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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594
|
"""
Compare 2 output files from ABINIT line by line with arithmetic
comparison of floating point substrings
The first character of each line in both files indicates the mode of
comparison; these first characters MUST be identical in corresponding lines.
By default, a floating point comparison with a tolerance of 1.01d-10 is done on
all floating point strings and a character comparison is done on all other
strings, disregarding multiple spaces. For compatibility reasons with
historical fldiff.pl a float is must contains a dot and at least one digit
after it to be recognized as float. In order for two numbers to be declared
different, BOTH the absolute difference and the relative difference (difference
divided by the sum of absolute values) must be bigger than the tolerance.
Some special characters at the beginning of lines require a different handling:
- mark lines as same regardless to their content (i.e. ignore lines)
(can be be present in the 2 files or not, but must begin with -)
_ mark lines as same regardless to their content
(must be present in the 2 files, but can begin with _ in only one of them)
+ mark lines as different regardless to their content
, handle as + if ignore option is False and as - else
P handle as + if ignoreP option is False and as - else
% floating point comparisons are done with a tolerance of 1.01e-2
; floating point comparisons are done irrespective of signs
: ignore floating point numbers and do a characters comparison
. do a characters comparison, but do not count this line in the Summary
) floating point comparison is done only for the FIRST numeric field
Both files should have the same number of non - starting lines.
The ignore options affects the treatment of the ',' special character in the
first column (see above)
The ignoreP options affects the treatment of the 'P' special character in the
first column (see above)
The label option, if specified, is appended at the end of the summary.
The tolerance option set the tolerance for comparison of floats, the default
is 1.01e-10. This modifications do not apply to the tolerance determined by
the '%',and '.' first-column special signs.
"""
from __future__ import print_function, division, unicode_literals
import re
from math import floor
from threading import Thread
from .data_extractor import DataExtractor
from .yaml_tools import is_available as has_yaml
if has_yaml:
from .yaml_tools.driver_test_conf import DriverTestConf as YDriverConf
from .yaml_tools.tester import Tester as YTester, Failure as YFailure
# Match floats. Minimal float is .0 for historical reasons.
# As a consequence, integers will be compared as strings
float_re = re.compile(r'([+-]?[0-9]*\.[0-9]+(?:[eEdDfF][+-]?[0-9]+)?)')
def norm_spaces(s):
r"""Normalize all blanks ( \\n\\r\\t)."""
# the join/split technique remove all blanks and put one space between non-blanks words
return ' '.join(s.split())
def relative_truncate(f, n):
"""
>>> rel_truncate(1.8367387367, 2)
1.83
>>> rel_truncate(1.8367387367e-5, 5)
1.83673e-05
>>> rel_truncate(1.8367387367e+7, 4)
18367000.0
"""
ten_n = 10.0**n
if f == 0.0:
return 0.0
elif abs(f) >= 1:
ten_p = 10.0
while abs(f) > ten_p:
ten_p *= 10.0
fact = 10 * ten_n / ten_p
return floor(f * fact) / fact
else:
ten_p = 0.1
p = -1
while abs(f) < ten_p:
ten_p *= 0.1
p -= 1
fact = ten_n / ten_p
return floor(f * fact) / fact
class NotDriverConf(object):
def __init__(self, has_yaml):
self.has_yaml = has_yaml
def extra_info(self):
if self.has_yaml:
return ('# YAML support is available, but is disabled for this test.',)
else:
return ('# YAML support is not available, YAML based tests will be ignored.',)
class LineDifference(object):
"""Base class representing a difference."""
def __init__(self, p1, p2, l1, l2):
self.lines = (p1 + 1, p2 + 1)
if l1 == '' or l1[-1] not in '\n\r':
l1 += '\n'
self.content = (l1, l2)
def __eq__(self, other):
"""Implement the == test."""
return self.lines == other.lines and self.content == other.content
def __ne__(self, other):
return not (self == other)
def __repr__(self):
"""Default representation of difference inspired by gnu diff tool."""
return (
'{}\n'.format(*self.lines)
+ '< ' + self.content[0]
+ '> ' + self.content[1]
)
class LineCountDifference(LineDifference):
"""Represents a difference between line counts."""
def __init__(self, more, less, line_count=(0, 0)):
"""
Args:
more: the name of the file with more lines
less: the name of the file with less lines
"""
LineDifference.__init__(self, 0, 0, '', '')
self.more = more
self.less = less
self.line_count = line_count
def __repr__(self):
if self.line_count != (0, 0):
return ('{} has more significant lines than {} ({} > {}).\n'
.format(self.more, self.less, *self.line_count))
return '{} has more significant lines than {}.\n'.format(self.more, self.less)
class MetaCharDifference(LineDifference):
"""Represent a difference between two lines with different meta characters."""
def __init__(self, p1, p2, m1, m2):
LineDifference.__init__(self, p1, p2, '', '')
self.metas = (m1, m2)
def __repr__(self):
return ('At line {} (in file 1), line {} (in file 2), different'
' leading characters: `{}` and `{}`.\n').format(*(self.lines + self.metas))
class FloatDifference(LineDifference):
"""Represent a difference between floating point values."""
def __init__(self, p1, p2, line1, line2, abs_err, rel_err):
LineDifference.__init__(self, p1, p2, line1, line2)
self.abs_err = abs_err
self.rel_err = rel_err
class TextDifference(LineDifference):
"""Represent a difference between text parts of a lines."""
def __init__(self, p1, p2, line1, line2, silent=False):
LineDifference.__init__(self, p1, p2, line1, line2)
self.silent = silent
class ForcedDifference(LineDifference):
"""A difference that is arbitrarly declared."""
class Result(object):
"""Analyse and summarize the set of differences found by a diff."""
def __init__(self, fl_diff, yaml_diff, extra_info=[], label=None, verbose=False):
"""
differences is expected to be a list of Difference instances
"""
self.fl_diff = fl_diff
self.yaml_diff = yaml_diff
self.extra_info = extra_info
self.fatal_error = False
self.yaml_error = False
self.success = True
self.max_abs_err = 0.0
self.max_rel_err = 0.0
self.max_abs_ln = 0
self.max_rel_ln = 0
self.ndiff_lines = 0
self.label = label
self.verbose = verbose
self.details = self._analyse()
def _analyse(self):
"""
Analyse a difference list and extract summary information and
details. Summary information is
- self.max_abs_err: maximum absolute difference
- self.max_rel_err: maximum relative difference
- self.max_abs_ln: line number where the maximum absolute
difference is reached for the first time
- self.max_rel_ln: line number where the maximum relative
difference is reached for the first time
- self.ndiff_lines: number of lines flagged as different (excluding
"silent" differences: line starting with '.' '+' and depending of Diff options ',' and 'P')
"""
details = []
error_lines = set()
if self.yaml_diff:
details.append('# Start YAML based comparison report\n')
for diff in self.yaml_diff:
if diff.is_fail():
self.success = False
self.yaml_error = True
details.append(repr(diff) + '\n\n')
elif self.verbose:
details.append(repr(diff) + '\n\n')
if self.fl_diff:
details.append('# Start legacy fldiff comparison report\n')
for diff in self.fl_diff:
if isinstance(diff, LineCountDifference) \
or isinstance(diff, MetaCharDifference):
self.fatal_error = True
self.success = False
details = str(diff)
elif isinstance(diff, ForcedDifference):
pass # Silent differences: not counted as different line
elif isinstance(diff, FloatDifference):
if diff.lines[0] not in error_lines:
self.ndiff_lines += 1
if diff.abs_err > self.max_abs_err:
self.max_abs_err = diff.abs_err
self.max_abs_ln = diff.lines[0]
if diff.rel_err > self.max_rel_err:
self.max_rel_err = diff.rel_err
self.max_rel_ln = diff.lines[0]
self.success = False
elif isinstance(diff, TextDifference):
if not diff.silent:
if diff.lines[0] not in error_lines:
self.ndiff_lines += 1
self.success = False
else: # any other Difference
assert isinstance(diff, LineDifference), 'Unknown type of Difference.'
if diff.lines[0] not in error_lines:
self.ndiff_lines += 1
self.success = False
if self.fatal_error:
break # stop analysing in case of fatal error
if diff.lines[0] not in error_lines:
details.append(str(diff))
error_lines.add(diff.lines[0])
return details
def get_summary(self):
"""Return a textual summary of the diff."""
if self.yaml_error:
summary = 'yaml_test errors.'
elif self.fatal_error:
summary = 'fldiff fatal error.'
elif self.success:
summary = 'no significant difference has been found.'
else:
summary = ('different lines={}, max abs_diff={:.3e} (l.{}),'
' max rel_diff={:.3e} (l.{}).').format(
self.ndiff_lines,
self.max_abs_err,
self.max_abs_ln,
self.max_rel_err,
self.max_rel_ln
)
if self.label is not None:
summary = 'Summary ' + self.label + ': ' + summary
else:
summary = 'Summary: ' + summary
return summary
def dump_details(self, file=None):
"""
Either return a string describing all detected differences
or write it into the given file (expected to be a writable stream).
"""
if file is None:
return ('\n'.join(self.extra_info) + '\n' + ''.join(self.details)
+ self.get_summary())
else:
file.write('\n'.join(self.extra_info) + '\n')
file.writelines(self.details)
file.write(self.get_summary() + '\n')
return None
def passed_within_tols(self, tolnlines, tolabs, tolrel):
"""
Check the result of the diff against the given tolerances.
"""
if self.yaml_error:
status = 'failed'
for diff in self.yaml_diff:
if diff.is_fail():
first_fail = diff
break
msg = 'yaml_test errors. First is:\n{}\n'.format(first_fail)
elif self.fatal_error:
status = 'failed'
msg = 'fldiff fatal error:\n' + self.details
elif self.success:
status = 'succeeded'
msg = 'succeeded'
else:
# truncate to prevent fldiff from printing 1.000 < 1.000
# compatibility fix, this may be removed later
abs_error = relative_truncate(self.max_abs_err, 3)
rel_error = relative_truncate(self.max_rel_err, 3)
ndiff_lines = self.ndiff_lines
status = 'failed'
fact = 1.0
locs = locals()
if ndiff_lines > tolnlines:
msg = 'failed: erroneous lines {ndiff_lines} > {tolnlines}'
elif abs_error > tolabs * fact and rel_error < tolrel:
msg = 'failed: abs error {abs_error:.4} > {tolabs}'
elif rel_error > tolrel * fact and abs_error < tolabs:
msg = 'failed: rel error {rel_error:.4} > {tolrel}'
elif abs_error > tolabs * fact and rel_error > tolrel * fact:
msg = ('failed: abs error {abs_error:.4} > {tolabs},'
' rel error {rel_error:.4} > {tolrel}')
else:
status = 'passed'
msg = ('passed: abs error {abs_error:.4} < {tolabs},'
' rel error {rel_error:.4} < {tolrel}')
msg = msg.format(**locs)
isok = status in ('succeeded', 'passed')
return isok, status, msg
def has_line_count_error(self):
return any(isinstance(diff, LineCountDifference)
for diff in self.fl_diff)
class Differ(object):
def __init__(self, yaml_test=None, **options):
"""
Init a differ with some parameters passed via options.
Known parameters are:
- ignore: bool (default True)
- ignoreP: bool (default True)
- tolerance: float (tolerance for both relative and absolute
difference)
- tolerance_abs: float (default 1.01e-10)
- tolerance_rel: float (default 1.01e-10)
- label: str (default None)
- use_yaml: bool (default True)
- use_fl: bool (default True)
- verbose: bool (default False) enable report of successful yaml tests too
"""
self.xml_mode = False # this is the first dirty fix.
self.options = {
'ignore': True,
'ignoreP': True,
'tolerance_abs': 1.01e-10,
'tolerance_rel': 1.01e-10,
'label': None,
'use_fl': True,
'use_yaml': False,
'verbose': False,
'debug': False
}
self.options.update(options)
if 'tolerance' in options:
self.options['tolerance_abs'] = options['tolerance']
self.options['tolerance_rel'] = options['tolerance']
self.use_fl = self.options['use_fl']
self.use_yaml = has_yaml and self.options['use_yaml']
if self.use_yaml:
if yaml_test and 'file' in yaml_test and yaml_test['file']:
self.yaml_conf = YDriverConf.from_file(yaml_test['file'])
elif yaml_test and 'yaml' in yaml_test and yaml_test['yaml']:
self.yaml_conf = YDriverConf(yaml_test['yaml'])
else:
self.yaml_conf = YDriverConf()
self.yaml_conf.debug = self.options['debug']
else:
self.yaml_conf = NotDriverConf(has_yaml)
def diff(self, file1, file2):
"""
Compute the diff of file 1 (reference) and file 2 (out)
and return a Result instance.
"""
if file1.endswith('.xml'):
self.xml_mode = True
with open(file1, 'rt') as f1, open(file2, 'rt') as f2:
line_diff, doc_diff = self._diff_lines(f1, f2)
return Result(line_diff, doc_diff,
extra_info=self.yaml_conf.extra_info(),
label=self.options['label'],
verbose=self.options['verbose'])
def _diff_lines(self, src1, src2):
lines = [None, None]
documents = [None, None]
corrupted = [None, None]
def extractor(src, i):
# Remark: self.options['use_yam'] -> explicit request for YAML
# self.use_yaml -> explicit request AND availability of YAML
dext = DataExtractor(self.options['use_yaml'], xml_mode=self.xml_mode,
ignore=self.options['ignore'], ignoreP=self.options['ignoreP'])
lines[i], documents[i], _ = dext.extract(src)
corrupted[i] = dext.corrupted_docs
t1 = Thread(target=extractor, args=(src1, 0))
t2 = Thread(target=extractor, args=(src2, 1))
t1.start()
t2.start()
t1.join()
t2.join()
if self.use_fl:
lines_differences = self._fldiff(*lines)
else:
lines_differences = []
if not self.use_yaml:
doc_differences = []
elif corrupted[0]:
doc_differences = [YFailure(
self.yaml_conf,
'Reference has corrupted YAML documents at line(s) {}.'
.format(', '.join(str(d.start + 1) for d in corrupted[0]))
)]
elif corrupted[1]:
doc_differences = [YFailure(
self.yaml_conf,
'Tested file has corrupted YAML documents at line(s) {}.'
.format(', '.join(str(d.start + 1) for d in corrupted[1]))
)]
else:
doc_differences = self._test_doc(*documents)
return lines_differences, doc_differences
def _test_doc(self, docs1, docs2):
"""Compare docs2 to docs1 and apply tests on docs2."""
return YTester(docs1, docs2, self.yaml_conf).run()
def _fldiff(self, lines1, lines2):
"""
Compute the effective comparison between two set of lines.
LineCountDifference and MetaCharDifference are both fatal so
they are returned alone if encountered.
"""
differences = []
if len(lines1) > len(lines2):
return [LineCountDifference('file 1', 'file 2', (len(lines1), len(lines2)))]
elif len(lines1) < len(lines2):
return [LineCountDifference('file 2', 'file 1', (len(lines2), len(lines1)))]
else:
for (i1, meta1, line1), (i2, meta2, line2) in zip(lines1, lines2):
if meta1 != meta2:
if meta1 != '_' and meta2 != '_':
return [MetaCharDifference(i1, i2, meta1, meta2)]
else:
if meta1 == '_':
# ignore these lines
pass
elif meta1 == '+':
# these lines are arbitrarily different
differences.append(ForcedDifference(
i1, i2, line1, line2
))
elif meta1 in {':', '.'}:
# do a character comparison
if norm_spaces(line1) != norm_spaces(line2):
differences.append(TextDifference(
i1, i2, line1, line2, silent=(meta1 == '.')
))
else:
# compare numerical values
splitted1 = float_re.split(line1)
splitted2 = float_re.split(line2)
# not the same number of floats on the line
if len(splitted1) != len(splitted2):
differences.append(TextDifference(i1, i2, line1, line2))
else:
if meta1 == '%':
# force tolerance
tol = 1.01e-2
tolrel = tol
else:
tol = self.options['tolerance_abs']
tolrel = self.options['tolerance_rel']
def to_float(f):
return float(f.lower().replace('d', 'e').replace('f', 'e'))
def pairs(seq1, seq2):
i = 0
n = len(seq1)
while i + 1 < n:
yield (seq1[i], seq1[i + 1],
seq2[i], seq2[i + 1])
i += 2
if i < n:
yield (seq1[i], None, seq2[i], None)
# si -> plain text separators
# fi -> floats
flag_compare = 1
for s1, f1, s2, f2 in pairs(splitted1, splitted2):
if norm_spaces(s1) != norm_spaces(s2):
differences.append(TextDifference(
i1, i2, line1, line2
))
if f1 is not None:
# reached the end
f1 = to_float(f1)
f2 = to_float(f2)
if meta1 == ';':
# compare absolute values
f1, f2 = abs(f1), abs(f2)
abs_sum = abs(f1) + abs(f2)
diff = abs(f1 - f2)
if abs_sum == 0.0:
diffrel = 0.0
else:
diffrel = diff / abs_sum
if diff > tol and diffrel > tolrel and flag_compare==1:
differences.append(
FloatDifference(
i1, i2, line1, line2,
diff, diffrel
)
)
if meta1 == ')':
# turn off further numerical comparison
flag_compare = 0
return differences
|