..652d5f6 100644
--- a/qdrant_demo/config.py
+++ b/qdrant_demo/config.py
@@ -18,8 +18,9 @@
 DENSE_VECTOR_NAME = os.environ.get("DENSE_VECTOR_NAME", "dense")
 SPARSE_VECTOR_NAME = os.environ.get("SPARSE_VECTOR_NAME", "sparse")
 
-# Embed the query with Qdrant Cloud server-side inference instead of downloading
-# the model locally. Parse leniently: some hosts keep the surrounding quotes.
-CLOUD_INFERENCE = os.environ.get("CLOUD_INFERENCE", "0").strip().strip('"').strip("'").lower() in ("1", "true", "yes")
+# Embed the query with Qdrant Cloud server-side inference. Defaults ON: the query
+# model is a 1024-d mxbai, too heavy to embed per-request on a small CPU box.
+# Parse leniently since some hosts keep the surrounding quotes on the value.
+CLOUD_INFERENCE = os.environ.get("CLOUD_INFERENCE", "1").strip().strip('"').strip("'").lower() in ("1", "true", "yes")
 RESULT_LIMIT = int(os.environ.get("RESULT_LIMIT", "20"))
 HYBRID_PREFETCH = int(os.environ.get("HYBRID_PREFETCH", "40"))
diff --git a/qdrant_demo/init_collection_startups.py b/qdrant_demo/init_collection_startups.py
index 176a3ce..e75648f 100644
--- a/qdrant_demo/init_collection_startups.py
+++ b/qdrant_demo/init_collection_startups.py
@@ -1,99 +1,90 @@
+"""Build the startups collection the search path expects: a named `dense` vector
+(mxbai) plus a `sparse` keyword vector with IDF, and a text index for keyword
+search. Documents are embedded with no prefix (mxbai is asymmetric; the query
+prefix is added at search time), so the same text drives dense, sparse, and
+keyword. Run:  python -m qdrant_demo.init_collection_startups
+"""
 import json
-import os.path
+import os
 from typing import Iterable
 
 from qdrant_client import QdrantClient, models
+from fastembed import TextEmbedding
 from tqdm import tqdm
 
-from qdrant_demo.config import DATA_DIR, QDRANT_URL, QDRANT_API_KEY, COLLECTION_NAME, TEXT_FIELD_NAME, EMBEDDINGS_MODEL
+from qdrant_demo.config import (
+    DATA_DIR, QDRANT_URL, QDRANT_API_KEY, COLLECTION_NAME, TEXT_FIELD_NAME,
+    EMBEDDINGS_MODEL, DENSE_VECTOR_NAME, SPARSE_VECTOR_NAME,
+)
+from qdrant_demo.sparse import to_sparse
 
+DENSE_DIM = 1024  # mxbai-embed-large-v1
 
-def read_points() -> Iterable[models.PointStruct]:
-    payload_path = os.path.join(DATA_DIR, 'startups_demo.json')
-    with open(payload_path) as fd:
-        for idx, line in enumerate(fd):
-            obj = json.loads(line)
 
-            # Rename fields to unified schema
-            obj["logo_url"] = obj.pop("images")
-            obj["homepage_url"] = obj.pop("link")
-            obj["document"] = obj.pop("description")
-            yield models.PointStruct(
-                id=idx,
-                vector=models.Document(
-                    text=obj["document"],
-                    model=EMBEDDINGS_MODEL,
-                ),
-                payload=obj,
-            )
+def _records() -> Iterable[dict]:
+    path = os.path.join(DATA_DIR, "startups_demo.json")
+    with open(path, encoding="utf-8") as fd:
+        for line in fd:
+            line = line.strip()
+            if line:
+                yield json.loads(line)
 
 
-def upload_embeddings():
-    client = QdrantClient(
-        url=QDRANT_URL,
-        api_key=QDRANT_API_KEY,
-        prefer_grpc=True,
-    )
+def _doc_text(obj: dict) -> str:
+    # Same text the searcher will match against: name + description.
+    return f"{obj.get('name', '')}. {obj.get(TEXT_FIELD_NAME) or obj.get('description', '')}".strip()
 
-    client.set_model(EMBEDDINGS_MODEL)
 
-    payload_path = os.path.join(DATA_DIR, 'startups_demo.json')
-    payload = []
-    documents = []
-
-    with open(payload_path) as fd:
-        for line in fd:
-            obj = json.loads(line)
-            # Rename fields to unified schema
-            documents.append(obj.pop('description'))
-            obj["logo_url"] = obj.pop("images")
-            obj["homepage_url"] = obj.pop("link")
-            payload.append(obj)
+def build():
+    client = QdrantClient(url=QDRANT_URL, api_key=QDRANT_API_KEY)
+    embedder = TextEmbedding(EMBEDDINGS_MODEL)  # local dense embedding for ingest
 
     if client.collection_exists(COLLECTION_NAME):
-        print(f"Collection {COLLECTION_NAME} already exists. Remove it first.")
+        print(f"{COLLECTION_NAME} exists, recreating.")
         client.delete_collection(COLLECTION_NAME)
 
     client.create_collection(
-        collection_name=COLLECTION_NAME,
-        vectors_config=models.VectorParams(
-            size=client.get_embedding_size(EMBEDDINGS_MODEL),
-            distance=models.Distance.COSINE,
-            on_disk=True,
-        ),
-        # Quantization is optional, but it can significantly reduce the memory usage
+        COLLECTION_NAME,
+        vectors_config={
+            DENSE_VECTOR_NAME: models.VectorParams(
+                size=DENSE_DIM, distance=models.Distance.COSINE, on_disk=True,
+            )
+        },
+        sparse_vectors_config={
+            # IDF is applied at query time; sparse values carry term frequency only.
+            SPARSE_VECTOR_NAME: models.SparseVectorParams(modifier=models.Modifier.IDF)
+        },
         quantization_config=models.ScalarQuantization(
             scalar=models.ScalarQuantizationConfig(
-                type=models.ScalarType.INT8,
-                quantile=0.99,
-                always_ram=True
+                type=models.ScalarType.INT8, quantile=0.99, always_ram=True,
             )
-        )
+        ),
     )
-
-    # Create a payload index for text field.
-    # This index enables text search by the TEXT_FIELD_NAME field.
     client.create_payload_index(
-        collection_name=COLLECTION_NAME,
-        field_name=TEXT_FIELD_NAME,
+        COLLECTION_NAME, field_name=TEXT_FIELD_NAME,
         field_schema=models.TextIndexParams(
-            type=models.TextIndexType.TEXT,
-            tokenizer=models.TokenizerType.WORD,
-            min_token_len=2,
-            max_token_len=20,
-            lowercase=True,
-        )
+            type=models.TextIndexType.TEXT, tokenizer=models.TokenizerType.WORD,
+            min_token_len=2, max_token_len=20, lowercase=True,
+        ),
     )
 
-    # Upload points to the collection
-    # Embeddings will be automatically generated from the Document model
-    client.upload_points(
-        collection_name=COLLECTION_NAME,
-        points=tqdm(read_points()),
-        parallel=4,
-        batch_size=16,
-    )
+    def points() -> Iterable[models.PointStruct]:
+        for idx, obj in enumerate(_records()):
+            text = _doc_text(obj)
+            dense = next(iter(embedder.embed([text]))).tolist()
+            s_idx, s_val = to_sparse(text)
+            yield models.PointStruct(
+                id=idx,
+                vector={
+                    DENSE_VECTOR_NAME: dense,
+                    SPARSE_VECTOR_NAME: models.SparseVector(indices=s_idx, values=s_val),
+                },
+                payload=obj,  # original keys; the API maps them to the frontend schema
+            )
+
+    client.upload_points(COLLECTION_NAME, points=tqdm(points()), batch_size=64)
+    print(f"built {COLLECTION_NAME}: {client.count(COLLECTION_NAME).count} points")
 
 
-if __name__ == '__main__':
-    upload_embeddings()
+if __name__ == "__main__":
+    build()
diff --git a/qdrant_demo/neural_searcher.py b/qdrant_demo/neural_searcher.py
index 5060db2..d84e8f2 100644
--- a/qdrant_demo/neural_searcher.py
+++ b/qdrant_demo/neural_searcher.py
@@ -25,19 +25,34 @@ def __init__(self, collection_name: str):
             cloud_inference=CLOUD_INFERENCE, timeout=timeout,
         )
 
+    # mxbai is an asymmetric retrieval model: the query gets a prompt prefix, the
+    # stored documents do not, so applying it only here needs no re-indexing.
+    # Note: Qdrant Cloud inference already applies this prompt server-side, so on
+    # the Cloud path the prefix is a verified no-op (identical scores). It matters
+    # for the self-hosted / local-embedding path, where nothing adds it otherwise.
+    QUERY_PREFIX = "Represent this sentence for searching relevant passages: "
+
     def _dense(self, text: str):
-        return models.Document(text=text, model=EMBEDDINGS_MODEL)
+        return models.Document(text=self.QUERY_PREFIX + text, model=EMBEDDINGS_MODEL)
+
+    def _dense_only(self, text: str):
+        return self.qdrant_client.query_points(
+            collection_name=self.collection_name,
+            query=self._dense(text),
+            using=DENSE_VECTOR_NAME or None,
+            limit=RESULT_LIMIT,
+        ).points
 
     def search(self, text: str, hybrid: bool = False) -> dict:
-        using = DENSE_VECTOR_NAME or None
         t0 = time.perf_counter()
+        mode = "hybrid" if hybrid else "semantic"
 
-        if hybrid:
-            idx, val = to_sparse(text)
+        idx, val = to_sparse(text) if hybrid else ([], [])
+        if hybrid and idx:
             hits = self.qdrant_client.query_points(
                 collection_name=self.collection_name,
                 prefetch=[
-                    models.Prefetch(query=self._dense(text), using=using, limit=HYBRID_PREFETCH),
+                    models.Prefetch(query=self._dense(text), using=DENSE_VECTOR_NAME or None, limit=HYBRID_PREFETCH),
                     models.Prefetch(
                         query=models.SparseVector(indices=idx, values=val),
                         using=SPARSE_VECTOR_NAME, limit=HYBRID_PREFETCH,
@@ -47,18 +62,18 @@ def search(self, text: str, hybrid: bool = False) -> dict:
                 limit=RESULT_LIMIT,
             ).points
         else:
-            hits = self.qdrant_client.query_points(
-                collection_name=self.collection_name,
-                query=self._dense(text),
-                using=using,
-                limit=RESULT_LIMIT,
-            ).points
+            # No usable keyword tokens (e.g. a stopword-only query) -> dense only.
+            hits = self._dense_only(text)
+            if hybrid:
+                mode = "semantic"  # honestly report what actually ran
 
         latency_ms = round((time.perf_counter() - t0) * 1000)
         results = [{**hit.payload, "score": hit.score} for hit in hits]
         stats = {
-            "mode": "hybrid" if hybrid else "semantic",
+            "mode": mode,
             "embedding_model": EMBEDDINGS_MODEL,
+            # RRF fusion scores (~1/60) are not on the same scale as cosine (~0..1).
+            "score_type": "rrf" if mode == "hybrid" else "cosine",
             "latency_ms": latency_ms,
             "results": len(results),
         }
diff --git a/qdrant_demo/service.py b/qdrant_demo/service.py
index a1e4fad..a679ac5 100644
--- a/qdrant_demo/service.py
+++ b/qdrant_demo/service.py
@@ -1,4 +1,5 @@
 import os
+import logging
 from typing import Optional
 
 from fastapi import FastAPI, HTTPException
@@ -12,6 +13,8 @@
 
 from fastapi.middleware.cors import CORSMiddleware
 
+logger = logging.getLogger(__name__)
+
 app = FastAPI()
 
 app.add_middleware(
@@ -26,24 +29,42 @@
 text_searcher = TextSearcher(collection_name=COLLECTION_NAME)
 
 
+def _present(items):
+    """Map stored payload keys to the schema the frontend reads, without renaming
+    anything in the collection. Keeps unknown keys too."""
+    out = []
+    for it in items:
+        r = dict(it)
+        if "document" not in r and "description" in r:
+            r["document"] = r.get("highlight", r["description"])
+        if "logo_url" not in r and "images" in r:
+            r["logo_url"] = r["images"]
+        if "homepage_url" not in r and "link" in r:
+            r["homepage_url"] = r["link"]
+        out.append(r)
+    return out
+
+
 @app.get("/api/search")
 async def read_item(q: str, mode: Optional[str] = None, neural: Optional[bool] = None):
     """mode = semantic (dense) | keyword (full-text) | hybrid (dense + keyword).
 
-    Back-compat: the older frontend passes `neural` (bool). Map it so that
-    neural=false is keyword and neural=true is hybrid (dense + keyword), which is
-    the stronger default. Explicit `mode` always wins."""
+    Back-compat with the older frontend, which passes `neural` (bool):
+    neural=true -> semantic (its original meaning), neural=false -> keyword.
+    When neither is given, default to hybrid. Explicit `mode` always wins."""
     if mode is None:
-        mode = "keyword" if neural is False else "hybrid"
+        mode = "semantic" if neural is True else "keyword" if neural is False else "hybrid"
     if not q.strip():
         return {"result": [], "stats": {"mode": mode}}
     try:
         if mode == "keyword":
-            return {"result": text_searcher.search(query=q, top=RESULT_LIMIT), "stats": {"mode": "keyword"}}
+            return {"result": _present(text_searcher.search(query=q, top=RESULT_LIMIT)),
+                    "stats": {"mode": "keyword"}}
         out = neural_searcher.search(text=q, hybrid=(mode == "hybrid"))
-        return {"result": out["results"], "stats": out["stats"]}
+        return {"result": _present(out["results"]), "stats": out["stats"]}
     except Exception as e:
-        raise HTTPException(status_code=500, detail=f"{type(e).__name__}: {str(e)[:300]}")
+        logger.exception("search failed for q=%r mode=%s", q, mode)
+        raise HTTPException(status_code=502, detail="Search is temporarily unavailable.")
 
 
 @app.get("/api/stats")
@@ -55,8 +76,9 @@ async def stats():
             "count": count, "collection": COLLECTION_NAME,
             "cloud_inference": CLOUD_INFERENCE, "model": EMBEDDINGS_MODEL,
         }
-    except Exception as e:
-        return {"count": None, "error": f"{type(e).__name__}: {str(e)[:120]}"}
+    except Exception:
+        logger.exception("stats failed")
+        raise HTTPException(status_code=502, detail="Stats are temporarily unavailable.")
 
 
 # Mount the static files directory once the search endpoint is defined
diff --git a/qdrant_demo/sparse.py b/qdrant_demo/sparse.py
index d24200a..d7b07b6 100644
--- a/qdrant_demo/sparse.py
+++ b/qdrant_demo/sparse.py
@@ -1,7 +1,12 @@
 """Keyword sparse encoder for hybrid search. Tokenize, term-frequency, stable
 hash to a u32 index. Qdrant applies IDF at query time (sparse index modifier=IDF),
 so index-time values carry the term-frequency component only. The same function
-runs for both documents and queries so tokenization stays consistent."""
+runs for both documents and queries so tokenization stays consistent.
+
+This is a small self-contained encoder so the demo has no extra model download.
+The idiomatic replacement is Qdrant's BM25 sparse model (`Qdrant/bm25`), which
+adds stemming and a tuned scoring; switching to it is a follow-up because it means
+re-indexing the collection."""
 import re
 import zlib
 import math
@@ -14,13 +19,16 @@
 
 
 def tokenize(text):
-    return [w for w in re.findall(r"[a-z0-9]+", (text or "").lower()) if len(w) > 1 and w not in _STOP]
+    # Keep single characters (languages like "c"/"r", versions like "3"); drop
+    # only stopwords. Dropping short tokens silently lost real query terms.
+    return [w for w in re.findall(r"[a-z0-9+#]+", (text or "").lower()) if w not in _STOP]
 
 
 def to_sparse(text):
     tf = Counter(tokenize(text))
     indices, values = [], []
     for tok, c in tf.items():
-        indices.append(zlib.crc32(tok.encode("utf-8")) & 0x7FFFFFFF)
+        # Qdrant sparse indices are unsigned 32-bit; use the full width.
+        indices.append(zlib.crc32(tok.encode("utf-8")) & 0xFFFFFFFF)
         values.append(1.0 + math.log(c))
     return indices, values

From 20c97b3c539173ca25450048f3d77dd7e5bb2f6a Mon Sep 17 00:00:00 2001
From: John Kupchanko <john.kupchanko@qdrant.com>
Date: Fri, 7 Aug 2026 08:05:55 -0700
Subject: [PATCH 3/5] Switch sparse search to Qdrant/bm25 and rename payload at
 ingest

Both the dense (mxbai) and sparse (bm25) vectors are now embedded server-side
via Cloud inference, so the ingest and query sides use the identical models by
construction. This removes the hand-rolled sparse encoder and its consistency
caveats.

init renames the payload once to the schema the frontend reads (document,
logo_url, homepage_url), and the search path returns it directly with no
per-request mapping. bm25 also handles degenerate queries (stopword-only,
punctuation) without the previous empty-vector crash.
---
 qdrant_demo/config.py                   |  5 ++-
 qdrant_demo/init_collection_startups.py | 41 ++++++++++++++-----------
 qdrant_demo/neural_searcher.py          | 31 ++++++++-----------
 qdrant_demo/service.py                  | 20 ++----------
 qdrant_demo/sparse.py                   | 34 --------------------
 5 files changed, 42 insertions(+), 89 deletions(-)
 delete mode 100644 qdrant_demo/sparse.py

diff --git a/qdrant_demo/config.py b/qdrant_demo/config.py
index 652d5f6..05cc2e6 100644
--- a/qdrant_demo/config.py
+++ b/qdrant_demo/config.py
@@ -10,8 +10,11 @@
 
 COLLECTION_NAME = os.environ.get("COLLECTION_NAME", "startups_hybrid")
 EMBEDDINGS_MODEL = os.environ.get("EMBEDDINGS_MODEL", "mixedbread-ai/mxbai-embed-large-v1")
+# Sparse keyword model. Qdrant/bm25 handles tokenization, stemming, and stopwords;
+# IDF is applied server-side via the collection's sparse modifier.
+SPARSE_EMBEDDINGS_MODEL = os.environ.get("SPARSE_EMBEDDINGS_MODEL", "Qdrant/bm25")
 
-TEXT_FIELD_NAME = os.environ.get("TEXT_FIELD_NAME", "description")
+TEXT_FIELD_NAME = os.environ.get("TEXT_FIELD_NAME", "document")
 
 # Named vectors on the hybrid collection. Leave DENSE_VECTOR_NAME empty for a
 # collection with a single unnamed vector.
diff --git a/qdrant_demo/init_collection_startups.py b/qdrant_demo/init_collection_startups.py
index e75648f..23d8280 100644
--- a/qdrant_demo/init_collection_startups.py
+++ b/qdrant_demo/init_collection_startups.py
@@ -1,43 +1,50 @@
 """Build the startups collection the search path expects: a named `dense` vector
-(mxbai) plus a `sparse` keyword vector with IDF, and a text index for keyword
-search. Documents are embedded with no prefix (mxbai is asymmetric; the query
-prefix is added at search time), so the same text drives dense, sparse, and
-keyword. Run:  python -m qdrant_demo.init_collection_startups
+(mxbai) plus a `sparse` bm25 keyword vector with IDF, and a text index for keyword
+search. Payload fields are renamed once here to the schema the frontend reads
+(`document`, `logo_url`, `homepage_url`). Both vectors are embedded by Qdrant Cloud
+inference, so the query and document sides use the identical models by construction.
+Documents get no mxbai prefix (the query prefix is added at search time).
+Run:  python -m qdrant_demo.init_collection_startups
 """
 import json
 import os
 from typing import Iterable
 
 from qdrant_client import QdrantClient, models
-from fastembed import TextEmbedding
 from tqdm import tqdm
 
 from qdrant_demo.config import (
     DATA_DIR, QDRANT_URL, QDRANT_API_KEY, COLLECTION_NAME, TEXT_FIELD_NAME,
-    EMBEDDINGS_MODEL, DENSE_VECTOR_NAME, SPARSE_VECTOR_NAME,
+    EMBEDDINGS_MODEL, SPARSE_EMBEDDINGS_MODEL, DENSE_VECTOR_NAME, SPARSE_VECTOR_NAME,
 )
-from qdrant_demo.sparse import to_sparse
 
 DENSE_DIM = 1024  # mxbai-embed-large-v1
 
 
+def _prepare(obj: dict) -> dict:
+    # Rename to the unified schema the frontend and search path read.
+    obj["logo_url"] = obj.pop("images", None)
+    obj["homepage_url"] = obj.pop("link", None)
+    obj[TEXT_FIELD_NAME] = obj.pop("description", "")
+    return obj
+
+
 def _records() -> Iterable[dict]:
     path = os.path.join(DATA_DIR, "startups_demo.json")
     with open(path, encoding="utf-8") as fd:
         for line in fd:
             line = line.strip()
             if line:
-                yield json.loads(line)
+                yield _prepare(json.loads(line))
 
 
 def _doc_text(obj: dict) -> str:
-    # Same text the searcher will match against: name + description.
-    return f"{obj.get('name', '')}. {obj.get(TEXT_FIELD_NAME) or obj.get('description', '')}".strip()
+    # Same text the searcher matches against: name + the document body.
+    return f"{obj.get('name', '')}. {obj.get(TEXT_FIELD_NAME, '')}".strip()
 
 
 def build():
-    client = QdrantClient(url=QDRANT_URL, api_key=QDRANT_API_KEY)
-    embedder = TextEmbedding(EMBEDDINGS_MODEL)  # local dense embedding for ingest
+    client = QdrantClient(url=QDRANT_URL, api_key=QDRANT_API_KEY, cloud_inference=True)
 
     if client.collection_exists(COLLECTION_NAME):
         print(f"{COLLECTION_NAME} exists, recreating.")
@@ -51,7 +58,7 @@ def build():
             )
         },
         sparse_vectors_config={
-            # IDF is applied at query time; sparse values carry term frequency only.
+            # bm25 values carry term frequency; IDF is applied at query time.
             SPARSE_VECTOR_NAME: models.SparseVectorParams(modifier=models.Modifier.IDF)
         },
         quantization_config=models.ScalarQuantization(
@@ -71,15 +78,13 @@ def build():
     def points() -> Iterable[models.PointStruct]:
         for idx, obj in enumerate(_records()):
             text = _doc_text(obj)
-            dense = next(iter(embedder.embed([text]))).tolist()
-            s_idx, s_val = to_sparse(text)
             yield models.PointStruct(
                 id=idx,
                 vector={
-                    DENSE_VECTOR_NAME: dense,
-                    SPARSE_VECTOR_NAME: models.SparseVector(indices=s_idx, values=s_val),
+                    DENSE_VECTOR_NAME: models.Document(text=text, model=EMBEDDINGS_MODEL),
+                    SPARSE_VECTOR_NAME: models.Document(text=text, model=SPARSE_EMBEDDINGS_MODEL),
                 },
-                payload=obj,  # original keys; the API maps them to the frontend schema
+                payload=obj,
             )
 
     client.upload_points(COLLECTION_NAME, points=tqdm(points()), batch_size=64)
diff --git a/qdrant_demo/neural_searcher.py b/qdrant_demo/neural_searcher.py
index d84e8f2..4e68eff 100644
--- a/qdrant_demo/neural_searcher.py
+++ b/qdrant_demo/neural_searcher.py
@@ -4,21 +4,20 @@
 from qdrant_client import QdrantClient, models
 
 from qdrant_demo.config import (
-    QDRANT_URL, QDRANT_API_KEY, EMBEDDINGS_MODEL,
+    QDRANT_URL, QDRANT_API_KEY, EMBEDDINGS_MODEL, SPARSE_EMBEDDINGS_MODEL,
     DENSE_VECTOR_NAME, SPARSE_VECTOR_NAME, RESULT_LIMIT, HYBRID_PREFETCH,
     CLOUD_INFERENCE,
 )
-from qdrant_demo.sparse import to_sparse
 
 
 class NeuralSearcher:
-    """Dense (semantic) and hybrid (dense + keyword, fused with RRF) search over a
-    collection with a named dense vector and a sparse keyword vector."""
+    """Dense (semantic) and hybrid (dense + bm25 keyword, fused with RRF) search
+    over a collection with a named dense vector and a bm25 sparse vector."""
 
     def __init__(self, collection_name: str):
         self.collection_name = collection_name
-        # Generous timeout: the first Cloud-inference call after idle has to load
-        # the model server-side, which can take longer than the default timeout.
+        # The first Cloud-inference call after idle loads the model server-side,
+        # which can take longer than the default client timeout.
         timeout = int(os.environ.get("QDRANT_TIMEOUT", "60"))
         self.qdrant_client = QdrantClient(
             url=QDRANT_URL, api_key=QDRANT_API_KEY,
@@ -27,14 +26,17 @@ def __init__(self, collection_name: str):
 
     # mxbai is an asymmetric retrieval model: the query gets a prompt prefix, the
     # stored documents do not, so applying it only here needs no re-indexing.
-    # Note: Qdrant Cloud inference already applies this prompt server-side, so on
-    # the Cloud path the prefix is a verified no-op (identical scores). It matters
-    # for the self-hosted / local-embedding path, where nothing adds it otherwise.
+    # Qdrant Cloud inference already applies this prompt server-side, so on the
+    # Cloud path the prefix is a verified no-op (identical scores). It matters for
+    # the self-hosted / local-embedding path, where nothing else adds it.
     QUERY_PREFIX = "Represent this sentence for searching relevant passages: "
 
     def _dense(self, text: str):
         return models.Document(text=self.QUERY_PREFIX + text, model=EMBEDDINGS_MODEL)
 
+    def _sparse(self, text: str):
+        return models.Document(text=text, model=SPARSE_EMBEDDINGS_MODEL)
+
     def _dense_only(self, text: str):
         return self.qdrant_client.query_points(
             collection_name=self.collection_name,
@@ -47,25 +49,18 @@ def search(self, text: str, hybrid: bool = False) -> dict:
         t0 = time.perf_counter()
         mode = "hybrid" if hybrid else "semantic"
 
-        idx, val = to_sparse(text) if hybrid else ([], [])
-        if hybrid and idx:
+        if hybrid:
             hits = self.qdrant_client.query_points(
                 collection_name=self.collection_name,
                 prefetch=[
                     models.Prefetch(query=self._dense(text), using=DENSE_VECTOR_NAME or None, limit=HYBRID_PREFETCH),
-                    models.Prefetch(
-                        query=models.SparseVector(indices=idx, values=val),
-                        using=SPARSE_VECTOR_NAME, limit=HYBRID_PREFETCH,
-                    ),
+                    models.Prefetch(query=self._sparse(text), using=SPARSE_VECTOR_NAME, limit=HYBRID_PREFETCH),
                 ],
                 query=models.FusionQuery(fusion=models.Fusion.RRF),
                 limit=RESULT_LIMIT,
             ).points
         else:
-            # No usable keyword tokens (e.g. a stopword-only query) -> dense only.
             hits = self._dense_only(text)
-            if hybrid:
-                mode = "semantic"  # honestly report what actually ran
 
         latency_ms = round((time.perf_counter() - t0) * 1000)
         results = [{**hit.payload, "score": hit.score} for hit in hits]
diff --git a/qdrant_demo/service.py b/qdrant_demo/service.py
index a679ac5..60f0735 100644
--- a/qdrant_demo/service.py
+++ b/qdrant_demo/service.py
@@ -29,22 +29,6 @@
 text_searcher = TextSearcher(collection_name=COLLECTION_NAME)
 
 
-def _present(items):
-    """Map stored payload keys to the schema the frontend reads, without renaming
-    anything in the collection. Keeps unknown keys too."""
-    out = []
-    for it in items:
-        r = dict(it)
-        if "document" not in r and "description" in r:
-            r["document"] = r.get("highlight", r["description"])
-        if "logo_url" not in r and "images" in r:
-            r["logo_url"] = r["images"]
-        if "homepage_url" not in r and "link" in r:
-            r["homepage_url"] = r["link"]
-        out.append(r)
-    return out
-
-
 @app.get("/api/search")
 async def read_item(q: str, mode: Optional[str] = None, neural: Optional[bool] = None):
     """mode = semantic (dense) | keyword (full-text) | hybrid (dense + keyword).
@@ -58,10 +42,10 @@ async def read_item(q: str, mode: Optional[str] = None, neural: Optional[bool] =
         return {"result": [], "stats": {"mode": mode}}
     try:
         if mode == "keyword":
-            return {"result": _present(text_searcher.search(query=q, top=RESULT_LIMIT)),
+            return {"result": text_searcher.search(query=q, top=RESULT_LIMIT),
                     "stats": {"mode": "keyword"}}
         out = neural_searcher.search(text=q, hybrid=(mode == "hybrid"))
-        return {"result": _present(out["results"]), "stats": out["stats"]}
+        return {"result": out["results"], "stats": out["stats"]}
     except Exception as e:
         logger.exception("search failed for q=%r mode=%s", q, mode)
         raise HTTPException(status_code=502, detail="Search is temporarily unavailable.")
diff --git a/qdrant_demo/sparse.py b/qdrant_demo/sparse.py
deleted file mode 100644
index d7b07b6..0000000
--- a/qdrant_demo/sparse.py
+++ /dev/null
@@ -1,34 +0,0 @@
-"""Keyword sparse encoder for hybrid search. Tokenize, term-frequency, stable
-hash to a u32 index. Qdrant applies IDF at query time (sparse index modifier=IDF),
-so index-time values carry the term-frequency component only. The same function
-runs for both documents and queries so tokenization stays consistent.
-
-This is a small self-contained encoder so the demo has no extra model download.
-The idiomatic replacement is Qdrant's BM25 sparse model (`Qdrant/bm25`), which
-adds stemming and a tuned scoring; switching to it is a follow-up because it means
-re-indexing the collection."""
-import re
-import zlib
-import math
-from collections import Counter
-
-_STOP = set(
-    "the a an and or of to in for on with is are was were be been being by at from as "
-    "it its this that these those i you he she we they your our their his her".split()
-)
-
-
-def tokenize(text):
-    # Keep single characters (languages like "c"/"r", versions like "3"); drop
-    # only stopwords. Dropping short tokens silently lost real query terms.
-    return [w for w in re.findall(r"[a-z0-9+#]+", (text or "").lower()) if w not in _STOP]
-
-
-def to_sparse(text):
-    tf = Counter(tokenize(text))
-    indices, values = [], []
-    for tok, c in tf.items():
-        # Qdrant sparse indices are unsigned 32-bit; use the full width.
-        indices.append(zlib.crc32(tok.encode("utf-8")) & 0xFFFFFFFF)
-        values.append(1.0 + math.log(c))
-    return indices, values

From c2c88daf121277caf2bac1c093abda875cc28dd1 Mon Sep 17 00:00:00 2001
From: John Kupchanko <john.kupchanko@qdrant.com>
Date: Fri, 7 Aug 2026 08:51:23 -0700
Subject: [PATCH 4/5] Bump qdrant-client to 1.19.0 and regenerate the lock

1.14.2 cannot talk to the current Qdrant Cloud server (1.19) for inference:
it fails parsing the inference response, so bm25/mxbai Document queries error.
Pin the client to 1.19.0 (matches the server) and regenerate poetry.lock so the
container is reproducible without the pip bump. Drop the fastembed extra: the app
embeds via Cloud inference and no longer imports fastembed. Widen the Python
constraint so the lock resolves on current interpreters; the image still runs 3.11.
---
 poetry.lock    | 725 ++++++-------------------------------------------
 pyproject.toml |   4 +-
 2 files changed, 80 insertions(+), 649 deletions(-)

diff --git a/poetry.lock b/poetry.lock
index 3b23638..f7aac5f 100644
--- a/poetry.lock
+++ b/poetry.lock
@@ -1,4 +1,4 @@
-# This file is automatically @generated by Poetry 2.0.1 and should not be changed by hand.
+# This file is automatically @generated by Poetry 2.4.1 and should not be changed by hand.
 
 [[package]]
 name = "annotated-types"
@@ -30,7 +30,7 @@ sniffio = ">=1.1"
 
 [package.extras]
 doc = ["Sphinx", "packaging", "sphinx-autodoc-typehints (>=1.2.0)", "sphinx-rtd-theme (>=1.2.2)", "sphinxcontrib-jquery"]
-test = ["anyio[trio]", "coverage[toml] (>=4.5)", "hypothesis (>=4.0)", "mock (>=4)", "psutil (>=5.9)", "pytest (>=7.0)", "pytest-mock (>=3.6.1)", "trustme", "uvloop (>=0.17)"]
+test = ["anyio[trio]", "coverage[toml] (>=4.5)", "hypothesis (>=4.0)", "mock (>=4) ; python_version < \"3.8\"", "psutil (>=5.9)", "pytest (>=7.0)", "pytest-mock (>=3.6.1)", "trustme", "uvloop (>=0.17) ; python_version < \"3.12\" and platform_python_implementation == \"CPython\" and platform_system != \"Windows\""]
 trio = ["trio (<0.22)"]
 
 [[package]]
@@ -175,24 +175,6 @@ files = [
     {file = "colorama-0.4.6.tar.gz", hash = "sha256:08695f5cb7ed6e0531a20572697297273c47b8cae5a63ffc6d6ed5c201be6e44"},
 ]
 
-[[package]]
-name = "coloredlogs"
-version = "15.0.1"
-description = "Colored terminal output for Python's logging module"
-optional = false
-python-versions = ">=2.7, !=3.0.*, !=3.1.*, !=3.2.*, !=3.3.*, !=3.4.*"
-groups = ["main"]
-files = [
-    {file = "coloredlogs-15.0.1-py2.py3-none-any.whl", hash = "sha256:612ee75c546f53e92e70049c9dbfcc18c935a2b9a53b66085ce9ef6a6e5c0934"},
-    {file = "coloredlogs-15.0.1.tar.gz", hash = "sha256:7c991aa71a4577af2f82600d8f8f3a89f936baeaf9b50a9c197da014e5bf16b0"},
-]
-
-[package.dependencies]
-humanfriendly = ">=9.1"
-
-[package.extras]
-cron = ["capturer (>=2.4)"]
-
 [[package]]
 name = "fastapi"
 version = "0.103.2"
@@ -214,99 +196,6 @@ typing-extensions = ">=4.5.0"
 [package.extras]
 all = ["email-validator (>=2.0.0)", "httpx (>=0.23.0)", "itsdangerous (>=1.1.0)", "jinja2 (>=2.11.2)", "orjson (>=3.2.1)", "pydantic-extra-types (>=2.0.0)", "pydantic-settings (>=2.0.0)", "python-multipart (>=0.0.5)", "pyyaml (>=5.3.1)", "ujson (>=4.0.1,!=4.0.2,!=4.1.0,!=4.2.0,!=4.3.0,!=5.0.0,!=5.1.0)", "uvicorn[standard] (>=0.12.0)"]
 
-[[package]]
-name = "fastembed"
-version = "0.6.1"
-description = "Fast, light, accurate library built for retrieval embedding generation"
-optional = false
-python-versions = ">=3.9.0"
-groups = ["main"]
-files = [
-    {file = "fastembed-0.6.1-py3-none-any.whl", hash = "sha256:89bfe07d3e12ae84474f0c85a6416ffbded25c72af43ac7ebf39acf309046044"},
-    {file = "fastembed-0.6.1.tar.gz", hash = "sha256:9c69ce389700eaa267eb9f488ecced94273d973be3fa933e6a9f807df3da5d96"},
-]
-
-[package.dependencies]
-huggingface-hub = ">=0.20,<1.0"
-loguru = ">=0.7.2,<0.8.0"
-mmh3 = ">=4.1.0,<6.0.0"
-numpy = {version = ">=1.21", markers = "python_version >= \"3.10\" and python_version < \"3.12\""}
-onnxruntime = {version = ">=1.17.0,<1.20.0 || >1.20.0", markers = "python_version >= \"3.10\" and python_version < \"3.13\""}
-pillow = ">=10.3.0,<12.0.0"
-py-rust-stemmers = ">=0.1.0,<0.2.0"
-requests = ">=2.31,<3.0"
-tokenizers = ">=0.15,<1.0"
-tqdm = ">=4.66,<5.0"
-
-[[package]]
-name = "filelock"
-version = "3.18.0"
-description = "A platform independent file lock."
-optional = false
-python-versions = ">=3.9"
-groups = ["main"]
-files = [
-    {file = "filelock-3.18.0-py3-none-any.whl", hash = "sha256:c401f4f8377c4464e6db25fff06205fd89bdd83b65eb0488ed1b160f780e21de"},
-    {file = "filelock-3.18.0.tar.gz", hash = "sha256:adbc88eabb99d2fec8c9c1b229b171f18afa655400173ddc653d5d01501fb9f2"},
-]
-
-[package.extras]
-docs = ["furo (>=2024.8.6)", "sphinx (>=8.1.3)", "sphinx-autodoc-typehints (>=3)"]
-testing = ["covdefaults (>=2.3)", "coverage (>=7.6.10)", "diff-cover (>=9.2.1)", "pytest (>=8.3.4)", "pytest-asyncio (>=0.25.2)", "pytest-cov (>=6)", "pytest-mock (>=3.14)", "pytest-timeout (>=2.3.1)", "virtualenv (>=20.28.1)"]
-typing = ["typing-extensions (>=4.12.2)"]
-
-[[package]]
-name = "flatbuffers"
-version = "25.2.10"
-description = "The FlatBuffers serialization format for Python"
-optional = false
-python-versions = "*"
-groups = ["main"]
-files = [
-    {file = "flatbuffers-25.2.10-py2.py3-none-any.whl", hash = "sha256:ebba5f4d5ea615af3f7fd70fc310636fbb2bbd1f566ac0a23d98dd412de50051"},
-    {file = "flatbuffers-25.2.10.tar.gz", hash = "sha256:97e451377a41262f8d9bd4295cc836133415cc03d8cb966410a4af92eb00d26e"},
-]
-
-[[package]]
-name = "fsspec"
-version = "2025.3.2"
-description = "File-system specification"
-optional = false
-python-versions = ">=3.9"
-groups = ["main"]
-files = [
-    {file = "fsspec-2025.3.2-py3-none-any.whl", hash = "sha256:2daf8dc3d1dfa65b6aa37748d112773a7a08416f6c70d96b264c96476ecaf711"},
-    {file = "fsspec-2025.3.2.tar.gz", hash = "sha256:e52c77ef398680bbd6a98c0e628fbc469491282981209907bbc8aea76a04fdc6"},
-]
-
-[package.extras]
-abfs = ["adlfs"]
-adl = ["adlfs"]
-arrow = ["pyarrow (>=1)"]
-dask = ["dask", "distributed"]
-dev = ["pre-commit", "ruff"]
-doc = ["numpydoc", "sphinx", "sphinx-design", "sphinx-rtd-theme", "yarl"]
-dropbox = ["dropbox", "dropboxdrivefs", "requests"]
-full = ["adlfs", "aiohttp (!=4.0.0a0,!=4.0.0a1)", "dask", "distributed", "dropbox", "dropboxdrivefs", "fusepy", "gcsfs", "libarchive-c", "ocifs", "panel", "paramiko", "pyarrow (>=1)", "pygit2", "requests", "s3fs", "smbprotocol", "tqdm"]
-fuse = ["fusepy"]
-gcs = ["gcsfs"]
-git = ["pygit2"]
-github = ["requests"]
-gs = ["gcsfs"]
-gui = ["panel"]
-hdfs = ["pyarrow (>=1)"]
-http = ["aiohttp (!=4.0.0a0,!=4.0.0a1)"]
-libarchive = ["libarchive-c"]
-oci = ["ocifs"]
-s3 = ["s3fs"]
-sftp = ["paramiko"]
-smb = ["smbprotocol"]
-ssh = ["paramiko"]
-test = ["aiohttp (!=4.0.0a0,!=4.0.0a1)", "numpy", "pytest", "pytest-asyncio (!=0.22.0)", "pytest-benchmark", "pytest-cov", "pytest-mock", "pytest-recording", "pytest-rerunfailures", "requests"]
-test-downstream = ["aiobotocore (>=2.5.4,<3.0.0)", "dask[dataframe,test]", "moto[server] (>4,<5)", "pytest-timeout", "xarray"]
-test-full = ["adlfs", "aiohttp (!=4.0.0a0,!=4.0.0a1)", "cloudpickle", "dask", "distributed", "dropbox", "dropboxdrivefs", "fastparquet", "fusepy", "gcsfs", "jinja2", "kerchunk", "libarchive-c", "lz4", "notebook", "numpy", "ocifs", "pandas", "panel", "paramiko", "pyarrow", "pyarrow (>=1)", "pyftpdlib", "pygit2", "pytest", "pytest-asyncio (!=0.22.0)", "pytest-benchmark", "pytest-cov", "pytest-mock", "pytest-recording", "pytest-rerunfailures", "python-snappy", "requests", "smbprotocol", "tqdm", "urllib3", "zarr", "zstandard"]
-tqdm = ["tqdm"]
-
 [[package]]
 name = "grpcio"
 version = "1.72.0"
@@ -453,63 +342,12 @@ httpcore = "==1.*"
 idna = "*"
 
 [package.extras]
-brotli = ["brotli", "brotlicffi"]
+brotli = ["brotli ; platform_python_implementation == \"CPython\"", "brotlicffi ; platform_python_implementation != \"CPython\""]
 cli = ["click (==8.*)", "pygments (==2.*)", "rich (>=10,<14)"]
 http2 = ["h2 (>=3,<5)"]
 socks = ["socksio (==1.*)"]
 zstd = ["zstandard (>=0.18.0)"]
 
-[[package]]
-name = "huggingface-hub"
-version = "0.30.2"
-description = "Client library to download and publish models, datasets and other repos on the huggingface.co hub"
-optional = false
-python-versions = ">=3.8.0"
-groups = ["main"]
-files = [
-    {file = "huggingface_hub-0.30.2-py3-none-any.whl", hash = "sha256:68ff05969927058cfa41df4f2155d4bb48f5f54f719dd0390103eefa9b191e28"},
-    {file = "huggingface_hub-0.30.2.tar.gz", hash = "sha256:9a7897c5b6fd9dad3168a794a8998d6378210f5b9688d0dfc180b1a228dc2466"},
-]
-
-[package.dependencies]
-filelock = "*"
-fsspec = ">=2023.5.0"
-packaging = ">=20.9"
-pyyaml = ">=5.1"
-requests = "*"
-tqdm = ">=4.42.1"
-typing-extensions = ">=3.7.4.3"
-
-[package.extras]
-all = ["InquirerPy (==0.3.4)", "Jinja2", "Pillow", "aiohttp", "fastapi", "gradio (>=4.0.0)", "jedi", "libcst (==1.4.0)", "mypy (==1.5.1)", "numpy", "pytest (>=8.1.1,<8.2.2)", "pytest-asyncio", "pytest-cov", "pytest-env", "pytest-mock", "pytest-rerunfailures", "pytest-vcr", "pytest-xdist", "ruff (>=0.9.0)", "soundfile", "types-PyYAML", "types-requests", "types-simplejson", "types-toml", "types-tqdm", "types-urllib3", "typing-extensions (>=4.8.0)", "urllib3 (<2.0)"]
-cli = ["InquirerPy (==0.3.4)"]
-dev = ["InquirerPy (==0.3.4)", "Jinja2", "Pillow", "aiohttp", "fastapi", "gradio (>=4.0.0)", "jedi", "libcst (==1.4.0)", "mypy (==1.5.1)", "numpy", "pytest (>=8.1.1,<8.2.2)", "pytest-asyncio", "pytest-cov", "pytest-env", "pytest-mock", "pytest-rerunfailures", "pytest-vcr", "pytest-xdist", "ruff (>=0.9.0)", "soundfile", "types-PyYAML", "types-requests", "types-simplejson", "types-toml", "types-tqdm", "types-urllib3", "typing-extensions (>=4.8.0)", "urllib3 (<2.0)"]
-fastai = ["fastai (>=2.4)", "fastcore (>=1.3.27)", "toml"]
-hf-transfer = ["hf-transfer (>=0.1.4)"]
-hf-xet = ["hf-xet (>=0.1.4)"]
-inference = ["aiohttp"]
-quality = ["libcst (==1.4.0)", "mypy (==1.5.1)", "ruff (>=0.9.0)"]
-tensorflow = ["graphviz", "pydot", "tensorflow"]
-tensorflow-testing = ["keras (<3.0)", "tensorflow"]
-testing = ["InquirerPy (==0.3.4)", "Jinja2", "Pillow", "aiohttp", "fastapi", "gradio (>=4.0.0)", "jedi", "numpy", "pytest (>=8.1.1,<8.2.2)", "pytest-asyncio", "pytest-cov", "pytest-env", "pytest-mock", "pytest-rerunfailures", "pytest-vcr", "pytest-xdist", "soundfile", "urllib3 (<2.0)"]
-torch = ["safetensors[torch]", "torch"]
-typing = ["types-PyYAML", "types-requests", "types-simplejson", "types-toml", "types-tqdm", "types-urllib3", "typing-extensions (>=4.8.0)"]
-
-[[package]]
-name = "humanfriendly"
-version = "10.0"
-description = "Human friendly output for text interfaces using Python"
-optional = false
-python-versions = ">=2.7, !=3.0.*, !=3.1.*, !=3.2.*, !=3.3.*, !=3.4.*"
-groups = ["main"]
-files = [
-    {file = "humanfriendly-10.0-py2.py3-none-any.whl", hash = "sha256:1697e1a8a8f550fd43c2865cd84542fc175a61dcb779b6fee18cf6b6ccba1477"},
-    {file = "humanfriendly-10.0.tar.gz", hash = "sha256:6b0b831ce8f15f7300721aa49829fc4e83921a9a301cc7f606be6686a2288ddc"},
-]
-
-[package.dependencies]
-pyreadline3 = {version = "*", markers = "sys_platform == \"win32\" and python_version >= \"3.8\""}
-
 [[package]]
 name = "hyperframe"
 version = "6.1.0"
@@ -554,124 +392,7 @@ colorama = {version = ">=0.3.4", markers = "sys_platform == \"win32\""}
 win32-setctime = {version = ">=1.0.0", markers = "sys_platform == \"win32\""}
 
 [package.extras]
-dev = ["Sphinx (==8.1.3)", "build (==1.2.2)", "colorama (==0.4.5)", "colorama (==0.4.6)", "exceptiongroup (==1.1.3)", "freezegun (==1.1.0)", "freezegun (==1.5.0)", "mypy (==v0.910)", "mypy (==v0.971)", "mypy (==v1.13.0)", "mypy (==v1.4.1)", "myst-parser (==4.0.0)", "pre-commit (==4.0.1)", "pytest (==6.1.2)", "pytest (==8.3.2)", "pytest-cov (==2.12.1)", "pytest-cov (==5.0.0)", "pytest-cov (==6.0.0)", "pytest-mypy-plugins (==1.9.3)", "pytest-mypy-plugins (==3.1.0)", "sphinx-rtd-theme (==3.0.2)", "tox (==3.27.1)", "tox (==4.23.2)", "twine (==6.0.1)"]
-
-[[package]]
-name = "mmh3"
-version = "5.1.0"
-description = "Python extension for MurmurHash (MurmurHash3), a set of fast and robust hash functions."
-optional = false
-python-versions = ">=3.9"
-groups = ["main"]
-files = [
-    {file = "mmh3-5.1.0-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:eaf4ac5c6ee18ca9232238364d7f2a213278ae5ca97897cafaa123fcc7bb8bec"},
-    {file = "mmh3-5.1.0-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:48f9aa8ccb9ad1d577a16104834ac44ff640d8de8c0caed09a2300df7ce8460a"},
-    {file = "mmh3-5.1.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:d4ba8cac21e1f2d4e436ce03a82a7f87cda80378691f760e9ea55045ec480a3d"},
-    {file = "mmh3-5.1.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d69281c281cb01994f054d862a6bb02a2e7acfe64917795c58934b0872b9ece4"},
-    {file = "mmh3-5.1.0-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:4d05ed3962312fbda2a1589b97359d2467f677166952f6bd410d8c916a55febf"},
-    {file = "mmh3-5.1.0-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:78ae6a03f4cff4aa92ddd690611168856f8c33a141bd3e5a1e0a85521dc21ea0"},
-    {file = "mmh3-5.1.0-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:95f983535b39795d9fb7336438faae117424c6798f763d67c6624f6caf2c4c01"},
-    {file = "mmh3-5.1.0-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d46fdd80d4c7ecadd9faa6181e92ccc6fe91c50991c9af0e371fdf8b8a7a6150"},
-    {file = "mmh3-5.1.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:0f16e976af7365ea3b5c425124b2a7f0147eed97fdbb36d99857f173c8d8e096"},
-    {file = "mmh3-5.1.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:6fa97f7d1e1f74ad1565127229d510f3fd65d931fdedd707c1e15100bc9e5ebb"},
-    {file = "mmh3-5.1.0-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:4052fa4a8561bd62648e9eb993c8f3af3bdedadf3d9687aa4770d10e3709a80c"},
-    {file = "mmh3-5.1.0-cp310-cp310-musllinux_1_2_s390x.whl", hash = "sha256:3f0e8ae9f961037f812afe3cce7da57abf734285961fffbeff9a4c011b737732"},
-    {file = "mmh3-5.1.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:99297f207db967814f1f02135bb7fe7628b9eacb046134a34e1015b26b06edce"},
-    {file = "mmh3-5.1.0-cp310-cp310-win32.whl", hash = "sha256:2e6c8dc3631a5e22007fbdb55e993b2dbce7985c14b25b572dd78403c2e79182"},
-    {file = "mmh3-5.1.0-cp310-cp310-win_amd64.whl", hash = "sha256:e4e8c7ad5a4dddcfde35fd28ef96744c1ee0f9d9570108aa5f7e77cf9cfdf0bf"},
-    {file = "mmh3-5.1.0-cp310-cp310-win_arm64.whl", hash = "sha256:45da549269883208912868a07d0364e1418d8292c4259ca11699ba1b2475bd26"},
-    {file = "mmh3-5.1.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:0b529dcda3f951ff363a51d5866bc6d63cf57f1e73e8961f864ae5010647079d"},
-    {file = "mmh3-5.1.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:4db1079b3ace965e562cdfc95847312f9273eb2ad3ebea983435c8423e06acd7"},
-    {file = "mmh3-5.1.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:22d31e3a0ff89b8eb3b826d6fc8e19532998b2aa6b9143698043a1268da413e1"},
-    {file = "mmh3-5.1.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2139bfbd354cd6cb0afed51c4b504f29bcd687a3b1460b7e89498329cc28a894"},
-    {file = "mmh3-5.1.0-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8c8105c6a435bc2cd6ea2ef59558ab1a2976fd4a4437026f562856d08996673a"},
-    {file = "mmh3-5.1.0-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:57730067174a7f36fcd6ce012fe359bd5510fdaa5fe067bc94ed03e65dafb769"},
-    {file = "mmh3-5.1.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:bde80eb196d7fdc765a318604ded74a4378f02c5b46c17aa48a27d742edaded2"},
-    {file = "mmh3-5.1.0-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e9c8eddcb441abddeb419c16c56fd74b3e2df9e57f7aa2903221996718435c7a"},
-    {file = "mmh3-5.1.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:99e07e4acafbccc7a28c076a847fb060ffc1406036bc2005acb1b2af620e53c3"},
-    {file = "mmh3-5.1.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:9e25ba5b530e9a7d65f41a08d48f4b3fedc1e89c26486361166a5544aa4cad33"},
-    {file = "mmh3-5.1.0-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:bb9bf7475b4d99156ce2f0cf277c061a17560c8c10199c910a680869a278ddc7"},
-    {file = "mmh3-5.1.0-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:2a1b0878dd281ea3003368ab53ff6f568e175f1b39f281df1da319e58a19c23a"},
-    {file = "mmh3-5.1.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:25f565093ac8b8aefe0f61f8f95c9a9d11dd69e6a9e9832ff0d293511bc36258"},
-    {file = "mmh3-5.1.0-cp311-cp311-win32.whl", hash = "sha256:1e3554d8792387eac73c99c6eaea0b3f884e7130eb67986e11c403e4f9b6d372"},
-    {file = "mmh3-5.1.0-cp311-cp311-win_amd64.whl", hash = "sha256:8ad777a48197882492af50bf3098085424993ce850bdda406a358b6ab74be759"},
-    {file = "mmh3-5.1.0-cp311-cp311-win_arm64.whl", hash = "sha256:f29dc4efd99bdd29fe85ed6c81915b17b2ef2cf853abf7213a48ac6fb3eaabe1"},
-    {file = "mmh3-5.1.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:45712987367cb9235026e3cbf4334670522a97751abfd00b5bc8bfa022c3311d"},
-    {file = "mmh3-5.1.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:b1020735eb35086ab24affbea59bb9082f7f6a0ad517cb89f0fc14f16cea4dae"},
-    {file = "mmh3-5.1.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:babf2a78ce5513d120c358722a2e3aa7762d6071cd10cede026f8b32452be322"},
-    {file = "mmh3-5.1.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d4f47f58cd5cbef968c84a7c1ddc192fef0a36b48b0b8a3cb67354531aa33b00"},
-    {file = "mmh3-5.1.0-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:2044a601c113c981f2c1e14fa33adc9b826c9017034fe193e9eb49a6882dbb06"},
-    {file = "mmh3-5.1.0-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:c94d999c9f2eb2da44d7c2826d3fbffdbbbbcde8488d353fee7c848ecc42b968"},
-    {file = "mmh3-5.1.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:a015dcb24fa0c7a78f88e9419ac74f5001c1ed6a92e70fd1803f74afb26a4c83"},
-    {file = "mmh3-5.1.0-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:457da019c491a2d20e2022c7d4ce723675e4c081d9efc3b4d8b9f28a5ea789bd"},
-    {file = "mmh3-5.1.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:71408579a570193a4ac9c77344d68ddefa440b00468a0b566dcc2ba282a9c559"},
-    {file = "mmh3-5.1.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:8b3a04bc214a6e16c81f02f855e285c6df274a2084787eeafaa45f2fbdef1b63"},
-    {file = "mmh3-5.1.0-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:832dae26a35514f6d3c1e267fa48e8de3c7b978afdafa0529c808ad72e13ada3"},
-    {file = "mmh3-5.1.0-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:bf658a61fc92ef8a48945ebb1076ef4ad74269e353fffcb642dfa0890b13673b"},
-    {file = "mmh3-5.1.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:3313577453582b03383731b66447cdcdd28a68f78df28f10d275d7d19010c1df"},
-    {file = "mmh3-5.1.0-cp312-cp312-win32.whl", hash = "sha256:1d6508504c531ab86c4424b5a5ff07c1132d063863339cf92f6657ff7a580f76"},
-    {file = "mmh3-5.1.0-cp312-cp312-win_amd64.whl", hash = "sha256:aa75981fcdf3f21759d94f2c81b6a6e04a49dfbcdad88b152ba49b8e20544776"},
-    {file = "mmh3-5.1.0-cp312-cp312-win_arm64.whl", hash = "sha256:a4c1a76808dfea47f7407a0b07aaff9087447ef6280716fd0783409b3088bb3c"},
-    {file = "mmh3-5.1.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:7a523899ca29cfb8a5239618474a435f3d892b22004b91779fcb83504c0d5b8c"},
-    {file = "mmh3-5.1.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:17cef2c3a6ca2391ca7171a35ed574b5dab8398163129a3e3a4c05ab85a4ff40"},
-    {file = "mmh3-5.1.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:52e12895b30110f3d89dae59a888683cc886ed0472dd2eca77497edef6161997"},
-    {file = "mmh3-5.1.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e0d6719045cda75c3f40397fc24ab67b18e0cb8f69d3429ab4c39763c4c608dd"},
-    {file = "mmh3-5.1.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:d19fa07d303a91f8858982c37e6939834cb11893cb3ff20e6ee6fa2a7563826a"},
-    {file = "mmh3-5.1.0-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:31b47a620d622fbde8ca1ca0435c5d25de0ac57ab507209245e918128e38e676"},
-    {file = "mmh3-5.1.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:00f810647c22c179b6821079f7aa306d51953ac893587ee09cf1afb35adf87cb"},
-    {file = "mmh3-5.1.0-cp313-cp313-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f6128b610b577eed1e89ac7177ab0c33d06ade2aba93f5c89306032306b5f1c6"},
-    {file = "mmh3-5.1.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:1e550a45d2ff87a1c11b42015107f1778c93f4c6f8e731bf1b8fa770321b8cc4"},
-    {file = "mmh3-5.1.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:785ae09276342f79fd8092633e2d52c0f7c44d56e8cfda8274ccc9b76612dba2"},
-    {file = "mmh3-5.1.0-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:0f4be3703a867ef976434afd3661a33884abe73ceb4ee436cac49d3b4c2aaa7b"},
-    {file = "mmh3-5.1.0-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:e513983830c4ff1f205ab97152a0050cf7164f1b4783d702256d39c637b9d107"},
-    {file = "mmh3-5.1.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:b9135c300535c828c0bae311b659f33a31c941572eae278568d1a953c4a57b59"},
-    {file = "mmh3-5.1.0-cp313-cp313-win32.whl", hash = "sha256:c65dbd12885a5598b70140d24de5839551af5a99b29f9804bb2484b29ef07692"},
-    {file = "mmh3-5.1.0-cp313-cp313-win_amd64.whl", hash = "sha256:10db7765201fc65003fa998faa067417ef6283eb5f9bba8f323c48fd9c33e91f"},
-    {file = "mmh3-5.1.0-cp313-cp313-win_arm64.whl", hash = "sha256:b22fe2e54be81f6c07dcb36b96fa250fb72effe08aa52fbb83eade6e1e2d5fd7"},
-    {file = "mmh3-5.1.0-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:166b67749a1d8c93b06f5e90576f1ba838a65c8e79f28ffd9dfafba7c7d0a084"},
-    {file = "mmh3-5.1.0-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:adba83c7ba5cc8ea201ee1e235f8413a68e7f7b8a657d582cc6c6c9d73f2830e"},
-    {file = "mmh3-5.1.0-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:a61f434736106804eb0b1612d503c4e6eb22ba31b16e6a2f987473de4226fa55"},
-    {file = "mmh3-5.1.0-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ba9ce59816b30866093f048b3312c2204ff59806d3a02adee71ff7bd22b87554"},
-    {file = "mmh3-5.1.0-cp39-cp39-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:cd51597bef1e503363b05cb579db09269e6e6c39d419486626b255048daf545b"},
-    {file = "mmh3-5.1.0-cp39-cp39-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:d51a1ed642d3fb37b8f4cab966811c52eb246c3e1740985f701ef5ad4cdd2145"},
-    {file = "mmh3-5.1.0-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:709bfe81c53bf8a3609efcbd65c72305ade60944f66138f697eefc1a86b6e356"},
-    {file = "mmh3-5.1.0-cp39-cp39-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:e01a9b0092b6f82e861137c8e9bb9899375125b24012eb5219e61708be320032"},
-    {file = "mmh3-5.1.0-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:27e46a2c13c9a805e03c9ec7de0ca8e096794688ab2125bdce4229daf60c4a56"},
-    {file = "mmh3-5.1.0-cp39-cp39-musllinux_1_2_i686.whl", hash = "sha256:5766299c1d26f6bfd0a638e070bd17dbd98d4ccb067d64db3745bf178e700ef0"},
-    {file = "mmh3-5.1.0-cp39-cp39-musllinux_1_2_ppc64le.whl", hash = "sha256:7785205e3e4443fdcbb73766798c7647f94c2f538b90f666688f3e757546069e"},
-    {file = "mmh3-5.1.0-cp39-cp39-musllinux_1_2_s390x.whl", hash = "sha256:8e574fbd39afb433b3ab95683b1b4bf18313dc46456fc9daaddc2693c19ca565"},
-    {file = "mmh3-5.1.0-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:1b6727a5a20e32cbf605743749f3862abe5f5e097cbf2afc7be5aafd32a549ae"},
-    {file = "mmh3-5.1.0-cp39-cp39-win32.whl", hash = "sha256:d6eaa711d4b9220fe5252032a44bf68e5dcfb7b21745a96efc9e769b0dd57ec2"},
-    {file = "mmh3-5.1.0-cp39-cp39-win_amd64.whl", hash = "sha256:49d444913f6c02980e5241a53fe9af2338f2043d6ce5b6f5ea7d302c52c604ac"},
-    {file = "mmh3-5.1.0-cp39-cp39-win_arm64.whl", hash = "sha256:0daaeaedd78773b70378f2413c7d6b10239a75d955d30d54f460fb25d599942d"},
-    {file = "mmh3-5.1.0.tar.gz", hash = "sha256:136e1e670500f177f49ec106a4ebf0adf20d18d96990cc36ea492c651d2b406c"},
-]
-
-[package.extras]
-benchmark = ["pymmh3 (==0.0.5)", "pyperf (==2.8.1)", "xxhash (==3.5.0)"]
-docs = ["myst-parser (==4.0.0)", "shibuya (==2024.12.21)", "sphinx (==8.1.3)", "sphinx-copybutton (==0.5.2)"]
-lint = ["black (==24.10.0)", "clang-format (==19.1.7)", "isort (==5.13.2)", "pylint (==3.3.3)"]
-plot = ["matplotlib (==3.10.0)", "pandas (==2.2.3)"]
-test = ["pytest (==8.3.4)", "pytest-sugar (==1.0.0)"]
-type = ["mypy (==1.14.1)"]
-
-[[package]]
-name = "mpmath"
-version = "1.3.0"
-description = "Python library for arbitrary-precision floating-point arithmetic"
-optional = false
-python-versions = "*"
-groups = ["main"]
-files = [
-    {file = "mpmath-1.3.0-py3-none-any.whl", hash = "sha256:a0b2b9fe80bbcd81a6647ff13108738cfb482d481d826cc0e02f5b35e5c88d2c"},
-    {file = "mpmath-1.3.0.tar.gz", hash = "sha256:7a28eb2a9774d00c7bc92411c19a89209d5da7c4c9a9e227be8330a23a25b91f"},
-]
-
-[package.extras]
-develop = ["codecov", "pycodestyle", "pytest (>=4.6)", "pytest-cov", "wheel"]
-docs = ["sphinx"]
-gmpy = ["gmpy2 (>=2.1.0a4)"]
-tests = ["pytest (>=4.6)"]
+dev = ["Sphinx (==8.1.3) ; python_version >= \"3.11\"", "build (==1.2.2) ; python_version >= \"3.11\"", "colorama (==0.4.5) ; python_version < \"3.8\"", "colorama (==0.4.6) ; python_version >= \"3.8\"", "exceptiongroup (==1.1.3) ; python_version >= \"3.7\" and python_version < \"3.11\"", "freezegun (==1.1.0) ; python_version < \"3.8\"", "freezegun (==1.5.0) ; python_version >= \"3.8\"", "mypy (==0.910) ; python_version < \"3.6\"", "mypy (==0.971) ; python_version == \"3.6\"", "mypy (==1.13.0) ; python_version >= \"3.8\"", "mypy (==1.4.1) ; python_version == \"3.7\"", "myst-parser (==4.0.0) ; python_version >= \"3.11\"", "pre-commit (==4.0.1) ; python_version >= \"3.9\"", "pytest (==6.1.2) ; python_version < \"3.8\"", "pytest (==8.3.2) ; python_version >= \"3.8\"", "pytest-cov (==2.12.1) ; python_version < \"3.8\"", "pytest-cov (==5.0.0) ; python_version == \"3.8\"", "pytest-cov (==6.0.0) ; python_version >= \"3.9\"", "pytest-mypy-plugins (==1.9.3) ; python_version >= \"3.6\" and python_version < \"3.8\"", "pytest-mypy-plugins (==3.1.0) ; python_version >= \"3.8\"", "sphinx-rtd-theme (==3.0.2) ; python_version >= \"3.11\"", "tox (==3.27.1) ; python_version < \"3.8\"", "tox (==4.23.2) ; python_version >= \"3.8\"", "twine (==6.0.1) ; python_version >= \"3.11\""]
 
 [[package]]
 name = "numpy"
@@ -680,6 +401,7 @@ description = "Fundamental package for array computing in Python"
 optional = false
 python-versions = ">=3.10"
 groups = ["main"]
+markers = "python_version <= \"3.13\""
 files = [
     {file = "numpy-2.2.5-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:1f4a922da1729f4c40932b2af4fe84909c7a6e167e6e99f71838ce3a29f3fe26"},
     {file = "numpy-2.2.5-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:b6f91524d31b34f4a5fee24f5bc16dcd1491b668798b6d85585d836c1e633a6a"},
@@ -739,51 +461,58 @@ files = [
 ]
 
 [[package]]
-name = "onnxruntime"
-version = "1.21.1"
-description = "ONNX Runtime is a runtime accelerator for Machine Learning models"
-optional = false
-python-versions = ">=3.10"
-groups = ["main"]
-files = [
-    {file = "onnxruntime-1.21.1-cp310-cp310-macosx_13_0_universal2.whl", hash = "sha256:daedb5d33d8963062a25f4a3c788262074587f685a19478ef759a911b4b12c25"},
-    {file = "onnxruntime-1.21.1-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:3a402f9bda0b1cc791d9cf31d23c471e8189a55369b49ef2b9d0854eb11d22c4"},
-    {file = "onnxruntime-1.21.1-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:15656a2d0126f4f66295381e39c8812a6d845ccb1bb1f7bf6dd0a46d7d602e7f"},
-    {file = "onnxruntime-1.21.1-cp310-cp310-win_amd64.whl", hash = "sha256:79bbedfd1263065532967a2132fb365a27ffe5f7ed962e16fec55cca741f72aa"},
-    {file = "onnxruntime-1.21.1-cp311-cp311-macosx_13_0_universal2.whl", hash = "sha256:8bee9b5ba7b88ae7bfccb4f97bbe1b4bae801b0fb05d686b28a722cb27c89931"},
-    {file = "onnxruntime-1.21.1-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:4b6a29a1767b92d543091349f5397a1c7619eaca746cd1bc47f8b4ec5a9f1a6c"},
-    {file = "onnxruntime-1.21.1-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:982dcc04a6688e1af9e3da1d4ef2bdeb11417cf3f8dde81f8f721043c1919a4f"},
-    {file = "onnxruntime-1.21.1-cp311-cp311-win_amd64.whl", hash = "sha256:2b6052c04b9125319293abb9bdcce40e806db3e097f15b82242d4cd72d81fd0c"},
-    {file = "onnxruntime-1.21.1-cp312-cp312-macosx_13_0_universal2.whl", hash = "sha256:f615c05869a523a94d0a4de1f0936d0199a473cf104d630fc26174bebd5759bd"},
-    {file = "onnxruntime-1.21.1-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:79dfb1f47386c4edd115b21015354b2f05f5566c40c98606251f15a64add3cbe"},
-    {file = "onnxruntime-1.21.1-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:2742935d6610fe0f58e1995018d9db7e8239d0201d9ebbdb7964a61386b5390a"},
-    {file = "onnxruntime-1.21.1-cp312-cp312-win_amd64.whl", hash = "sha256:a7afdb3fcb162f5536225e13c2b245018068964b1d0eee05303ea6823ca6785e"},
-    {file = "onnxruntime-1.21.1-cp313-cp313-macosx_13_0_universal2.whl", hash = "sha256:ed4f9771233a92edcab9f11f537702371d450fe6cd79a727b672d37b9dab0cde"},
-    {file = "onnxruntime-1.21.1-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:1bc100fd1f4f95258e7d0f7068ec69dec2a47cc693f745eec9cf4561ee8d952a"},
-    {file = "onnxruntime-1.21.1-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:0fea0d2b98eecf4bebe01f7ce9a265a5d72b3050e9098063bfe65fa2b0633a8e"},
-    {file = "onnxruntime-1.21.1-cp313-cp313-win_amd64.whl", hash = "sha256:da606061b9ed1b05b63a37be38c2014679a3e725903f58036ffd626df45c0e47"},
-    {file = "onnxruntime-1.21.1-cp313-cp313t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:94674315d40d521952bfc28007ce9b6728e87753e1f18d243c8cd953f25903b8"},
-    {file = "onnxruntime-1.21.1-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:5c9e4571ff5b2a5d377d414bc85cd9450ba233a9a92f766493874f1093976453"},
-]
-
-[package.dependencies]
-coloredlogs = "*"
-flatbuffers = "*"
-numpy = ">=1.21.6"
-packaging = "*"
-protobuf = "*"
-sympy = "*"
-
-[[package]]
-name = "packaging"
-version = "25.0"
-description = "Core utilities for Python packages"
+name = "numpy"
+version = "2.5.1"
+description = "Fundamental package for array computing in Python"
 optional = false
-python-versions = ">=3.8"
+python-versions = ">=3.12"
 groups = ["main"]
+markers = "python_version == \"3.14\""
 files = [
-    {file = "packaging-25.0-py3-none-any.whl", hash = "sha256:29572ef2b1f17581046b3a2227d5c611fb25ec70ca1ba8554b24b0e69331a484"},
-    {file = "packaging-25.0.tar.gz", hash = "sha256:d443872c98d677bf60f6a1f2f8c1cb748e8fe762d2bf9d3148b5599295b0fc4f"},
+    {file = "numpy-2.5.1-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:2c889b56fe48b1018f764b0eec8df59ab654e9148aa91faa12596043500de277"},
+    {file = "numpy-2.5.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:ab451b59c5643c570974c43aef780703ef1d3b4965d2be07afd530615a9358d1"},
+    {file = "numpy-2.5.1-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:78798bd5b9ad744056af8efa90e3b9ddaa53272a0848a483084a1cc0a13b2dc0"},
+    {file = "numpy-2.5.1-cp312-cp312-macosx_14_0_x86_64.whl", hash = "sha256:2ae0ca40bcb22d6ba59c1dfd5446f49940b0f2d821fde133f10dda11f816b84e"},
+    {file = "numpy-2.5.1-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:61ac47e772e6b8ea489e1d2f441a34c5c3ac17327e7ce294cbdf535795ad4e75"},
+    {file = "numpy-2.5.1-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:59fda5e192b570217ec2580c96f00e9a7e12ef6866a900eb089b62c1a32545ca"},
+    {file = "numpy-2.5.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:f7119ebff1a9829e9f431a4f9d28e703023bb6b9fe7c8f724467dbfc27c94ab3"},
+    {file = "numpy-2.5.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:e824c2acf8862052246be5a44c15da1777940c60d010dd2aab897824d9c430f9"},
+    {file = "numpy-2.5.1-cp312-cp312-win32.whl", hash = "sha256:08d60c810432eb83360958dea0999ac4cfb94531ea8efcbf0b7f277c2068aeb2"},
+    {file = "numpy-2.5.1-cp312-cp312-win_amd64.whl", hash = "sha256:f7d60026c0bdb1380e83bfa7a0419c4577ee4b9a08880afcb6dadeb74c649fa2"},
+    {file = "numpy-2.5.1-cp312-cp312-win_arm64.whl", hash = "sha256:17a25e09640602e10bc8de0e6fa2b3fd68eedd84ba6d7842dc8f32f9ab87bd0b"},
+    {file = "numpy-2.5.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:0bfebd8695f9863592fe744be833a258120b14a9f39da255e8aa8fade2c0ddd1"},
+    {file = "numpy-2.5.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:30b44a6b53a7ae63c54c089a8726e5563ed302716c5b7ccc85afade40b0e7ff6"},
+    {file = "numpy-2.5.1-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:6165343f81b56ef8f514f396989e529b61d9dc709b99421b07e9f3e698e2287d"},
+    {file = "numpy-2.5.1-cp313-cp313-macosx_14_0_x86_64.whl", hash = "sha256:4939237038ada79308dda3204ac6462df056b5672b2e25db1149cf873668b3e1"},
+    {file = "numpy-2.5.1-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:1c6759f538fb912fc46de0a6b1758ccf7b57bc7c7ebebc23974fdac3de8db0cd"},
+    {file = "numpy-2.5.1-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:9726558e8db4a5bf7929a70ae50f63abda4daf0efe810e3bfbab95976f75fc1a"},
+    {file = "numpy-2.5.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:3935f3b419b244a02732676fa5317a9193cc596a4c0646db07e5b421229ac9f7"},
+    {file = "numpy-2.5.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:dc932a65ded7ce9013d120845a2514dcccb1a67bfc8deb8d37633762951904a6"},
+    {file = "numpy-2.5.1-cp313-cp313-win32.whl", hash = "sha256:4b4ff1608417eb7a59da7b967bbb798cacfe071d2caf526a24281cd562072ed9"},
+    {file = "numpy-2.5.1-cp313-cp313-win_amd64.whl", hash = "sha256:6c3fe51bc6a16453d452997053454f309e8e0ed7b42d6b361ce4ac8c32913d74"},
+    {file = "numpy-2.5.1-cp313-cp313-win_arm64.whl", hash = "sha256:f7feb014281029e628ba2d5a007407443b06e418b6fe451d1e2adcbc8eba0107"},
+    {file = "numpy-2.5.1-cp314-cp314-macosx_10_15_x86_64.whl", hash = "sha256:7c786fe9a5bbe360022e584c5a34cf6b54265c71bd7ec8ac3d8fec38968071f8"},
+    {file = "numpy-2.5.1-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:32985c896d897419ef8da6917872d80b78ad0ea26d85b23245c7366ffde76d75"},
+    {file = "numpy-2.5.1-cp314-cp314-macosx_14_0_arm64.whl", hash = "sha256:efd736408cc97c79b9e6917338dfc8f06013b2274f992e96b1d9a81a71e2a2c2"},
+    {file = "numpy-2.5.1-cp314-cp314-macosx_14_0_x86_64.whl", hash = "sha256:ab84dc6b074fa881cae55bea94cc4f68e285181ba7f32497bf7dee6b1496165b"},
+    {file = "numpy-2.5.1-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:caf3e317d33d60c37986b452613f4ab51246d0691350c03d0cb4a898627f4a95"},
+    {file = "numpy-2.5.1-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:54ad769f17bc2d833b620851989f62054fb9ab93c969d9e1dc3c8e3d56beea21"},
+    {file = "numpy-2.5.1-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:c12afb53450fa976d4c681c50a7423729a4c51c0465ed9f32b8a9cabbc472373"},
+    {file = "numpy-2.5.1-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:e8c11c405efc5ff6816d5983c96cdfa215bab3428961243af3ff59b228490438"},
+    {file = "numpy-2.5.1-cp314-cp314-win32.whl", hash = "sha256:f2479a47f8d5932d1718168a681ad6e536a9df484c83cfcf9de365e164537ace"},
+    {file = "numpy-2.5.1-cp314-cp314-win_amd64.whl", hash = "sha256:24d0eb82c0541d3415a33425db64ae439dffccd7b4dbcb30e7c35120205c506a"},
+    {file = "numpy-2.5.1-cp314-cp314-win_arm64.whl", hash = "sha256:5a4c988b38d261deeeaad9954e3deb091ad905c94e8bb6708654ef1d97f286b0"},
+    {file = "numpy-2.5.1-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:a33276be12fa045805f477f22482088b66bb758ffbe89a9d21457de863a32e22"},
+    {file = "numpy-2.5.1-cp314-cp314t-macosx_14_0_arm64.whl", hash = "sha256:f089d7b00756190aacf1f5d34bdf38c3c430ac82b4f868f8cede73380460fce7"},
+    {file = "numpy-2.5.1-cp314-cp314t-macosx_14_0_x86_64.whl", hash = "sha256:09e9bfd8d2cf479c7d174804fb3811c53a8e9f20a37444008606b57d6b7a826d"},
+    {file = "numpy-2.5.1-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:e68d8dd1e7eba712948f2053a29ec86917bc70ba1358df869d9f06649ef9cf09"},
+    {file = "numpy-2.5.1-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:99d5095fa265a0c4152e7bb12759e14381ef5496152f1ce58f44bdf55c44beb4"},
+    {file = "numpy-2.5.1-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:ab87a91b3cc3382b8956095bd8f95e00cf679bb81554339be1a2ba404a1473c1"},
+    {file = "numpy-2.5.1-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:224ca51130ef7da85bea2191625181cb4f337f9cb64b471f10c1a12aa8b60077"},
+    {file = "numpy-2.5.1-cp314-cp314t-win32.whl", hash = "sha256:6eab239876581b2b3c5a242281b6007bbdbcd1c7085d7709bb57c5929b11e6bf"},
+    {file = "numpy-2.5.1-cp314-cp314t-win_amd64.whl", hash = "sha256:83ce9c80d5b521b0d77ddcbe5447c218d247929b6cc056ca5351342accfff0af"},
+    {file = "numpy-2.5.1-cp314-cp314t-win_arm64.whl", hash = "sha256:5a6db61f9aaa57e369905c67d852045d3c4f7126405b29d09b19dec118e9c9cb"},
+    {file = "numpy-2.5.1.tar.gz", hash = "sha256:a48a113e6afea91f5608793bafa7ef2ad481fefbda87ec5069f483de61cb9fa3"},
 ]
 
 [[package]]
@@ -839,7 +568,10 @@ files = [
 ]
 
 [package.dependencies]
-numpy = {version = ">=1.23.2", markers = "python_version == \"3.11\""}
+numpy = [
+    {version = ">=1.23.2", markers = "python_version == \"3.11\""},
+    {version = ">=1.26.0", markers = "python_version >= \"3.12\""},
+]
 python-dateutil = ">=2.8.2"
 pytz = ">=2020.1"
 tzdata = ">=2022.7"
@@ -869,106 +601,6 @@ sql-other = ["SQLAlchemy (>=2.0.0)", "adbc-driver-postgresql (>=0.8.0)", "adbc-d
 test = ["hypothesis (>=6.46.1)", "pytest (>=7.3.2)", "pytest-xdist (>=2.2.0)"]
 xml = ["lxml (>=4.9.2)"]
 
-[[package]]
-name = "pillow"
-version = "11.2.1"
-description = "Python Imaging Library (Fork)"
-optional = false
-python-versions = ">=3.9"
-groups = ["main"]
-files = [
-    {file = "pillow-11.2.1-cp310-cp310-macosx_10_10_x86_64.whl", hash = "sha256:d57a75d53922fc20c165016a20d9c44f73305e67c351bbc60d1adaf662e74047"},
-    {file = "pillow-11.2.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:127bf6ac4a5b58b3d32fc8289656f77f80567d65660bc46f72c0d77e6600cc95"},
-    {file = "pillow-11.2.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b4ba4be812c7a40280629e55ae0b14a0aafa150dd6451297562e1764808bbe61"},
-    {file = "pillow-11.2.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:c8bd62331e5032bc396a93609982a9ab6b411c05078a52f5fe3cc59234a3abd1"},
-    {file = "pillow-11.2.1-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:562d11134c97a62fe3af29581f083033179f7ff435f78392565a1ad2d1c2c45c"},
-    {file = "pillow-11.2.1-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:c97209e85b5be259994eb5b69ff50c5d20cca0f458ef9abd835e262d9d88b39d"},
-    {file = "pillow-11.2.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:0c3e6d0f59171dfa2e25d7116217543310908dfa2770aa64b8f87605f8cacc97"},
-    {file = "pillow-11.2.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:cc1c3bc53befb6096b84165956e886b1729634a799e9d6329a0c512ab651e579"},
-    {file = "pillow-11.2.1-cp310-cp310-win32.whl", hash = "sha256:312c77b7f07ab2139924d2639860e084ec2a13e72af54d4f08ac843a5fc9c79d"},
-    {file = "pillow-11.2.1-cp310-cp310-win_amd64.whl", hash = "sha256:9bc7ae48b8057a611e5fe9f853baa88093b9a76303937449397899385da06fad"},
-    {file = "pillow-11.2.1-cp310-cp310-win_arm64.whl", hash = "sha256:2728567e249cdd939f6cc3d1f049595c66e4187f3c34078cbc0a7d21c47482d2"},
-    {file = "pillow-11.2.1-cp311-cp311-macosx_10_10_x86_64.whl", hash = "sha256:35ca289f712ccfc699508c4658a1d14652e8033e9b69839edf83cbdd0ba39e70"},
-    {file = "pillow-11.2.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:e0409af9f829f87a2dfb7e259f78f317a5351f2045158be321fd135973fff7bf"},
-    {file = "pillow-11.2.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d4e5c5edee874dce4f653dbe59db7c73a600119fbea8d31f53423586ee2aafd7"},
-    {file = "pillow-11.2.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b93a07e76d13bff9444f1a029e0af2964e654bfc2e2c2d46bfd080df5ad5f3d8"},
-    {file = "pillow-11.2.1-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:e6def7eed9e7fa90fde255afaf08060dc4b343bbe524a8f69bdd2a2f0018f600"},
-    {file = "pillow-11.2.1-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:8f4f3724c068be008c08257207210c138d5f3731af6c155a81c2b09a9eb3a788"},
-    {file = "pillow-11.2.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:a0a6709b47019dff32e678bc12c63008311b82b9327613f534e496dacaefb71e"},
-    {file = "pillow-11.2.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:f6b0c664ccb879109ee3ca702a9272d877f4fcd21e5eb63c26422fd6e415365e"},
-    {file = "pillow-11.2.1-cp311-cp311-win32.whl", hash = "sha256:cc5d875d56e49f112b6def6813c4e3d3036d269c008bf8aef72cd08d20ca6df6"},
-    {file = "pillow-11.2.1-cp311-cp311-win_amd64.whl", hash = "sha256:0f5c7eda47bf8e3c8a283762cab94e496ba977a420868cb819159980b6709193"},
-    {file = "pillow-11.2.1-cp311-cp311-win_arm64.whl", hash = "sha256:4d375eb838755f2528ac8cbc926c3e31cc49ca4ad0cf79cff48b20e30634a4a7"},
-    {file = "pillow-11.2.1-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:78afba22027b4accef10dbd5eed84425930ba41b3ea0a86fa8d20baaf19d807f"},
-    {file = "pillow-11.2.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:78092232a4ab376a35d68c4e6d5e00dfd73454bd12b230420025fbe178ee3b0b"},
-    {file = "pillow-11.2.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:25a5f306095c6780c52e6bbb6109624b95c5b18e40aab1c3041da3e9e0cd3e2d"},
-    {file = "pillow-11.2.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:0c7b29dbd4281923a2bfe562acb734cee96bbb129e96e6972d315ed9f232bef4"},
-    {file = "pillow-11.2.1-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:3e645b020f3209a0181a418bffe7b4a93171eef6c4ef6cc20980b30bebf17b7d"},
-    {file = "pillow-11.2.1-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:b2dbea1012ccb784a65349f57bbc93730b96e85b42e9bf7b01ef40443db720b4"},
-    {file = "pillow-11.2.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:da3104c57bbd72948d75f6a9389e6727d2ab6333c3617f0a89d72d4940aa0443"},
-    {file = "pillow-11.2.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:598174aef4589af795f66f9caab87ba4ff860ce08cd5bb447c6fc553ffee603c"},
-    {file = "pillow-11.2.1-cp312-cp312-win32.whl", hash = "sha256:1d535df14716e7f8776b9e7fee118576d65572b4aad3ed639be9e4fa88a1cad3"},
-    {file = "pillow-11.2.1-cp312-cp312-win_amd64.whl", hash = "sha256:14e33b28bf17c7a38eede290f77db7c664e4eb01f7869e37fa98a5aa95978941"},
-    {file = "pillow-11.2.1-cp312-cp312-win_arm64.whl", hash = "sha256:21e1470ac9e5739ff880c211fc3af01e3ae505859392bf65458c224d0bf283eb"},
-    {file = "pillow-11.2.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:fdec757fea0b793056419bca3e9932eb2b0ceec90ef4813ea4c1e072c389eb28"},
-    {file = "pillow-11.2.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:b0e130705d568e2f43a17bcbe74d90958e8a16263868a12c3e0d9c8162690830"},
-    {file = "pillow-11.2.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:7bdb5e09068332578214cadd9c05e3d64d99e0e87591be22a324bdbc18925be0"},
-    {file = "pillow-11.2.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d189ba1bebfbc0c0e529159631ec72bb9e9bc041f01ec6d3233d6d82eb823bc1"},
-    {file = "pillow-11.2.1-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:191955c55d8a712fab8934a42bfefbf99dd0b5875078240943f913bb66d46d9f"},
-    {file = "pillow-11.2.1-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:ad275964d52e2243430472fc5d2c2334b4fc3ff9c16cb0a19254e25efa03a155"},
-    {file = "pillow-11.2.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:750f96efe0597382660d8b53e90dd1dd44568a8edb51cb7f9d5d918b80d4de14"},
-    {file = "pillow-11.2.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:fe15238d3798788d00716637b3d4e7bb6bde18b26e5d08335a96e88564a36b6b"},
-    {file = "pillow-11.2.1-cp313-cp313-win32.whl", hash = "sha256:3fe735ced9a607fee4f481423a9c36701a39719252a9bb251679635f99d0f7d2"},
-    {file = "pillow-11.2.1-cp313-cp313-win_amd64.whl", hash = "sha256:74ee3d7ecb3f3c05459ba95eed5efa28d6092d751ce9bf20e3e253a4e497e691"},
-    {file = "pillow-11.2.1-cp313-cp313-win_arm64.whl", hash = "sha256:5119225c622403afb4b44bad4c1ca6c1f98eed79db8d3bc6e4e160fc6339d66c"},
-    {file = "pillow-11.2.1-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:8ce2e8411c7aaef53e6bb29fe98f28cd4fbd9a1d9be2eeea434331aac0536b22"},
-    {file = "pillow-11.2.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:9ee66787e095127116d91dea2143db65c7bb1e232f617aa5957c0d9d2a3f23a7"},
-    {file = "pillow-11.2.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9622e3b6c1d8b551b6e6f21873bdcc55762b4b2126633014cea1803368a9aa16"},
-    {file = "pillow-11.2.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:63b5dff3a68f371ea06025a1a6966c9a1e1ee452fc8020c2cd0ea41b83e9037b"},
-    {file = "pillow-11.2.1-cp313-cp313t-manylinux_2_28_aarch64.whl", hash = "sha256:31df6e2d3d8fc99f993fd253e97fae451a8db2e7207acf97859732273e108406"},
-    {file = "pillow-11.2.1-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:062b7a42d672c45a70fa1f8b43d1d38ff76b63421cbbe7f88146b39e8a558d91"},
-    {file = "pillow-11.2.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:4eb92eca2711ef8be42fd3f67533765d9fd043b8c80db204f16c8ea62ee1a751"},
-    {file = "pillow-11.2.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:f91ebf30830a48c825590aede79376cb40f110b387c17ee9bd59932c961044f9"},
-    {file = "pillow-11.2.1-cp313-cp313t-win32.whl", hash = "sha256:e0b55f27f584ed623221cfe995c912c61606be8513bfa0e07d2c674b4516d9dd"},
-    {file = "pillow-11.2.1-cp313-cp313t-win_amd64.whl", hash = "sha256:36d6b82164c39ce5482f649b437382c0fb2395eabc1e2b1702a6deb8ad647d6e"},
-    {file = "pillow-11.2.1-cp313-cp313t-win_arm64.whl", hash = "sha256:225c832a13326e34f212d2072982bb1adb210e0cc0b153e688743018c94a2681"},
-    {file = "pillow-11.2.1-cp39-cp39-macosx_10_10_x86_64.whl", hash = "sha256:7491cf8a79b8eb867d419648fff2f83cb0b3891c8b36da92cc7f1931d46108c8"},
-    {file = "pillow-11.2.1-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:8b02d8f9cb83c52578a0b4beadba92e37d83a4ef11570a8688bbf43f4ca50909"},
-    {file = "pillow-11.2.1-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:014ca0050c85003620526b0ac1ac53f56fc93af128f7546623cc8e31875ab928"},
-    {file = "pillow-11.2.1-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3692b68c87096ac6308296d96354eddd25f98740c9d2ab54e1549d6c8aea9d79"},
-    {file = "pillow-11.2.1-cp39-cp39-manylinux_2_28_aarch64.whl", hash = "sha256:f781dcb0bc9929adc77bad571b8621ecb1e4cdef86e940fe2e5b5ee24fd33b35"},
-    {file = "pillow-11.2.1-cp39-cp39-manylinux_2_28_x86_64.whl", hash = "sha256:2b490402c96f907a166615e9a5afacf2519e28295f157ec3a2bb9bd57de638cb"},
-    {file = "pillow-11.2.1-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:dd6b20b93b3ccc9c1b597999209e4bc5cf2853f9ee66e3fc9a400a78733ffc9a"},
-    {file = "pillow-11.2.1-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:4b835d89c08a6c2ee7781b8dd0a30209a8012b5f09c0a665b65b0eb3560b6f36"},
-    {file = "pillow-11.2.1-cp39-cp39-win32.whl", hash = "sha256:b10428b3416d4f9c61f94b494681280be7686bda15898a3a9e08eb66a6d92d67"},
-    {file = "pillow-11.2.1-cp39-cp39-win_amd64.whl", hash = "sha256:6ebce70c3f486acf7591a3d73431fa504a4e18a9b97ff27f5f47b7368e4b9dd1"},
-    {file = "pillow-11.2.1-cp39-cp39-win_arm64.whl", hash = "sha256:c27476257b2fdcd7872d54cfd119b3a9ce4610fb85c8e32b70b42e3680a29a1e"},
-    {file = "pillow-11.2.1-pp310-pypy310_pp73-macosx_10_15_x86_64.whl", hash = "sha256:9b7b0d4fd2635f54ad82785d56bc0d94f147096493a79985d0ab57aedd563156"},
-    {file = "pillow-11.2.1-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:aa442755e31c64037aa7c1cb186e0b369f8416c567381852c63444dd666fb772"},
-    {file = "pillow-11.2.1-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f0d3348c95b766f54b76116d53d4cb171b52992a1027e7ca50c81b43b9d9e363"},
-    {file = "pillow-11.2.1-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:85d27ea4c889342f7e35f6d56e7e1cb345632ad592e8c51b693d7b7556043ce0"},
-    {file = "pillow-11.2.1-pp310-pypy310_pp73-manylinux_2_28_aarch64.whl", hash = "sha256:bf2c33d6791c598142f00c9c4c7d47f6476731c31081331664eb26d6ab583e01"},
-    {file = "pillow-11.2.1-pp310-pypy310_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:e616e7154c37669fc1dfc14584f11e284e05d1c650e1c0f972f281c4ccc53193"},
-    {file = "pillow-11.2.1-pp310-pypy310_pp73-win_amd64.whl", hash = "sha256:39ad2e0f424394e3aebc40168845fee52df1394a4673a6ee512d840d14ab3013"},
-    {file = "pillow-11.2.1-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:80f1df8dbe9572b4b7abdfa17eb5d78dd620b1d55d9e25f834efdbee872d3aed"},
-    {file = "pillow-11.2.1-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:ea926cfbc3957090becbcbbb65ad177161a2ff2ad578b5a6ec9bb1e1cd78753c"},
-    {file = "pillow-11.2.1-pp311-pypy311_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:738db0e0941ca0376804d4de6a782c005245264edaa253ffce24e5a15cbdc7bd"},
-    {file = "pillow-11.2.1-pp311-pypy311_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9db98ab6565c69082ec9b0d4e40dd9f6181dab0dd236d26f7a50b8b9bfbd5076"},
-    {file = "pillow-11.2.1-pp311-pypy311_pp73-manylinux_2_28_aarch64.whl", hash = "sha256:036e53f4170e270ddb8797d4c590e6dd14d28e15c7da375c18978045f7e6c37b"},
-    {file = "pillow-11.2.1-pp311-pypy311_pp73-manylinux_2_28_x86_64.whl", hash = "sha256:14f73f7c291279bd65fda51ee87affd7c1e097709f7fdd0188957a16c264601f"},
-    {file = "pillow-11.2.1-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:208653868d5c9ecc2b327f9b9ef34e0e42a4cdd172c2988fd81d62d2bc9bc044"},
-    {file = "pillow-11.2.1.tar.gz", hash = "sha256:a64dd61998416367b7ef979b73d3a85853ba9bec4c2925f74e588879a58716b6"},
-]
-
-[package.extras]
-docs = ["furo", "olefile", "sphinx (>=8.2)", "sphinx-copybutton", "sphinx-inline-tabs", "sphinxext-opengraph"]
-fpx = ["olefile"]
-mic = ["olefile"]
-test-arrow = ["pyarrow"]
-tests = ["check-manifest", "coverage (>=7.4.2)", "defusedxml", "markdown2", "olefile", "packaging", "pyroma", "pytest", "pytest-cov", "pytest-timeout", "trove-classifiers (>=2024.10.12)"]
-typing = ["typing-extensions"]
-xmp = ["defusedxml"]
-
 [[package]]
 name = "portalocker"
 version = "2.10.1"
@@ -1035,82 +667,7 @@ files = [
 ]
 
 [package.extras]
-test = ["enum34", "ipaddress", "mock", "pywin32", "wmi"]
-
-[[package]]
-name = "py-rust-stemmers"
-version = "0.1.5"
-description = "Fast and parallel snowball stemmer"
-optional = false
-python-versions = "*"
-groups = ["main"]
-files = [
-    {file = "py_rust_stemmers-0.1.5-cp310-cp310-macosx_10_12_x86_64.whl", hash = "sha256:bfbd9034ae00419ff2154e33b8f5b4c4d99d1f9271f31ed059e5c7e9fa005844"},
-    {file = "py_rust_stemmers-0.1.5-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:c7162ae66df2bb0fc39b350c24a049f5f5151c03c046092ba095c2141ec223a2"},
-    {file = "py_rust_stemmers-0.1.5-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:da6de2b694af6227ba8c5a0447d4e0ef69991e63ee558b969f90c415f33e54d0"},
-    {file = "py_rust_stemmers-0.1.5-cp310-cp310-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:a3abbd6d26722951a04550fff55460c0f26819169c23286e11ea25c645be6140"},
-    {file = "py_rust_stemmers-0.1.5-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:019221c57a7bcc51097fa3f124b62d0577b5b6167184ee51abd3aea822d78f69"},
-    {file = "py_rust_stemmers-0.1.5-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:8dd5824194c279ee07f2675a55b3d728dfeec69a4b3c27329fab9b2ff5063c91"},
-    {file = "py_rust_stemmers-0.1.5-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:7cf4d69bf20cec373ba0e89df3d98549b1a0cfb130dbd859a50ed772dd044546"},
-    {file = "py_rust_stemmers-0.1.5-cp310-cp310-musllinux_1_2_armv7l.whl", hash = "sha256:b42eb52609ac958e7fcc441395457dc5183397e8014e954f4aed78de210837b9"},
-    {file = "py_rust_stemmers-0.1.5-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:c836aeb53409a44f38b153106374fe780099a7c976c582c5ae952061ff5d2fed"},
-    {file = "py_rust_stemmers-0.1.5-cp310-none-win_amd64.whl", hash = "sha256:39550089f7a021a3a97fec2ff0d4ad77e471f0a65c0f100919555e60a4daabf0"},
-    {file = "py_rust_stemmers-0.1.5-cp311-cp311-macosx_10_12_x86_64.whl", hash = "sha256:e644987edaf66919f5a9e4693336930f98d67b790857890623a431bb77774c84"},
-    {file = "py_rust_stemmers-0.1.5-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:910d87d39ba75da1fe3d65df88b926b4b454ada8d73893cbd36e258a8a648158"},
-    {file = "py_rust_stemmers-0.1.5-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:31ff4fb9417cec35907c18a6463e3d5a4941a5aa8401f77fbb4156b3ada69e3f"},
-    {file = "py_rust_stemmers-0.1.5-cp311-cp311-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:07b3b8582313ef8a7f544acf2c887f27c3dd48c5ddca028fa0f498de7380e24f"},
-    {file = "py_rust_stemmers-0.1.5-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:804944eeb5c5559443d81f30c34d6e83c6292d72423f299e42f9d71b9d240941"},
-    {file = "py_rust_stemmers-0.1.5-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:c52c5c326de78c70cfc71813fa56818d1bd4894264820d037d2be0e805b477bd"},
-    {file = "py_rust_stemmers-0.1.5-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:d8f374c0f26ef35fb87212686add8dff394bcd9a1364f14ce40fe11504e25e30"},
-    {file = "py_rust_stemmers-0.1.5-cp311-cp311-musllinux_1_2_armv7l.whl", hash = "sha256:0ae0540453843bc36937abb54fdbc0d5d60b51ef47aa9667afd05af9248e09eb"},
-    {file = "py_rust_stemmers-0.1.5-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:85944262c248ea30444155638c9e148a3adc61fe51cf9a3705b4055b564ec95d"},
-    {file = "py_rust_stemmers-0.1.5-cp311-none-win_amd64.whl", hash = "sha256:147234020b3eefe6e1a962173e41d8cf1dbf5d0689f3cd60e3022d1ac5c2e203"},
-    {file = "py_rust_stemmers-0.1.5-cp312-cp312-macosx_10_12_x86_64.whl", hash = "sha256:29772837126a28263bf54ecd1bc709dd569d15a94d5e861937813ce51e8a6df4"},
-    {file = "py_rust_stemmers-0.1.5-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:4d62410ada44a01e02974b85d45d82f4b4c511aae9121e5f3c1ba1d0bea9126b"},
-    {file = "py_rust_stemmers-0.1.5-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b28ef729a4c83c7d9418be3c23c0372493fcccc67e86783ff04596ef8a208cdf"},
-    {file = "py_rust_stemmers-0.1.5-cp312-cp312-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:a979c3f4ff7ad94a0d4cf566ca7bfecebb59e66488cc158e64485cf0c9a7879f"},
-    {file = "py_rust_stemmers-0.1.5-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1c3593d895453fa06bf70a7b76d6f00d06def0f91fc253fe4260920650c5e078"},
-    {file = "py_rust_stemmers-0.1.5-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:96ccc7fd042ffc3f7f082f2223bb7082ed1423aa6b43d5d89ab23e321936c045"},
-    {file = "py_rust_stemmers-0.1.5-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:ef18cfced2c9c676e0d7d172ba61c3fab2aa6969db64cc8f5ca33a7759efbefe"},
-    {file = "py_rust_stemmers-0.1.5-cp312-cp312-musllinux_1_2_armv7l.whl", hash = "sha256:541d4b5aa911381e3d37ec483abb6a2cf2351b4f16d5e8d77f9aa2722956662a"},
-    {file = "py_rust_stemmers-0.1.5-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:ffd946a36e9ac17ca96821963663012e04bc0ee94d21e8b5ae034721070b436c"},
-    {file = "py_rust_stemmers-0.1.5-cp312-none-win_amd64.whl", hash = "sha256:6ed61e1207f3b7428e99b5d00c055645c6415bb75033bff2d06394cbe035fd8e"},
-    {file = "py_rust_stemmers-0.1.5-cp313-cp313-macosx_10_12_x86_64.whl", hash = "sha256:398b3a843a9cd4c5d09e726246bc36f66b3d05b0a937996814e91f47708f5db5"},
-    {file = "py_rust_stemmers-0.1.5-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:4e308fc7687901f0c73603203869908f3156fa9c17c4ba010a7fcc98a7a1c5f2"},
-    {file = "py_rust_stemmers-0.1.5-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1f9efc4da5e734bdd00612e7506de3d0c9b7abc4b89d192742a0569d0d1fe749"},
-    {file = "py_rust_stemmers-0.1.5-cp313-cp313-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:cc2cc8d2b36bc05b8b06506199ac63d437360ae38caefd98cd19e479d35afd42"},
-    {file = "py_rust_stemmers-0.1.5-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a231dc6f0b2a5f12a080dfc7abd9e6a4ea0909290b10fd0a4620e5a0f52c3d17"},
-    {file = "py_rust_stemmers-0.1.5-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:5845709d48afc8b29e248f42f92431155a3d8df9ba30418301c49c6072b181b0"},
-    {file = "py_rust_stemmers-0.1.5-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:e48bfd5e3ce9d223bfb9e634dc1425cf93ee57eef6f56aa9a7120ada3990d4be"},
-    {file = "py_rust_stemmers-0.1.5-cp313-cp313-musllinux_1_2_armv7l.whl", hash = "sha256:35d32f6e7bdf6fd90e981765e32293a8be74def807147dea9fdc1f65d6ce382f"},
-    {file = "py_rust_stemmers-0.1.5-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:191ea8bf922c984631ffa20bf02ef0ad7eec0465baeaed3852779e8f97c7e7a3"},
-    {file = "py_rust_stemmers-0.1.5-cp313-none-win_amd64.whl", hash = "sha256:e564c9efdbe7621704e222b53bac265b0e4fbea788f07c814094f0ec6b80adcf"},
-    {file = "py_rust_stemmers-0.1.5-cp38-cp38-macosx_10_12_x86_64.whl", hash = "sha256:7720c4d472653f7301537fb289d10f827b25c9b998d1b58403181180097212ee"},
-    {file = "py_rust_stemmers-0.1.5-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:53e2ad505d15959fd86a8b204e55fd73290cf5fdba0020fd0d9323d7fe225962"},
-    {file = "py_rust_stemmers-0.1.5-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:fa7d6bd88cb99933178bc1e9f803d921c13274a2fe52325d40f8d35046e929c3"},
-    {file = "py_rust_stemmers-0.1.5-cp38-cp38-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:f240be882e9afcc8eeb5860c13bf37cf666d99179da49b0cd19ac3cbb4871423"},
-    {file = "py_rust_stemmers-0.1.5-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:52c397e314a02319d7a0f7d2cd845557a3ba145f8318a6f2e119b46056bf4379"},
-    {file = "py_rust_stemmers-0.1.5-cp38-cp38-manylinux_2_28_x86_64.whl", hash = "sha256:b5ae7a3d1d015e72817f8fe6b35bb762681b7197fee2d80232528fff68be7576"},
-    {file = "py_rust_stemmers-0.1.5-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:f161278168a7ad5ace2909454f2ee87d2c815bfd1bfec5a80fac36de31bf96e4"},
-    {file = "py_rust_stemmers-0.1.5-cp38-cp38-musllinux_1_2_armv7l.whl", hash = "sha256:918ce5252570febdf8accded210046e7f3edb933eac5599fa40e773ff42c7a8f"},
-    {file = "py_rust_stemmers-0.1.5-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:0db3d2fff5775060c78a67e72132fa8802ace50175da26bf4661e4071aa10094"},
-    {file = "py_rust_stemmers-0.1.5-cp38-none-win_amd64.whl", hash = "sha256:9982b5f915e8e5b7ca83104f57cc4e7668b900d87232eea2c6f432d2009f0d18"},
-    {file = "py_rust_stemmers-0.1.5-cp39-cp39-macosx_10_12_x86_64.whl", hash = "sha256:67b2753cadb0bdb827ae8088ef5ead44408b8ff92d2bc8926231c7e056810e4a"},
-    {file = "py_rust_stemmers-0.1.5-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:68f1fd369c4be5d63c2f697a07511c868f41875a464aab0f564071131d84bab4"},
-    {file = "py_rust_stemmers-0.1.5-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ec840a009de79a9a8231090222a8fdc07043d337e2ef38843daae88f0479ae7f"},
-    {file = "py_rust_stemmers-0.1.5-cp39-cp39-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:1816ade2c5148337cec16c115e210c8e3341c4cde5b591bf43224bd2168e5eda"},
-    {file = "py_rust_stemmers-0.1.5-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d095b359d6e14e219b6fbff074f7b53e845f70e6d7fc9e63b50f192677c94cf5"},
-    {file = "py_rust_stemmers-0.1.5-cp39-cp39-manylinux_2_28_x86_64.whl", hash = "sha256:5bc04bd5f72ee5d9cfb1530c56e450de7d80ed4504d5013f02f522cf8ea9b474"},
-    {file = "py_rust_stemmers-0.1.5-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:7687586cf4e34f230ee154b5f034b9c568409aaef997c888d4749e4bdbab79fd"},
-    {file = "py_rust_stemmers-0.1.5-cp39-cp39-musllinux_1_2_armv7l.whl", hash = "sha256:544e3e5e0174924f2dce3248094d9632c3106c1619a03141dbfb8bfdb29b0925"},
-    {file = "py_rust_stemmers-0.1.5-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:77b96f1141de6db8b550052ef7f90099e3c3e10bb0d472a4af7a30af776252f8"},
-    {file = "py_rust_stemmers-0.1.5-cp39-none-win_amd64.whl", hash = "sha256:c3078476318a4697dc216bb6c615b59d1ee02058202c2fa91ccfb4bebf412e22"},
-    {file = "py_rust_stemmers-0.1.5-pp310-pypy310_pp73-macosx_10_12_x86_64.whl", hash = "sha256:f8c6596f04e7a6df2a5cc18854d31b133d2a69a8c494fa49853fe174d8739d14"},
-    {file = "py_rust_stemmers-0.1.5-pp310-pypy310_pp73-macosx_11_0_arm64.whl", hash = "sha256:154c27f5d576fabf2bacf53620f014562af4c6cf9eb09ba7477830f2be868902"},
-    {file = "py_rust_stemmers-0.1.5-pp310-pypy310_pp73-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ec42b66927b62fd57328980b6c7004fe85e8fad89c952e8718da68b805a119e3"},
-    {file = "py_rust_stemmers-0.1.5-pp310-pypy310_pp73-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:57b061c3b4af9e409d009d729b21bc53dabe47116c955ccf0b642a5a2d438f93"},
-    {file = "py_rust_stemmers-0.1.5.tar.gz", hash = "sha256:e9c310cfb5c2470d7c7c8a0484725965e7cab8b1237e106a0863d5741da3e1f7"},
-]
+test = ["enum34 ; python_version <= \"3.4\"", "ipaddress ; python_version < \"3.0\"", "mock ; python_version < \"3.0\"", "pywin32 ; sys_platform == \"win32\"", "wmi ; sys_platform == \"win32\""]
 
 [[package]]
 name = "pydantic"
@@ -1132,7 +689,7 @@ typing-inspection = ">=0.4.0"
 
 [package.extras]
 email = ["email-validator (>=2.0.0)"]
-timezone = ["tzdata"]
+timezone = ["tzdata ; python_version >= \"3.9\" and platform_system == \"Windows\""]
 
 [[package]]
 name = "pydantic-core"
@@ -1246,22 +803,6 @@ files = [
 [package.dependencies]
 typing-extensions = ">=4.6.0,<4.7.0 || >4.7.0"
 
-[[package]]
-name = "pyreadline3"
-version = "3.5.4"
-description = "A python implementation of GNU readline."
-optional = false
-python-versions = ">=3.8"
-groups = ["main"]
-markers = "sys_platform == \"win32\""
-files = [
-    {file = "pyreadline3-3.5.4-py3-none-any.whl", hash = "sha256:eaf8e6cc3c49bcccf145fc6067ba8643d1df34d604a1ec0eccbf7a18e6d3fae6"},
-    {file = "pyreadline3-3.5.4.tar.gz", hash = "sha256:8d57d53039a1c75adba8e50dd3d992b28143480816187ea5efbd5c78e6c885b7"},
-]
-
-[package.extras]
-dev = ["build", "flake8", "mypy", "pytest", "twine"]
-
 [[package]]
 name = "python-dateutil"
 version = "2.9.0.post0"
@@ -1316,94 +857,35 @@ files = [
     {file = "pywin32-310-cp39-cp39-win_amd64.whl", hash = "sha256:96867217335559ac619f00ad70e513c0fcf84b8a3af9fc2bba3b59b97da70475"},
 ]
 
-[[package]]
-name = "pyyaml"
-version = "6.0.2"
-description = "YAML parser and emitter for Python"
-optional = false
-python-versions = ">=3.8"
-groups = ["main"]
-files = [
-    {file = "PyYAML-6.0.2-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:0a9a2848a5b7feac301353437eb7d5957887edbf81d56e903999a75a3d743086"},
-    {file = "PyYAML-6.0.2-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:29717114e51c84ddfba879543fb232a6ed60086602313ca38cce623c1d62cfbf"},
-    {file = "PyYAML-6.0.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:8824b5a04a04a047e72eea5cec3bc266db09e35de6bdfe34c9436ac5ee27d237"},
-    {file = "PyYAML-6.0.2-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:7c36280e6fb8385e520936c3cb3b8042851904eba0e58d277dca80a5cfed590b"},
-    {file = "PyYAML-6.0.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ec031d5d2feb36d1d1a24380e4db6d43695f3748343d99434e6f5f9156aaa2ed"},
-    {file = "PyYAML-6.0.2-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:936d68689298c36b53b29f23c6dbb74de12b4ac12ca6cfe0e047bedceea56180"},
-    {file = "PyYAML-6.0.2-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:23502f431948090f597378482b4812b0caae32c22213aecf3b55325e049a6c68"},
-    {file = "PyYAML-6.0.2-cp310-cp310-win32.whl", hash = "sha256:2e99c6826ffa974fe6e27cdb5ed0021786b03fc98e5ee3c5bfe1fd5015f42b99"},
-    {file = "PyYAML-6.0.2-cp310-cp310-win_amd64.whl", hash = "sha256:a4d3091415f010369ae4ed1fc6b79def9416358877534caf6a0fdd2146c87a3e"},
-    {file = "PyYAML-6.0.2-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:cc1c1159b3d456576af7a3e4d1ba7e6924cb39de8f67111c735f6fc832082774"},
-    {file = "PyYAML-6.0.2-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:1e2120ef853f59c7419231f3bf4e7021f1b936f6ebd222406c3b60212205d2ee"},
-    {file = "PyYAML-6.0.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:5d225db5a45f21e78dd9358e58a98702a0302f2659a3c6cd320564b75b86f47c"},
-    {file = "PyYAML-6.0.2-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:5ac9328ec4831237bec75defaf839f7d4564be1e6b25ac710bd1a96321cc8317"},
-    {file = "PyYAML-6.0.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3ad2a3decf9aaba3d29c8f537ac4b243e36bef957511b4766cb0057d32b0be85"},
-    {file = "PyYAML-6.0.2-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:ff3824dc5261f50c9b0dfb3be22b4567a6f938ccce4587b38952d85fd9e9afe4"},
-    {file = "PyYAML-6.0.2-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:797b4f722ffa07cc8d62053e4cff1486fa6dc094105d13fea7b1de7d8bf71c9e"},
-    {file = "PyYAML-6.0.2-cp311-cp311-win32.whl", hash = "sha256:11d8f3dd2b9c1207dcaf2ee0bbbfd5991f571186ec9cc78427ba5bd32afae4b5"},
-    {file = "PyYAML-6.0.2-cp311-cp311-win_amd64.whl", hash = "sha256:e10ce637b18caea04431ce14fabcf5c64a1c61ec9c56b071a4b7ca131ca52d44"},
-    {file = "PyYAML-6.0.2-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:c70c95198c015b85feafc136515252a261a84561b7b1d51e3384e0655ddf25ab"},
-    {file = "PyYAML-6.0.2-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:ce826d6ef20b1bc864f0a68340c8b3287705cae2f8b4b1d932177dcc76721725"},
-    {file = "PyYAML-6.0.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1f71ea527786de97d1a0cc0eacd1defc0985dcf6b3f17bb77dcfc8c34bec4dc5"},
-    {file = "PyYAML-6.0.2-cp312-cp312-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:9b22676e8097e9e22e36d6b7bda33190d0d400f345f23d4065d48f4ca7ae0425"},
-    {file = "PyYAML-6.0.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:80bab7bfc629882493af4aa31a4cfa43a4c57c83813253626916b8c7ada83476"},
-    {file = "PyYAML-6.0.2-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:0833f8694549e586547b576dcfaba4a6b55b9e96098b36cdc7ebefe667dfed48"},
-    {file = "PyYAML-6.0.2-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:8b9c7197f7cb2738065c481a0461e50ad02f18c78cd75775628afb4d7137fb3b"},
-    {file = "PyYAML-6.0.2-cp312-cp312-win32.whl", hash = "sha256:ef6107725bd54b262d6dedcc2af448a266975032bc85ef0172c5f059da6325b4"},
-    {file = "PyYAML-6.0.2-cp312-cp312-win_amd64.whl", hash = "sha256:7e7401d0de89a9a855c839bc697c079a4af81cf878373abd7dc625847d25cbd8"},
-    {file = "PyYAML-6.0.2-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:efdca5630322a10774e8e98e1af481aad470dd62c3170801852d752aa7a783ba"},
-    {file = "PyYAML-6.0.2-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:50187695423ffe49e2deacb8cd10510bc361faac997de9efef88badc3bb9e2d1"},
-    {file = "PyYAML-6.0.2-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0ffe8360bab4910ef1b9e87fb812d8bc0a308b0d0eef8c8f44e0254ab3b07133"},
-    {file = "PyYAML-6.0.2-cp313-cp313-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:17e311b6c678207928d649faa7cb0d7b4c26a0ba73d41e99c4fff6b6c3276484"},
-    {file = "PyYAML-6.0.2-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:70b189594dbe54f75ab3a1acec5f1e3faa7e8cf2f1e08d9b561cb41b845f69d5"},
-    {file = "PyYAML-6.0.2-cp313-cp313-musllinux_1_1_aarch64.whl", hash = "sha256:41e4e3953a79407c794916fa277a82531dd93aad34e29c2a514c2c0c5fe971cc"},
-    {file = "PyYAML-6.0.2-cp313-cp313-musllinux_1_1_x86_64.whl", hash = "sha256:68ccc6023a3400877818152ad9a1033e3db8625d899c72eacb5a668902e4d652"},
-    {file = "PyYAML-6.0.2-cp313-cp313-win32.whl", hash = "sha256:bc2fa7c6b47d6bc618dd7fb02ef6fdedb1090ec036abab80d4681424b84c1183"},
-    {file = "PyYAML-6.0.2-cp313-cp313-win_amd64.whl", hash = "sha256:8388ee1976c416731879ac16da0aff3f63b286ffdd57cdeb95f3f2e085687563"},
-    {file = "PyYAML-6.0.2-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:24471b829b3bf607e04e88d79542a9d48bb037c2267d7927a874e6c205ca7e9a"},
-    {file = "PyYAML-6.0.2-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d7fded462629cfa4b685c5416b949ebad6cec74af5e2d42905d41e257e0869f5"},
-    {file = "PyYAML-6.0.2-cp38-cp38-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:d84a1718ee396f54f3a086ea0a66d8e552b2ab2017ef8b420e92edbc841c352d"},
-    {file = "PyYAML-6.0.2-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9056c1ecd25795207ad294bcf39f2db3d845767be0ea6e6a34d856f006006083"},
-    {file = "PyYAML-6.0.2-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:82d09873e40955485746739bcb8b4586983670466c23382c19cffecbf1fd8706"},
-    {file = "PyYAML-6.0.2-cp38-cp38-win32.whl", hash = "sha256:43fa96a3ca0d6b1812e01ced1044a003533c47f6ee8aca31724f78e93ccc089a"},
-    {file = "PyYAML-6.0.2-cp38-cp38-win_amd64.whl", hash = "sha256:01179a4a8559ab5de078078f37e5c1a30d76bb88519906844fd7bdea1b7729ff"},
-    {file = "PyYAML-6.0.2-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:688ba32a1cffef67fd2e9398a2efebaea461578b0923624778664cc1c914db5d"},
-    {file = "PyYAML-6.0.2-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:a8786accb172bd8afb8be14490a16625cbc387036876ab6ba70912730faf8e1f"},
-    {file = "PyYAML-6.0.2-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d8e03406cac8513435335dbab54c0d385e4a49e4945d2909a581c83647ca0290"},
-    {file = "PyYAML-6.0.2-cp39-cp39-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:f753120cb8181e736c57ef7636e83f31b9c0d1722c516f7e86cf15b7aa57ff12"},
-    {file = "PyYAML-6.0.2-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3b1fdb9dc17f5a7677423d508ab4f243a726dea51fa5e70992e59a7411c89d19"},
-    {file = "PyYAML-6.0.2-cp39-cp39-musllinux_1_1_aarch64.whl", hash = "sha256:0b69e4ce7a131fe56b7e4d770c67429700908fc0752af059838b1cfb41960e4e"},
-    {file = "PyYAML-6.0.2-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:a9f8c2e67970f13b16084e04f134610fd1d374bf477b17ec1599185cf611d725"},
-    {file = "PyYAML-6.0.2-cp39-cp39-win32.whl", hash = "sha256:6395c297d42274772abc367baaa79683958044e5d3835486c16da75d2a694631"},
-    {file = "PyYAML-6.0.2-cp39-cp39-win_amd64.whl", hash = "sha256:39693e1f8320ae4f43943590b49779ffb98acb81f788220ea932a6b6c51004d8"},
-    {file = "pyyaml-6.0.2.tar.gz", hash = "sha256:d584d9ec91ad65861cc08d42e834324ef890a082e591037abe114850ff7bbc3e"},
-]
-
 [[package]]
 name = "qdrant-client"
-version = "1.14.2"
+version = "1.19.0"
 description = "Client library for the Qdrant vector search engine"
 optional = false
-python-versions = ">=3.9"
+python-versions = ">=3.10"
 groups = ["main"]
 files = [
-    {file = "qdrant_client-1.14.2-py3-none-any.whl", hash = "sha256:7c283b1f0e71db9c21b85d898fb395791caca2a6d56ee751da96d797b001410c"},
-    {file = "qdrant_client-1.14.2.tar.gz", hash = "sha256:da5cab4d367d099d1330b6f30d45aefc8bd76f8b8f9d8fa5d4f813501b93af0d"},
+    {file = "qdrant_client-1.19.0-py3-none-any.whl", hash = "sha256:13602a2b3478a95ecdf42f97b93d7f703b63a3361cd912a04495a33a5ac14121"},
+    {file = "qdrant_client-1.19.0.tar.gz", hash = "sha256:365395a04b0a26c309b25b7d8b1c99ef2071ec9a2b74bc8a5fd3b7a3642fe963"},
 ]
 
 [package.dependencies]
-fastembed = {version = "0.6.1", optional = true, markers = "extra == \"fastembed\""}
 grpcio = ">=1.41.0"
 httpx = {version = ">=0.20.0", extras = ["http2"]}
-numpy = {version = ">=1.21", markers = "python_version >= \"3.10\" and python_version < \"3.12\""}
-portalocker = ">=2.7.0,<3.0.0"
+numpy = [
+    {version = ">=1.21", markers = "python_version == \"3.11\""},
+    {version = ">=1.26", markers = "python_version == \"3.12\""},
+    {version = ">=2.1.0", markers = "python_version == \"3.13\""},
+    {version = ">=2.3.0", markers = "python_version >= \"3.14\""},
+]
+portalocker = ">=2.7.0,<4.0"
 protobuf = ">=3.20.0"
 pydantic = ">=1.10.8,<2.0.dev0 || >2.2.0"
 urllib3 = ">=1.26.14,<3"
 
 [package.extras]
-fastembed = ["fastembed (==0.6.1)"]
-fastembed-gpu = ["fastembed-gpu (==0.6.1)"]
+fastembed = ["fastembed (>=0.8,<0.9)"]
+fastembed-gpu = ["fastembed-gpu (>=0.8,<0.9)"]
 
 [[package]]
 name = "requests"
@@ -1469,57 +951,6 @@ anyio = ">=3.4.0,<5"
 [package.extras]
 full = ["httpx (>=0.22.0)", "itsdangerous", "jinja2", "python-multipart", "pyyaml"]
 
-[[package]]
-name = "sympy"
-version = "1.13.3"
-description = "Computer algebra system (CAS) in Python"
-optional = false
-python-versions = ">=3.8"
-groups = ["main"]
-files = [
-    {file = "sympy-1.13.3-py3-none-any.whl", hash = "sha256:54612cf55a62755ee71824ce692986f23c88ffa77207b30c1368eda4a7060f73"},
-    {file = "sympy-1.13.3.tar.gz", hash = "sha256:b27fd2c6530e0ab39e275fc9b683895367e51d5da91baa8d3d64db2565fec4d9"},
-]
-
-[package.dependencies]
-mpmath = ">=1.1.0,<1.4"
-
-[package.extras]
-dev = ["hypothesis (>=6.70.0)", "pytest (>=7.1.0)"]
-
-[[package]]
-name = "tokenizers"
-version = "0.21.1"
-description = ""
-optional = false
-python-versions = ">=3.9"
-groups = ["main"]
-files = [
-    {file = "tokenizers-0.21.1-cp39-abi3-macosx_10_12_x86_64.whl", hash = "sha256:e78e413e9e668ad790a29456e677d9d3aa50a9ad311a40905d6861ba7692cf41"},
-    {file = "tokenizers-0.21.1-cp39-abi3-macosx_11_0_arm64.whl", hash = "sha256:cd51cd0a91ecc801633829fcd1fda9cf8682ed3477c6243b9a095539de4aecf3"},
-    {file = "tokenizers-0.21.1-cp39-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:28da6b72d4fb14ee200a1bd386ff74ade8992d7f725f2bde2c495a9a98cf4d9f"},
-    {file = "tokenizers-0.21.1-cp39-abi3-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:34d8cfde551c9916cb92014e040806122295a6800914bab5865deb85623931cf"},
-    {file = "tokenizers-0.21.1-cp39-abi3-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:aaa852d23e125b73d283c98f007e06d4595732104b65402f46e8ef24b588d9f8"},
-    {file = "tokenizers-0.21.1-cp39-abi3-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:a21a15d5c8e603331b8a59548bbe113564136dc0f5ad8306dd5033459a226da0"},
-    {file = "tokenizers-0.21.1-cp39-abi3-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:2fdbd4c067c60a0ac7eca14b6bd18a5bebace54eb757c706b47ea93204f7a37c"},
-    {file = "tokenizers-0.21.1-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:2dd9a0061e403546f7377df940e866c3e678d7d4e9643d0461ea442b4f89e61a"},
-    {file = "tokenizers-0.21.1-cp39-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:db9484aeb2e200c43b915a1a0150ea885e35f357a5a8fabf7373af333dcc8dbf"},
-    {file = "tokenizers-0.21.1-cp39-abi3-musllinux_1_2_armv7l.whl", hash = "sha256:ed248ab5279e601a30a4d67bdb897ecbe955a50f1e7bb62bd99f07dd11c2f5b6"},
-    {file = "tokenizers-0.21.1-cp39-abi3-musllinux_1_2_i686.whl", hash = "sha256:9ac78b12e541d4ce67b4dfd970e44c060a2147b9b2a21f509566d556a509c67d"},
-    {file = "tokenizers-0.21.1-cp39-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:e5a69c1a4496b81a5ee5d2c1f3f7fbdf95e90a0196101b0ee89ed9956b8a168f"},
-    {file = "tokenizers-0.21.1-cp39-abi3-win32.whl", hash = "sha256:1039a3a5734944e09de1d48761ade94e00d0fa760c0e0551151d4dd851ba63e3"},
-    {file = "tokenizers-0.21.1-cp39-abi3-win_amd64.whl", hash = "sha256:0f0dcbcc9f6e13e675a66d7a5f2f225a736745ce484c1a4e07476a89ccdad382"},
-    {file = "tokenizers-0.21.1.tar.gz", hash = "sha256:a1bb04dc5b448985f86ecd4b05407f5a8d97cb2c0532199b2a302a604a0165ab"},
-]
-
-[package.dependencies]
-huggingface-hub = ">=0.16.4,<1.0"
-
-[package.extras]
-dev = ["tokenizers[testing]"]
-docs = ["setuptools-rust", "sphinx", "sphinx-rtd-theme"]
-testing = ["black (==22.3)", "datasets", "numpy", "pytest", "requests", "ruff"]
-
 [[package]]
 name = "tqdm"
 version = "4.67.1"
@@ -1594,7 +1025,7 @@ files = [
 ]
 
 [package.extras]
-brotli = ["brotli (>=1.0.9)", "brotlicffi (>=0.8.0)"]
+brotli = ["brotli (>=1.0.9) ; platform_python_implementation == \"CPython\"", "brotlicffi (>=0.8.0) ; platform_python_implementation != \"CPython\""]
 h2 = ["h2 (>=4,<5)"]
 socks = ["pysocks (>=1.5.6,!=1.5.7,<2.0)"]
 zstd = ["zstandard (>=0.18.0)"]
@@ -1616,7 +1047,7 @@ click = ">=7.0"
 h11 = ">=0.8"
 
 [package.extras]
-standard = ["colorama (>=0.4)", "httptools (>=0.4.0)", "python-dotenv (>=0.13)", "pyyaml (>=5.1)", "uvloop (>=0.14.0,!=0.15.0,!=0.15.1)", "watchfiles (>=0.13)", "websockets (>=10.0)"]
+standard = ["colorama (>=0.4) ; sys_platform == \"win32\"", "httptools (>=0.4.0)", "python-dotenv (>=0.13)", "pyyaml (>=5.1)", "uvloop (>=0.14.0,!=0.15.0,!=0.15.1) ; sys_platform != \"win32\" and sys_platform != \"cygwin\" and platform_python_implementation != \"PyPy\"", "watchfiles (>=0.13)", "websockets (>=10.0)"]
 
 [[package]]
 name = "win32-setctime"
@@ -1632,9 +1063,9 @@ files = [
 ]
 
 [package.extras]
-dev = ["black (>=19.3b0)", "pytest (>=4.6.2)"]
+dev = ["black (>=19.3b0) ; python_version >= \"3.6\"", "pytest (>=4.6.2)"]
 
 [metadata]
 lock-version = "2.1"
-python-versions = "~3.11"
-content-hash = "79fdcfd8d9d08853fa0ce0715040843b463fb8d6a057a92e5e643739001a2659"
+python-versions = ">=3.11,<3.15"
+content-hash = "a966cbd5062cf781034f457b858ff28484e19907780395ce9fb3dc124bfc4728"
diff --git a/pyproject.toml b/pyproject.toml
index 09ddc65..1db6765 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -5,7 +5,7 @@ description = "Qdrant vector similarity engine demo"
 authors = ["Andrey Vasnetsov <andrey@vasnetsov.com>"]
 
 [tool.poetry.dependencies]
-python = "~3.11"
+python = ">=3.11,<3.15"
 fastapi = "^0.103.1"
 uvicorn = "^0.18.3"
 psutil = "^5.7.3"
@@ -13,7 +13,7 @@ pandas = "^2.2.3"
 loguru = ">=0.7.2"
 requests = "^2.25.1"
 tqdm = "^4.66.1"
-qdrant-client = { extras = ["fastembed"], version = "1.14.2" }
+qdrant-client = "1.19.0"
 
 [tool.poetry.dev-dependencies]
 

From c51895a09ff5db3e9564c5d05f7a50c8c7f0b331 Mon Sep 17 00:00:00 2001
From: John Kupchanko <john.kupchanko@qdrant.com>
Date: Fri, 7 Aug 2026 08:54:19 -0700
Subject: [PATCH 5/5] Lower the client timeout to 15s so slow calls fail loudly

---
 qdrant_demo/neural_searcher.py | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/qdrant_demo/neural_searcher.py b/qdrant_demo/neural_searcher.py
index 4e68eff..1372ade 100644
--- a/qdrant_demo/neural_searcher.py
+++ b/qdrant_demo/neural_searcher.py
@@ -16,9 +16,10 @@ class NeuralSearcher:
 
     def __init__(self, collection_name: str):
         self.collection_name = collection_name
-        # The first Cloud-inference call after idle loads the model server-side,
-        # which can take longer than the default client timeout.
-        timeout = int(os.environ.get("QDRANT_TIMEOUT", "60"))
+        # Fail loudly instead of hanging: measured novel-query latency is well
+        # under a second, so 15s is a generous ceiling that still surfaces a real
+        # problem quickly rather than masking it for a minute.
+        timeout = int(os.environ.get("QDRANT_TIMEOUT", "15"))
         self.qdrant_client = QdrantClient(
             url=QDRANT_URL, api_key=QDRANT_API_KEY,
             cloud_inference=CLOUD_INFERENCE, timeout=timeout,