1# es.yaml 2# ------- 3# Note: make Latin-American conventions by default (country overrides for Spain 4# as well as any other country-specific norms) 5 6components: 7 level: 8 # If no floor number is specified 9 null_probability: 0.6 10 alphanumeric_probability: 0.35 11 standalone_probability: 0.05 12 13 staircase: 14 null_probability: 0.99 15 alphanumeric_probability: 0.01 16 17 entrance: 18 null_probability: 0.999 19 alphanumeric_probability: 0.001 20 21 unit: 22 # If no unit number is specified 23 null_probability: 0.3 24 alphanumeric_probability: 0.65 25 standalone_probability: 0.05 26 27numbers: 28 default: &numero 29 canonical: número 30 abbreviated: "nº" 31 sample: true 32 canonical_probability: 0.1 33 abbreviated_probability: 0.4 34 sample_probability: 0.5 35 numeric: 36 direction: left 37 numeric_affix: 38 affix: "#" # e.g. #3, #2F, etc. 39 probability: 0.5 40 alternatives: 41 - alternative: 42 direction: left # affix goes on the number's left 43 44 # Probabilities for numbers 45 numeric_probability: 0.8 46 numeric_affix_probability: 0.2 47 48and: 49 default: &y 50 canonical: y 51 abbreviated: "&" 52 sample: true 53 canonical_probability: 0.5 54 abbreviated_probability: 0.4 55 sample_probability: 0.1 56 57 58house_numbers: 59 # sin número (s/n) addresses 60 no_number: 61 default: 62 canonical: sin número 63 abbreviated: s/n 64 sample: true 65 canonical_probability: 0.1 66 abbreviated_probability: 0.7 67 sample_probability: 0.2 68 alphanumeric: 69 default: *numero 70 probability: 0.99 71 alternatives: 72 - alternative: 73 canonical: número exterior 74 abbreviated: "nº ext" 75 sample: true 76 canonical_probability: 0.1 77 abbreviated_probability: 0.6 78 sample_probability: 0.3 79 sample_exclude: 80 - "numero exterior" # Exclude the unicode normalized version 81 numeric: 82 direction: left 83 probability: 0.01 84 85 alphanumeric_phrase_probability: 0.1 86 no_number_probability: 0.1 # With this probability, use sin número if no house_number is specified 87 88levels: 89 # Everywhere except Spain 90 floor: &piso 91 canonical: piso 92 abbreviated: p 93 sample: true 94 canonical_probability: 0.8 95 abbreviated_probability: 0.1 96 sample_probability: 0.1 97 numeric: 98 direction: left 99 add_number_phrase: true # Occasionally add variation of "number", e.g. Piso No 2 100 add_number_phrase_probability: 0.05 101 numeric_affix: 102 affix: p 103 direction: left # P2 104 # e.g. 2o piso 105 ordinal: 106 direction: right 107 direction_probability: 0.95 # Let it vary occasionally e.g. Piso 2o 108 standalone_probability: 0.2 # Let e.g. 5º be the entire floor string 109 digits: 110 ascii_probability: 0.8 111 spellout_probability: 0.2 112 numeric_probability: 0.6 113 numeric_affix_probability: 0.05 114 ordinal_probability: 0.35 115 # Ground floor 116 bajos: &bajos 117 canonical: bajos 118 abbreviated: bjs 119 sample: true 120 canonical_probability: 0.6 121 abbreviated_probability: 0.3 122 sample_probability: 0.1 123 piso_bajo: &piso_bajo 124 canonical: piso bajo 125 abbreviated: pb 126 sample: true 127 canonical_probability: 0.4 128 abbreviated_probability: 0.5 129 sample_probability: 0.1 130 bajo: &bajo 131 canonical: bajo 132 abbreviated: bjo 133 sample: true 134 canonical_probability: 0.8 135 abbreviated_probability: 0.1 136 sample_probability: 0.1 137 # Used when floor number is < 0 (starts at -1 in all countries) 138 sotano: &sotano 139 canonical: sótano 140 abbreviated: so 141 sample: true 142 canonical_probability: 0.5 143 abbreviated_probability: 0.3 144 sample_probability: 0.2 145 # e.g. sótano 1 146 numeric: 147 direction: left 148 numeric_affix: 149 affix: so 150 direction: left 151 # e.g. segundo sótano 152 ordinal: 153 direction: right 154 standalone_probability: 0.985 155 number_abs_value: true 156 number_min_abs_value: 1 157 numeric_probability: 0.005 158 numeric_affix_probability: 0.005 159 ordinal_probability: 0.005 160 sub_sotano: &sub_sotano 161 canonical: sub sótano 162 abbreviated: ss 163 sample: true 164 # e.g. sub sótano 1 165 numeric: 166 direction: left 167 numeric_affix: 168 affix: ss 169 direction: left 170 # e.g. segundo sub sótano 171 ordinal: 172 direction: right 173 number_abs_value: true 174 number_min_abs_value: 2 175 # Sotano 2 == Sub-sotano 1 176 number_subtract_abs_value: 1 177 standalone_probability: 0.985 178 numeric_probability: 0.005 179 numeric_affix_probability: 0.005 180 ordinal_probability: 0.005 181 entresuelo: &entresuelo 182 canonical: entresuelo 183 abbreviated: entlo 184 half_floors: true 185 canonical_probability: 0.7 186 abbreviated_probability: 0.2 187 sample_probability: 0.1 188 sample: true 189 # e.g. entresuelo 2 190 numeric: 191 direction: left 192 # e.g. ent2 193 numeric_affix: 194 affix: ent 195 direction: left 196 # e.g. segundo entresuelo 197 ordinal: 198 direction: right 199 numeric_probability: 0.1 200 numeric_affix_probability: 0.1 201 ordinal_probability: 0.2 202 standalone_probability: 0.6 203 piso_principal: &piso_principal 204 canonical: piso principal 205 abbreviated: piso pral 206 sample: true 207 canonical_probability: 0.2 208 abbreviated_probability: 0.3 209 sample_probability: 0.5 210 principal: &principal 211 canonical: principal 212 abbreviated: pral 213 sample: true 214 canonical_probability: 0.2 215 abbreviated_probability: 0.6 216 sample_probability: 0.2 217 atico: &atico 218 canonical: ático 219 abbreviated: át 220 sample: true 221 canonical_probability: 0.7 222 abbreviated_probability: 0.1 223 sample_probability: 0.2 224 sobreatico: &sobreatico 225 canonical: sobreatico 226 aliases: 227 "<-1": 228 default: *sotano 229 probability: 0.6 230 alternatives: 231 - alternative: *sub_sotano 232 probability: 0.3995 233 - alternative: *piso 234 probability: 0.0005 235 "-1": 236 default: *sotano 237 probability: 0.9995 238 alternatives: 239 - alternative: *piso 240 probability: 0.0005 241 # Special token for half-floors 242 half_floors: 243 default: *entresuelo 244 "0": 245 default: *bajos 246 probability: 0.495 247 alternatives: 248 - alternative: *piso_bajo 249 probability: 0.395 250 - alternative: *bajo 251 probability: 0.1 252 - alternative: *piso 253 # Piso 0 is uncommon 254 probability: 0.01 255 top: 256 default: *piso 257 probability: 0.85 258 alternatives: 259 - alternative: *atico 260 probability: 0.1 261 - alternative: *sobreatico 262 probability: 0.05 263 264 numbering_starts_at: 0 265 266 alphanumeric: 267 default: *piso 268 add_number_phrase: true 269 add_number_phrase_probability: 0.05 270 numeric_probability: 0.99 271 alpha_probability: 0.01 272 273blocks: 274 alphanumeric: 275 default: 276 canonical: bloque 277 abbreviated: blq 278 sample: true 279 canonical_probability: 0.6 280 abbreviated_probability: 0.2 281 sample_probability: 0.2 282 numeric: 283 direction: left 284 285categories: 286 near: 287 default: 288 canonical: cerca de 289 probability: 0.8 290 alternatives: 291 - alternative: 292 canonical: cerca 293 probability: 0.2 294 nearby: 295 default: 296 canonical: cerca 297 probability: 0.5 298 alternatives: 299 - alternative: 300 canonical: próximo 301 probability: 0.05 302 - alternative: 303 canonical: proximo 304 probability: 0.05 305 - alternative: 306 canonical: cerca de aquí 307 probability: 0.05 308 - alternative: 309 canonical: cerca de aqui 310 probability: 0.05 311 - alternative: 312 canonical: acá 313 probability: 0.05 314 - alternative: 315 canonical: aca 316 probability: 0.05 317 - alternative: 318 canonical: cerca de acá 319 probability: 0.05 320 - alternative: 321 canonical: cerca de aca 322 probability: 0.05 323 - alternative: 324 canonical: por aquí 325 probability: 0.05 326 - alternative: 327 canonical: por aqui 328 probability: 0.05 329 near_me: 330 default: 331 canonical: cerca de mì 332 in: 333 default: 334 canonical: en 335 # Probabilities of each phrase 336 near_probability: 0.35 337 nearby_probability: 0.2 338 near_me_probability: 0.1 339 in_probability: 0.35 340 341cross_streets: 342 and: *y 343 con: &con 344 canonical: con 345 en: &en 346 canonical: en 347 x: &x 348 canonical: x 349 corner_of: &esquina_de 350 canonical: esquina de 351 abbreviated: esq de 352 sample: true 353 canonical_probability: 0.5 354 abbreviated_probability: 0.2 355 sample_probability: 0.3 356 at_the_corner_of: &en_la_esquina_de 357 canonical: en la esquina de 358 abbreviated: en la esq de 359 sample: true 360 canonical_probability: 0.5 361 abbreviated_probability: 0.2 362 sample_probability: 0.3 363 corner: &esquina 364 canonical: esquina 365 abbreviated: esq 366 sample: true 367 canonical_probability: 0.5 368 abbreviated_probability: 0.2 369 sample_probability: 0.3 370 intersection: 371 default: *y 372 probability: 0.55 373 alternatives: 374 - alternative: *con 375 probability: 0.2 376 - alternative: *en 377 probability: 0.1 378 - alternative: *x 379 probability: 0.075 380 - alternative: *esquina_de 381 probability: 0.05 382 - alternative: *en_la_esquina_de 383 probability: 0.025 384 385 between: 386 canonical: entre 387 sample: true 388 canonical_probability: 0.8 389 sample_probability: 0.2 390 parentheses_probabililty: 0.5 391 392po_boxes: 393 apartado: &apartado 394 canonical: apartado 395 abbreviated: apdo 396 sample: true 397 canonical_probability: 0.5 398 abbreviated_probability: 0.3 399 sample_probability: 0.2 400 numeric: 401 direction: left 402 add_number_phrase: true 403 add_number_phrase_probability: 0.4 # Apdo No 1234 404 numeric_probability: 1.0 405 alphanumeric: 406 sample: false 407 default: *apartado 408 numeric_probability: 0.9 # Apdo 123 409 alpha_probability: 0.05 # Apdo A 410 numeric_plus_alpha_probability: 0.04 # Apdo 123G 411 alpha_plus_numeric_probability: 0.01 # Apdo A123 412 alpha_plus_numeric: 413 whitespace_probability: 0.1 414 numeric_plus_alpha: 415 whitespace_probability: 0.1 416 417 digits: 418 - length: 1 419 probability: 0.05 420 - length: 2 421 probability: 0.1 422 - length: 3 423 probability: 0.2 424 - length: 4 425 probability: 0.5 426 - length: 5 427 probability: 0.1 428 - length: 6 429 probability: 0.05 430 431postcodes: 432 alphanumeric: 433 default: 434 canonical: codigo postal 435 abbreviated: cp 436 sample: true 437 canonical_probability: 0.01 438 abbreviated_probability: 0.95 439 sample_probability: 0.04 440 441 numeric: 442 # Postcodes in Spain and Latin America are sometimes prefixed by CP 443 direction: left 444 445 numeric_affix: 446 affix: cp 447 direction: left 448 # null_probability means the chance of doing nothing e.g. just the postal code 449 null_probability: 0.7 450 numeric_probability: 0.18 451 numeric_affix_probability: 0.12 452 strict_numeric: true 453 454directions: 455 right: &derecha 456 canonical: derecha 457 abbreviated: dcha 458 sample: true 459 canonical_probability: 0.3 460 abbreviated_probability: 0.4 461 sample_probability: 0.3 462 numeric: 463 direction: right 464 numeric_affix: 465 affix: d 466 direction: right 467 whitespace_probability: 0.1 468 numeric_probability: 0.4 469 numeric_affix_probability: 0.6 470 left: &izquierda 471 canonical: izquierda 472 abbreviated: izq 473 sample: true 474 canonical_probability: 0.3 475 abbreviated_probability: 0.4 476 sample_probability: 0.3 477 numeric: 478 direction: right 479 numeric_affix: 480 affix: i 481 direction: right 482 whitespace_probability: 0.1 483 numeric_probability: 0.4 484 numeric_affix_probability: 0.6 485 rear: &trasera 486 canonical: trasera 487 abbreviated: tras 488 sample: true 489 canonical_probability: 0.6 490 abbreviated_probability: 0.2 491 sample_probability: 0.2 492 numeric: 493 direction: right 494 front: &frente 495 canonical: frente 496 abbreviated: fren 497 sample: true 498 canonical_probability: 0.6 499 abbreviated_probability: 0.2 500 sample_probability: 0.2 501 numeric: 502 direction: right 503 alternatives: 504 - alternative: *derecha 505 probability: 0.45 506 - alternative: *izquierda 507 probability: 0.45 508 - alternative: *trasera 509 probability: 0.05 510 - alternative: *frente 511 probability: 0.05 512 513 anteroposterior: 514 alternatives: 515 - alternative: *frente 516 probability: 0.5 517 - alternative: *trasera 518 probability: 0.5 519 520 lateral: 521 alternatives: 522 - alternative: *derecha 523 probability: 0.5 524 - alternative: *izquierda 525 probability: 0.5 526 527cardinal_directions: 528 east: &este 529 canonical: este 530 abbreviated: e 531 canonical_probability: 0.4 532 abbreviated_probability: 0.6 533 numeric: 534 direction: right 535 numeric_affix: 536 affix: e 537 direction: right 538 numeric_probability: 0.5 539 numeric_affix_probability: 0.5 540 541 west: &oeste 542 canonical: oeste 543 abbreviated: w 544 canonical_probability: 0.4 545 abbreviated_probability: 0.6 546 numeric: 547 direction: right 548 numeric_affix: 549 affix: w 550 direction: right 551 numeric_probability: 0.5 552 numeric_affix_probability: 0.5 553 554 north: &norte 555 canonical: norte 556 abbreviated: n 557 canonical_probability: 0.4 558 abbreviated_probability: 0.6 559 numeric: 560 direction: right 561 numeric_affix: 562 affix: n 563 direction: right 564 numeric_probability: 0.5 565 numeric_affix_probability: 0.5 566 567 south: &sur 568 canonical: sur 569 abbreviated: s 570 canonical_probability: 0.4 571 abbreviated_probability: 0.6 572 numeric: 573 direction: right 574 numeric_affix: 575 affix: s 576 direction: right 577 numeric_probability: 0.5 578 numeric_affix_probability: 0.5 579 580 alternatives: 581 - alternative: *norte 582 probability: 0.25 583 - alternative: *este 584 probability: 0.25 585 - alternative: *sur 586 probability: 0.25 587 - alternative: *oeste 588 probability: 0.25 589 590entrances: 591 entrada: &entrada 592 canonical: entrada 593 abbreviated: entr 594 sample: true 595 canonical_probability: 0.5 596 abbreviated_probability: 0.2 597 sample_probability: 0.3 598 numeric: 599 direction: left 600 601 # Entrance 1, Entrance A, etc. 602 alphanumeric: 603 default: *entrada 604 numeric_probability: 0.1 # e.g. Entrance 1 605 alpha_probability: 0.85 # e.g. Entrnace A 606 numeric_plus_alpha_probability: 0.025 # e.g. 1A 607 alpha_plus_numeric_probability: 0.025 # e.g. A1 608 609 alpha_plus_numeric: 610 whitespace_probability: 0.1 611 612 numeric_plus_alpha: 613 whitespace_probability: 0.1 614 615 directional: 616 modifier: 617 alternatives: 618 - alternative: *norte 619 - alternative: *sur 620 - alternative: *este 621 - alternative: *oeste 622 - alternative: *derecha 623 - alternative: *izquierda 624 - alternative: *trasera 625 - alternative: *frente 626 627staircases: 628 escalera: &escalera 629 canonical: escalera 630 abbreviated: esc 631 sample: true 632 canonical_probability: 0.3 633 abbreviated_probability: 0.4 634 sample_probability: 0.3 635 numeric: 636 direction: left 637 638 alphanumeric: 639 # For alphanumerics, Stair A, Stair 1, etc. 640 default: *escalera 641 numeric_probability: 0.6 # e.g. Escalera 1 642 alpha_probability: 0.35 # e.g. Escalera A 643 numeric_plus_alpha_probability: 0.025 # e.g. 1A 644 alpha_plus_numeric_probability: 0.025 # e.g. A1 645 646 alpha_plus_numeric: 647 whitespace_probability: 0.1 648 649 numeric_plus_alpha: 650 whitespace_probability: 0.1 651 652 directional: 653 direction: right # e.g. Escalera Izq 654 direction_probability: 0.8 655 modifier: 656 alternatives: 657 - alternative: *norte 658 - alternative: *sur 659 - alternative: *este 660 - alternative: *oeste 661 - alternative: *derecha 662 - alternative: *izquierda 663 - alternative: *trasera 664 - alternative: *frente 665 666units: 667 apartment: &apartamento 668 canonical: apartamento 669 abbreviated: apto 670 sample: true 671 canonical_probability: 0.3 672 abbreviated_probability: 0.4 673 sample_probability: 0.3 674 numeric: 675 direction: left 676 suite: &suite 677 canonical: suite 678 sample: true 679 canonical_probability: 0.7 680 sample_probability: 0.3 681 numeric: 682 direction: left 683 door: &puerta 684 canonical: puerta 685 abbreviated: pta 686 sample: true 687 canonical_probability: 0.4 688 abbreviated_probability: 0.4 689 sample_probability: 0.2 690 numeric: 691 direction: left 692 ordinal: 693 direction: right 694 gender: f 695 direction_probability: 0.95 # Let it vary occasionally e.g. Puerta 2a 696 numeric_probability: 0.45 697 ordinal_probability: 0.55 698 letra: &letra 699 canonical: letra 700 sample: true 701 canonical_probability: 0.9 702 sample_probability: 0.1 703 numeric: 704 direction: left 705 flat: &departamento 706 canonical: departamento 707 abbreviated: dpto 708 sample: true 709 canonical_probability: 0.2 710 abbreviated_probability: 0.5 711 sample_probability: 0.3 712 numeric: 713 direction: left 714 office: &oficina 715 canonical: oficina 716 abbreviated: of 717 sample: true 718 canonical_probability: 0.4 719 abbreviated_probability: 0.3 720 sample_probability: 0.3 721 numeric: 722 direction: left 723 # By default use departamento when e.g. addr:unit is encountered 724 unit: *departamento 725 # Another word for unit, used more in Colombia 726 unidad: &unidad 727 canonical: unidad 728 abbreviated: un 729 sample: true 730 canonical_probability: 0.4 731 abbreviated_probability: 0.4 732 sample_probability: 0.2 733 lot: &lote 734 canonical: lote 735 abbreviated: lt 736 sample: true 737 plural: 738 canonical: lotes 739 abbreviated: lts 740 canonical_probability: 0.7 741 abbreviated_probability: 0.3 742 canonical_probability: 0.6 743 abbreviated_probability: 0.2 744 sample_probability: 0.2 745 parcel: &parcela 746 canonical: parcela 747 plural: 748 canonical: parcelas 749 casa: &casa 750 canonical: casa 751 plural: 752 canonical: casas 753 numeric: 754 direction: left 755 room: &sala 756 canonical: sala 757 plural: 758 canonical: salas 759 numeric: 760 direction: left 761 762 alphanumeric: &unit_alphanumeric 763 default: *departamento 764 probability: 0.8 765 sample: true 766 alternatives: 767 - alternative: *apartamento 768 probability: 0.1 769 - alternative: *casa 770 probability: 0.05 771 - alternative: *puerta 772 probability: 0.05 773 774 # Separate random probability for adding directions like 2o Izq, 2 Dcha, etc. 775 add_direction: true 776 add_direction_probability: 0.1 777 add_direction_numeric: true # Only for numbers 778 add_direction_standalone: true # A unit can be as simple as "D" 779 780 numeric_probability: 0.9 # e.g. Dpto 1 781 numeric_plus_alpha_probability: 0.01 # e.g. Dpto 1A 782 alpha_plus_numeric_probability: 0.01 # e.g. Dpto A1 783 alpha_probability: 0.08 # e.g. Dpto A 784 alpha_plus_numeric: 785 whitespace_probability: 0.1 786 numeric_plus_alpha: 787 whitespace_probability: 0.1 788 789 alpha: 790 default: *departamento 791 probability: 0.8 792 alternatives: 793 - alternative: *apartamento 794 probability: 0.1 795 - alternative: *casa 796 probability: 0.03 797 - alternative: *puerta 798 probability: 0.05 799 - alternative: *letra 800 probability: 0.02 801 802 alphanumeric_apartamento: &unit_alphanumeric_apartamento 803 <<: *unit_alphanumeric 804 default: *apartamento 805 probability: 0.8 806 alternatives: 807 - alternative: *departamento 808 probability: 0.1 809 - alternative: *casa 810 probability: 0.05 811 - alternative: *puerta 812 probability: 0.05 813 814 alpha: 815 default: *apartamento 816 probability: 0.8 817 alternatives: 818 - alternative: *departamento 819 probability: 0.1 820 - alternative: *casa 821 probability: 0.03 822 - alternative: *puerta 823 probability: 0.05 824 - alternative: *letra 825 probability: 0.02 826 827 alphanumeric_apartamento_exclusive: &unit_alphanumeric_apartamento_exclusive 828 <<: *unit_alphanumeric 829 default: *apartamento 830 probability: 0.9 831 alternatives: 832 - alternative: *casa 833 probability: 0.05 834 - alternative: *puerta 835 probability: 0.05 836 837 alpha: 838 default: *apartamento 839 probability: 0.9 840 alternatives: 841 - alternative: *casa 842 probability: 0.03 843 - alternative: *puerta 844 probability: 0.05 845 - alternative: *letra 846 probability: 0.02 847 848 849 zones: 850 residential: *unit_alphanumeric 851 commercial: 852 default: *oficina 853 probability: 0.8 854 alternatives: 855 - alternative: *suite 856 probability: 0.2 857 858 numeric_probability: 0.9 # e.g. Oficina 1 859 numeric_plus_alpha_probability: 0.01 # e.g. Oficina 1A 860 alpha_plus_numeric_probability: 0.01 # e.g. Oficina A1 861 alpha_probability: 0.08 # e.g. Oficina A 862 alpha_plus_numeric: 863 whitespace_probability: 0.1 864 numeric_plus_alpha: 865 whitespace_probability: 0.1 866 867 alpha: 868 default: *oficina 869 probability: 0.8 870 alternatives: 871 - alternative: *suite 872 probability: 0.15 873 - alternative: *letra 874 probability: 0.05 875 876 industrial: 877 default: *lote 878 probability: 0.5 879 alternatives: 880 - alternative: *oficina 881 probability: 0.3 882 - alternative: *unidad 883 probability: 0.19 884 - alternative: *parcela 885 probability: 0.01 886 887 numeric_probability: 0.9 # e.g. Lote 1 888 numeric_plus_alpha_probability: 0.01 # e.g. Lote 1A 889 alpha_plus_numeric_probability: 0.01 # e.g. Lote A1 890 alpha_probability: 0.08 # e.g. Lote A 891 alpha_plus_numeric: 892 whitespace_probability: 0.1 893 numeric_plus_alpha: 894 whitespace_probability: 0.1 895 896 university: 897 default: *sala 898 probability: 0.9 899 alternatives: 900 - alternative: *puerta 901 probability: 0.1 902 903 numeric_probability: 0.9 # e.g. Sala 1 904 numeric_plus_alpha_probability: 0.01 # e.g. Sala 1A 905 alpha_plus_numeric_probability: 0.01 # e.g. Sala A1 906 alpha_probability: 0.08 # e.g. Sala A 907 alpha_plus_numeric: 908 whitespace_probability: 0.1 909 numeric_plus_alpha: 910 whitespace_probability: 0.1 911 912 alpha: 913 default: *sala 914 probability: 0.9 915 alternatives: 916 - alternative: *puerta 917 probability: 0.08 918 - alternative: *letra 919 probability: 0.02 920 921 allotments: 922 lot: 923 default: *lote 924 numeric_probability: 0.8 925 alphanumeric_probability: 0.1 926 alpha_probability: 0.1 927 parcel: 928 default: *parcela 929 numeric_probability: 0.3 930 alphanumeric_probability: 0.3 931 alpha_probability: 0.4 932 lot_probability: 0.9 933 parcel_probability: 0.06 934 lot_plus_parcel_probability: 0.02 935 parcel_plus_lot_probability: 0.02 936 937 938countries: 939 # España / Spain 940 es: 941 components: 942 staircase: 943 null_probability: 0.97 944 alphanumeric_probability: 0.03 945 levels: 946 planta: &planta 947 # Everywhere except Spain 948 canonical: planta 949 abbreviated: pl 950 sample: true 951 # Numeric version e.g. Planta 1 952 numeric: 953 direction: left 954 add_number_phrase: true 955 add_number_phrase_probability: 0.05 956 numeric_affix: 957 affix: p 958 direction: left 959 # Ordinal como 2a planta 960 ordinal: 961 direction: right 962 gender: f 963 digits: 964 ascii_probability: 0.8 965 spellout_probability: 0.2 966 canonical_probability: 0.6 967 abbreviated_probability: 0.2 968 sample_probability: 0.2 969 numeric_probability: 0.475 970 numeric_affix_probability: 0.05 971 ordinal_probability: 0.475 972 baja: &baja 973 canonical: baja 974 abbreviated: bja 975 canonical_probability: 0.8 976 abbreviated_probability: 0.2 977 planta_baja: &planta_baja 978 canonical: planta baja 979 abbreviated: pb 980 prefer_abbreviated: true 981 sample: true 982 entreplanta: &entreplanta 983 canonical: entreplanta 984 principal: &planta_principal 985 canonical: planta principal 986 abbreviated: ppl 987 sample: true 988 aliases: 989 "0": 990 default: *bajos 991 probability: 0.4 992 alternatives: 993 - alternative: *planta_baja 994 probability: 0.2 995 - alternative: *baja 996 probability: 0.09 997 - alternative: *bajo 998 probability: 0.09 999 - alternative: *piso_bajo 1000 probability: 0.2 1001 # Planta/Piso 0 is uncommon 1002 - alternative: *planta 1003 probability: 0.01 1004 - alternative: *piso 1005 probability: 0.01 1006 "1": 1007 default: *planta 1008 probability: 0.4 1009 alternatives: 1010 - alternative: *entresuelo 1011 probability: 0.2 1012 - alternative: *piso 1013 probability: 0.4 1014 "2": 1015 default: *planta 1016 probability: 0.4 1017 alternatives: 1018 - alternative: *planta_principal 1019 probability: 0.1 1020 - alternative: *principal 1021 probability: 0.1 1022 - alternative: *piso 1023 probability: 0.4 1024 top: 1025 default: *planta 1026 probability: 0.425 1027 alternatives: 1028 - alternative: *piso 1029 probability: 0.425 1030 - alternative: *atico 1031 probability: 0.1 1032 - alternative: *sobreatico 1033 probability: 0.05 1034 1035 alphanumeric: 1036 default: *planta 1037 probability: 0.5 1038 alternatives: 1039 - alternative: *piso 1040 probability: 0.5 1041 numeric_probability: 0.99 # With this probability, pick an integer 1042 alpha_probability: 0.01 # With this probability, pick a letter e.g. Floor A 1043 1044 units: 1045 door: &puerta_espana 1046 <<: *puerta 1047 numeric: 1048 direction: left 1049 # If it's just puerta B, many times it's just e.g. 3o B for "tercero piso puerta B" 1050 null_phrase_probability: 0.15 1051 null_phrase_alpha_only: true 1052 ordinal: 1053 direction: right 1054 gender: f 1055 direction_probability: 0.95 1056 null_phrase_probability: 0.8 # Let e.g. 5a be the entire unit string 1057 # These sum to 1 1058 numeric_probability: 0.25 1059 ordinal_probability: 0.75 1060 1061 alphanumeric: &unit_alphanumeric_puerta 1062 <<: *unit_alphanumeric 1063 default: *puerta_espana 1064 probability: 0.8 1065 alternatives: 1066 - alternative: *apartamento 1067 probability: 0.1 1068 - alternative: *casa 1069 probability: 0.1 1070 1071 numeric_probability: 0.7 # e.g. Puerta 1a 1072 numeric_plus_alpha_probability: 0.01 # e.g. Puerta 1A 1073 alpha_plus_numeric_probability: 0.01 # e.g. Puerta A1 1074 alpha_probability: 0.28 # e.g. Puerta A 1075 1076 alpha: 1077 default: *puerta_espana 1078 probability: 0.8 1079 alternatives: 1080 - alternative: *letra 1081 probability: 0.15 1082 - alternative: *apartamento 1083 probability: 0.04 1084 - alternative: *casa 1085 probability: 0.01 1086 1087 zones: 1088 residential: *unit_alphanumeric_puerta 1089 1090 # Argentina 1091 ar: 1092 po_boxes: 1093 alphanumeric: 1094 sample: false 1095 default: 1096 canonical: casilla de correos 1097 abbreviated: c.c. 1098 sample: true 1099 canonical_probability: 0.1 1100 abbreviated_probability: 0.8 1101 sample_probability: 0.1 1102 numeric: 1103 direction: left 1104 1105 # Chile 1106 cl: 1107 po_boxes: 1108 alphanumeric: 1109 sample: false 1110 default: 1111 canonical: casilla de correos 1112 abbreviated: c.c. 1113 sample: true 1114 canonical_probability: 0.1 1115 abbreviated_probability: 0.8 1116 sample_probability: 0.1 1117 numeric: 1118 direction: left 1119 1120 # Colombia 1121 co: 1122 numbers: 1123 default: 1124 <<: *numero 1125 abbreviated: "no" 1126 1127 house_numbers: 1128 alphanumeric_phrase_probability: 0.4 1129 1130 units: 1131 alphanumeric: *unit_alphanumeric_apartamento_exclusive 1132 1133 zones: 1134 residential: *unit_alphanumeric_apartamento_exclusive 1135 1136 1137 # Dominican Republic 1138 dr: 1139 units: 1140 alphanumeric: *unit_alphanumeric_apartamento_exclusive 1141 1142 zones: 1143 residential: *unit_alphanumeric_apartamento_exclusive 1144 1145 # México 1146 mx: 1147 # México uses the North American convention of starting floors at 1 1148 levels: 1149 numbering_starts_at: 1 1150 1151 # Panamá 1152 pa: 1153 units: 1154 alphanumeric: *unit_alphanumeric_apartamento_exclusive 1155 1156 zones: 1157 residential: *unit_alphanumeric_apartamento_exclusive 1158 1159 # Puerto Rico 1160 pr: 1161 units: 1162 alphanumeric: *unit_alphanumeric_apartamento_exclusive 1163 1164 zones: 1165 residential: *unit_alphanumeric_apartamento_exclusive 1166 1167 # United States - libpostal recognizes Spanish in the US, even if the US does not 1168 us: 1169 units: 1170 alphanumeric: *unit_alphanumeric_apartamento_exclusive 1171 1172 zones: 1173 residential: *unit_alphanumeric_apartamento_exclusive 1174 1175 # Venezuela 1176 ve: 1177 units: 1178 alphanumeric: *unit_alphanumeric_apartamento_exclusive 1179 1180 zones: 1181 residential: *unit_alphanumeric_apartamento_exclusive 1182 1183 # Uruguay 1184 uy: 1185 units: 1186 alphanumeric: *unit_alphanumeric_apartamento 1187 1188 zones: 1189 residential: *unit_alphanumeric_apartamento 1190