Release embcli-jina 0.1.3

.github/workflows/ci-jina.yml

@@ -8,6 +8,7 @@ jobs:
 
     strategy:
       fail-fast: false
+      max-parallel: 1
       matrix:
         os: [ubuntu-latest, macos-latest]
         python-version: ["3.10", "3.11", "3.12", "3.13"]

packages/embcli-jina/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "embcli-jina"
-version = "0.1.3-dev"
+version = "0.1.4-dev"
 description = "jina plugin for embcli"
 readme = "README.md"
 authors = [{ name = "Tomoko Uchida", email = "tomoko.uchida.1111@gmail.com" }]

packages/embcli-jina/src/embcli_jina/jina.py

@@ -5,6 +5,7 @@
 import httpx
 from embcli_core.models import EmbeddingModel, ModelOption, ModelOptionType
 
+TIMEOUT_SEC = 3  # Default timeout for embedding requests
 COLBERT_TIMEOUT_SEC = 5  # Timeout for ColBERT model requests
 
 
@@ -81,7 +82,7 @@ def _embed_one_batch(self, input: list[str], **kwargs) -> Iterator[list[float] |
         if "embedding_type" in kwargs:
             data["embedding_type"] = kwargs["embedding_type"]
 
-        timeout = COLBERT_TIMEOUT_SEC if self.model_id == "jina-colbert-v2" else None
+        timeout = COLBERT_TIMEOUT_SEC if self.model_id == "jina-colbert-v2" else TIMEOUT_SEC
         response = httpx.post(self.endpoint, headers=headers, json=data, timeout=timeout)
         response.raise_for_status()
         for item in response.json().get("data", []):

packages/embcli-jina/src/embcli_jina/jina_multimodal.py

@@ -6,6 +6,8 @@
 import httpx
 from embcli_core.models import Modality, ModelOption, ModelOptionType, MultimodalEmbeddingModel
 
+TIMEOUT_SEC = 3  # Default timeout for embedding requests
+
 
 def image_to_base64(image_path: str) -> str:
     """Encodes an image file to a base64 string."""
@@ -74,7 +76,7 @@ def _embed_one_batch_multimodal(
         if "embedding_type" in kwargs:
             data["embedding_type"] = kwargs["embedding_type"]
 
-        response = httpx.post(self.endpoint, headers=headers, json=data)
+        response = httpx.post(self.endpoint, headers=headers, json=data, timeout=TIMEOUT_SEC)
         response.raise_for_status()
         for item in response.json().get("data", []):
             if "embedding" in item:

packages/embcli-jina/tests/embcli_jina/test_cli_embed.py

@@ -14,26 +14,28 @@
 
 @skip_if_no_api_key
 def test_embed_command_text(plugin_manager, mocker):
+    mocker.patch("embcli_jina.jina.TIMEOUT_SEC", 30)
     mocker.patch("embcli_core.cli._pm", plugin_manager)
     runner = CliRunner()
-    result = runner.invoke(embed, ["--model", "jina-v4", "flying cat"])
+    result = runner.invoke(embed, ["--model", "jina-v3", "flying cat"])
     assert result.exit_code == 0
 
     embeddings = json.loads(result.output)
     assert isinstance(embeddings, list)
-    assert len(embeddings) == 2048
+    assert len(embeddings) == 1024
     assert all(isinstance(val, float) for val in embeddings)
 
 
 @skip_if_no_api_key
 def test_embed_command_image(plugin_manager, mocker):
+    mocker.patch("embcli_jina.jina_multimodal.TIMEOUT_SEC", 60)
     mocker.patch("embcli_core.cli._pm", plugin_manager)
     runner = CliRunner()
     image_path = files("tests.embcli_jina").joinpath("flying_cat.jpeg")
-    result = runner.invoke(embed, ["--model", "jina-v4", "--image", str(image_path)])
+    result = runner.invoke(embed, ["--model", "jina-clip-v2", "--image", str(image_path)])
     assert result.exit_code == 0
 
     embeddings = json.loads(result.output)
     assert isinstance(embeddings, list)
-    assert len(embeddings) == 2048
+    assert len(embeddings) == 1024
     assert all(isinstance(val, float) for val in embeddings)

packages/embcli-jina/tests/embcli_jina/test_jina.py

@@ -26,7 +26,9 @@ def test_factory_create_invalid_model():
 
 
 @skip_if_no_api_key
-def test_embed_one_batch_yields_embeddings(jina_models):
+def test_embed_one_batch_yields_embeddings(jina_models, mocker):
+    mocker.patch("embcli_jina.jina.TIMEOUT_SEC", 30)
+    mocker.patch("embcli_jina.jina.COLBERT_TIMEOUT_SEC", 30)
     for model in jina_models:
         print(f"Testing model: {model.model_id}")
         input_data = ["hello", "world"]
@@ -43,6 +45,7 @@ def test_embed_one_batch_yields_embeddings(jina_models):
 
 @skip_if_no_api_key
 def test_embed_batch_with_options(jina_models, mocker):
+    mocker.patch("embcli_jina.jina.TIMEOUT_SEC", 30)
     mocker.patch("embcli_jina.jina.COLBERT_TIMEOUT_SEC", 30)
     input_data = ["hello", "world"]
     for model in jina_models:
@@ -68,7 +71,9 @@ def test_embed_batch_with_options(jina_models, mocker):
 
 
 @skip_if_no_api_key
-def test_embed_batch_embedding_types(jina_models):
+def test_embed_batch_embedding_types(jina_models, mocker):
+    mocker.patch("embcli_jina.jina.TIMEOUT_SEC", 30)
+    mocker.patch("embcli_jina.jina.COLBERT_TIMEOUT_SEC", 30)
     input_data = ["hello", "world"]
     for model in jina_models:
         # Test binary embedding type
@@ -90,6 +95,7 @@ def test_embed_batch_embedding_types(jina_models):
 
 @skip_if_no_api_key
 def test_embed_batch_for_ingest(jina_models, mocker):
+    mocker.patch("embcli_jina.jina.TIMEOUT_SEC", 30)
     mocker.patch("embcli_jina.jina.COLBERT_TIMEOUT_SEC", 30)
     for model in jina_models:
         input_data = ["hello", "world"]
@@ -106,6 +112,7 @@ def test_embed_batch_for_ingest(jina_models, mocker):
 
 @skip_if_no_api_key
 def test_embed_for_search(jina_models, mocker):
+    mocker.patch("embcli_jina.jina.TIMEOUT_SEC", 30)
     mocker.patch("embcli_jina.jina.COLBERT_TIMEOUT_SEC", 30)
     for model in jina_models:
         input = "hello world"

packages/embcli-jina/tests/embcli_jina/test_jina_multimodal.py

@@ -28,7 +28,8 @@ def test_factory_create_invalid_model():
 
 
 @skip_if_no_api_key
-def test_embed_one_batch_multimodal(jina_multimodal_models):
+def test_embed_one_batch_multimodal(jina_multimodal_models, mocker):
+    mocker.patch("embcli_jina.jina_multimodal.TIMEOUT_SEC", 60)
     for model in jina_multimodal_models:
         print(f"Testing model: {model.model_id}")
         input_data = ["hello", "world"]
@@ -42,7 +43,8 @@ def test_embed_one_batch_multimodal(jina_multimodal_models):
 
 
 @skip_if_no_api_key
-def test_embed_one_batch_multimodal_image(jina_multimodal_models):
+def test_embed_one_batch_multimodal_image(jina_multimodal_models, mocker):
+    mocker.patch("embcli_jina.jina_multimodal.TIMEOUT_SEC", 60)
     for model in jina_multimodal_models:
         image_paths = [
             files("tests.embcli_jina").joinpath("flying_cat.jpeg"),
@@ -61,7 +63,8 @@ def test_embed_one_batch_multimodal_image(jina_multimodal_models):
 
 
 @skip_if_no_api_key
-def test_embed_batch_with_options(jina_multimodal_models):
+def test_embed_batch_with_options(jina_multimodal_models, mocker):
+    mocker.patch("embcli_jina.jina_multimodal.TIMEOUT_SEC", 30)
     input_data = ["hello", "world"]
     for model in jina_multimodal_models:
         options = {"task": "retrieval.query", "dimensions": 512}
@@ -74,7 +77,8 @@ def test_embed_batch_with_options(jina_multimodal_models):
 
 
 @skip_if_no_api_key
-def test_embed_batch_embedding_types(jina_multimodal_models):
+def test_embed_batch_embedding_types(jina_multimodal_models, mocker):
+    mocker.patch("embcli_jina.jina_multimodal.TIMEOUT_SEC", 30)
     input_data = ["hello", "world"]
     for model in jina_multimodal_models:
         # Test binary embedding type