awgraph 1.0.0__tar.gz
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.
- awgraph-1.0.0/LICENSE +202 -0
- awgraph-1.0.0/PKG-INFO +141 -0
- awgraph-1.0.0/README.md +121 -0
- awgraph-1.0.0/awgraph/__init__.py +70 -0
- awgraph-1.0.0/awgraph/base.py +139 -0
- awgraph-1.0.0/awgraph/degradation.py +48 -0
- awgraph-1.0.0/awgraph/graph.py +4866 -0
- awgraph-1.0.0/awgraph/logging.py +11 -0
- awgraph-1.0.0/awgraph/registry.py +780 -0
- awgraph-1.0.0/awgraph/store.py +658 -0
- awgraph-1.0.0/awgraph.egg-info/PKG-INFO +141 -0
- awgraph-1.0.0/awgraph.egg-info/SOURCES.txt +16 -0
- awgraph-1.0.0/awgraph.egg-info/dependency_links.txt +1 -0
- awgraph-1.0.0/awgraph.egg-info/requires.txt +8 -0
- awgraph-1.0.0/awgraph.egg-info/top_level.txt +1 -0
- awgraph-1.0.0/pyproject.toml +44 -0
- awgraph-1.0.0/setup.cfg +4 -0
- awgraph-1.0.0/tests/test_standalone.py +125 -0
awgraph-1.0.0/LICENSE
ADDED
|
@@ -0,0 +1,202 @@
|
|
|
1
|
+
|
|
2
|
+
Apache License
|
|
3
|
+
Version 2.0, January 2004
|
|
4
|
+
http://www.apache.org/licenses/
|
|
5
|
+
|
|
6
|
+
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
|
|
7
|
+
|
|
8
|
+
1. Definitions.
|
|
9
|
+
|
|
10
|
+
"License" shall mean the terms and conditions for use, reproduction,
|
|
11
|
+
and distribution as defined by Sections 1 through 9 of this document.
|
|
12
|
+
|
|
13
|
+
"Licensor" shall mean the copyright owner or entity authorized by
|
|
14
|
+
the copyright owner that is granting the License.
|
|
15
|
+
|
|
16
|
+
"Legal Entity" shall mean the union of the acting entity and all
|
|
17
|
+
other entities that control, are controlled by, or are under common
|
|
18
|
+
control with that entity. For the purposes of this definition,
|
|
19
|
+
"control" means (i) the power, direct or indirect, to cause the
|
|
20
|
+
direction or management of such entity, whether by contract or
|
|
21
|
+
otherwise, or (ii) ownership of fifty percent (50%) or more of the
|
|
22
|
+
outstanding shares, or (iii) beneficial ownership of such entity.
|
|
23
|
+
|
|
24
|
+
"You" (or "Your") shall mean an individual or Legal Entity
|
|
25
|
+
exercising permissions granted by this License.
|
|
26
|
+
|
|
27
|
+
"Source" form shall mean the preferred form for making modifications,
|
|
28
|
+
including but not limited to software source code, documentation
|
|
29
|
+
source, and configuration files.
|
|
30
|
+
|
|
31
|
+
"Object" form shall mean any form resulting from mechanical
|
|
32
|
+
transformation or translation of a Source form, including but
|
|
33
|
+
not limited to compiled object code, generated documentation,
|
|
34
|
+
and conversions to other media types.
|
|
35
|
+
|
|
36
|
+
"Work" shall mean the work of authorship, whether in Source or
|
|
37
|
+
Object form, made available under the License, as indicated by a
|
|
38
|
+
copyright notice that is included in or attached to the work
|
|
39
|
+
(an example is provided in the Appendix below).
|
|
40
|
+
|
|
41
|
+
"Derivative Works" shall mean any work, whether in Source or Object
|
|
42
|
+
form, that is based on (or derived from) the Work and for which the
|
|
43
|
+
editorial revisions, annotations, elaborations, or other modifications
|
|
44
|
+
represent, as a whole, an original work of authorship. For the purposes
|
|
45
|
+
of this License, Derivative Works shall not include works that remain
|
|
46
|
+
separable from, or merely link (or bind by name) to the interfaces of,
|
|
47
|
+
the Work and Derivative Works thereof.
|
|
48
|
+
|
|
49
|
+
"Contribution" shall mean any work of authorship, including
|
|
50
|
+
the original version of the Work and any modifications or additions
|
|
51
|
+
to that Work or Derivative Works thereof, that is intentionally
|
|
52
|
+
submitted to Licensor for inclusion in the Work by the copyright owner
|
|
53
|
+
or by an individual or Legal Entity authorized to submit on behalf of
|
|
54
|
+
the copyright owner. For the purposes of this definition, "submitted"
|
|
55
|
+
means any form of electronic, verbal, or written communication sent
|
|
56
|
+
to the Licensor or its representatives, including but not limited to
|
|
57
|
+
communication on electronic mailing lists, source code control systems,
|
|
58
|
+
and issue tracking systems that are managed by, or on behalf of, the
|
|
59
|
+
Licensor for the purpose of discussing and improving the Work, but
|
|
60
|
+
excluding communication that is conspicuously marked or otherwise
|
|
61
|
+
designated in writing by the copyright owner as "Not a Contribution."
|
|
62
|
+
|
|
63
|
+
"Contributor" shall mean Licensor and any individual or Legal Entity
|
|
64
|
+
on behalf of whom a Contribution has been received by Licensor and
|
|
65
|
+
subsequently incorporated within the Work.
|
|
66
|
+
|
|
67
|
+
2. Grant of Copyright License. Subject to the terms and conditions of
|
|
68
|
+
this License, each Contributor hereby grants to You a perpetual,
|
|
69
|
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
|
70
|
+
copyright license to reproduce, prepare Derivative Works of,
|
|
71
|
+
publicly display, publicly perform, sublicense, and distribute the
|
|
72
|
+
Work and such Derivative Works in Source or Object form.
|
|
73
|
+
|
|
74
|
+
3. Grant of Patent License. Subject to the terms and conditions of
|
|
75
|
+
this License, each Contributor hereby grants to You a perpetual,
|
|
76
|
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
|
77
|
+
(except as stated in this section) patent license to make, have made,
|
|
78
|
+
use, offer to sell, sell, import, and otherwise transfer the Work,
|
|
79
|
+
where such license applies only to those patent claims licensable
|
|
80
|
+
by such Contributor that are necessarily infringed by their
|
|
81
|
+
Contribution(s) alone or by combination of their Contribution(s)
|
|
82
|
+
with the Work to which such Contribution(s) was submitted. If You
|
|
83
|
+
institute patent litigation against any entity (including a
|
|
84
|
+
cross-claim or counterclaim in a lawsuit) alleging that the Work
|
|
85
|
+
or a Contribution incorporated within the Work constitutes direct
|
|
86
|
+
or contributory patent infringement, then any patent licenses
|
|
87
|
+
granted to You under this License for that Work shall terminate
|
|
88
|
+
as of the date such litigation is filed.
|
|
89
|
+
|
|
90
|
+
4. Redistribution. You may reproduce and distribute copies of the
|
|
91
|
+
Work or Derivative Works thereof in any medium, with or without
|
|
92
|
+
modifications, and in Source or Object form, provided that You
|
|
93
|
+
meet the following conditions:
|
|
94
|
+
|
|
95
|
+
(a) You must give any other recipients of the Work or
|
|
96
|
+
Derivative Works a copy of this License; and
|
|
97
|
+
|
|
98
|
+
(b) You must cause any modified files to carry prominent notices
|
|
99
|
+
stating that You changed the files; and
|
|
100
|
+
|
|
101
|
+
(c) You must retain, in the Source form of any Derivative Works
|
|
102
|
+
that You distribute, all copyright, patent, trademark, and
|
|
103
|
+
attribution notices from the Source form of the Work,
|
|
104
|
+
excluding those notices that do not pertain to any part of
|
|
105
|
+
the Derivative Works; and
|
|
106
|
+
|
|
107
|
+
(d) If the Work includes a "NOTICE" text file as part of its
|
|
108
|
+
distribution, then any Derivative Works that You distribute must
|
|
109
|
+
include a readable copy of the attribution notices contained
|
|
110
|
+
within such NOTICE file, excluding those notices that do not
|
|
111
|
+
pertain to any part of the Derivative Works, in at least one
|
|
112
|
+
of the following places: within a NOTICE text file distributed
|
|
113
|
+
as part of the Derivative Works; within the Source form or
|
|
114
|
+
documentation, if provided along with the Derivative Works; or,
|
|
115
|
+
within a display generated by the Derivative Works, if and
|
|
116
|
+
wherever such third-party notices normally appear. The contents
|
|
117
|
+
of the NOTICE file are for informational purposes only and
|
|
118
|
+
do not modify the License. You may add Your own attribution
|
|
119
|
+
notices within Derivative Works that You distribute, alongside
|
|
120
|
+
or as an addendum to the NOTICE text from the Work, provided
|
|
121
|
+
that such additional attribution notices cannot be construed
|
|
122
|
+
as modifying the License.
|
|
123
|
+
|
|
124
|
+
You may add Your own copyright statement to Your modifications and
|
|
125
|
+
may provide additional or different license terms and conditions
|
|
126
|
+
for use, reproduction, or distribution of Your modifications, or
|
|
127
|
+
for any such Derivative Works as a whole, provided Your use,
|
|
128
|
+
reproduction, and distribution of the Work otherwise complies with
|
|
129
|
+
the conditions stated in this License.
|
|
130
|
+
|
|
131
|
+
5. Submission of Contributions. Unless You explicitly state otherwise,
|
|
132
|
+
any Contribution intentionally submitted for inclusion in the Work
|
|
133
|
+
by You to the Licensor shall be under the terms and conditions of
|
|
134
|
+
this License, without any additional terms or conditions.
|
|
135
|
+
Notwithstanding the above, nothing herein shall supersede or modify
|
|
136
|
+
the terms of any separate license agreement you may have executed
|
|
137
|
+
with Licensor regarding such Contributions.
|
|
138
|
+
|
|
139
|
+
6. Trademarks. This License does not grant permission to use the trade
|
|
140
|
+
names, trademarks, service marks, or product names of the Licensor,
|
|
141
|
+
except as required for reasonable and customary use in describing the
|
|
142
|
+
origin of the Work and reproducing the content of the NOTICE file.
|
|
143
|
+
|
|
144
|
+
7. Disclaimer of Warranty. Unless required by applicable law or
|
|
145
|
+
agreed to in writing, Licensor provides the Work (and each
|
|
146
|
+
Contributor provides its Contributions) on an "AS IS" BASIS,
|
|
147
|
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
|
|
148
|
+
implied, including, without limitation, any warranties or conditions
|
|
149
|
+
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
|
|
150
|
+
PARTICULAR PURPOSE. You are solely responsible for determining the
|
|
151
|
+
appropriateness of using or redistributing the Work and assume any
|
|
152
|
+
risks associated with Your exercise of permissions under this License.
|
|
153
|
+
|
|
154
|
+
8. Limitation of Liability. In no event and under no legal theory,
|
|
155
|
+
whether in tort (including negligence), contract, or otherwise,
|
|
156
|
+
unless required by applicable law (such as deliberate and grossly
|
|
157
|
+
negligent acts) or agreed to in writing, shall any Contributor be
|
|
158
|
+
liable to You for damages, including any direct, indirect, special,
|
|
159
|
+
incidental, or consequential damages of any character arising as a
|
|
160
|
+
result of this License or out of the use or inability to use the
|
|
161
|
+
Work (including but not limited to damages for loss of goodwill,
|
|
162
|
+
work stoppage, computer failure or malfunction, or any and all
|
|
163
|
+
other commercial damages or losses), even if such Contributor
|
|
164
|
+
has been advised of the possibility of such damages.
|
|
165
|
+
|
|
166
|
+
9. Accepting Warranty or Additional Liability. While redistributing
|
|
167
|
+
the Work or Derivative Works thereof, You may choose to offer,
|
|
168
|
+
and charge a fee for, acceptance of support, warranty, indemnity,
|
|
169
|
+
or other liability obligations and/or rights consistent with this
|
|
170
|
+
License. However, in accepting such obligations, You may act only
|
|
171
|
+
on Your own behalf and on Your sole responsibility, not on behalf
|
|
172
|
+
of any other Contributor, and only if You agree to indemnify,
|
|
173
|
+
defend, and hold each Contributor harmless for any liability
|
|
174
|
+
incurred by, or claims asserted against, such Contributor by reason
|
|
175
|
+
of your accepting any such warranty or additional liability.
|
|
176
|
+
|
|
177
|
+
END OF TERMS AND CONDITIONS
|
|
178
|
+
|
|
179
|
+
APPENDIX: How to apply the Apache License to your work.
|
|
180
|
+
|
|
181
|
+
To apply the Apache License to your work, attach the following
|
|
182
|
+
boilerplate notice, with the fields enclosed by brackets "[]"
|
|
183
|
+
replaced with your own identifying information. (Don't include
|
|
184
|
+
the brackets!) The text should be enclosed in the appropriate
|
|
185
|
+
comment syntax for the file format. We also recommend that a
|
|
186
|
+
file or class name and description of purpose be included on the
|
|
187
|
+
same "printed page" as the copyright notice for easier
|
|
188
|
+
identification within third-party archives.
|
|
189
|
+
|
|
190
|
+
Copyright [yyyy] [name of copyright owner]
|
|
191
|
+
|
|
192
|
+
Licensed under the Apache License, Version 2.0 (the "License");
|
|
193
|
+
you may not use this file except in compliance with the License.
|
|
194
|
+
You may obtain a copy of the License at
|
|
195
|
+
|
|
196
|
+
http://www.apache.org/licenses/LICENSE-2.0
|
|
197
|
+
|
|
198
|
+
Unless required by applicable law or agreed to in writing, software
|
|
199
|
+
distributed under the License is distributed on an "AS IS" BASIS,
|
|
200
|
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
201
|
+
See the License for the specific language governing permissions and
|
|
202
|
+
limitations under the License.
|
awgraph-1.0.0/PKG-INFO
ADDED
|
@@ -0,0 +1,141 @@
|
|
|
1
|
+
Metadata-Version: 2.4
|
|
2
|
+
Name: awgraph
|
|
3
|
+
Version: 1.0.0
|
|
4
|
+
Summary: Semantic Python code graph — AST indexing, call graphs, and hybrid search for Python codebases.
|
|
5
|
+
License: Apache-2.0
|
|
6
|
+
Project-URL: Homepage, https://github.com/Aitherium/awgraph
|
|
7
|
+
Project-URL: Documentation, https://github.com/Aitherium/awgraph#readme
|
|
8
|
+
Project-URL: Repository, https://github.com/Aitherium/awgraph.git
|
|
9
|
+
Project-URL: Issues, https://github.com/Aitherium/awgraph/issues
|
|
10
|
+
Requires-Python: >=3.10
|
|
11
|
+
Description-Content-Type: text/markdown
|
|
12
|
+
License-File: LICENSE
|
|
13
|
+
Requires-Dist: httpx>=0.25.0
|
|
14
|
+
Provides-Extra: numpy
|
|
15
|
+
Requires-Dist: numpy>=1.21.0; extra == "numpy"
|
|
16
|
+
Provides-Extra: dev
|
|
17
|
+
Requires-Dist: pytest>=7.0; extra == "dev"
|
|
18
|
+
Requires-Dist: pytest-asyncio>=0.21.0; extra == "dev"
|
|
19
|
+
Dynamic: license-file
|
|
20
|
+
|
|
21
|
+
# awgraph — the code graph your agent reads instead of grepping
|
|
22
|
+
|
|
23
|
+
An agent asked to fix a bug does not know which files matter, so it greps, opens
|
|
24
|
+
whatever matched, and spends most of its context window on code it will not
|
|
25
|
+
change. awgraph indexes the repository into a graph of symbols — functions,
|
|
26
|
+
methods, classes, their calls and callers — and answers a natural-language task
|
|
27
|
+
with the handful of chunks that task actually needs.
|
|
28
|
+
|
|
29
|
+
```bash
|
|
30
|
+
pip install awgraph
|
|
31
|
+
```
|
|
32
|
+
|
|
33
|
+
Python 3.10+. Nothing else is required to index and query.
|
|
34
|
+
|
|
35
|
+
## What it costs, measured
|
|
36
|
+
|
|
37
|
+
The interesting question is not "is a graph better than grep" — it is *what does
|
|
38
|
+
each cost*. Measured on a corpus of real commits, where the task is a commit
|
|
39
|
+
message with the answer filenames stripped out and the truth is the set of files
|
|
40
|
+
that commit actually modified:
|
|
41
|
+
|
|
42
|
+
| retriever | recall@10 | context tokens per task |
|
|
43
|
+
|---|---|---|
|
|
44
|
+
| ranked multi-term grep | 0.933 | 351,369 |
|
|
45
|
+
| **awgraph** | **0.800** | **1,503** |
|
|
46
|
+
|
|
47
|
+
**86% of grep's recall for 0.43% of the context — a 234x reduction.**
|
|
48
|
+
|
|
49
|
+
Read that honestly: **grep still wins on recall.** awgraph is not a better
|
|
50
|
+
*finder*; it is a dramatically cheaper one, and on a long agent loop the context
|
|
51
|
+
budget is usually what runs out first. If you need maximum recall and cost is no
|
|
52
|
+
object, grep. If you are paying per token across thousands of turns, this is a
|
|
53
|
+
different order of magnitude.
|
|
54
|
+
|
|
55
|
+
Caveats, because a benchmark without them is marketing: n=15 tasks, and only
|
|
56
|
+
33.3% of chunks carried embeddings on that run, so the semantic half was working
|
|
57
|
+
at partial strength — the recall number understates it.
|
|
58
|
+
|
|
59
|
+
A naive fusion (run the graph, fall back to grep when it returns few files) was
|
|
60
|
+
also measured: it reaches grep's 0.933 recall at 352,872 tokens — *more* than
|
|
61
|
+
grep alone, because the fallback fires on nearly every task and pays both bills.
|
|
62
|
+
Reported because it did not work; a smarter trigger is future work.
|
|
63
|
+
|
|
64
|
+
## Setup: index once, embed lazily
|
|
65
|
+
|
|
66
|
+
Two costs, and only one of them scales with repo size.
|
|
67
|
+
|
|
68
|
+
| step | 2,400 chunks | 43,730 chunks |
|
|
69
|
+
|---|---|---|
|
|
70
|
+
| parse + index | 49.8s | 75.5s |
|
|
71
|
+
| embed (CPU) | — | ~97 min at ~450 vectors/min |
|
|
72
|
+
|
|
73
|
+
**Indexing is close to size-insensitive** — 27x the files for 1.5x the time,
|
|
74
|
+
because parsing runs across workers. Embedding is the part that hurts on CPU, so
|
|
75
|
+
it is optional, cached and incremental: re-indexing reuses stored vectors and
|
|
76
|
+
only embeds what changed.
|
|
77
|
+
|
|
78
|
+
Without any embedding backend, queries fall back to keyword scoring and still
|
|
79
|
+
work. **That fallback is silent by design and dangerous by nature** — a graph
|
|
80
|
+
with no vectors looks like a working graph that is merely worse. Check coverage
|
|
81
|
+
rather than assuming it:
|
|
82
|
+
|
|
83
|
+
```python
|
|
84
|
+
embedded = sum(1 for c in graph.chunks.values() if c.embedding is not None)
|
|
85
|
+
print(f"{embedded}/{len(graph.chunks)} chunks carry vectors")
|
|
86
|
+
```
|
|
87
|
+
|
|
88
|
+
## Use it
|
|
89
|
+
|
|
90
|
+
```python
|
|
91
|
+
import asyncio
|
|
92
|
+
from awgraph import CodeGraph
|
|
93
|
+
|
|
94
|
+
async def main():
|
|
95
|
+
graph = CodeGraph(root_path="/abs/path/to/repo", auto_index=False)
|
|
96
|
+
await graph.index_codebase("/abs/path/to/repo") # absolute path required
|
|
97
|
+
|
|
98
|
+
for chunk in await graph.hybrid_query("retry with exponential backoff", max_results=5):
|
|
99
|
+
print(chunk.name, chunk.source_path, chunk.start_line)
|
|
100
|
+
|
|
101
|
+
asyncio.run(main())
|
|
102
|
+
```
|
|
103
|
+
|
|
104
|
+
`index_codebase` needs an **absolute** path. Given a relative one it walks
|
|
105
|
+
nothing, indexes zero chunks, and returns successfully — so assert on
|
|
106
|
+
`len(graph.chunks)` rather than on the absence of an exception.
|
|
107
|
+
|
|
108
|
+
The query does not need to contain the symbol name. Asking for "backoff policy
|
|
109
|
+
for flaky calls" against a class documented as *"Backoff policy for flaky
|
|
110
|
+
calls"* returns it by meaning, not by string match.
|
|
111
|
+
|
|
112
|
+
## Where it sits
|
|
113
|
+
|
|
114
|
+
Three packages, three different questions about the same repository:
|
|
115
|
+
|
|
116
|
+
- **[awgit](https://github.com/Aitherium/awgit)** — semantic version control.
|
|
117
|
+
Stable node ids, semantic edit-ops, leases so concurrent agents do not
|
|
118
|
+
overwrite each other, stacked commits with one PR each. It knows **what
|
|
119
|
+
changed and who is editing it**.
|
|
120
|
+
- **awgraph** — code intelligence. Symbols, call paths, dependencies, blast
|
|
121
|
+
radius. It knows **what the code is and what depends on what**.
|
|
122
|
+
- **[aither-adk](https://github.com/Aitherium/aither-adk)** — the agent runtime
|
|
123
|
+
that consumes both.
|
|
124
|
+
|
|
125
|
+
The seam is the useful part: awgit tells you a commit touched
|
|
126
|
+
`RetryPolicy.next_delay`; awgraph tells you what calls it and which tests cover
|
|
127
|
+
it; the agent reads that instead of the repository.
|
|
128
|
+
|
|
129
|
+
## Related work
|
|
130
|
+
|
|
131
|
+
[GitNexus](https://github.com/abhigyanpatwari/GitNexus) is the closest analogue
|
|
132
|
+
and worth reading. Its recommended mode augments grep with graph context rather
|
|
133
|
+
than replacing grep — a conclusion these measurements independently reach. Note
|
|
134
|
+
its licence is PolyForm Noncommercial (source-available, commercial use
|
|
135
|
+
forbidden), where awgraph is Apache 2.0. Its published figures measure SWE-bench
|
|
136
|
+
task resolution; the numbers above measure retrieval recall. Those are different
|
|
137
|
+
axes and should not be compared directly.
|
|
138
|
+
|
|
139
|
+
## Licence
|
|
140
|
+
|
|
141
|
+
Apache 2.0.
|
awgraph-1.0.0/README.md
ADDED
|
@@ -0,0 +1,121 @@
|
|
|
1
|
+
# awgraph — the code graph your agent reads instead of grepping
|
|
2
|
+
|
|
3
|
+
An agent asked to fix a bug does not know which files matter, so it greps, opens
|
|
4
|
+
whatever matched, and spends most of its context window on code it will not
|
|
5
|
+
change. awgraph indexes the repository into a graph of symbols — functions,
|
|
6
|
+
methods, classes, their calls and callers — and answers a natural-language task
|
|
7
|
+
with the handful of chunks that task actually needs.
|
|
8
|
+
|
|
9
|
+
```bash
|
|
10
|
+
pip install awgraph
|
|
11
|
+
```
|
|
12
|
+
|
|
13
|
+
Python 3.10+. Nothing else is required to index and query.
|
|
14
|
+
|
|
15
|
+
## What it costs, measured
|
|
16
|
+
|
|
17
|
+
The interesting question is not "is a graph better than grep" — it is *what does
|
|
18
|
+
each cost*. Measured on a corpus of real commits, where the task is a commit
|
|
19
|
+
message with the answer filenames stripped out and the truth is the set of files
|
|
20
|
+
that commit actually modified:
|
|
21
|
+
|
|
22
|
+
| retriever | recall@10 | context tokens per task |
|
|
23
|
+
|---|---|---|
|
|
24
|
+
| ranked multi-term grep | 0.933 | 351,369 |
|
|
25
|
+
| **awgraph** | **0.800** | **1,503** |
|
|
26
|
+
|
|
27
|
+
**86% of grep's recall for 0.43% of the context — a 234x reduction.**
|
|
28
|
+
|
|
29
|
+
Read that honestly: **grep still wins on recall.** awgraph is not a better
|
|
30
|
+
*finder*; it is a dramatically cheaper one, and on a long agent loop the context
|
|
31
|
+
budget is usually what runs out first. If you need maximum recall and cost is no
|
|
32
|
+
object, grep. If you are paying per token across thousands of turns, this is a
|
|
33
|
+
different order of magnitude.
|
|
34
|
+
|
|
35
|
+
Caveats, because a benchmark without them is marketing: n=15 tasks, and only
|
|
36
|
+
33.3% of chunks carried embeddings on that run, so the semantic half was working
|
|
37
|
+
at partial strength — the recall number understates it.
|
|
38
|
+
|
|
39
|
+
A naive fusion (run the graph, fall back to grep when it returns few files) was
|
|
40
|
+
also measured: it reaches grep's 0.933 recall at 352,872 tokens — *more* than
|
|
41
|
+
grep alone, because the fallback fires on nearly every task and pays both bills.
|
|
42
|
+
Reported because it did not work; a smarter trigger is future work.
|
|
43
|
+
|
|
44
|
+
## Setup: index once, embed lazily
|
|
45
|
+
|
|
46
|
+
Two costs, and only one of them scales with repo size.
|
|
47
|
+
|
|
48
|
+
| step | 2,400 chunks | 43,730 chunks |
|
|
49
|
+
|---|---|---|
|
|
50
|
+
| parse + index | 49.8s | 75.5s |
|
|
51
|
+
| embed (CPU) | — | ~97 min at ~450 vectors/min |
|
|
52
|
+
|
|
53
|
+
**Indexing is close to size-insensitive** — 27x the files for 1.5x the time,
|
|
54
|
+
because parsing runs across workers. Embedding is the part that hurts on CPU, so
|
|
55
|
+
it is optional, cached and incremental: re-indexing reuses stored vectors and
|
|
56
|
+
only embeds what changed.
|
|
57
|
+
|
|
58
|
+
Without any embedding backend, queries fall back to keyword scoring and still
|
|
59
|
+
work. **That fallback is silent by design and dangerous by nature** — a graph
|
|
60
|
+
with no vectors looks like a working graph that is merely worse. Check coverage
|
|
61
|
+
rather than assuming it:
|
|
62
|
+
|
|
63
|
+
```python
|
|
64
|
+
embedded = sum(1 for c in graph.chunks.values() if c.embedding is not None)
|
|
65
|
+
print(f"{embedded}/{len(graph.chunks)} chunks carry vectors")
|
|
66
|
+
```
|
|
67
|
+
|
|
68
|
+
## Use it
|
|
69
|
+
|
|
70
|
+
```python
|
|
71
|
+
import asyncio
|
|
72
|
+
from awgraph import CodeGraph
|
|
73
|
+
|
|
74
|
+
async def main():
|
|
75
|
+
graph = CodeGraph(root_path="/abs/path/to/repo", auto_index=False)
|
|
76
|
+
await graph.index_codebase("/abs/path/to/repo") # absolute path required
|
|
77
|
+
|
|
78
|
+
for chunk in await graph.hybrid_query("retry with exponential backoff", max_results=5):
|
|
79
|
+
print(chunk.name, chunk.source_path, chunk.start_line)
|
|
80
|
+
|
|
81
|
+
asyncio.run(main())
|
|
82
|
+
```
|
|
83
|
+
|
|
84
|
+
`index_codebase` needs an **absolute** path. Given a relative one it walks
|
|
85
|
+
nothing, indexes zero chunks, and returns successfully — so assert on
|
|
86
|
+
`len(graph.chunks)` rather than on the absence of an exception.
|
|
87
|
+
|
|
88
|
+
The query does not need to contain the symbol name. Asking for "backoff policy
|
|
89
|
+
for flaky calls" against a class documented as *"Backoff policy for flaky
|
|
90
|
+
calls"* returns it by meaning, not by string match.
|
|
91
|
+
|
|
92
|
+
## Where it sits
|
|
93
|
+
|
|
94
|
+
Three packages, three different questions about the same repository:
|
|
95
|
+
|
|
96
|
+
- **[awgit](https://github.com/Aitherium/awgit)** — semantic version control.
|
|
97
|
+
Stable node ids, semantic edit-ops, leases so concurrent agents do not
|
|
98
|
+
overwrite each other, stacked commits with one PR each. It knows **what
|
|
99
|
+
changed and who is editing it**.
|
|
100
|
+
- **awgraph** — code intelligence. Symbols, call paths, dependencies, blast
|
|
101
|
+
radius. It knows **what the code is and what depends on what**.
|
|
102
|
+
- **[aither-adk](https://github.com/Aitherium/aither-adk)** — the agent runtime
|
|
103
|
+
that consumes both.
|
|
104
|
+
|
|
105
|
+
The seam is the useful part: awgit tells you a commit touched
|
|
106
|
+
`RetryPolicy.next_delay`; awgraph tells you what calls it and which tests cover
|
|
107
|
+
it; the agent reads that instead of the repository.
|
|
108
|
+
|
|
109
|
+
## Related work
|
|
110
|
+
|
|
111
|
+
[GitNexus](https://github.com/abhigyanpatwari/GitNexus) is the closest analogue
|
|
112
|
+
and worth reading. Its recommended mode augments grep with graph context rather
|
|
113
|
+
than replacing grep — a conclusion these measurements independently reach. Note
|
|
114
|
+
its licence is PolyForm Noncommercial (source-available, commercial use
|
|
115
|
+
forbidden), where awgraph is Apache 2.0. Its published figures measure SWE-bench
|
|
116
|
+
task resolution; the numbers above measure retrieval recall. Those are different
|
|
117
|
+
axes and should not be compared directly.
|
|
118
|
+
|
|
119
|
+
## Licence
|
|
120
|
+
|
|
121
|
+
Apache 2.0.
|
|
@@ -0,0 +1,70 @@
|
|
|
1
|
+
"""awgraph — Semantic Python Code Graph.
|
|
2
|
+
|
|
3
|
+
A lightweight, portable Python AST indexer with call graphs, embeddings support,
|
|
4
|
+
and hybrid keyword+semantic search. Originally developed as part of AitherOS's
|
|
5
|
+
CodeGraph faculty, now available as a public package.
|
|
6
|
+
|
|
7
|
+
## What it does
|
|
8
|
+
|
|
9
|
+
- Real AST parsing (not regex)
|
|
10
|
+
- Extracts functions, classes, and methods into chunks
|
|
11
|
+
- Builds call graphs (what calls what, what's called by what)
|
|
12
|
+
- Hybrid search: keyword (FTS5) + semantic (embeddings)
|
|
13
|
+
- SQLite storage with WAL mode for concurrent access
|
|
14
|
+
- Optional NumPy for fast vector operations
|
|
15
|
+
|
|
16
|
+
## What it doesn't do
|
|
17
|
+
|
|
18
|
+
- PDFs, web pages, or "universal" anything — just Python code
|
|
19
|
+
- Embeddings by default — bring your own or use plugin hooks
|
|
20
|
+
- Network access by default — embeddings API must be provided externally
|
|
21
|
+
- Persistence across instances — in-memory by design; SQLite for durability
|
|
22
|
+
|
|
23
|
+
## Usage
|
|
24
|
+
|
|
25
|
+
```python
|
|
26
|
+
from awgraph import CodeGraph
|
|
27
|
+
|
|
28
|
+
# Create an indexer
|
|
29
|
+
graph = CodeGraph()
|
|
30
|
+
|
|
31
|
+
# Index a directory
|
|
32
|
+
await graph.index_codebase("/path/to/code")
|
|
33
|
+
|
|
34
|
+
# Query
|
|
35
|
+
chunks = await graph.query("rate limiter", limit=10)
|
|
36
|
+
|
|
37
|
+
for chunk in chunks:
|
|
38
|
+
print(f"{chunk.name}: {chunk.calls}")
|
|
39
|
+
```
|
|
40
|
+
|
|
41
|
+
## Architecture
|
|
42
|
+
|
|
43
|
+
- **graph.py**: CodeGraph engine, AST parsing, call graphs, chunking
|
|
44
|
+
- **store.py**: SQLite+FTS5 backend for storage and search
|
|
45
|
+
- **registry.py**: Multi-root manager for indexing external repos
|
|
46
|
+
- **base.py**: Minimal faculty graph base class
|
|
47
|
+
- **logging.py**: Standard Python logging shim
|
|
48
|
+
- **degradation.py**: Optional dependency tracking
|
|
49
|
+
|
|
50
|
+
Plugin support via awgraph.plugins (not yet public).
|
|
51
|
+
"""
|
|
52
|
+
|
|
53
|
+
from __future__ import annotations
|
|
54
|
+
|
|
55
|
+
from awgraph.graph import CodeGraph, CodeChunk
|
|
56
|
+
from awgraph.registry import CodeGraphRegistry, get_codegraph_registry
|
|
57
|
+
from awgraph.store import CodeGraphStore
|
|
58
|
+
from awgraph.base import BaseFacultyGraph, GraphSyncConfig
|
|
59
|
+
|
|
60
|
+
__version__ = "1.0.0"
|
|
61
|
+
|
|
62
|
+
__all__ = [
|
|
63
|
+
"CodeGraph",
|
|
64
|
+
"CodeChunk",
|
|
65
|
+
"CodeGraphRegistry",
|
|
66
|
+
"CodeGraphStore",
|
|
67
|
+
"BaseFacultyGraph",
|
|
68
|
+
"GraphSyncConfig",
|
|
69
|
+
"get_codegraph_registry",
|
|
70
|
+
]
|
|
@@ -0,0 +1,139 @@
|
|
|
1
|
+
"""
|
|
2
|
+
Minimal base class for awgraph.
|
|
3
|
+
|
|
4
|
+
Provides the core interface a faculty graph implements, without the AitherOS
|
|
5
|
+
integrations (sync, provenance, integrity). This is what the public package
|
|
6
|
+
exposes.
|
|
7
|
+
"""
|
|
8
|
+
|
|
9
|
+
from __future__ import annotations
|
|
10
|
+
|
|
11
|
+
from dataclasses import dataclass
|
|
12
|
+
from typing import Any, Dict, List
|
|
13
|
+
|
|
14
|
+
|
|
15
|
+
@dataclass
|
|
16
|
+
class GraphSyncConfig:
|
|
17
|
+
"""Stub for compatibility; sync is not available in the public package."""
|
|
18
|
+
enabled: bool = False
|
|
19
|
+
domain: str = ""
|
|
20
|
+
batch_size: int = 20
|
|
21
|
+
flush_interval: float = 5.0
|
|
22
|
+
source_graph: str = ""
|
|
23
|
+
provenance: bool = False
|
|
24
|
+
provenance_node_type: str = "claim"
|
|
25
|
+
|
|
26
|
+
|
|
27
|
+
class BaseFacultyGraph:
|
|
28
|
+
"""
|
|
29
|
+
Minimal base class for a faculty graph.
|
|
30
|
+
|
|
31
|
+
In AitherOS, this also provides sync to AitherKnowledgeGraph, provenance
|
|
32
|
+
emission, and integrity tracking. The public package drops those features,
|
|
33
|
+
keeping only the core interface: query(), stats(), and graph_query().
|
|
34
|
+
"""
|
|
35
|
+
|
|
36
|
+
_sync_config: GraphSyncConfig = GraphSyncConfig()
|
|
37
|
+
_scope_level: str = "platform"
|
|
38
|
+
|
|
39
|
+
def __init__(self):
|
|
40
|
+
self._sync_config = GraphSyncConfig()
|
|
41
|
+
|
|
42
|
+
def _queue_sync(self, node_data: Dict[str, Any], tenant_id: str = "platform") -> None:
|
|
43
|
+
"""No-op: sync to AitherKnowledgeGraph is not available in public package."""
|
|
44
|
+
pass
|
|
45
|
+
|
|
46
|
+
def _queue_deletion(self, node_id: str, tenant_id: str = "platform") -> None:
|
|
47
|
+
"""No-op: sync is not available in public package."""
|
|
48
|
+
pass
|
|
49
|
+
|
|
50
|
+
# SYNC on purpose: every call site is `self._flush_to_bus()` with no await.
|
|
51
|
+
# Declaring it async made each call return a coroutine nobody awaits, which
|
|
52
|
+
# Python reports only as a RuntimeWarning — so the no-op would have looked
|
|
53
|
+
# like it ran while doing nothing, in a package whose whole promise is that
|
|
54
|
+
# the stub is inert rather than silently broken.
|
|
55
|
+
def _flush_to_bus(self, *args: Any, **kwargs: Any) -> None:
|
|
56
|
+
"""No-op: there is no event bus in the standalone package.
|
|
57
|
+
|
|
58
|
+
This one is not decoration. Indexing calls it, so omitting it made the
|
|
59
|
+
package import cleanly and then die with AttributeError on the FIRST
|
|
60
|
+
index_codebase() call — the shape where a vendored package looks fine to
|
|
61
|
+
every import check and is broken for the person who pip-installed it.
|
|
62
|
+
An import test cannot catch this; only actually indexing something can.
|
|
63
|
+
|
|
64
|
+
Subclasses in a host application may override it to re-attach their own
|
|
65
|
+
event plumbing; nothing here needs to know that they exist.
|
|
66
|
+
"""
|
|
67
|
+
return None
|
|
68
|
+
|
|
69
|
+
def _emit_provenance(
|
|
70
|
+
self,
|
|
71
|
+
node_id: str,
|
|
72
|
+
name: str,
|
|
73
|
+
properties: Dict[str, Any],
|
|
74
|
+
tenant_id: str = "platform",
|
|
75
|
+
) -> None:
|
|
76
|
+
"""No-op: provenance emission is not available in public package."""
|
|
77
|
+
pass
|
|
78
|
+
|
|
79
|
+
def graph_stats(self) -> Dict[str, Any]:
|
|
80
|
+
"""Inventory: how many nodes does this graph hold?
|
|
81
|
+
|
|
82
|
+
Subclasses may override. Default derives counts from public attributes.
|
|
83
|
+
"""
|
|
84
|
+
entities: Dict[str, int] = {}
|
|
85
|
+
for name, val in vars(self).items():
|
|
86
|
+
if name.startswith("_") or name.startswith("by_"):
|
|
87
|
+
continue
|
|
88
|
+
if not isinstance(val, (dict, list, set, tuple)):
|
|
89
|
+
continue
|
|
90
|
+
entities[name] = len(val)
|
|
91
|
+
return {"nodes": sum(entities.values()), "containers": entities}
|
|
92
|
+
|
|
93
|
+
async def graph_query(
|
|
94
|
+
self, text: str, limit: int = 10,
|
|
95
|
+
) -> List[Dict[str, Any]]:
|
|
96
|
+
"""Uniform cross-graph query.
|
|
97
|
+
|
|
98
|
+
Adapts this graph's search() or query() method by parameter name.
|
|
99
|
+
Subclasses may override for a better native path.
|
|
100
|
+
"""
|
|
101
|
+
import asyncio
|
|
102
|
+
import inspect
|
|
103
|
+
|
|
104
|
+
fn = getattr(self, "search", None) or getattr(self, "query", None)
|
|
105
|
+
if not callable(fn):
|
|
106
|
+
return []
|
|
107
|
+
|
|
108
|
+
try:
|
|
109
|
+
sig = inspect.signature(fn)
|
|
110
|
+
except (TypeError, ValueError):
|
|
111
|
+
return []
|
|
112
|
+
|
|
113
|
+
# Bind by parameter name (not position)
|
|
114
|
+
kwargs = {}
|
|
115
|
+
for p in sig.parameters.values():
|
|
116
|
+
if p.name in ("query", "query_str", "text", "prompt", "search"):
|
|
117
|
+
kwargs[p.name] = text
|
|
118
|
+
elif p.name in ("limit", "max_results", "top_k", "k"):
|
|
119
|
+
kwargs[p.name] = limit
|
|
120
|
+
|
|
121
|
+
# Call sync or async
|
|
122
|
+
try:
|
|
123
|
+
if asyncio.iscoroutinefunction(fn):
|
|
124
|
+
return await fn(**kwargs)
|
|
125
|
+
else:
|
|
126
|
+
loop = asyncio.get_event_loop()
|
|
127
|
+
return await loop.run_in_executor(None, fn, **kwargs)
|
|
128
|
+
except TypeError:
|
|
129
|
+
# Fall back to no kwargs if binding failed
|
|
130
|
+
try:
|
|
131
|
+
if asyncio.iscoroutinefunction(fn):
|
|
132
|
+
return await fn(text, limit)
|
|
133
|
+
else:
|
|
134
|
+
loop = asyncio.get_event_loop()
|
|
135
|
+
return await loop.run_in_executor(None, lambda: fn(text, limit))
|
|
136
|
+
except Exception:
|
|
137
|
+
return []
|
|
138
|
+
except Exception:
|
|
139
|
+
return []
|