OLD | NEW |
(Empty) | |
| 1 #!/usr/bin/env python |
| 2 |
| 3 # Copyright 2014 The Chromium Authors. All rights reserved. |
| 4 # Use of this source code is governed by a BSD-style license that can be |
| 5 # found in the LICENSE file. |
| 6 |
| 7 |
| 8 import sys |
| 9 import unittest |
| 10 import make_dafsa |
| 11 |
| 12 |
| 13 class ParseGperfTest(unittest.TestCase): |
| 14 def testMalformedKey(self): |
| 15 """Tests exception is thrown at bad format.""" |
| 16 infile1 = [ '%%', '', '%%' ] |
| 17 self.assertRaises(make_dafsa.InputError, make_dafsa.parse_gperf, infile1) |
| 18 |
| 19 infile2 = [ '%%', 'apa,1', '%%' ] |
| 20 self.assertRaises(make_dafsa.InputError, make_dafsa.parse_gperf, infile2) |
| 21 |
| 22 infile3 = [ '%%', 'apa, 1', '%%' ] |
| 23 self.assertRaises(make_dafsa.InputError, make_dafsa.parse_gperf, infile3) |
| 24 |
| 25 def testBadValues(self): |
| 26 """Tests exception is thrown when value is out of range.""" |
| 27 infile1 = [ '%%', 'a, -1', '%%' ] |
| 28 self.assertRaises(make_dafsa.InputError, make_dafsa.parse_gperf, infile1) |
| 29 |
| 30 infile2 = [ '%%', 'a, x', '%%' ] |
| 31 self.assertRaises(make_dafsa.InputError, make_dafsa.parse_gperf, infile2) |
| 32 |
| 33 infile3 = [ '%%', 'a, 3', '%%' ] |
| 34 self.assertRaises(make_dafsa.InputError, make_dafsa.parse_gperf, infile3) |
| 35 |
| 36 infile4 = [ '%%', 'a, 6', '%%' ] |
| 37 self.assertRaises(make_dafsa.InputError, make_dafsa.parse_gperf, infile4) |
| 38 |
| 39 infile5 = [ '%%', 'a, 12', '%%' ] |
| 40 self.assertRaises(make_dafsa.InputError, make_dafsa.parse_gperf, infile5) |
| 41 |
| 42 def testValues(self): |
| 43 """Tests legal values are accepted.""" |
| 44 infile1 = [ '%%', 'a, 0', '%%' ] |
| 45 words1 = [ 'a0' ] |
| 46 self.assertEqual(make_dafsa.parse_gperf(infile1), words1) |
| 47 |
| 48 infile2 = [ '%%', 'a, 1', '%%' ] |
| 49 words2 = [ 'a1' ] |
| 50 self.assertEqual(make_dafsa.parse_gperf(infile2), words2) |
| 51 |
| 52 infile3 = [ '%%', 'a, 2', '%%' ] |
| 53 words3 = [ 'a2' ] |
| 54 self.assertEqual(make_dafsa.parse_gperf(infile3), words3) |
| 55 |
| 56 infile4 = [ '%%', 'a, 4', '%%' ] |
| 57 words4 = [ 'a4' ] |
| 58 self.assertEqual(make_dafsa.parse_gperf(infile4), words4) |
| 59 |
| 60 def testOneWord(self): |
| 61 """Tests a single key can be parsed.""" |
| 62 infile = [ '%%', 'apa, 1', '%%' ] |
| 63 words = [ 'apa1' ] |
| 64 self.assertEqual(make_dafsa.parse_gperf(infile), words) |
| 65 |
| 66 def testTwoWords(self): |
| 67 """Tests a sequence of keys can be parsed.""" |
| 68 infile = [ '%%', 'apa, 1', 'bepa.com, 2', '%%' ] |
| 69 words = [ 'apa1', 'bepa.com2' ] |
| 70 self.assertEqual(make_dafsa.parse_gperf(infile), words) |
| 71 |
| 72 |
| 73 class ToDafsaTest(unittest.TestCase): |
| 74 def testEmptyInput(self): |
| 75 """Tests exception is thrown at empty input.""" |
| 76 words = () |
| 77 self.assertRaises(make_dafsa.InputError, make_dafsa.to_dafsa, words) |
| 78 |
| 79 def testNonASCII(self): |
| 80 """Tests exception is thrown if illegal characters are used.""" |
| 81 words1 = ( chr(0x1F) + 'a1', ) |
| 82 self.assertRaises(make_dafsa.InputError, make_dafsa.to_dafsa, words1) |
| 83 |
| 84 words2 = ( 'a' + chr(0x1F) + '1', ) |
| 85 self.assertRaises(make_dafsa.InputError, make_dafsa.to_dafsa, words2) |
| 86 |
| 87 words3 = ( chr(0x80) + 'a1', ) |
| 88 self.assertRaises(make_dafsa.InputError, make_dafsa.to_dafsa, words3) |
| 89 |
| 90 words4 = ( 'a' + chr(0x80) + '1', ) |
| 91 self.assertRaises(make_dafsa.InputError, make_dafsa.to_dafsa, words4) |
| 92 |
| 93 def testChar(self): |
| 94 """Tests a DAFSA can be created from a single character domain name.""" |
| 95 words = [ 'a0' ] |
| 96 node2 = ( chr(0), [ None ] ) |
| 97 node1 = ( 'a', [ node2 ] ) |
| 98 source = [ node1 ] |
| 99 self.assertEqual(make_dafsa.to_dafsa(words), source) |
| 100 |
| 101 def testChars(self): |
| 102 """Tests a DAFSA can be created from a multi character domain name.""" |
| 103 words = [ 'ab0' ] |
| 104 node3 = ( chr(0), [ None ] ) |
| 105 node2 = ( 'b', [ node3 ] ) |
| 106 node1 = ( 'a', [ node2 ] ) |
| 107 source = [ node1 ] |
| 108 self.assertEqual(make_dafsa.to_dafsa(words), source) |
| 109 |
| 110 def testWords(self): |
| 111 """Tests a DAFSA can be created from a sequence of domain names.""" |
| 112 words = [ 'a0', 'b1' ] |
| 113 node4 = ( chr(1), [ None ] ) |
| 114 node3 = ( 'b', [ node4 ] ) |
| 115 node2 = ( chr(0), [ None ] ) |
| 116 node1 = ( 'a', [ node2 ] ) |
| 117 source = [ node1, node3 ] |
| 118 self.assertEqual(make_dafsa.to_dafsa(words), source) |
| 119 |
| 120 |
| 121 class ToWordsTest(unittest.TestCase): |
| 122 def testSink(self): |
| 123 """Tests the sink is exapnded to a list with an empty string.""" |
| 124 node1 = None |
| 125 words = [ '' ] |
| 126 self.assertEqual(make_dafsa.to_words(node1), words) |
| 127 |
| 128 def testSingleNode(self): |
| 129 """Tests a single node is expanded to a list with the label string.""" |
| 130 |
| 131 # 'ab' -> [ 'ab' ] |
| 132 |
| 133 node1 = ( 'ab', [ None ] ) |
| 134 words = [ 'ab' ] |
| 135 self.assertEqual(make_dafsa.to_words(node1), words) |
| 136 |
| 137 def testChain(self): |
| 138 """Tests a sequence of nodes are preoperly expanded.""" |
| 139 |
| 140 # 'ab' -> 'cd' => [ 'abcd' ] |
| 141 |
| 142 node2 = ( 'cd', [ None ] ) |
| 143 node1 = ( 'ab', [ node2 ] ) |
| 144 words = [ 'abcd' ] |
| 145 self.assertEqual(make_dafsa.to_words(node1), words) |
| 146 |
| 147 def testInnerTerminator(self): |
| 148 """Tests a sequence with an inner terminator is expanded to two strings.""" |
| 149 |
| 150 # 'a' -> 'b' |
| 151 # \ => [ 'ab', 'a' ] |
| 152 # {sink} |
| 153 |
| 154 node2 = ( 'b', [ None ] ) |
| 155 node1 = ( 'a', [ node2, None ] ) |
| 156 words = [ 'ab', 'a' ] |
| 157 self.assertEqual(make_dafsa.to_words(node1), words) |
| 158 |
| 159 def testDiamond(self): |
| 160 """Tests a diamond can be expanded to a word list.""" |
| 161 |
| 162 # 'cd' |
| 163 # / \ |
| 164 # 'ab' 'gh' |
| 165 # \ / |
| 166 # 'ef' |
| 167 |
| 168 node4 = ( 'gh', [ None ] ) |
| 169 node3 = ( 'ef', [ node4 ] ) |
| 170 node2 = ( 'cd', [ node4 ] ) |
| 171 node1 = ( 'ab', [ node2, node3 ] ) |
| 172 words = [ 'abcdgh', 'abefgh' ] |
| 173 self.assertEqual(make_dafsa.to_words(node1), words) |
| 174 |
| 175 |
| 176 class JoinLabelsTest(unittest.TestCase): |
| 177 def testLabel(self): |
| 178 """Tests a single label passes unchanged.""" |
| 179 |
| 180 # 'a' => 'a' |
| 181 |
| 182 node1 = ( 'a', [ None ] ) |
| 183 source = [ node1 ] |
| 184 self.assertEqual(make_dafsa.join_labels(source), source) |
| 185 |
| 186 def testInnerTerminator(self): |
| 187 """Tests a sequence with an inner terminator passes unchanged.""" |
| 188 |
| 189 # 'a' -> 'b' 'a' -> 'b' |
| 190 # \ => \ |
| 191 # {sink} {sink} |
| 192 |
| 193 node2 = ( 'b', [ None ] ) |
| 194 node1 = ( 'a', [ node2, None ] ) |
| 195 source = [ node1 ] |
| 196 self.assertEqual(make_dafsa.join_labels(source), source) |
| 197 |
| 198 def testLabels(self): |
| 199 """Tests a sequence of labels can be joined.""" |
| 200 |
| 201 # 'a' -> 'b' => 'ab' |
| 202 |
| 203 node2 = ( 'b', [ None ] ) |
| 204 node1 = ( 'a', [ node2 ] ) |
| 205 source1 = [ node1 ] |
| 206 node3 = ( 'ab', [ None ] ) |
| 207 source2 = [ node3 ] |
| 208 self.assertEqual(make_dafsa.join_labels(source1), source2) |
| 209 |
| 210 def testCompositeLabels(self): |
| 211 """Tests a sequence of multi character labels can be joined.""" |
| 212 |
| 213 # 'ab' -> 'cd' => 'abcd' |
| 214 |
| 215 node2 = ( 'cd', [ None ] ) |
| 216 node1 = ( 'ab', [ node2 ] ) |
| 217 source1 = [ node1 ] |
| 218 node3 = ( 'abcd', [ None ] ) |
| 219 source2 = [ node3 ] |
| 220 self.assertEqual(make_dafsa.join_labels(source1), source2) |
| 221 |
| 222 def testAtomicTrie(self): |
| 223 """Tests a trie formed DAFSA with atomic labels passes unchanged.""" |
| 224 |
| 225 # 'b' 'b' |
| 226 # / / |
| 227 # 'a' => 'a' |
| 228 # \ \ |
| 229 # 'c' 'c' |
| 230 |
| 231 node3 = ( 'c', [ None ] ) |
| 232 node2 = ( 'b', [ None ] ) |
| 233 node1 = ( 'a', [ node2, node3 ] ) |
| 234 source = [ node1 ] |
| 235 self.assertEqual(make_dafsa.join_labels(source), source) |
| 236 |
| 237 def testReverseAtomicTrie(self): |
| 238 """Tests a reverse trie formed DAFSA with atomic labels passes unchanged.""" |
| 239 |
| 240 # 'a' 'a' |
| 241 # \ \ |
| 242 # 'c' => 'c' |
| 243 # / / |
| 244 # 'b' 'b' |
| 245 |
| 246 node3 = ( 'c', [ None ] ) |
| 247 node2 = ( 'b', [ node3 ] ) |
| 248 node1 = ( 'a', [ node3 ] ) |
| 249 source = [ node1, node2 ] |
| 250 self.assertEqual(make_dafsa.join_labels(source), source) |
| 251 |
| 252 def testChainedTrie(self): |
| 253 """Tests a trie formed DAFSA with chained labels can be joined.""" |
| 254 |
| 255 # 'c' -> 'd' 'cd' |
| 256 # / / |
| 257 # 'a' -> 'b' => 'ab' |
| 258 # \ \ |
| 259 # 'e' -> 'f' 'ef' |
| 260 |
| 261 node6 = ( 'f', [ None ] ) |
| 262 node5 = ( 'e', [ node6 ] ) |
| 263 node4 = ( 'd', [ None ] ) |
| 264 node3 = ( 'c', [ node4 ] ) |
| 265 node2 = ( 'b', [ node3, node5 ] ) |
| 266 node1 = ( 'a', [ node2 ] ) |
| 267 source1 = [ node1 ] |
| 268 node9 = ( 'ef', [ None ] ) |
| 269 node8 = ( 'cd', [ None ] ) |
| 270 node7 = ( 'ab', [ node8, node9 ] ) |
| 271 source2 = [ node7 ] |
| 272 self.assertEqual(make_dafsa.join_labels(source1), source2) |
| 273 |
| 274 def testReverseChainedTrie(self): |
| 275 """Tests a reverse trie formed DAFSA with chained labels can be joined.""" |
| 276 |
| 277 # 'a' -> 'b' 'ab' |
| 278 # \ \ |
| 279 # 'e' -> 'f' => 'ef' |
| 280 # / / |
| 281 # 'c' -> 'd' 'cd' |
| 282 |
| 283 node6 = ( 'f', [ None ] ) |
| 284 node5 = ( 'e', [ node6 ] ) |
| 285 node4 = ( 'd', [ node5 ] ) |
| 286 node3 = ( 'c', [ node4 ] ) |
| 287 node2 = ( 'b', [ node5 ] ) |
| 288 node1 = ( 'a', [ node2 ] ) |
| 289 source1 = [ node1, node3 ] |
| 290 node9 = ( 'ef', [ None ] ) |
| 291 node8 = ( 'cd', [ node9 ] ) |
| 292 node7 = ( 'ab', [ node9 ] ) |
| 293 source2 = [ node7, node8 ] |
| 294 self.assertEqual(make_dafsa.join_labels(source1), source2) |
| 295 |
| 296 |
| 297 class JoinSuffixesTest(unittest.TestCase): |
| 298 def testSingleLabel(self): |
| 299 """Tests a single label passes unchanged.""" |
| 300 |
| 301 # 'a' => 'a' |
| 302 |
| 303 node1 = ( 'a', [ None ] ) |
| 304 source = [ node1 ] |
| 305 self.assertEqual(make_dafsa.join_suffixes(source), source) |
| 306 |
| 307 def testInnerTerminator(self): |
| 308 """Tests a sequence with an inner terminator passes unchanged.""" |
| 309 |
| 310 # 'a' -> 'b' 'a' -> 'b' |
| 311 # \ => \ |
| 312 # {sink} {sink} |
| 313 |
| 314 node2 = ( 'b', [ None ] ) |
| 315 node1 = ( 'a', [ node2, None ] ) |
| 316 source = [ node1 ] |
| 317 self.assertEqual(make_dafsa.join_suffixes(source), source) |
| 318 |
| 319 def testDistinctTrie(self): |
| 320 """Tests a trie formed DAFSA with distinct labels passes unchanged.""" |
| 321 |
| 322 # 'b' 'b' |
| 323 # / / |
| 324 # 'a' => 'a' |
| 325 # \ \ |
| 326 # 'c' 'c' |
| 327 |
| 328 node3 = ( 'c', [ None ] ) |
| 329 node2 = ( 'b', [ None ] ) |
| 330 node1 = ( 'a', [ node2, node3 ] ) |
| 331 source = [ node1 ] |
| 332 self.assertEqual(make_dafsa.join_suffixes(source), source) |
| 333 |
| 334 def testReverseDistinctTrie(self): |
| 335 """Tests a reverse trie formed DAFSA with distinct labels passes unchanged. |
| 336 """ |
| 337 |
| 338 # 'a' 'a' |
| 339 # \ \ |
| 340 # 'c' => 'c' |
| 341 # / / |
| 342 # 'b' 'b' |
| 343 |
| 344 node3 = ( 'c', [ None ] ) |
| 345 node2 = ( 'b', [ node3 ] ) |
| 346 node1 = ( 'a', [ node3 ] ) |
| 347 source = [ node1, node2 ] |
| 348 self.assertEqual(make_dafsa.join_suffixes(source), source) |
| 349 |
| 350 def testJoinTwoHeads(self): |
| 351 """Tests two heads can be joined even if there is something else between.""" |
| 352 |
| 353 # 'a' ------'a' |
| 354 # / |
| 355 # 'b' => 'b' / |
| 356 # / |
| 357 # 'a' --- |
| 358 # |
| 359 # The picture above should shows that the new version should have just one |
| 360 # instance of the node with label 'a'. |
| 361 |
| 362 node3 = ( 'a', [ None ] ) |
| 363 node2 = ( 'b', [ None ] ) |
| 364 node1 = ( 'a', [ None ] ) |
| 365 source1 = [ node1, node2, node3 ] |
| 366 source2 = make_dafsa.join_suffixes(source1) |
| 367 |
| 368 # Both versions should expand to the same content. |
| 369 self.assertEqual(source1, source2) |
| 370 # But the new version should have just one instance of 'a'. |
| 371 self.assertIs(source2[0], source2[2]) |
| 372 |
| 373 def testJoinTails(self): |
| 374 """Tests tails can be joined.""" |
| 375 |
| 376 # 'a' -> 'c' 'a' |
| 377 # \ |
| 378 # => 'c' |
| 379 # / |
| 380 # 'b' -> 'c' 'b' |
| 381 |
| 382 node4 = ( 'c', [ None ] ) |
| 383 node3 = ( 'b', [ node4 ] ) |
| 384 node2 = ( 'c', [ None ] ) |
| 385 node1 = ( 'a', [ node2 ] ) |
| 386 source1 = [ node1, node3 ] |
| 387 source2 = make_dafsa.join_suffixes(source1) |
| 388 |
| 389 # Both versions should expand to the same content. |
| 390 self.assertEqual(source1, source2) |
| 391 # But the new version should have just one tail. |
| 392 self.assertIs(source2[0][1][0], source2[1][1][0]) |
| 393 |
| 394 def testMakeRecursiveTrie(self): |
| 395 """Tests recursive suffix join.""" |
| 396 |
| 397 # 'a' -> 'e' -> 'g' 'a' |
| 398 # \ |
| 399 # 'e' |
| 400 # / \ |
| 401 # 'b' -> 'e' -> 'g' 'b' \ |
| 402 # \ |
| 403 # => 'g' |
| 404 # / |
| 405 # 'c' -> 'f' -> 'g' 'c' / |
| 406 # \ / |
| 407 # 'f' |
| 408 # / |
| 409 # 'd' -> 'f' -> 'g' 'd' |
| 410 |
| 411 node7 = ( 'g', [ None ] ) |
| 412 node6 = ( 'f', [ node7 ] ) |
| 413 node5 = ( 'e', [ node7 ] ) |
| 414 node4 = ( 'd', [ node6 ] ) |
| 415 node3 = ( 'c', [ node6 ] ) |
| 416 node2 = ( 'b', [ node5 ] ) |
| 417 node1 = ( 'a', [ node5 ] ) |
| 418 source1 = [ node1, node2, node3, node4 ] |
| 419 source2 = make_dafsa.join_suffixes(source1) |
| 420 |
| 421 # Both versions should expand to the same content. |
| 422 self.assertEqual(source1, source2) |
| 423 # But the new version should have just one 'e'. |
| 424 self.assertIs(source2[0][1][0], source2[1][1][0]) |
| 425 # And one 'f'. |
| 426 self.assertIs(source2[2][1][0], source2[3][1][0]) |
| 427 # And one 'g'. |
| 428 self.assertIs(source2[0][1][0][1][0], source2[2][1][0][1][0]) |
| 429 |
| 430 def testMakeDiamond(self): |
| 431 """Test we can join suffixes of a trie.""" |
| 432 |
| 433 # 'b' -> 'd' 'b' |
| 434 # / / \ |
| 435 # 'a' => 'a' 'd' |
| 436 # \ \ / |
| 437 # 'c' -> 'd' 'c' |
| 438 |
| 439 node5 = ( 'd', [ None ] ) |
| 440 node4 = ( 'c', [ node5 ] ) |
| 441 node3 = ( 'd', [ None ] ) |
| 442 node2 = ( 'b', [ node3 ] ) |
| 443 node1 = ( 'a', [ node2, node4 ] ) |
| 444 source1 = [ node1 ] |
| 445 source2 = make_dafsa.join_suffixes(source1) |
| 446 |
| 447 # Both versions should expand to the same content. |
| 448 self.assertEqual(source1, source2) |
| 449 # But the new version should have just one 'd'. |
| 450 self.assertIs(source2[0][1][0][1][0], source2[0][1][1][1][0]) |
| 451 |
| 452 def testJoinOneChild(self): |
| 453 """Tests that we can join some children but not all.""" |
| 454 |
| 455 # 'c' ----'c' |
| 456 # / / / |
| 457 # 'a' 'a' / |
| 458 # \ \ / |
| 459 # 'd' 'd'/ |
| 460 # => / |
| 461 # 'c' / |
| 462 # / / |
| 463 # 'b' 'b' |
| 464 # \ \ |
| 465 # 'e' 'e' |
| 466 |
| 467 node6 = ( 'e', [ None ] ) |
| 468 node5 = ( 'c', [ None ] ) |
| 469 node4 = ( 'b', [ node5, node6 ] ) |
| 470 node3 = ( 'd', [ None ] ) |
| 471 node2 = ( 'c', [ None ] ) |
| 472 node1 = ( 'a', [ node2, node3 ] ) |
| 473 source1 = [ node1, node4 ] |
| 474 source2 = make_dafsa.join_suffixes(source1) |
| 475 |
| 476 # Both versions should expand to the same content. |
| 477 self.assertEqual(source1, source2) |
| 478 # But the new version should have just one 'c'. |
| 479 self.assertIs(source2[0][1][0], source2[1][1][0]) |
| 480 |
| 481 |
| 482 class ReverseTest(unittest.TestCase): |
| 483 def testAtomicLabel(self): |
| 484 """Tests an atomic label passes unchanged.""" |
| 485 |
| 486 # 'a' => 'a' |
| 487 |
| 488 node1 = ( 'a', [ None ] ) |
| 489 source = [ node1 ] |
| 490 self.assertEqual(make_dafsa.reverse(source), source) |
| 491 |
| 492 def testLabel(self): |
| 493 """Tests that labels are reversed.""" |
| 494 |
| 495 # 'ab' => 'ba' |
| 496 |
| 497 node1 = ( 'ab', [ None ] ) |
| 498 source1 = [ node1 ] |
| 499 node2 = ( 'ba', [ None ] ) |
| 500 source2 = [ node2 ] |
| 501 self.assertEqual(make_dafsa.reverse(source1), source2) |
| 502 |
| 503 def testChain(self): |
| 504 """Tests that edges are reversed.""" |
| 505 |
| 506 # 'a' -> 'b' => 'b' -> 'a' |
| 507 |
| 508 node2 = ( 'b', [ None ] ) |
| 509 node1 = ( 'a', [ node2 ] ) |
| 510 source1 = [ node1 ] |
| 511 node4 = ( 'a', [ None ] ) |
| 512 node3 = ( 'b', [ node4 ] ) |
| 513 source2 = [ node3 ] |
| 514 self.assertEqual(make_dafsa.reverse(source1), source2) |
| 515 |
| 516 def testInnerTerminator(self): |
| 517 """Tests a sequence with an inner terminator can be reversed.""" |
| 518 |
| 519 # 'a' -> 'b' 'b' -> 'a' |
| 520 # \ => / |
| 521 # {sink} ------ |
| 522 |
| 523 node2 = ( 'b', [ None ] ) |
| 524 node1 = ( 'a', [ node2, None ] ) |
| 525 source1 = [ node1 ] |
| 526 node4 = ( 'a', [ None ] ) |
| 527 node3 = ( 'b', [ node4 ] ) |
| 528 source2 = [ node3, node4 ] |
| 529 self.assertEqual(make_dafsa.reverse(source1), source2) |
| 530 |
| 531 def testAtomicTrie(self): |
| 532 """Tests a trie formed DAFSA can be reversed.""" |
| 533 |
| 534 # 'b' 'b' |
| 535 # / \ |
| 536 # 'a' => 'a' |
| 537 # \ / |
| 538 # 'c' 'c' |
| 539 |
| 540 node3 = ( 'c', [ None ] ) |
| 541 node2 = ( 'b', [ None ] ) |
| 542 node1 = ( 'a', [ node2, node3 ] ) |
| 543 source1 = [ node1 ] |
| 544 node6 = ( 'a', [ None ] ) |
| 545 node5 = ( 'c', [ node6 ] ) |
| 546 node4 = ( 'b', [ node6 ] ) |
| 547 source2 = [ node4, node5 ] |
| 548 self.assertEqual(make_dafsa.reverse(source1), source2) |
| 549 |
| 550 def testReverseAtomicTrie(self): |
| 551 """Tests a reverse trie formed DAFSA can be reversed.""" |
| 552 |
| 553 # 'a' 'a' |
| 554 # \ / |
| 555 # 'c' => 'c' |
| 556 # / \ |
| 557 # 'b' 'b' |
| 558 |
| 559 node3 = ( 'c', [ None ] ) |
| 560 node2 = ( 'b', [ node3 ] ) |
| 561 node1 = ( 'a', [ node3 ] ) |
| 562 source1 = [ node1, node2 ] |
| 563 node6 = ( 'b', [ None ] ) |
| 564 node5 = ( 'a', [ None ] ) |
| 565 node4 = ( 'c', [ node5, node6 ] ) |
| 566 source2 = [ node4 ] |
| 567 self.assertEqual(make_dafsa.reverse(source1), source2) |
| 568 |
| 569 def testDiamond(self): |
| 570 """Tests we can reverse both edges and nodes in a diamond.""" |
| 571 |
| 572 # 'cd' 'dc' |
| 573 # / \ / \ |
| 574 # 'ab' 'gh' => 'hg' 'ba' |
| 575 # \ / \ / |
| 576 # 'ef' 'fe' |
| 577 |
| 578 node4 = ( 'gh', [ None ] ) |
| 579 node3 = ( 'ef', [ node4 ] ) |
| 580 node2 = ( 'cd', [ node4 ] ) |
| 581 node1 = ( 'ab', [ node2, node3 ] ) |
| 582 source1 = [ node1 ] |
| 583 node8 = ( 'ba', [ None ] ) |
| 584 node7 = ( 'fe', [ node8 ] ) |
| 585 node6 = ( 'dc', [ node8 ] ) |
| 586 node5 = ( 'hg', [ node6, node7 ] ) |
| 587 source2 = [ node5 ] |
| 588 self.assertEqual(make_dafsa.reverse(source1), source2) |
| 589 |
| 590 |
| 591 class TopSortTest(unittest.TestCase): |
| 592 def testNode(self): |
| 593 """Tests a DAFSA with one node can be sorted.""" |
| 594 |
| 595 # 'a' => [ 'a' ] |
| 596 |
| 597 node1 = ( 'a', [ None ] ) |
| 598 source = [ node1 ] |
| 599 nodes = [ node1 ] |
| 600 self.assertEqual(make_dafsa.top_sort(source), nodes) |
| 601 |
| 602 def testDiamond(self): |
| 603 """Tests nodes in a diamond can be sorted.""" |
| 604 |
| 605 # 'b' |
| 606 # / \ |
| 607 # 'a' 'd' |
| 608 # \ / |
| 609 # 'c' |
| 610 |
| 611 node4 = ( 'd', [ None ] ) |
| 612 node3 = ( 'c', [ node4 ] ) |
| 613 node2 = ( 'b', [ node4 ] ) |
| 614 node1 = ( 'a', [ node2, node3 ] ) |
| 615 source = [ node1 ] |
| 616 nodes = make_dafsa.top_sort(source) |
| 617 self.assertLess(nodes.index(node1), nodes.index(node2)) |
| 618 self.assertLess(nodes.index(node2), nodes.index(node4)) |
| 619 self.assertLess(nodes.index(node3), nodes.index(node4)) |
| 620 |
| 621 |
| 622 class EncodePrefixTest(unittest.TestCase): |
| 623 def testChar(self): |
| 624 """Tests to encode a single character prefix.""" |
| 625 label = 'a' |
| 626 bytes = [ ord('a') ] |
| 627 self.assertEqual(make_dafsa.encode_prefix(label), bytes) |
| 628 |
| 629 def testChars(self): |
| 630 """Tests to encode a multi character prefix.""" |
| 631 label = 'ab' |
| 632 bytes = [ ord('b'), ord('a') ] |
| 633 self.assertEqual(make_dafsa.encode_prefix(label), bytes) |
| 634 |
| 635 |
| 636 class EncodeLabelTest(unittest.TestCase): |
| 637 def testChar(self): |
| 638 """Tests to encode a single character label.""" |
| 639 label = 'a' |
| 640 bytes = [ ord('a') + 0x80 ] |
| 641 self.assertEqual(make_dafsa.encode_label(label), bytes) |
| 642 |
| 643 def testChars(self): |
| 644 """Tests to encode a multi character label.""" |
| 645 label = 'ab' |
| 646 bytes = [ ord('b') + 0x80, ord('a') ] |
| 647 self.assertEqual(make_dafsa.encode_label(label), bytes) |
| 648 |
| 649 |
| 650 class EncodeLinksTest(unittest.TestCase): |
| 651 def testEndLabel(self): |
| 652 """Tests to encode link to the sink.""" |
| 653 children = [ None ] |
| 654 offsets = {} |
| 655 bytes = 0 |
| 656 output = [] |
| 657 self.assertEqual(make_dafsa.encode_links(children, offsets, bytes), |
| 658 output) |
| 659 |
| 660 def testOneByteOffset(self): |
| 661 """Tests to encode a single one byte offset.""" |
| 662 node = ( '', [ None ] ) |
| 663 children = [ node ] |
| 664 offsets = { id(node) : 2 } |
| 665 bytes = 5 |
| 666 output = [ 132 ] |
| 667 self.assertEqual(make_dafsa.encode_links(children, offsets, bytes), |
| 668 output) |
| 669 |
| 670 def testOneByteOffsets(self): |
| 671 """Tests to encode a sequence of one byte offsets.""" |
| 672 node1 = ( '', [ None ] ) |
| 673 node2 = ( '', [ None ] ) |
| 674 children = [ node1, node2 ] |
| 675 offsets = { id(node1) : 2, id(node2) : 1 } |
| 676 bytes = 5 |
| 677 output = [ 129, 5 ] |
| 678 self.assertEqual(make_dafsa.encode_links(children, offsets, bytes), |
| 679 output) |
| 680 |
| 681 def testTwoBytesOffset(self): |
| 682 """Tests to encode a single two byte offset.""" |
| 683 node = ( '', [ None ] ) |
| 684 children = [ node ] |
| 685 offsets = { id(node) : 2 } |
| 686 bytes = 1005 |
| 687 output = [ 237, 195] |
| 688 self.assertEqual(make_dafsa.encode_links(children, offsets, bytes), |
| 689 output) |
| 690 |
| 691 def testTwoBytesOffsets(self): |
| 692 """Tests to encode a sequence of two byte offsets.""" |
| 693 node1 = ( '', [ None ] ) |
| 694 node2 = ( '', [ None ] ) |
| 695 node3 = ( '', [ None ] ) |
| 696 children = [ node1, node2, node3 ] |
| 697 offsets = { id(node1) : 1002, id(node2) : 2, id(node3) : 2002 } |
| 698 bytes = 3005 |
| 699 output = [ 232, 195, 232, 67, 241, 67 ] |
| 700 self.assertEqual(make_dafsa.encode_links(children, offsets, bytes), |
| 701 output) |
| 702 |
| 703 def testThreeBytesOffset(self): |
| 704 """Tests to encode a single three byte offset.""" |
| 705 node = ( '', [ None ] ) |
| 706 children = [ node ] |
| 707 offsets = { id(node) : 2 } |
| 708 bytes = 100005 |
| 709 output = [ 166, 134, 225 ] |
| 710 self.assertEqual(make_dafsa.encode_links(children, offsets, bytes), |
| 711 output) |
| 712 |
| 713 def testThreeBytesOffsets(self): |
| 714 """Tests to encode a sequence of three byte offsets.""" |
| 715 node1 = ( '', [ None ] ) |
| 716 node2 = ( '', [ None ] ) |
| 717 node3 = ( '', [ None ] ) |
| 718 children = [ node1, node2, node3 ] |
| 719 offsets = { id(node1) : 100002, id(node2) : 2, id(node3) : 200002 } |
| 720 bytes = 300005 |
| 721 output = [ 160, 134, 225, 160, 134, 97, 172, 134, 97 ] |
| 722 self.assertEqual(make_dafsa.encode_links(children, offsets, bytes), |
| 723 output) |
| 724 |
| 725 def testOneTwoThreeBytesOffsets(self): |
| 726 """Tests to encode offsets of different sizes.""" |
| 727 node1 = ( '', [ None ] ) |
| 728 node2 = ( '', [ None ] ) |
| 729 node3 = ( '', [ None ] ) |
| 730 children = [ node1, node2, node3 ] |
| 731 offsets = { id(node1) : 10003, id(node2) : 10002, id(node3) : 100002 } |
| 732 bytes = 300005 |
| 733 output = [ 129, 143, 95, 97, 74, 13, 99 ] |
| 734 self.assertEqual(make_dafsa.encode_links(children, offsets, bytes), |
| 735 output) |
| 736 |
| 737 |
| 738 class ExamplesTest(unittest.TestCase): |
| 739 def testExample1(self): |
| 740 """Tests Example 1 from make_dafsa.py.""" |
| 741 infile = [ '%%', 'aa, 1', 'a, 2', '%%' ] |
| 742 bytes = [ 0x81, 0xE1, 0x02, 0x81, 0x82, 0x61, 0x81 ] |
| 743 outfile = make_dafsa.to_cxx(bytes) |
| 744 self.assertEqual(make_dafsa.words_to_cxx(make_dafsa.parse_gperf(infile)), |
| 745 outfile) |
| 746 |
| 747 def testExample2(self): |
| 748 """Tests Example 2 from make_dafsa.py.""" |
| 749 infile = [ '%%', 'aa, 1', 'bbb, 2', 'baa, 1', '%%' ] |
| 750 bytes = [ 0x02, 0x83, 0xE2, 0x02, 0x83, 0x61, 0x61, 0x81, 0x62, 0x62, |
| 751 0x82 ] |
| 752 outfile = make_dafsa.to_cxx(bytes) |
| 753 self.assertEqual(make_dafsa.words_to_cxx(make_dafsa.parse_gperf(infile)), |
| 754 outfile) |
| 755 |
| 756 |
| 757 if __name__ == '__main__': |
| 758 unittest.main() |
OLD | NEW |