cecli/cecli/coders/udiff_coder.py at main · cecli-dev/cecli · GitHub

Repository navigation

History

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

404

405

406

407

408

409

410

411

412

413

414

415

416

417

418

419

420

421

422

423

424

425

426

427

428

import difflib

from itertools import groupby

from pathlib import Path

from ..dump import dump # noqa: F401

from .base_coder import Coder

from .search_replace import (

SearchTextNotUnique,

all_preprocs,

diff_lines,

flexible_search_and_replace,

search_and_replace,

)

no_match_error = """UnifiedDiffNoMatch: hunk failed to apply!

{path} does not contain lines that match the diff you provided!

Try again.

DO NOT skip blank lines, comments, docstrings, etc!

The diff needs to apply cleanly to the lines in {path}!

{path} does not contain these {num_lines} exact lines in a row:

```

{original}```

"""

not_unique_error = """UnifiedDiffNotUnique: hunk failed to apply!

{path} contains multiple sets of lines that match the diff you provided!

Try again.

Use additional ` ` lines to provide context that uniquely indicates which code needs to be changed.

The diff needs to apply to a unique set of lines in {path}!

{path} contains multiple copies of these {num_lines} lines:

```

{original}```

"""

other_hunks_applied = (

"Note: some hunks did apply successfully. See the updated source code shown above.\n\n"

)

class UnifiedDiffCoder(Coder):

"""A coder that uses unified diff format for code modifications."""

edit_format = "udiff"

prompt_format = "udiff"

def get_edits(self):

content = self.partial_response_content

# might raise ValueError for malformed ORIG/UPD blocks

raw_edits = list(find_diffs(content))

last_path = None

edits = []

for path, hunk in raw_edits:

if path:

last_path = path

else:

path = last_path

edits.append((path, hunk))

return edits

def apply_edits(self, edits):

seen = set()

uniq = []

for path, hunk in edits:

hunk = normalize_hunk(hunk)

if not hunk:

continue

this = [path + "\n"] + hunk

this = "".join(this)

if this in seen:

continue

seen.add(this)

uniq.append((path, hunk))

errors = []

for path, hunk in uniq:

full_path = self.abs_root_path(path)

content = self.io.read_text(full_path)

original, _ = hunk_to_before_after(hunk)

try:

content = do_replace(full_path, content, hunk)

except SearchTextNotUnique:

errors.append(

not_unique_error.format(

path=path, original=original, num_lines=len(original.splitlines())

)

continue

if not content:

errors.append(

no_match_error.format(

path=path, original=original, num_lines=len(original.splitlines())

)

continue

# SUCCESS!

self.io.write_text(full_path, content)

if errors:

errors = "\n\n".join(errors)

if len(errors) < len(uniq):

errors += other_hunks_applied

raise ValueError(errors)

def do_replace(fname, content, hunk):

fname = Path(fname)

before_text, after_text = hunk_to_before_after(hunk)

# does it want to make a new file?

if not fname.exists() and not before_text.strip():

fname.touch()

content = ""

if content is None:

return

# TODO: handle inserting into new file

if not before_text.strip():

# append to existing file, or start a new file

new_content = content + after_text

return new_content

new_content = None

new_content = apply_hunk(content, hunk)

if new_content:

return new_content

def collapse_repeats(s):

return "".join(k for k, g in groupby(s))

def apply_hunk(content, hunk):

before_text, after_text = hunk_to_before_after(hunk)

res = directly_apply_hunk(content, hunk)

if res:

return res

hunk = make_new_lines_explicit(content, hunk)

# just consider space vs not-space

ops = "".join([line[0] for line in hunk])

ops = ops.replace("-", "x")

ops = ops.replace("+", "x")

ops = ops.replace("\n", " ")

cur_op = " "

section = []

sections = []

for i in range(len(ops)):

op = ops[i]

if op != cur_op:

sections.append(section)

section = []

cur_op = op

section.append(hunk[i])

sections.append(section)

if cur_op != " ":

sections.append([])

all_done = True

for i in range(2, len(sections), 2):

preceding_context = sections[i - 2]

changes = sections[i - 1]

following_context = sections[i]

res = apply_partial_hunk(content, preceding_context, changes, following_context)

if res:

content = res

else:

all_done = False

# FAILED!

# this_hunk = preceding_context + changes + following_context

break

if all_done:

return content

def flexi_just_search_and_replace(texts):

strategies = [

(search_and_replace, all_preprocs),

]

return flexible_search_and_replace(texts, strategies)

def make_new_lines_explicit(content, hunk):

before, after = hunk_to_before_after(hunk)

diff = diff_lines(before, content)

back_diff = []

for line in diff:

if line[0] == "+":

continue

# if line[0] == "-":

# line = "+" + line[1:]

back_diff.append(line)

new_before = directly_apply_hunk(before, back_diff)

if not new_before:

return hunk

if len(new_before.strip()) < 10:

return hunk

before = before.splitlines(keepends=True)

new_before = new_before.splitlines(keepends=True)

after = after.splitlines(keepends=True)

if len(new_before) < len(before) * 0.66:

return hunk

new_hunk = difflib.unified_diff(new_before, after, n=max(len(new_before), len(after)))

new_hunk = list(new_hunk)[3:]

return new_hunk

def cleanup_pure_whitespace_lines(lines):

res = [

line if line.strip() else line[-(len(line) - len(line.rstrip("\r\n")))] for line in lines

]

return res

def normalize_hunk(hunk):

before, after = hunk_to_before_after(hunk, lines=True)

before = cleanup_pure_whitespace_lines(before)

after = cleanup_pure_whitespace_lines(after)

diff = difflib.unified_diff(before, after, n=max(len(before), len(after)))

diff = list(diff)[3:]

return diff

def directly_apply_hunk(content, hunk):

before, after = hunk_to_before_after(hunk)

if not before:

return

before_lines, _ = hunk_to_before_after(hunk, lines=True)

before_lines = "".join([line.strip() for line in before_lines])

# Refuse to do a repeated search and replace on a tiny bit of non-whitespace context

if len(before_lines) < 10 and content.count(before) > 1:

return

try:

new_content = flexi_just_search_and_replace([before, after, content])

except SearchTextNotUnique:

new_content = None

return new_content

def apply_partial_hunk(content, preceding_context, changes, following_context):

len_prec = len(preceding_context)

len_foll = len(following_context)

use_all = len_prec + len_foll

# if there is a - in the hunk, we can go all the way to `use=0`

for drop in range(use_all + 1):

use = use_all - drop

for use_prec in range(len_prec, -1, -1):

if use_prec > use:

continue

use_foll = use - use_prec

if use_foll > len_foll:

continue

if use_prec:

this_prec = preceding_context[-use_prec:]

else:

this_prec = []

this_foll = following_context[:use_foll]

res = directly_apply_hunk(content, this_prec + changes + this_foll)

if res:

return res

def find_diffs(content):

# We can always fence with triple-quotes, because all the udiff content

# is prefixed with +/-/space.

if not content.endswith("\n"):

content = content + "\n"

lines = content.splitlines(keepends=True)

line_num = 0

edits = []

while line_num < len(lines):

line = lines[line_num]

if line.startswith("```diff"):

line_num, these_edits = process_fenced_block(lines, line_num + 1)

edits += these_edits

break

line_num += 1

# For now, just take 1!

# edits = edits[:1]

return edits

def process_fenced_block(lines, start_line_num):

for line_num in range(start_line_num, len(lines)):

line = lines[line_num]

if line.startswith("```"):

break

block = lines[start_line_num:line_num]

block.append("@@ @@")

if block[0].startswith("--- ") and block[1].startswith("+++ "):

# Extract the file path, considering that it might contain spaces

a_fname = block[0][4:].strip()

b_fname = block[1][4:].strip()

# Check if standard git diff prefixes are present (or /dev/null) and strip them

if (a_fname.startswith("a/") or a_fname == "/dev/null") and b_fname.startswith("b/"):

fname = b_fname[2:]

else:

# Otherwise, assume the path is as intended

fname = b_fname

block = block[2:]

else:

fname = None

edits = []

keeper = False

hunk = []

op = " "

for line in block:

hunk.append(line)

if len(line) < 2:

continue

if line.startswith("+++ ") and hunk[-2].startswith("--- "):

if hunk[-3] == "\n":

hunk = hunk[:-3]

else:

hunk = hunk[:-2]

edits.append((fname, hunk))

hunk = []

keeper = False

fname = line[4:].strip()

continue

op = line[0]

if op in "-+":

keeper = True

continue

if op != "@":

continue

if not keeper:

hunk = []

continue

hunk = hunk[:-1]

edits.append((fname, hunk))

hunk = []

keeper = False

return line_num + 1, edits

def hunk_to_before_after(hunk, lines=False):

before = []

after = []

op = " "

for line in hunk:

if len(line) < 2:

op = " "

line = line

else:

op = line[0]

line = line[1:]

if op == " ":

before.append(line)

after.append(line)

elif op == "-":

before.append(line)

elif op == "+":

after.append(line)

if lines:

return before, after

before = "".join(before)

after = "".join(after)

return before, after

cecli/cecli/coders/udiff_coder.py at main · cecli-dev/cecli · GitHub

Repository navigation

Expand file tree

Breadcrumbs

udiff_coder.py

Latest commit

History

Breadcrumbs

udiff_coder.py

File metadata and controls