Coverage for nlp_manager/regex

1"""

2crate_anon/nlp_manager/regex_units.py

4===============================================================================

7 Created by Rudolf Cardinal (rnc1001@cam.ac.uk).

9 This file is part of CRATE.

11 CRATE is free software: you can redistribute it and/or modify

12 it under the terms of the GNU General Public License as published by

13 the Free Software Foundation, either version 3 of the License, or

14 (at your option) any later version.

16 CRATE is distributed in the hope that it will be useful,

17 but WITHOUT ANY WARRANTY; without even the implied warranty of

18 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the

19 GNU General Public License for more details.

21 You should have received a copy of the GNU General Public License

22 along with CRATE. If not, see <https://www.gnu.org/licenses/>.

24===============================================================================

26**Regular expressions to detect physical units.**

28"""

30from typing import List, Optional, Tuple

32from crate_anon.nlp_manager.regex_numbers import (

33 BILLION,

34 MULTIPLY_OR_SPACE,

35 PLAIN_INTEGER,

36 POWER,

37 TRILLION,

38)

41# =============================================================================

42# Physical units

43# =============================================================================

45OUT_OF_SEPARATOR = r"(?: \/ | \b out \s+ of \b )"

48def per(

49 numerator: str,

50 denominator: str,

51 include_power_minus1: bool = True,

52 numerator_optional: bool = False,

53) -> str:

54 """

55 Returns regex text representing "X per Y"; e.g. "millimoles per litre",

56 "cells per cubic millimetre".

58 Args:

59 numerator: regex representing the numerator

60 denominator: regex representing the denominator

61 include_power_minus1: include the "n d -1" format for "n/d"

62 numerator_optional: presence of the numerator is optional

63 """

64 if numerator:

65 if numerator_optional:

66 # ensure that the optional whitespace is captured as part of the

67 # "optional" bit, so there is no leftover whitespace that can

68 # remain

69 numerator_part = rf"(?: {numerator} \s* )?"

70 else:

71 # numerator, optional whitespace

72 numerator_part = rf"{numerator} \s*"

73 # Use of "\s* \b" rather than "\s+" is so we can have a BLANK

74 # numerator.

75 else:

76 # Blank numerator

77 numerator_part = ""

78 options = [

79 rf"{numerator_part} (?: \/ | \b per \b) \s* {denominator}",

80 ]

81 if include_power_minus1:

82 options.append(rf"{numerator_part} \b {denominator} \s* -1")

83 return r"(?: {} )".format(" | ".join(options))

86def _out_of_str(n_as_regex: str) -> str:

87 """

88 Returns regex text representing "out of N".

90 Args:

91 n_as_regex: the "N", as a regular expression

92 """

93 # / n

94 # out of n

95 return rf"(?: {OUT_OF_SEPARATOR} \s* {n_as_regex} \b)"

98def out_of(n: int) -> str:

99 """

100 Returns regex text representing "out of N".

101

102 Args:

103 n: the number N

104 """

105 return _out_of_str(str(n))

106

107

108def out_of_anything() -> str:

109 """

110 Returns:

111 regex representing "out of N" where N is any number

112 """

113 return _out_of_str(PLAIN_INTEGER)

114

115

116def power(x: str, n: int, allow_no_operator: bool = False) -> str:

117 """

118 Returns regex text representing "x to the power n".

119

120 Args:

121 x: base

122 n: exponent

123 allow_no_operator: make the operator (like ``^`` or ``**``) optional?

124 """

125 return r"(?: {x} \s* {power}{optional} \s* {n})".format(

126 x=x,

127 power=POWER,

128 optional="?" if allow_no_operator else "",

129 n=n,

130 )

131

132

133def units_times(*args: str) -> str:

134 """

135 Returns regular expression text combining all its inputs with optional

136 multiplication.

137

138 For units, where they are notionally multiplied.

139 """

140 multiply = MULTIPLY_OR_SPACE + "?"

141 joined = multiply.join(args)

142 return rf"(?: {joined} )"

143

144

145def units_by_dimension(

146 *args: Tuple[str, int], # specify type of *one* arg!

147 allow_no_operator: bool = False,

148) -> str:

149 """

150 Returns regex text for a unit where we specify them by their dimensions.

151

152 Args:

153 *args: each is a tuple ``unit, power``

154 allow_no_operator: make the operator (like ``^`` or ``**``) optional?

155 """

156 multiply = " " + MULTIPLY_OR_SPACE + " "

157 power_elements = [] # type: List[str]

158 for i, unit_exponent in enumerate(args):

159 unit, exponent = unit_exponent

160 assert exponent != 0

161 power_elements.append(

162 power(unit, exponent, allow_no_operator=allow_no_operator)

163 )

164 joined_power_elements = multiply.join(power_elements)

165 power_style = rf"(?: {joined_power_elements} )"

166 options = [power_style]

167 # noinspection PyChainedComparisons

168 if len(args) == 2 and args[0][1] > 0 and args[1][1] < 0:

169 # x per y

170 options.append(per(args[0][0], args[1][0], include_power_minus1=False))

171 return r"(?: {} )".format(r" | ".join(options))

172

173

174# -----------------------------------------------------------------------------

175# Distance

176# -----------------------------------------------------------------------------

177

178M = r"(?: met(?:re|er)s? | m )" # m, metre(s), meter(s)

179CM = r"(?: cm | centimet(?:re|er)s? )" # cm, centimetre(s), centimeter(s)

180MM = r"(?: mm | millimet(?:re|er)s? )" # mm, millimetre(s), millimeter(s)

181

182FEET = r"""(?: f(?:ee|oo)?t | \' | ’ | ′ )"""

183# ... feet, foot, ft

184# ... apostrophe, right single quote (U+2019), prime (U+2032)

185INCHES = r"""(?: in(?:ch(?:e)?)?s? | \" | ” | ″)"""

186# ... in, ins, inch, inches, [inchs = typo but clear]

187# ... ", right double quote (U+2014), double prime (U+2033)

188

189# -----------------------------------------------------------------------------

190# Mass

191# -----------------------------------------------------------------------------

192

193MCG = r"(?: mcg | microgram(?:me)?s? | [μu]g )" # you won't stop people using ug... # noqa: E501

194MG = r"(?: mg | milligram(?:me)?s? )" # mg, milligram, milligrams, milligramme, milligrammes # noqa: E501

195G = r"(?: gram(?:me)?s? | g )" # g, gram, grams, gramme, grammes

196KG = r"(?: kgs? | kilo(?:gram(?:me)?)?s? )" # kg, kgs, kilos ... kilogrammes etc. # noqa: E501

197LB = r"(?: pounds? | lbs? )" # pound(s), lb(s)

198STONES = r"(?: stones? | st\.? )" # stone(s), st, st.

199

200# -----------------------------------------------------------------------------

201# Volume

202# -----------------------------------------------------------------------------

203

204L = r"(?: lit(?:re|er)s? | L )" # L, litre(s), liter(s)

205DL = rf"(?: d(?:eci)?{L} )" # 10^-1

206ML = rf"(?: m(?:illi)?{L} )" # 10^-3

207MICROLITRE = rf"(?: micro{L} | [μu]L )" # 10^-6: microL, microliter(s), microlitre(s), μL, uL # noqa: E501

208NANOLITRE = rf"(?: nano{L} | nL )" # 10^-9: nanoL, nanoliter(s), nanolitre(s), nL # noqa: E501

209PICOLITRE = rf"(?: pico{L} | pL )" # 10^-12: picoL, picoliter(s), picolitre(s), pL # noqa: E501

210FEMTOLITRE = rf"(?: femto{L} | fL )" # 10^-15: femtoL, femtoliter(s), femtolitre(s), fL # noqa: E501

211# CUBIC_MM = r"""(?: (?:\b cubic \s+ {mm}) | {mm_cubed} )""".format(

212CUBIC_MM = (

213 r"""(?: (?:\b cubic \s+ {mm}) | {mm_cubed} | (?: \b cmm \b ) )"""

214).format(mm=MM, mm_cubed=power(MM, 3, allow_no_operator=True))

215# cubic mm, etc. | mm^3, mm3, mm 3, etc. | cmm

216# "cmm" added 2018-09-07 having seen this in the wild (albeit urinary results).

217

218# A microlitre is of course the same as a cubic millimetre:

219CUBIC_MM_OR_MICROLITRE = rf"(?: {MICROLITRE} | {CUBIC_MM} )"

220

221# -----------------------------------------------------------------------------

222# Inverse (reciprocal) volume

223# -----------------------------------------------------------------------------

224

225PER_CUBIC_MM = per("", CUBIC_MM, numerator_optional=True)

226

227# -----------------------------------------------------------------------------

228# Time

229# -----------------------------------------------------------------------------

230

231HOUR = r"(?: \b h(?:rs?|ours?)? \b)" # h, hr, hrs, hour, hours

232DAY = r"(?: \b d(?:y?|ay?)? \b )" # d, dy, day

233WEEK = r"(?: \b w(?:k?|eek?)? \b)" # w, wk, week

234MONTH = r"(?:\b month \b)" # month

235YEAR = r"(?:\b y(?:(?:ea)?r)? \b)" # y, yr, year

236

237DAYS_PER_WEEK = 7

238

239# The mean month (across a normal 4-year cycle ignoring century non-leap years)

240# is 30.4375 days:

241# n <- c(28, rep(30, 4), rep(31, 7)) # mean 30.41667

242# l <- c(29, rep(30, 4), rep(31, 7)) # mean 30.5

243# fouryearcycle <- c(n, n, n, l) # mean 30.4375

244# century <- c(rep(n, 76), rep(l, 24)) # mean 30.43667

245# mean(n) / 7 # 4.345238

246# mean(fouryearcycle) / 7 # 4.348214

247# mean(century) / 7 # 4.348095

248# ... the Google answer for weeks per month is 4.34524, i.e. a normal year.

249# But let's not be spuriouly precise:

250WEEKS_PER_MONTH_APPROX = 4.35

251WEEKS_PER_YEAR_APPROX = 52

252

253# -----------------------------------------------------------------------------

254# Proportions

255# -----------------------------------------------------------------------------

256

257PERCENT = r"""(?:%|pe?r?\s?ce?n?t)"""

258# "%" or some subset of "percent" -- for the latter, must have "pct", other

259# characters optional

260

261# -----------------------------------------------------------------------------

262# Arbitrary count things

263# -----------------------------------------------------------------------------

264

265CELLS = r"(?:\b cells? \b)"

266

267UNITS = r"(?: (?:I\.?)? U(?:nits?|\.)? )" # U, IU, I.U., unit, units...

268# (IU for international units)

269MICROUNITS = rf"(?: (?:micro|μ|u) {UNITS} )"

270MILLIUNITS = rf"(?: m(?:illi)? {UNITS} )"

271

272UK = r"(?: U(?:nited\s+|\.\s*)? K(?:ingdom|\.)? )"

273ALCOHOL = r"(?: \b(?:alcohol|ethanol|EtOH)\b )"

274UK_ALCOHOL_UNITS = rf"(?: (?: {UK} \s+)? ({ALCOHOL} \s+)? {UNITS} )"

275# U, unit, units, UK units, UK alcohol units...

276# I thought not "IU" as they are not international units; however, RS used that

277# term, so whether correct or in error, that's sufficient for me to include it!

278UK_ALCOHOL_UNITS_PER_DAY = per(

279 UK_ALCOHOL_UNITS, DAY, include_power_minus1=False

280)

281UK_ALCOHOL_UNITS_PER_WEEK = per(

282 UK_ALCOHOL_UNITS, WEEK, include_power_minus1=False

283)

284UK_ALCOHOL_UNITS_PER_MONTH = per(

285 UK_ALCOHOL_UNITS, MONTH, include_power_minus1=False

286)

287UK_ALCOHOL_UNITS_PER_YEAR = per(

288 UK_ALCOHOL_UNITS, YEAR, include_power_minus1=False

289)

290

291SCORE = r"(?:scored?)" # score(d)

292

293# -----------------------------------------------------------------------------

294# Moles

295# -----------------------------------------------------------------------------

296

297MOLES = r"(?:\b mole?s? \b)" # mol, mole, mols, moles

298MICROMOLES = r"(?: (?:micro|μ|u)mole?s? )"

299MILLIMOLES = r"(?: m(?:illi)?mole?s? )"

300

301MICROEQ = r"(?: (?:micro|μ|u)Eq )"

302MILLIEQ = r"(?: m(?:illi)?Eq )"

303

304# -----------------------------------------------------------------------------

305# Concentration (molarity)

306# -----------------------------------------------------------------------------

307

308MICROMOLAR = r"(?:[μu]M | micromolar)"

309MILLIMOLAR = r"(?:mM)" # NB case-insensitive... confusable with millimetres

310

311MICROEQ_PER_L = per(MICROEQ, L)

312MICROMOLES_PER_L = per(MICROMOLES, L)

313MILLIEQ_PER_L = per(MILLIEQ, L)

314MILLIMOLES_PER_L = per(MILLIMOLES, L)

315

316# -----------------------------------------------------------------------------

317# Concentration (mass)

318# -----------------------------------------------------------------------------

319

320G_PER_DL = per(G, DL)

321G_PER_L = per(G, L)

322L_PER_L = per(L, L)

323MG_PER_DL = per(MG, DL)

324MG_PER_L = per(MG, L)

325

326# -----------------------------------------------------------------------------

327# Concentration (arbitrary count and dimensionless things)

328# -----------------------------------------------------------------------------

329

330BILLION_PER_L = per(BILLION, L)

331TRILLION_PER_L = per(TRILLION, L)

332

333CELLS_PER_CUBIC_MM = per(CELLS, CUBIC_MM, numerator_optional=True)

334CELLS_PER_CUBIC_MM_OR_MICROLITRE = per(

335 CELLS, CUBIC_MM_OR_MICROLITRE, numerator_optional=True

336)

337

338MICROUNITS_PER_ML = per(MICROUNITS, ML)

339MILLIUNITS_PER_L = per(MILLIUNITS, L)

340UNITS_PER_L = per(UNITS, L)

341

342MILLIMOLES_PER_MOL = per(MILLIMOLES, MOLES)

343

344# -----------------------------------------------------------------------------

345# Speed

346# -----------------------------------------------------------------------------

347

348MM_PER_H = per(MM, HOUR)

349

350# -----------------------------------------------------------------------------

351# Pressure

352# -----------------------------------------------------------------------------

353

354MM_HG = r"(?: mm \s* Hg )" # mmHg, mm Hg

355# ... likelihood of "millimetres of mercury" quite small?

356

357# -----------------------------------------------------------------------------

358# Area and related

359# -----------------------------------------------------------------------------

360

361SQ_M = r"""

362 (?: # square metres

363 (?: sq(?:uare)? \s+ {m} ) # sq m, square metres, etc.

364 | (?: {m} \s+ sq(?:uared?)? ) # m sq, metres square(d), etc.

365 | {m_sq} # m ^ 2, etc.

366 )

367""".format(

368 m=M, m_sq=power(M, 2)

369)

370

371# BMI

372KG_PER_SQ_M = r"(?: {kg_per_sqm} | {kg_sqm_pow_minus2} )".format(

373 kg_per_sqm=per(KG, SQ_M, include_power_minus1=False),

374 kg_sqm_pow_minus2=units_times(KG, power(M, -2)),

375)

376

377

378# =============================================================================

379# Generic conversion functions

380# =============================================================================

381

382

383def kg_from_st_lb_oz(

384 stones: float = 0, pounds: float = 0, ounces: float = 0

385) -> Optional[float]:

386 """

387 Convert Imperial to metric mass.

388

389 Returns:

390 mass in kg

391

392 """

393 # 16 ounces in a pound

394 # 14 pounds in a stone

395 # 1 avoirdupois pound = 0.45359237 kg

396 # https://en.wikipedia.org/wiki/Pound_(mass)

397 # Have you the peas? "Goods of weight"; aveir de peis (OFr.; see OED).

398 try:

399 total_pounds = (stones * 14) + pounds + (ounces / 16)

400 return 0.45359237 * total_pounds

401 except (TypeError, ValueError):

402 return None

403

404

405def m_from_ft_in(feet: float = 0, inches: float = 0) -> Optional[float]:

406 """

407 Converts Imperial to metric length.

408

409 Returns:

410 length in m

411

412 """

413 # 12 inches in a foot

414 # 1 inch = 25.4 mm

415 try:

416 total_inches = (feet * 12) + inches

417 return total_inches * 25.4 / 1000

418 except (TypeError, ValueError):

419 return None

420

421

422def m_from_m_cm(metres: float = 0, centimetres: float = 0) -> Optional[float]:

423 """

424 Converts metres/centimetres to metres.

425 """

426 try:

427 return metres + (centimetres / 100)

428 except (TypeError, ValueError):

429 return None

430

431

432def assemble_units(components: List[Optional[str]]) -> str:

433 """

434 Takes e.g. ``["ft", "in"]`` and makes ``"ft in"``.

435 """

436 active_components = [c for c in components if c]

437 return " ".join(active_components)

438

439

440def factor_millimolar_from_mg_per_dl(molecular_mass_g_per_mol: float) -> float:

441 """

442 Returns the conversion factor that you should multiple a "mg/dL" number by

443 to get a "mM" (mmol/L) number.

444

445 Principle:

446

447 .. code-block:: none

448

449 mmol_per_L

450 = 0.001 * mol_per_L

451 = 0.001 * (g_per_L / g_per_mol)

452 = 0.001 * ((10 * g_per_dL) / g_per_mol)

453 = 0.001 * ((10 * 1000 * mg_per_dL) / g_per_mol)

454 = (0.001 * 10 * 1000 / g_per_mol) * mg_per_dL

455 = (10 / g_per_mol) * mg_per_dl

456

457 Example:

458 glucose, molecular mass 180.156 g/mol

459 => conversion factor is (10 / 180.156)

460 90 mg/dL -> (10 / 180.156) * 90 mM = 5.0 mM

461

462 Args:

463 molecular_mass_g_per_mol: molecular mass in g/mol

464

465 Returns:

466 conversion factor

467

468 """

469 return 10 / molecular_mass_g_per_mol

470

471

472def factor_micromolar_from_mg_per_dl(molecular_mass_g_per_mol: float) -> float:

473 """

474 Returns the conversion factor that you should multiple a "mg/dL" number by

475 to get a "μM" (μmol/L) number.

476

477 Args:

478 molecular_mass_g_per_mol: molecular mass in g/mol

479

480 Returns:

481 conversion factor

482

483 """

484 return 1000 * factor_millimolar_from_mg_per_dl(molecular_mass_g_per_mol)

485

486

487def millimolar_from_mg_per_dl(

488 mg_per_dl: float, molecular_mass_g_per_mol: float

489) -> float:

490 """

491 Converts a concentration from mg/dL to mM (mmol/L).

492

493 Args:

494 mg_per_dl: value in mg/dL

495 molecular_mass_g_per_mol: molecular mass in g/mol

496

497 Returns:

498 value in mM = mmol/L

499

500 """

501 return mg_per_dl * factor_millimolar_from_mg_per_dl(

502 molecular_mass_g_per_mol

503 )

504

505

506def micromolar_from_mg_per_dl(

507 mg_per_dl: float, molecular_mass_g_per_mol: float

508) -> float:

509 """

510 Converts a concentration from mg/dL to μM (μmol/L).

511

512 Args:

513 mg_per_dl: value in mg/dL

514 molecular_mass_g_per_mol: molecular mass in g/mol

515

516 Returns:

517 value in μM = μmol/L

518

519 """

520 return mg_per_dl * factor_micromolar_from_mg_per_dl(

521 molecular_mass_g_per_mol

522 )

Coverage for nlp_manager/regex_units.py: 85%

136 statements