A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 16, 05:45 UTC
[Doc] Add github links for source code references (#10672)
Signed-off-by: Russell Bryant <rbryant@redhat.com> Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk> Co-authored-by: DarkLight1337 <tlleungac@connect.ust.hk>
Russell Bryant committed
Dec 3, 2024 at 01:46 UTC
ef51831ee8dbd64833b25e042d4e984d169202f9
2 files changed
+68
-1
docs/requirements-docs.txt
+2
-1
@@ -16,4 +16,5 @@ mistral_common >= 1.5.0
16
aiohttp
17
starlette
18
openai # Required by docs/source/serving/openai_compatible_server.md's vllm.entrypoints.openai.cli_args
19
-partial-json-parser # Required by docs/source/serving/openai_compatible_server.md's vllm.entrypoints.openai.cli_args
\ No newline at end of file
19
+partial-json-parser # Required by docs/source/serving/openai_compatible_server.md's vllm.entrypoints.openai.cli_args
20
+requests
docs/source/conf.py
+66
@@ -10,11 +10,13 @@
10
# add these directories to sys.path here. If the directory is relative to the
11
# documentation root, use os.path.abspath to make it absolute, like shown here.
12
13
+import inspect
14
import logging
15
import os
16
import sys
17
from typing import List
18
19
+import requests
20
from sphinx.ext import autodoc
21
22
logger = logging.getLogger(__name__)
@@ -34,6 +36,7 @@ author = 'the vLLM Team'
36
extensions = [
37
"sphinx.ext.napoleon",
38
"sphinx.ext.viewcode",
39
+ "sphinx.ext.linkcode",
40
"sphinx.ext.intersphinx",
41
"sphinx_copybutton",
42
"sphinx.ext.autodoc",
@@ -94,6 +97,69 @@ def setup(app):
97
generate_examples()
98
99
100
+_cached_base: str = ""
101
+_cached_branch: str = ""
102
+
103
+
104
+def get_repo_base_and_branch(pr_number):
105
+ global _cached_base, _cached_branch
106
+ if _cached_base and _cached_branch:
107
+ return _cached_base, _cached_branch
108
+
109
+ url = f"https://api.github.com/repos/vllm-project/vllm/pulls/{pr_number}"
110
+ response = requests.get(url)
111
+ if response.status_code == 200:
112
+ data = response.json()
113
+ _cached_base = data['head']['repo']['full_name']
114
+ _cached_branch = data['head']['ref']
115
+ return _cached_base, _cached_branch
116
+ else:
117
+ logger.error("Failed to fetch PR details: %s", response)
118
+ return None, None
119
+
120
+
121
+def linkcode_resolve(domain, info):
122
+ if domain != 'py':
123
+ return None
124
+ if not info['module']:
125
+ return None
126
+ filename = info['module'].replace('.', '/')
127
+ module = info['module']
128
+
129
+ # try to determine the correct file and line number to link to
130
+ obj = sys.modules[module]
131
+
132
+ # get as specific as we can
133
+ lineno: int = 0
134
+ filename: str = ""
135
+ try:
136
+ for part in info['fullname'].split('.'):
137
+ obj = getattr(obj, part)
138
+
139
+ if not (inspect.isclass(obj) or inspect.isfunction(obj)
140
+ or inspect.ismethod(obj)):
141
+ obj = obj.__class__ # Get the class of the instance
142
+
143
+ lineno = inspect.getsourcelines(obj)[1]
144
+ filename = (inspect.getsourcefile(obj)
145
+ or f"{filename}.py").split("vllm/", 1)[1]
146
+ except Exception:
147
+ # For some things, like a class member, won't work, so
148
+ # we'll use the line number of the parent (the class)
149
+ pass
150
+
151
+ if filename.startswith("checkouts/"):
152
+ # a PR build on readthedocs
153
+ pr_number = filename.split("/")[1]
154
+ filename = filename.split("/", 2)[2]
155
+ base, branch = get_repo_base_and_branch(pr_number)
156
+ if base and branch:
157
+ return f"https://github.com/{base}/blob/{branch}/{filename}#L{lineno}"
158
+
159
+ # Otherwise, link to the source file on the main branch
160
+ return f"https://github.com/vllm-project/vllm/blob/main/{filename}#L{lineno}"
161
+
162
+
163
# Mock out external dependencies here, otherwise the autodoc pages may be blank.
164
autodoc_mock_imports = [
165
"compressed_tensors",