Repository navigation
Expand file tree
/
Copy pathmodular_auto_docstring.py
More file actions
403 lines (321 loc) · 14.6 KB
/
Copy pathmodular_auto_docstring.py
File metadata and controls
403 lines (321 loc) · 14.6 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
# coding=utf-8
# Copyright 2025 The HuggingFace Inc. team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Auto Docstring Generator for Modular Pipeline Blocks
This script scans Python files for classes that have `# auto_docstring` comment above them
and inserts/updates the docstring from the class's `doc` property.
Run from the root of the repo:
python utils/modular_auto_docstring.py [path] [--fix_and_overwrite]
Examples:
# Check mode (will error if a marked class's docstring is missing or out of date — the check regenerates
# each docstring the same way --fix_and_overwrite would and compares the result with the file)
python utils/modular_auto_docstring.py
# Check specific directory
python utils/modular_auto_docstring.py src/diffusers/modular_pipelines/
# Fix and overwrite the docstrings
python utils/modular_auto_docstring.py --fix_and_overwrite
Usage in code:
# auto_docstring
class QwenImageAutoVaeEncoderStep(AutoPipelineBlocks):
# docstring will be automatically inserted here
@property
def doc(self):
return "Your docstring content..."
"""
import argparse
import ast
import glob
import importlib
import os
import re
import subprocess
import sys
import tempfile
# All paths are set with the intent you should run this script from the root of the repo
DIFFUSERS_PATH = "src/diffusers"
REPO_PATH = "."
# Pattern to match the auto_docstring comment
AUTO_DOCSTRING_PATTERN = re.compile(r"^\s*#\s*auto_docstring\s*$")
def setup_diffusers_import():
"""Setup import path to use the local diffusers module."""
src_path = os.path.join(REPO_PATH, "src")
if src_path not in sys.path:
sys.path.insert(0, src_path)
def get_module_from_filepath(filepath: str) -> str:
"""Convert a filepath to a module name."""
filepath = os.path.normpath(filepath)
if filepath.startswith("src" + os.sep):
filepath = filepath[4:]
if filepath.endswith(".py"):
filepath = filepath[:-3]
module_name = filepath.replace(os.sep, ".")
return module_name
def load_module(filepath: str):
"""Load a module from filepath."""
setup_diffusers_import()
module_name = get_module_from_filepath(filepath)
try:
module = importlib.import_module(module_name)
return module
except Exception as e:
print(f"Warning: Could not import module {module_name}: {e}")
return None
def get_doc_from_class(module, class_name: str) -> str:
"""Get the doc property from an instantiated class."""
if module is None:
return None
cls = getattr(module, class_name, None)
if cls is None:
return None
try:
instance = cls()
if hasattr(instance, "doc"):
return instance.doc
except Exception as e:
print(f"Warning: Could not instantiate {class_name}: {e}")
return None
def find_auto_docstring_classes(filepath: str) -> list:
"""
Find all classes in a file that have # auto_docstring comment above them.
Returns list of (class_name, class_line_number, has_existing_docstring, docstring_end_line)
"""
with open(filepath, "r", encoding="utf-8", newline="\n") as f:
lines = f.readlines()
# Parse AST to find class locations and their docstrings
content = "".join(lines)
try:
tree = ast.parse(content)
except SyntaxError as e:
print(f"Syntax error in {filepath}: {e}")
return []
# Build a map of class_name -> (class_line, has_docstring, docstring_end_line)
class_info = {}
for node in ast.walk(tree):
if isinstance(node, ast.ClassDef):
has_docstring = False
docstring_end_line = node.lineno # default to class line
if node.body and isinstance(node.body[0], ast.Expr):
first_stmt = node.body[0]
if isinstance(first_stmt.value, ast.Constant) and isinstance(first_stmt.value.value, str):
has_docstring = True
docstring_end_line = first_stmt.end_lineno or first_stmt.lineno
class_info[node.name] = (node.lineno, has_docstring, docstring_end_line)
# Now scan for # auto_docstring comments
classes_to_update = []
for i, line in enumerate(lines):
if AUTO_DOCSTRING_PATTERN.match(line):
# Found the marker, look for class definition on next non-empty, non-comment line
j = i + 1
while j < len(lines):
next_line = lines[j].strip()
if next_line and not next_line.startswith("#"):
break
j += 1
if j < len(lines) and lines[j].strip().startswith("class "):
# Extract class name
match = re.match(r"class\s+(\w+)", lines[j].strip())
if match:
class_name = match.group(1)
if class_name in class_info:
class_line, has_docstring, docstring_end_line = class_info[class_name]
classes_to_update.append((class_name, class_line, has_docstring, docstring_end_line))
return classes_to_update
def strip_class_name_line(doc: str, class_name: str) -> str:
"""Remove the 'class ClassName' line from the doc if present."""
lines = doc.strip().split("\n")
if lines and lines[0].strip() == f"class {class_name}":
# Remove the class line and any blank line following it
lines = lines[1:]
while lines and not lines[0].strip():
lines = lines[1:]
return "\n".join(lines)
def format_docstring(doc: str, indent: str = " ") -> str:
"""Format a doc string as a properly indented docstring."""
lines = doc.strip().split("\n")
if len(lines) == 1:
return f'{indent}"""{lines[0]}"""\n'
else:
result = [f'{indent}"""\n']
for line in lines:
if line.strip():
result.append(f"{indent}{line}\n")
else:
result.append("\n")
result.append(f'{indent}"""\n')
return "".join(result)
def run_ruff_format(filepath: str, quiet: bool = False):
"""Run ruff check --fix, ruff format, and doc-builder style on a file to ensure consistent formatting."""
try:
# First run ruff check --fix to fix any linting issues (including line length)
subprocess.run(
["ruff", "check", "--fix", filepath],
check=False, # Don't fail if there are unfixable issues
capture_output=True,
text=True,
)
# Then run ruff format for code formatting
subprocess.run(
["ruff", "format", filepath],
check=True,
capture_output=True,
text=True,
)
# Finally run doc-builder style for docstring formatting
subprocess.run(
["doc-builder", "style", filepath, "--max_len", "119"],
check=False, # Don't fail if doc-builder has issues
capture_output=True,
text=True,
)
if not quiet:
print(f"Formatted {filepath}")
except subprocess.CalledProcessError as e:
print(f"Warning: formatting failed for {filepath}: {e.stderr}")
except FileNotFoundError as e:
print(f"Warning: tool not found ({e}). Skipping formatting.")
except Exception as e:
print(f"Warning: unexpected error formatting {filepath}: {e}")
def format_content(content: str, filepath: str) -> str:
"""Run the same formatters fix mode uses on a temporary copy of `content` and return the result."""
fd, tmp_path = tempfile.mkstemp(suffix=".py", dir=os.path.dirname(filepath) or ".")
try:
with os.fdopen(fd, "w", encoding="utf-8", newline="\n") as f:
f.write(content)
run_ruff_format(tmp_path, quiet=True)
with open(tmp_path, "r", encoding="utf-8", newline="\n") as f:
return f.read()
finally:
os.remove(tmp_path)
def get_existing_docstring(lines: list, class_line: int, docstring_end_line: int) -> str:
"""Extract the existing docstring content from lines."""
# class_line is 1-indexed, docstring starts at class_line (0-indexed: class_line)
# and ends at docstring_end_line (1-indexed, inclusive)
docstring_lines = lines[class_line:docstring_end_line]
return "".join(docstring_lines)
def build_updated_lines(filepath: str, classes_to_update: list) -> tuple:
"""
Return the file's lines with every marked class's docstring regenerated (pre-formatting), plus the names of the
classes that were regenerated.
"""
module = load_module(filepath)
with open(filepath, "r", encoding="utf-8", newline="\n") as f:
lines = f.readlines()
# Process in reverse order to maintain line numbers
updated_names = []
for class_name, class_line, has_docstring, docstring_end_line in reversed(classes_to_update):
doc = get_doc_from_class(module, class_name)
if doc is None:
print(f"Warning: Could not get doc for {class_name} in {filepath}")
continue
# Remove the "class ClassName" line since it's redundant in a docstring
doc = strip_class_name_line(doc, class_name)
# Format the new docstring with 4-space indent
new_docstring = format_docstring(doc, " ")
if has_docstring:
# Replace existing docstring (line after class definition to docstring_end_line)
# class_line is 1-indexed, we want to replace from class_line+1 to docstring_end_line
lines = lines[:class_line] + [new_docstring] + lines[docstring_end_line:]
else:
# Insert new docstring right after class definition line
# class_line is 1-indexed, so lines[class_line-1] is the class line
# Insert at position class_line (which is right after the class line)
lines = lines[:class_line] + [new_docstring] + lines[class_line:]
updated_names.append(class_name)
return lines, updated_names
def _class_docstrings(content: str) -> dict:
"""Map class name -> raw docstring value for every class in `content`."""
tree = ast.parse(content)
return {
node.name: ast.get_docstring(node, clean=False) for node in ast.walk(tree) if isinstance(node, ast.ClassDef)
}
def process_file(filepath: str, overwrite: bool = False) -> list:
"""
Process a file and find/insert docstrings for # auto_docstring marked classes.
In fix mode, regenerates every marked docstring in place. In check mode, regenerates them on a temporary copy
through the same formatting pipeline and returns the classes whose checked-in docstring differs from the
regenerated one (missing docstrings included).
Returns list of classes that need updating.
"""
classes_to_update = find_auto_docstring_classes(filepath)
if not classes_to_update:
return []
lines, updated_names = build_updated_lines(filepath, classes_to_update)
if overwrite:
if updated_names:
with open(filepath, "w", encoding="utf-8", newline="\n") as f:
f.writelines(lines)
# Run ruff format to ensure consistent line wrapping
run_ruff_format(filepath)
for class_name in reversed(updated_names):
print(f"Updated docstring for {class_name} in {filepath}")
return [(filepath, cls_name, line) for cls_name, line, _, _ in classes_to_update]
# Check mode: regenerate through the same formatting pipeline and compare with the checked-in file
with open(filepath, "r", encoding="utf-8", newline="\n") as f:
current_content = f.read()
new_content = format_content("".join(lines), filepath)
if new_content == current_content:
return []
current_docs = _class_docstrings(current_content)
new_docs = _class_docstrings(new_content)
stale = [
(filepath, class_name, class_line)
for class_name, class_line, _, _ in classes_to_update
if current_docs.get(class_name) != new_docs.get(class_name)
]
if not stale:
# the regenerated file differs outside the marked docstrings (e.g. the file itself is not
# ruff/doc-builder clean); attribute it to the file so the mismatch is not silently ignored
stale = [(filepath, classes_to_update[0][0], classes_to_update[0][1])]
return stale
def check_auto_docstrings(path: str = None, overwrite: bool = False):
"""
Check all files for # auto_docstring markers and optionally fix them.
"""
if path is None:
path = DIFFUSERS_PATH
if os.path.isfile(path):
all_files = [path]
else:
all_files = glob.glob(os.path.join(path, "**/*.py"), recursive=True)
all_markers = []
for filepath in all_files:
markers = process_file(filepath, overwrite)
all_markers.extend(markers)
if not overwrite and len(all_markers) > 0:
message = "\n".join([f"- {f}: {cls} at line {line}" for f, cls, line in all_markers])
raise ValueError(
f"The following # auto_docstring docstrings are missing or out of date:\n{message}\n\n"
f"These docstrings are generated from the block declarations (which may live in other files), so a "
f"change elsewhere can make them stale. Run `python utils/modular_auto_docstring.py --fix_and_overwrite` "
f"from the root of the repo to regenerate them."
)
if overwrite and len(all_markers) > 0:
print(f"\nProcessed {len(all_markers)} docstring(s).")
elif not overwrite and len(all_markers) == 0:
print("All # auto_docstring markers have up-to-date docstrings.")
elif len(all_markers) == 0:
print("No # auto_docstring markers found.")
if __name__ == "__main__":
parser = argparse.ArgumentParser(
description="Check and fix # auto_docstring markers in modular pipeline blocks",
)
parser.add_argument("path", nargs="?", default=None, help="File or directory to process (default: src/diffusers)")
parser.add_argument(
"--fix_and_overwrite",
action="store_true",
help="Whether to fix the docstrings by inserting them from doc property.",
)
args = parser.parse_args()
check_auto_docstrings(args.path, args.fix_and_overwrite)