fastapi-semcache 0.2.7__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- fastapi_semcache-0.2.7.dist-info/METADATA +160 -0
- fastapi_semcache-0.2.7.dist-info/RECORD +26 -0
- fastapi_semcache-0.2.7.dist-info/WHEEL +5 -0
- fastapi_semcache-0.2.7.dist-info/licenses/LICENSE +201 -0
- fastapi_semcache-0.2.7.dist-info/top_level.txt +1 -0
- semanticcache/__init__.py +17 -0
- semanticcache/cache.py +204 -0
- semanticcache/config.py +105 -0
- semanticcache/embedders/__init__.py +46 -0
- semanticcache/embedders/_base.py +29 -0
- semanticcache/embedders/cohere.py +14 -0
- semanticcache/embedders/ollama.py +10 -0
- semanticcache/embedders/openai.py +246 -0
- semanticcache/embedders/sbert.py +142 -0
- semanticcache/embedders/voyage.py +10 -0
- semanticcache/exceptions.py +16 -0
- semanticcache/middleware/__init__.py +8 -0
- semanticcache/middleware/fastapi.py +548 -0
- semanticcache/proxy.py +232 -0
- semanticcache/stores/__init__.py +6 -0
- semanticcache/stores/response/__init__.py +5 -0
- semanticcache/stores/response/redis_store.py +116 -0
- semanticcache/stores/vector/__init__.py +5 -0
- semanticcache/stores/vector/pgvector.py +234 -0
- semanticcache/stores/vector/storage_ids.py +25 -0
- semanticcache/types.py +31 -0
|
@@ -0,0 +1,160 @@
|
|
|
1
|
+
Metadata-Version: 2.4
|
|
2
|
+
Name: fastapi-semcache
|
|
3
|
+
Version: 0.2.7
|
|
4
|
+
Summary: Drop in FastAPI middleware/reverse proxy with semantic caching for APIs & LLMs
|
|
5
|
+
License-Expression: Apache-2.0
|
|
6
|
+
Project-URL: Homepage, https://github.com/axm1647/fastapi-semcache
|
|
7
|
+
Project-URL: Repository, https://github.com/axm1647/fastapi-semcache
|
|
8
|
+
Project-URL: Issues, https://github.com/axm1647/fastapi-semcache/issues
|
|
9
|
+
Keywords: cache,embedding,llm,pgvector,redis,semantic-cache,vector
|
|
10
|
+
Classifier: Development Status :: 3 - Alpha
|
|
11
|
+
Classifier: Intended Audience :: Developers
|
|
12
|
+
Classifier: Programming Language :: Python :: 3
|
|
13
|
+
Classifier: Programming Language :: Python :: 3.12
|
|
14
|
+
Classifier: Programming Language :: Python :: 3.13
|
|
15
|
+
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
|
|
16
|
+
Classifier: Topic :: Software Development :: Libraries :: Python Modules
|
|
17
|
+
Requires-Python: >=3.12
|
|
18
|
+
Description-Content-Type: text/markdown
|
|
19
|
+
License-File: LICENSE
|
|
20
|
+
Requires-Dist: fastapi>=0.136.1
|
|
21
|
+
Requires-Dist: httpx>=0.27.0
|
|
22
|
+
Requires-Dist: psycopg[binary]>=3.3.3
|
|
23
|
+
Requires-Dist: psycopg-pool>=3.3.0
|
|
24
|
+
Requires-Dist: pydantic-settings>=2.14.0
|
|
25
|
+
Requires-Dist: redis>=7.4.0
|
|
26
|
+
Provides-Extra: embed-huggingface-cpu
|
|
27
|
+
Requires-Dist: sentence-transformers>=3.0; extra == "embed-huggingface-cpu"
|
|
28
|
+
Requires-Dist: torch>=2.1.0; extra == "embed-huggingface-cpu"
|
|
29
|
+
Provides-Extra: embed-huggingface-gpu
|
|
30
|
+
Requires-Dist: sentence-transformers>=3.0; extra == "embed-huggingface-gpu"
|
|
31
|
+
Requires-Dist: torch>=2.1.0; extra == "embed-huggingface-gpu"
|
|
32
|
+
Provides-Extra: embed-huggingface
|
|
33
|
+
Requires-Dist: sentence-transformers>=3.0; extra == "embed-huggingface"
|
|
34
|
+
Requires-Dist: torch>=2.1.0; extra == "embed-huggingface"
|
|
35
|
+
Provides-Extra: embed-openai
|
|
36
|
+
Requires-Dist: openai>=1.40; extra == "embed-openai"
|
|
37
|
+
Requires-Dist: tiktoken>=0.7; extra == "embed-openai"
|
|
38
|
+
Provides-Extra: test
|
|
39
|
+
Requires-Dist: numpy>=1.24; extra == "test"
|
|
40
|
+
Requires-Dist: pytest>=8.0; extra == "test"
|
|
41
|
+
Requires-Dist: pytest-asyncio>=0.24; extra == "test"
|
|
42
|
+
Dynamic: license-file
|
|
43
|
+
|
|
44
|
+
# fastapi-semcache
|
|
45
|
+
|
|
46
|
+
Semantic caching middleware and reverse proxy for APIs and LLMs, with embeddings, pgvector similarity search, and Redis-backed response caching.
|
|
47
|
+
|
|
48
|
+
The PyPI distribution and GitHub repository are **`fastapi-semcache`** (the import package remains **`semanticcache`**).
|
|
49
|
+
|
|
50
|
+
## Why fastapi-semcache?
|
|
51
|
+
|
|
52
|
+
This package is designed for direct integration into modern Python API stacks with minimal refactoring needed. It keeps the caching path simple and gives you explicit control over embeddings, vector search, and cache behavior.
|
|
53
|
+
|
|
54
|
+
It includes **FastAPI** middleware as a first-class integration path and can also run as a reverse proxy in front of an upstream API or LLM service. **Django** and **Flask** middleware are planned for a future release so you can hook semantic caching into those stacks the same way as FastAPI.
|
|
55
|
+
|
|
56
|
+
## What is implemented
|
|
57
|
+
|
|
58
|
+
- **Huggingface embeddings** via Sentence Transformers (`embedder_type="huggingface"`).
|
|
59
|
+
- **OpenAI embeddings** via the official async client (`embedder_type="openai"`; install
|
|
60
|
+
`embed-openai` and set `OPENAI_API_KEY`). Use
|
|
61
|
+
`OpenAIEmbedder(..., send_dimensions_to_api=False)` when the model has a fixed
|
|
62
|
+
output size and the API must not get a `dimensions` field.
|
|
63
|
+
- **PostgreSQL + pgvector** for semantic similarity lookup. The library creates a
|
|
64
|
+
dedicated cache table per embedder configuration (derived from model id and vector
|
|
65
|
+
dimension) on first use, so you are not tied to a single hard-coded vector width.
|
|
66
|
+
- **Redis** for response caching (keys include an embedder-specific prefix so separate
|
|
67
|
+
models do not collide).
|
|
68
|
+
|
|
69
|
+
- **FastAPI middleware** for in-app semantic caching.
|
|
70
|
+
- **Reverse proxy mode** via `create_semantic_cache_proxy_app()`.
|
|
71
|
+
|
|
72
|
+
## Future support
|
|
73
|
+
|
|
74
|
+
- **Django** and **Flask** middleware for in-app semantic caching (not yet shipped. same role as the FastAPI middleware).
|
|
75
|
+
|
|
76
|
+
Embeddings from the following providers are planned:
|
|
77
|
+
|
|
78
|
+
- **Ollama** (HTTP embedding API against a configurable base URL, so the server can run locally or on another host).
|
|
79
|
+
- **Cohere**
|
|
80
|
+
- **Voyage**
|
|
81
|
+
|
|
82
|
+
## Quick start
|
|
83
|
+
|
|
84
|
+
```python
|
|
85
|
+
from semanticcache import SemanticCache, create_semantic_cache_proxy_app
|
|
86
|
+
|
|
87
|
+
cache = SemanticCache()
|
|
88
|
+
app = create_semantic_cache_proxy_app(
|
|
89
|
+
upstream="http://127.0.0.1:11434",
|
|
90
|
+
cache=cache,
|
|
91
|
+
)
|
|
92
|
+
```
|
|
93
|
+
|
|
94
|
+
Run with:
|
|
95
|
+
|
|
96
|
+
```bash
|
|
97
|
+
uvicorn mymodule:app --host 0.0.0.0 --port 8080
|
|
98
|
+
```
|
|
99
|
+
|
|
100
|
+
## Reverse proxy
|
|
101
|
+
|
|
102
|
+
Point clients at the proxy and configure Postgres, Redis, and the upstream base URL.
|
|
103
|
+
|
|
104
|
+
This repository includes a small ASGI app at `app/main.py` (import `app` for uvicorn). Set **`SEMANTIC_CACHE_PROXY_UPSTREAM`** to the backend base URL; the default is `http://127.0.0.1:11434`.
|
|
105
|
+
|
|
106
|
+
```bash
|
|
107
|
+
uv run uvicorn app.main:app --host 0.0.0.0 --port 8080
|
|
108
|
+
```
|
|
109
|
+
|
|
110
|
+
See `create_semantic_cache_proxy_app` in `semanticcache.proxy` for timeout, TLS verification, `httpx_client_kwargs`, and middleware options such as `path_prefix` and `extract_query`.
|
|
111
|
+
|
|
112
|
+
## Install
|
|
113
|
+
|
|
114
|
+
```bash
|
|
115
|
+
pip install fastapi-semcache
|
|
116
|
+
```
|
|
117
|
+
|
|
118
|
+
**Custom embedders:** subclass `BaseEmbedder` from `semanticcache.embedders` and pass it to `SemanticCache(embedder=...)` to skip the optional embedding extras. See [docs/embedders.md](docs/embedders.md).
|
|
119
|
+
|
|
120
|
+
Optional extras:
|
|
121
|
+
|
|
122
|
+
- `embed-huggingface` / `embed-huggingface-cpu`: Sentence Transformers with **CPU** PyTorch.
|
|
123
|
+
- `embed-huggingface-gpu`: Sentence Transformers with a CUDA-enabled PyTorch install.
|
|
124
|
+
- `embed-openai`: OpenAI embeddings (`openai`, `tiktoken`).
|
|
125
|
+
|
|
126
|
+
### CPU
|
|
127
|
+
|
|
128
|
+
```bash
|
|
129
|
+
pip install "fastapi-semcache[embed-huggingface-cpu]"
|
|
130
|
+
# or: pip install "fastapi-semcache[embed-huggingface]"
|
|
131
|
+
```
|
|
132
|
+
|
|
133
|
+
### GPU
|
|
134
|
+
|
|
135
|
+
Pick a CUDA version that matches your system from [PyTorch Get Started](https://pytorch.org/get-started/locally/), then install with that index so pip selects CUDA wheels.
|
|
136
|
+
|
|
137
|
+
```bash
|
|
138
|
+
pip install "fastapi-semcache[embed-huggingface-gpu]" \
|
|
139
|
+
--extra-index-url https://download.pytorch.org/whl/cu124
|
|
140
|
+
```
|
|
141
|
+
|
|
142
|
+
### OpenAI embeddings
|
|
143
|
+
|
|
144
|
+
Install the OpenAI extra so `embedder_type="openai"` works (pulls `openai` and `tiktoken`). Set `OPENAI_API_KEY` in your environment.
|
|
145
|
+
|
|
146
|
+
```bash
|
|
147
|
+
pip install "fastapi-semcache[embed-openai]"
|
|
148
|
+
```
|
|
149
|
+
|
|
150
|
+
## Requirements
|
|
151
|
+
|
|
152
|
+
Python 3.12+.
|
|
153
|
+
|
|
154
|
+
## Links
|
|
155
|
+
|
|
156
|
+
- Repository: [fastapi-semcache](https://github.com/axm1647/fastapi-semcache)
|
|
157
|
+
|
|
158
|
+
## License
|
|
159
|
+
|
|
160
|
+
Apache-2.0. See `LICENSE`.
|
|
@@ -0,0 +1,26 @@
|
|
|
1
|
+
fastapi_semcache-0.2.7.dist-info/licenses/LICENSE,sha256=39-iA6zk2IdFerGyiQ1AuBq7_R71HeMjab5yU5tqijM,11344
|
|
2
|
+
semanticcache/__init__.py,sha256=rysIpve9XLx7LHcIqZj1-xfLS5PpmbdreRvqwJP2FPY,487
|
|
3
|
+
semanticcache/cache.py,sha256=M8BUiUykNQzlXNmzBoc49kezbrHEd5nandxB4eU4-S0,7697
|
|
4
|
+
semanticcache/config.py,sha256=JguBgSD4wGie1umAyykPfFRl-3j8KqqJoRP-Tw8Kjos,3080
|
|
5
|
+
semanticcache/exceptions.py,sha256=ZIlsVeIA9P-Bl_oRCgrjZFRqFRlBsrneLfLIVtEwqLI,518
|
|
6
|
+
semanticcache/proxy.py,sha256=8NQMTYKXMCVmEMHRSGNWly_la7BEBiUfiHPxgeehZ8A,7275
|
|
7
|
+
semanticcache/types.py,sha256=z6QUcjw0Y21Eg-LD1uQ4b1AyQ9AFq5zbSLVYbf-pVHM,642
|
|
8
|
+
semanticcache/embedders/__init__.py,sha256=N8TaLpzFFTtvxPX3uTXghtfm_Z9_XRZz_FL_71wpzaQ,1556
|
|
9
|
+
semanticcache/embedders/_base.py,sha256=fNkdXTIvQ1vRCL1Vt4dc4qqfO3FRtBNdf0ZyoQQU7fE,918
|
|
10
|
+
semanticcache/embedders/cohere.py,sha256=KsqEKl6sF092sICwEP6Mhf1wy2iGFDZrQNAmoa0iVds,369
|
|
11
|
+
semanticcache/embedders/ollama.py,sha256=7HeoZac-Q4tReWG4GIfEsZ2keJMNpcBy7hut85IOD6U,216
|
|
12
|
+
semanticcache/embedders/openai.py,sha256=WsWc-If5DRnnyavVNVFddD5bkTmCP10hY5xNs3zsaiE,8488
|
|
13
|
+
semanticcache/embedders/sbert.py,sha256=uASJP4TDM3BBhR3_0K-8H9R1OtarUcBSdAmwDxtmKas,4532
|
|
14
|
+
semanticcache/embedders/voyage.py,sha256=QhKuI-fNvHeLtKw1_TJgz3oJQnUeJYYvw-4b16KTdW8,216
|
|
15
|
+
semanticcache/middleware/__init__.py,sha256=ENNkVAmjumdW-NLyfPUCHT0nf0BVkRDHKTjWRDFThkc,201
|
|
16
|
+
semanticcache/middleware/fastapi.py,sha256=fbsVT3SZifDTXF-OLHWXLqal6tnQKH5U4WRBJyNKy78,20571
|
|
17
|
+
semanticcache/stores/__init__.py,sha256=-1e6OgC3Udjm015fgxeddg3a5TGmNq9nY333d5C8ke4,222
|
|
18
|
+
semanticcache/stores/response/__init__.py,sha256=ffkXcr8LCr3GBFtZNdFMZz0cJiVgxASbsvEHa5ngUjY,137
|
|
19
|
+
semanticcache/stores/response/redis_store.py,sha256=Ip1XyQP4u6Dviq0cGkwjnEcua0BJ7lAEXfXRCLhV5O4,3807
|
|
20
|
+
semanticcache/stores/vector/__init__.py,sha256=O4OOY0QpEgsPgthrY0KHs3_KtJVCFgacnGH8-ZRHsGo,134
|
|
21
|
+
semanticcache/stores/vector/pgvector.py,sha256=tUlozF3TiT62EhbBh3IdNafZcjtYp00jiafDav3WkT0,8357
|
|
22
|
+
semanticcache/stores/vector/storage_ids.py,sha256=KuDjzN3bp8lLnK41L2wnxotYIqTCGAiLWWs9TBXLZtQ,850
|
|
23
|
+
fastapi_semcache-0.2.7.dist-info/METADATA,sha256=B1K81ZiD_GUzgZ8AeykhibKNwyiW7uar5OZyZ3LYoIY,6200
|
|
24
|
+
fastapi_semcache-0.2.7.dist-info/WHEEL,sha256=aeYiig01lYGDzBgS8HxWXOg3uV61G9ijOsup-k9o1sk,91
|
|
25
|
+
fastapi_semcache-0.2.7.dist-info/top_level.txt,sha256=i3oQUtN9nM9e4caDV-PTBDo9Npa1o8w4rF77S_Uo3I8,14
|
|
26
|
+
fastapi_semcache-0.2.7.dist-info/RECORD,,
|
|
@@ -0,0 +1,201 @@
|
|
|
1
|
+
Apache License
|
|
2
|
+
Version 2.0, January 2004
|
|
3
|
+
http://www.apache.org/licenses/
|
|
4
|
+
|
|
5
|
+
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
|
|
6
|
+
|
|
7
|
+
1. Definitions.
|
|
8
|
+
|
|
9
|
+
"License" shall mean the terms and conditions for use, reproduction,
|
|
10
|
+
and distribution as defined by Sections 1 through 9 of this document.
|
|
11
|
+
|
|
12
|
+
"Licensor" shall mean the copyright owner or entity authorized by
|
|
13
|
+
the copyright owner that is granting the License.
|
|
14
|
+
|
|
15
|
+
"Legal Entity" shall mean the union of the acting entity and all
|
|
16
|
+
other entities that control, are controlled by, or are under common
|
|
17
|
+
control with that entity. For the purposes of this definition,
|
|
18
|
+
"control" means (i) the power, direct or indirect, to cause the
|
|
19
|
+
direction or management of such entity, whether by contract or
|
|
20
|
+
otherwise, or (ii) ownership of fifty percent (50%) or more of the
|
|
21
|
+
outstanding shares, or (iii) beneficial ownership of such entity.
|
|
22
|
+
|
|
23
|
+
"You" (or "Your") shall mean an individual or Legal Entity
|
|
24
|
+
exercising permissions granted by this License.
|
|
25
|
+
|
|
26
|
+
"Source" form shall mean the preferred form for making modifications,
|
|
27
|
+
including but not limited to software source code, documentation
|
|
28
|
+
source, and configuration files.
|
|
29
|
+
|
|
30
|
+
"Object" form shall mean any form resulting from mechanical
|
|
31
|
+
transformation or translation of a Source form, including but
|
|
32
|
+
not limited to compiled object code, generated documentation,
|
|
33
|
+
and conversions to other media types.
|
|
34
|
+
|
|
35
|
+
"Work" shall mean the work of authorship, whether in Source or
|
|
36
|
+
Object form, made available under the License, as indicated by a
|
|
37
|
+
copyright notice that is included in or attached to the work
|
|
38
|
+
(an example is provided in the Appendix below).
|
|
39
|
+
|
|
40
|
+
"Derivative Works" shall mean any work, whether in Source or Object
|
|
41
|
+
form, that is based on (or derived from) the Work and for which the
|
|
42
|
+
editorial revisions, annotations, elaborations, or other modifications
|
|
43
|
+
represent, as a whole, an original work of authorship. For the purposes
|
|
44
|
+
of this License, Derivative Works shall not include works that remain
|
|
45
|
+
separable from, or merely link (or bind by name) to the interfaces of,
|
|
46
|
+
the Work and Derivative Works thereof.
|
|
47
|
+
|
|
48
|
+
"Contribution" shall mean any work of authorship, including
|
|
49
|
+
the original version of the Work and any modifications or additions
|
|
50
|
+
to that Work or Derivative Works thereof, that is intentionally
|
|
51
|
+
submitted to Licensor for inclusion in the Work by the copyright owner
|
|
52
|
+
or by an individual or Legal Entity authorized to submit on behalf of
|
|
53
|
+
the copyright owner. For the purposes of this definition, "submitted"
|
|
54
|
+
means any form of electronic, verbal, or written communication sent
|
|
55
|
+
to the Licensor or its representatives, including but not limited to
|
|
56
|
+
communication on electronic mailing lists, source code control systems,
|
|
57
|
+
and issue tracking systems that are managed by, or on behalf of, the
|
|
58
|
+
Licensor for the purpose of discussing and improving the Work, but
|
|
59
|
+
excluding communication that is conspicuously marked or otherwise
|
|
60
|
+
designated in writing by the copyright owner as "Not a Contribution."
|
|
61
|
+
|
|
62
|
+
"Contributor" shall mean Licensor and any individual or Legal Entity
|
|
63
|
+
on behalf of whom a Contribution has been received by Licensor and
|
|
64
|
+
subsequently incorporated within the Work.
|
|
65
|
+
|
|
66
|
+
2. Grant of Copyright License. Subject to the terms and conditions of
|
|
67
|
+
this License, each Contributor hereby grants to You a perpetual,
|
|
68
|
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
|
69
|
+
copyright license to reproduce, prepare Derivative Works of,
|
|
70
|
+
publicly display, publicly perform, sublicense, and distribute the
|
|
71
|
+
Work and such Derivative Works in Source or Object form.
|
|
72
|
+
|
|
73
|
+
3. Grant of Patent License. Subject to the terms and conditions of
|
|
74
|
+
this License, each Contributor hereby grants to You a perpetual,
|
|
75
|
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
|
76
|
+
(except as stated in this section) patent license to make, have made,
|
|
77
|
+
use, offer to sell, sell, import, and otherwise transfer the Work,
|
|
78
|
+
where such license applies only to those patent claims licensable
|
|
79
|
+
by such Contributor that are necessarily infringed by their
|
|
80
|
+
Contribution(s) alone or by combination of their Contribution(s)
|
|
81
|
+
with the Work to which such Contribution(s) was submitted. If You
|
|
82
|
+
institute patent litigation against any entity (including a
|
|
83
|
+
cross-claim or counterclaim in a lawsuit) alleging that the Work
|
|
84
|
+
or a Contribution incorporated within the Work constitutes direct
|
|
85
|
+
or contributory patent infringement, then any patent licenses
|
|
86
|
+
granted to You under this License for that Work shall terminate
|
|
87
|
+
as of the date such litigation is filed.
|
|
88
|
+
|
|
89
|
+
4. Redistribution. You may reproduce and distribute copies of the
|
|
90
|
+
Work or Derivative Works thereof in any medium, with or without
|
|
91
|
+
modifications, and in Source or Object form, provided that You
|
|
92
|
+
meet the following conditions:
|
|
93
|
+
|
|
94
|
+
(a) You must give any other recipients of the Work or
|
|
95
|
+
Derivative Works a copy of this License; and
|
|
96
|
+
|
|
97
|
+
(b) You must cause any modified files to carry prominent notices
|
|
98
|
+
stating that You changed the files; and
|
|
99
|
+
|
|
100
|
+
(c) You must retain, in the Source form of any Derivative Works
|
|
101
|
+
that You distribute, all copyright, patent, trademark, and
|
|
102
|
+
attribution notices from the Source form of the Work,
|
|
103
|
+
excluding those notices that do not pertain to any part of
|
|
104
|
+
the Derivative Works; and
|
|
105
|
+
|
|
106
|
+
(d) If the Work includes a "NOTICE" text file as part of its
|
|
107
|
+
distribution, then any Derivative Works that You distribute must
|
|
108
|
+
include a readable copy of the attribution notices contained
|
|
109
|
+
within such NOTICE file, excluding those notices that do not
|
|
110
|
+
pertain to any part of the Derivative Works, in at least one
|
|
111
|
+
of the following places: within a NOTICE text file distributed
|
|
112
|
+
as part of the Derivative Works; within the Source form or
|
|
113
|
+
documentation, if provided along with the Derivative Works; or,
|
|
114
|
+
within a display generated by the Derivative Works, if and
|
|
115
|
+
wherever such third-party notices normally appear. The contents
|
|
116
|
+
of the NOTICE file are for informational purposes only and
|
|
117
|
+
do not modify the License. You may add Your own attribution
|
|
118
|
+
notices within Derivative Works that You distribute, alongside
|
|
119
|
+
or as an addendum to the NOTICE text from the Work, provided
|
|
120
|
+
that such additional attribution notices cannot be construed
|
|
121
|
+
as modifying the License.
|
|
122
|
+
|
|
123
|
+
You may add Your own copyright statement to Your modifications and
|
|
124
|
+
may provide additional or different license terms and conditions
|
|
125
|
+
for use, reproduction, or distribution of Your modifications, or
|
|
126
|
+
for any such Derivative Works as a whole, provided Your use,
|
|
127
|
+
reproduction, and distribution of the Work otherwise complies with
|
|
128
|
+
the conditions stated in this License.
|
|
129
|
+
|
|
130
|
+
5. Submission of Contributions. Unless You explicitly state otherwise,
|
|
131
|
+
any Contribution intentionally submitted for inclusion in the Work
|
|
132
|
+
by You to the Licensor shall be under the terms and conditions of
|
|
133
|
+
this License, without any additional terms or conditions.
|
|
134
|
+
Notwithstanding the above, nothing herein shall supersede or modify
|
|
135
|
+
the terms of any separate license agreement you may have executed
|
|
136
|
+
with Licensor regarding such Contributions.
|
|
137
|
+
|
|
138
|
+
6. Trademarks. This License does not grant permission to use the trade
|
|
139
|
+
names, trademarks, service marks, or product names of the Licensor,
|
|
140
|
+
except as required for reasonable and customary use in describing the
|
|
141
|
+
origin of the Work and reproducing the content of the NOTICE file.
|
|
142
|
+
|
|
143
|
+
7. Disclaimer of Warranty. Unless required by applicable law or
|
|
144
|
+
agreed to in writing, Licensor provides the Work (and each
|
|
145
|
+
Contributor provides its Contributions) on an "AS IS" BASIS,
|
|
146
|
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
|
|
147
|
+
implied, including, without limitation, any warranties or conditions
|
|
148
|
+
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
|
|
149
|
+
PARTICULAR PURPOSE. You are solely responsible for determining the
|
|
150
|
+
appropriateness of using or redistributing the Work and assume any
|
|
151
|
+
risks associated with Your exercise of permissions under this License.
|
|
152
|
+
|
|
153
|
+
8. Limitation of Liability. In no event and under no legal theory,
|
|
154
|
+
whether in tort (including negligence), contract, or otherwise,
|
|
155
|
+
unless required by applicable law (such as deliberate and grossly
|
|
156
|
+
negligent acts) or agreed to in writing, shall any Contributor be
|
|
157
|
+
liable to You for damages, including any direct, indirect, special,
|
|
158
|
+
incidental, or consequential damages of any character arising as a
|
|
159
|
+
result of this License or out of the use or inability to use the
|
|
160
|
+
Work (including but not limited to damages for loss of goodwill,
|
|
161
|
+
work stoppage, computer failure or malfunction, or any and all
|
|
162
|
+
other commercial damages or losses), even if such Contributor
|
|
163
|
+
has been advised of the possibility of such damages.
|
|
164
|
+
|
|
165
|
+
9. Accepting Warranty or Additional Liability. While redistributing
|
|
166
|
+
the Work or Derivative Works thereof, You may choose to offer,
|
|
167
|
+
and charge a fee for, acceptance of support, warranty, indemnity,
|
|
168
|
+
or other liability obligations and/or rights consistent with this
|
|
169
|
+
License. However, in accepting such obligations, You may act only
|
|
170
|
+
on Your own behalf and on Your sole responsibility, not on behalf
|
|
171
|
+
of any other Contributor, and only if You agree to indemnify,
|
|
172
|
+
defend, and hold each Contributor harmless for any liability
|
|
173
|
+
incurred by, or claims asserted against, such Contributor by reason
|
|
174
|
+
of your accepting any such warranty or additional liability.
|
|
175
|
+
|
|
176
|
+
END OF TERMS AND CONDITIONS
|
|
177
|
+
|
|
178
|
+
APPENDIX: How to apply the Apache License to your work.
|
|
179
|
+
|
|
180
|
+
To apply the Apache License to your work, attach the following
|
|
181
|
+
boilerplate notice, with the fields enclosed by brackets "[]"
|
|
182
|
+
replaced with your own identifying information. (Don't include
|
|
183
|
+
the brackets!) The text should be enclosed in the appropriate
|
|
184
|
+
comment syntax for the file format. We also recommend that a
|
|
185
|
+
file or class name and description of purpose be included on the
|
|
186
|
+
same "printed page" as the copyright notice for easier
|
|
187
|
+
identification within third-party archives.
|
|
188
|
+
|
|
189
|
+
Copyright [2026] [Adam Mackey]
|
|
190
|
+
|
|
191
|
+
Licensed under the Apache License, Version 2.0 (the "License");
|
|
192
|
+
you may not use this file except in compliance with the License.
|
|
193
|
+
You may obtain a copy of the License at
|
|
194
|
+
|
|
195
|
+
http://www.apache.org/licenses/LICENSE-2.0
|
|
196
|
+
|
|
197
|
+
Unless required by applicable law or agreed to in writing, software
|
|
198
|
+
distributed under the License is distributed on an "AS IS" BASIS,
|
|
199
|
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
200
|
+
See the License for the specific language governing permissions and
|
|
201
|
+
limitations under the License.
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
semanticcache
|
|
@@ -0,0 +1,17 @@
|
|
|
1
|
+
"""Semantic caching: embeddings, pgvector, optional Redis."""
|
|
2
|
+
|
|
3
|
+
from .cache import SemanticCache
|
|
4
|
+
from .config import get_cache_settings
|
|
5
|
+
from .middleware import SemanticCacheMiddleware
|
|
6
|
+
from .proxy import create_semantic_cache_proxy_app
|
|
7
|
+
from .types import CacheEntry, CacheQuery, CacheResult
|
|
8
|
+
|
|
9
|
+
__all__: list[str] = [
|
|
10
|
+
"CacheEntry",
|
|
11
|
+
"CacheQuery",
|
|
12
|
+
"CacheResult",
|
|
13
|
+
"SemanticCache",
|
|
14
|
+
"SemanticCacheMiddleware",
|
|
15
|
+
"create_semantic_cache_proxy_app",
|
|
16
|
+
"get_cache_settings",
|
|
17
|
+
]
|
semanticcache/cache.py
ADDED
|
@@ -0,0 +1,204 @@
|
|
|
1
|
+
"""High-level semantic cache orchestrating embedder, pgvector, and optional Redis."""
|
|
2
|
+
|
|
3
|
+
from __future__ import annotations
|
|
4
|
+
|
|
5
|
+
from typing import TYPE_CHECKING, Literal
|
|
6
|
+
|
|
7
|
+
from .config import get_cache_settings
|
|
8
|
+
from .embedders import BaseEmbedder, get_embedder
|
|
9
|
+
from .stores import AsyncPgVectorStore, RedisResponseStore
|
|
10
|
+
from .stores.vector.storage_ids import embedding_storage_ids
|
|
11
|
+
from .types import CacheResult
|
|
12
|
+
|
|
13
|
+
if TYPE_CHECKING:
|
|
14
|
+
from semanticcache.config import CacheSettings
|
|
15
|
+
|
|
16
|
+
|
|
17
|
+
def _embed_source(
|
|
18
|
+
settings: "CacheSettings",
|
|
19
|
+
) -> Literal[
|
|
20
|
+
"embedders.sbert",
|
|
21
|
+
"embedders.openai",
|
|
22
|
+
"embedders.cohere",
|
|
23
|
+
"embedders.voyage",
|
|
24
|
+
"embedders.ollama",
|
|
25
|
+
"none",
|
|
26
|
+
]:
|
|
27
|
+
"""Map configured embedder type to ``CacheResult.source``."""
|
|
28
|
+
if settings.embedder_type == "huggingface":
|
|
29
|
+
return "embedders.sbert"
|
|
30
|
+
if settings.embedder_type == "openai":
|
|
31
|
+
return "embedders.openai"
|
|
32
|
+
if settings.embedder_type == "cohere":
|
|
33
|
+
return "embedders.cohere"
|
|
34
|
+
if settings.embedder_type == "voyage":
|
|
35
|
+
return "embedders.voyage"
|
|
36
|
+
if settings.embedder_type == "ollama":
|
|
37
|
+
return "embedders.ollama"
|
|
38
|
+
return "none"
|
|
39
|
+
|
|
40
|
+
|
|
41
|
+
class SemanticCache:
|
|
42
|
+
"""Orchestrate embedding, pgvector ANN search, and optional Redis response cache."""
|
|
43
|
+
|
|
44
|
+
threshold: float
|
|
45
|
+
pg_uri: str
|
|
46
|
+
redis_uri: str
|
|
47
|
+
_embedding_dim: int
|
|
48
|
+
_redis_key_prefix: str
|
|
49
|
+
_settings: "CacheSettings"
|
|
50
|
+
_embedder: BaseEmbedder
|
|
51
|
+
_vector_store: AsyncPgVectorStore
|
|
52
|
+
_pg_open: bool
|
|
53
|
+
_closed: bool
|
|
54
|
+
|
|
55
|
+
def __init__(
|
|
56
|
+
self,
|
|
57
|
+
threshold: float | None = None,
|
|
58
|
+
pg_uri: str | None = None,
|
|
59
|
+
redis_uri: str | None = None,
|
|
60
|
+
*,
|
|
61
|
+
embedder: BaseEmbedder | None = None,
|
|
62
|
+
embedding_dim: int | None = None,
|
|
63
|
+
settings: "CacheSettings | None" = None,
|
|
64
|
+
) -> None:
|
|
65
|
+
"""Build stores and embedder from explicit args or ``CacheSettings``.
|
|
66
|
+
|
|
67
|
+
Args:
|
|
68
|
+
threshold: Minimum cosine similarity in ``[0, 1]`` for a vector hit.
|
|
69
|
+
pg_uri: PostgreSQL URI with pgvector; defaults to settings.
|
|
70
|
+
redis_uri: Redis URI for TTL response cache. Empty or whitespace-only
|
|
71
|
+
disables Redis (Postgres only).
|
|
72
|
+
embedder: Custom embedder; defaults to ``get_embedder(settings)``.
|
|
73
|
+
embedding_dim: When set, must equal ``embedder.embedding_dim`` (safety
|
|
74
|
+
check). The embedder defines the vector width and storage namespace.
|
|
75
|
+
settings: Base settings object; defaults to ``get_cache_settings()``.
|
|
76
|
+
"""
|
|
77
|
+
self._settings = settings if settings is not None else get_cache_settings()
|
|
78
|
+
self.threshold = (
|
|
79
|
+
threshold if threshold is not None else self._settings.threshold
|
|
80
|
+
)
|
|
81
|
+
self.pg_uri = pg_uri if pg_uri is not None else self._settings.pg_uri
|
|
82
|
+
resolved_redis = (
|
|
83
|
+
redis_uri if redis_uri is not None else self._settings.redis_uri
|
|
84
|
+
)
|
|
85
|
+
self.redis_uri = resolved_redis
|
|
86
|
+
self._embedder = (
|
|
87
|
+
embedder if embedder is not None else get_embedder(self._settings)
|
|
88
|
+
)
|
|
89
|
+
resolved_dim = self._embedder.embedding_dim
|
|
90
|
+
if embedding_dim is not None and embedding_dim != resolved_dim:
|
|
91
|
+
msg = (
|
|
92
|
+
f"embedding_dim={embedding_dim} does not match "
|
|
93
|
+
f"embedder.embedding_dim={resolved_dim}"
|
|
94
|
+
)
|
|
95
|
+
raise ValueError(msg)
|
|
96
|
+
self._embedding_dim = resolved_dim
|
|
97
|
+
table_name, redis_prefix = embedding_storage_ids(
|
|
98
|
+
self._embedder.cache_namespace,
|
|
99
|
+
self._embedding_dim,
|
|
100
|
+
)
|
|
101
|
+
self._redis_key_prefix = redis_prefix
|
|
102
|
+
max_pg = self._settings.pg_pool_size + self._settings.pg_pool_max_overflow
|
|
103
|
+
self._vector_store = AsyncPgVectorStore(
|
|
104
|
+
self.pg_uri,
|
|
105
|
+
table_name=table_name,
|
|
106
|
+
embedding_dim=self._embedding_dim,
|
|
107
|
+
min_pool_size=self._settings.pg_pool_size,
|
|
108
|
+
max_pool_size=max_pg,
|
|
109
|
+
)
|
|
110
|
+
self._redis_store: RedisResponseStore | None = (
|
|
111
|
+
RedisResponseStore(
|
|
112
|
+
resolved_redis.strip(),
|
|
113
|
+
default_ttl_seconds=self._settings.redis_ttl_seconds,
|
|
114
|
+
)
|
|
115
|
+
if resolved_redis.strip()
|
|
116
|
+
else None
|
|
117
|
+
)
|
|
118
|
+
self._pg_open = False
|
|
119
|
+
self._closed = False
|
|
120
|
+
|
|
121
|
+
async def _ensure_open(self) -> None:
|
|
122
|
+
"""Open the pgvector pool on first use."""
|
|
123
|
+
if self._closed:
|
|
124
|
+
msg = "SemanticCache is closed"
|
|
125
|
+
raise RuntimeError(msg)
|
|
126
|
+
if not self._pg_open:
|
|
127
|
+
await self._vector_store.open()
|
|
128
|
+
await self._vector_store.ensure_schema()
|
|
129
|
+
self._pg_open = True
|
|
130
|
+
|
|
131
|
+
async def get(self, query: str, model: str | None = None) -> CacheResult:
|
|
132
|
+
"""Embed the query, search vectors, then optionally resolve Redis by row id.
|
|
133
|
+
|
|
134
|
+
On a vector hit, the response body prefers Redis (keyed by embedder namespace
|
|
135
|
+
plus row id) when present; otherwise the JSON from Postgres is returned.
|
|
136
|
+
|
|
137
|
+
Args:
|
|
138
|
+
query: Text to embed and match semantically.
|
|
139
|
+
model: Reserved for future embedder routing (unused).
|
|
140
|
+
|
|
141
|
+
Returns:
|
|
142
|
+
``CacheResult`` with ``is_hit`` False on vector miss, else True with
|
|
143
|
+
similarity and response payload.
|
|
144
|
+
"""
|
|
145
|
+
_ = model
|
|
146
|
+
src = _embed_source(self._settings)
|
|
147
|
+
await self._ensure_open()
|
|
148
|
+
vectors = await self._embedder.embed([query])
|
|
149
|
+
if not vectors:
|
|
150
|
+
return CacheResult(is_hit=False, similarity=None, source=src, response=None)
|
|
151
|
+
query_embedding = vectors[0]
|
|
152
|
+
entry = await self._vector_store.similarity_search(
|
|
153
|
+
query_embedding, self.threshold
|
|
154
|
+
)
|
|
155
|
+
if entry is None:
|
|
156
|
+
return CacheResult(is_hit=False, similarity=None, source=src, response=None)
|
|
157
|
+
response: dict[str, object] = entry.response
|
|
158
|
+
if self._redis_store is not None:
|
|
159
|
+
from_redis = await self._redis_store.get(
|
|
160
|
+
f"{self._redis_key_prefix}:{entry.id}"
|
|
161
|
+
)
|
|
162
|
+
if from_redis is not None:
|
|
163
|
+
response = from_redis
|
|
164
|
+
return CacheResult(
|
|
165
|
+
is_hit=True,
|
|
166
|
+
similarity=entry.similarity,
|
|
167
|
+
source=src,
|
|
168
|
+
response=response,
|
|
169
|
+
)
|
|
170
|
+
|
|
171
|
+
async def put(
|
|
172
|
+
self, query: str, response: dict[str, object], model: str | None = None
|
|
173
|
+
) -> None:
|
|
174
|
+
"""Embed and persist the response in Postgres and optionally Redis.
|
|
175
|
+
|
|
176
|
+
Redis stores the same payload under a key scoped to this embedder plus
|
|
177
|
+
``row_id``, with configured TTL.
|
|
178
|
+
|
|
179
|
+
Args:
|
|
180
|
+
query: Query text stored alongside the embedding in the pgvector table.
|
|
181
|
+
response: JSON-serializable payload (mirrored in Redis when enabled).
|
|
182
|
+
model: Reserved for future embedder routing (unused).
|
|
183
|
+
"""
|
|
184
|
+
_ = model
|
|
185
|
+
await self._ensure_open()
|
|
186
|
+
vectors = await self._embedder.embed([query])
|
|
187
|
+
if not vectors:
|
|
188
|
+
msg = "embedder returned no vectors for a non-empty query list"
|
|
189
|
+
raise RuntimeError(msg)
|
|
190
|
+
embedding = vectors[0]
|
|
191
|
+
row_id = await self._vector_store.upsert(query, embedding, response)
|
|
192
|
+
if self._redis_store is not None:
|
|
193
|
+
await self._redis_store.put(f"{self._redis_key_prefix}:{row_id}", response)
|
|
194
|
+
|
|
195
|
+
async def close(self) -> None:
|
|
196
|
+
"""Close the Postgres pool and Redis client if they were opened."""
|
|
197
|
+
if self._closed:
|
|
198
|
+
return
|
|
199
|
+
self._closed = True
|
|
200
|
+
if self._pg_open:
|
|
201
|
+
await self._vector_store.close()
|
|
202
|
+
self._pg_open = False
|
|
203
|
+
if self._redis_store is not None:
|
|
204
|
+
await self._redis_store.close()
|