Hide keyboard shortcuts

Hot-keys on this page

r m x p   toggle line displays

j k   next/prev highlighted chunk

0   (zero) top of page

1   (one) first highlighted chunk

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

# see https://pypi.org/project/setuptools-github 

# copy of setuptools_github.tools 

from __future__ import annotations 

 

import argparse 

import functools 

import json 

import logging 

import re 

from pathlib import Path 

from typing import Any 

 

import jinja2 

 

from . import code, common, fileos, scm, text 

 

log = logging.getLogger(__name__) 

 

 

class ToolsError(Exception): 

pass 

 

 

class ValidationError(ToolsError): 

pass 

 

 

class InvalidVersionError(ToolsError): 

pass 

 

 

class MissingVariableError(ToolsError): 

pass 

 

 

class _NO: 

pass 

 

 

class _Context(argparse.Namespace): 

def items(self): 

for name, value in self.__dict__.items(): 

if name.startswith("_"): 

continue 

yield (name, value) 

 

 

def list_of_paths(paths: str | Path | list[str | Path] | None) -> list[Path]: 

if not paths: 

return [] 

return [Path(s) for s in ([paths] if isinstance(paths, (str, Path)) else paths)] 

 

 

backup = functools.partial(fileos.backup, ext=common.BACKUP_SUFFIX) 

unbackup = functools.partial(fileos.unbackup, ext=common.BACKUP_SUFFIX) 

 

 

def get_option( 

config: dict[str, str], var: str, typ: Any = _NO, fallback: Any = _NO 

) -> Any: 

"""extract an option value from the hatch config dict 

 

Eg. 

get_options(self.config, "key") 

""" 

value = config.get(var, fallback) 

if value is _NO: 

raise ValidationError(f"cannot find variable '{var}' for plugin 'ci'") 

try: 

new_value = typ(value) if typ is not _NO else value 

except Exception as exc: 

raise ValidationError(f"cannot convert to {typ=} the {value=}") from exc 

return new_value 

 

 

def replace(txt: str, replacements: dict[str, str] | None = None) -> str: 

result = txt 

for src, dst in (replacements or {}).items(): 

if src.startswith("re:"): 

result = re.sub(src[3:], dst, result) 

else: 

result = result.replace(src, dst) 

return result 

 

 

def bump_version(version: str, mode: str) -> str: 

"""given a version str will bump it according to mode 

 

Arguments: 

version: text in the N.M.O form 

mode: major, minor or micro 

 

Returns: 

increased text 

 

>>> bump_version("1.0.3", "micro") 

"1.0.4" 

>>> bump_version("1.0.3", "minor") 

"1.1.0" 

""" 

newver = [int(n) for n in version.split(".")] 

if mode == "major": 

newver[-3] += 1 

newver[-2] = 0 

newver[-1] = 0 

elif mode == "minor": 

newver[-2] += 1 

newver[-1] = 0 

elif mode == "micro": 

newver[-1] += 1 

return ".".join(str(v) for v in newver) 

 

 

def _validate_gdata( 

gdata: dict[str, Any], abort: bool = True 

) -> tuple[set[str], set[str]]: 

keys = { 

"ref", 

"sha", 

"run_id", 

"run_number", 

} 

missing = keys - set(gdata) 

extra = set(gdata) - keys 

if abort and missing: 

raise ToolsError( 

f"missing keys from gdata '{','.join(missing)}'", missing, extra, gdata 

) 

return missing, extra 

 

 

def get_data( 

version_file: str | Path, 

github_dump: str | None = None, 

record_path: Path | None = None, 

abort: bool = True, 

) -> dict[str, str | None]: 

"""extracts version information from github_dump and updates version_file in-place 

 

This gather version, branch, commit and other info from: 

 

- version_file (eg. the __init__.py file containing __version__) 

- github_dump *a json encoded dictionary in GITHUB_DUMP 

- a record file (eg. _build.py) 

- the git checkout (using git) 

 

Args: 

version_file (str, Path): path to a file with a __version__ variable 

github_dump (str): the os.getenv("GITHUB_DUMP") value (a json structure) 

record: pull data from a _build.py file 

 

Returns: 

dict[str,str|None]: a dict with the current config 

dict[str,str|None]: a dict with the github dump data 

 

Example: 

for github data: 

{ 

"ref": "refs/heads/beta/0.3.10", 

"run_id": "5904313530", 

"run_number": "98", 

"sha": "507c657056d1a66520ec6b219a64706e70b0ff15", 

} 

for data: 

{ 

"branch": "beta/0.3.10", 

"build": "98", 

"current": "0.3.10", 

"ref": "refs/heads/beta/0.3.10", 

"runid": "5904313530", 

"sha": "507c657056d1a66520ec6b219a64706e70b0ff15", 

"version": "0.3.10b98", 

"workflow": "beta", 

} 

""" 

 

# we fill this structure getting data from these sources: 

# github_dump -> record_path (eg. _build.py) -> repo infos (eg. the git checkout) 

data = { 

"version": code.get_module_var(version_file, "__version__"), 

"current": code.get_module_var(version_file, "__version__"), 

"ref": None, 

"branch": None, 

"sha": None, 

"build": None, 

"runid": None, 

"workflow": None, 

} 

 

path = Path(version_file) 

repo = scm.lookup(path) 

record = record_path.exists() if record_path else None 

 

if not (repo or github_dump or record): 

if abort: 

raise scm.InvalidGitRepoError( 

f"cannot figure out settings (no repo in {path}, " 

f"a GITHUB_DUMP or a _build.py file)" 

) 

return data 

 

dirty = False 

if github_dump: 

gdata = json.loads(github_dump) if isinstance(github_dump, str) else github_dump 

elif record_path and record_path.exists(): 

mod = fileos.loadmod(record_path) 

gdata = { 

"ref": mod.ref, 

"sha": mod.sha, 

"run_number": mod.build, 

"run_id": mod.runid, 

} 

elif repo: 

gdata = { 

"ref": repo.head.name, 

"sha": repo.head.target.hex, 

"run_number": 0, 

"run_id": 0, 

} 

dirty = repo.dirty() 

else: 

raise RuntimeError( 

"Unable to retrieve data from GITHUB_DUMP or " "{record_path=} or {repo=}" 

) 

 

# make sure we have all keys 

_validate_gdata(gdata) 

 

expr = re.compile(r"/(?P<what>beta|release)/(?P<version>\d+([.]\d+)*)$") 

expr1 = re.compile(r"(?P<version>\d+([.]\d+)*)(?P<num>b\d+)?$") 

 

# update/fill the data values 

data["ref"] = gdata["ref"] 

data["sha"] = gdata["sha"] + ("*" if dirty else "") 

data["build"] = gdata["run_number"] # unless in GITHUB_DUMP these are likely None 

data["runid"] = gdata["run_id"] # unless in GITHUB_DUMP these are likely None 

 

data["branch"] = text.lstrip(gdata["ref"], ["refs/heads/", "refs/tags/"]) 

data["workflow"] = data["branch"] 

 

current = data["current"] 

if match := expr.search(gdata["ref"]): 

# setuptools double calls the update_version, 

# this fixes the issue 

match1 = expr1.search(current or "") 

if not match1: 

raise InvalidVersionError(f"cannot parse current version '{current}'") 

if match1.group("version") != match.group("version"): 

raise InvalidVersionError( 

f"building package for {current} from '{gdata['ref']}' " 

f"branch ({match.groupdict()} mismatch {match1.groupdict()})" 

) 

if match.group("what") == "beta": 

data["version"] = f"{match1.group('version')}b{gdata['run_number']}" 

data["workflow"] = "beta" 

else: 

data["workflow"] = "tags" 

 

assert len(data) == 8, "cannot change the data structure shape" # noqa: S101 

return data 

 

 

def update_version( 

version_file: str | Path, github_dump: str | None = None, abort: bool = True 

) -> str | None: 

"""extracts version information from github_dump and updates version_file in-place 

 

Args: 

version_file (str, Path): path to a file with a __version__ variable 

github_dump (str): the os.getenv("GITHUB_DUMP") value 

 

Returns: 

str: the new version for the package 

""" 

 

data = get_data(version_file, github_dump, abort=abort) 

code.set_module_var(version_file, "__version__", data["version"]) 

code.set_module_var(version_file, "__hash__", (data["sha"] or "")[:7]) 

return data["version"] 

 

 

def get_environment( 

version_file: str | Path, 

github_dump: str | None = None, 

record_path: Path | None = None, 

abort: bool = True, 

) -> jinja2.Environment: 

"""returns a context object""" 

 

from urllib.parse import quote 

 

env = jinja2.Environment(autoescape=True) 

env.filters["urlquote"] = functools.partial(quote, safe="") 

env.globals = { 

"dir": dir, 

"len": len, 

"sorted": sorted, 

"reversed": reversed, 

} 

data = get_data(version_file, github_dump, record_path, abort) 

env.globals["ctx"] = _Context(**data) 

return env 

 

 

def generate_build_record(record_path: Path, data: dict[str, Any]) -> Path: 

record_path.parent.mkdir(parents=True, exist_ok=True) 

with record_path.open("w") as fp: 

print("# autogenerate build file", file=fp) 

for key, value in sorted((data or {}).items()): 

value = f"'{value}'" if isinstance(value, str) else value 

print(f"{key} = {value}", file=fp) 

return record_path 

 

 

# def process( 

# version_file: str | Path, 

# github_dump: str | None = None, 

# record: str | Path = "_build.py", 

# paths: str | Path | list[str | Path] | None = None, 

# fixers: dict[str, str] | None = None, 

# backup: Callable[[Path | str], None] | None = None, 

# abort: bool = True, 

# ) -> dict[str, str | None]: 

# """get version from github_dump and updates version_file/paths 

# 

# Args: 

# version_file (str, Path): path to a file with __version__ variable 

# github_dump (str): the os.getenv("GITHUB_DUMP") value 

# paths (str, Path): path(s) to files jinja2 processeable 

# fixers (dict[str,str]): fixer dictionary 

# record: set to True will generate a _build.py sibling of version_file 

# 

# Returns: 

# str: the new version for the package 

# 

# Example: 

# {'branch': 'beta/0.3.1', 

# 'build': 0, 

# 'current': '0.3.1', 

# 'hash': 'c9e484a*', 

# 'version': '0.3.1b0', 

# 'runid': 0 

# } 

# """ 

# from argparse import Namespace 

# from functools import partial 

# from urllib.parse import quote 

# 

# from jinja2 import Environment 

# 

# class Context(Namespace): 

# def items(self): 

# for name, value in self.__dict__.items(): 

# if name.startswith("_"): 

# continue 

# yield (name, value) 

# 

# record_path = (Path(version_file).parent / record).absolute() if record else None 

# data = get_data(version_file, github_dump, record_path, abort) 

# 

# if not backup: 

# 

# def backup(_: Path | str) -> None: 

# pass 

# 

# breakpoint() 

# # backup(version_file) 

# 

# # set_module_var(version_file, "__version__", data["version"]) 

# # set_module_var(version_file, "__hash__", (data["sha"] or "")[:7]) 

# 

# # env = Environment(autoescape=True) 

# # env.filters["urlquote"] = partial(quote, safe="") 

# # for path in list_of_paths(paths): 

# # txt = replace(path.read_text(), fixers) 

# # tmpl = env.from_string(txt) 

# # backup(path) 

# # path.write_text(tmpl.render(ctx=Context(**data))) 

# # 

# # if record_path: 

# # generate_build_record(record_path, data) 

# 

# return data 

# 

#