source:
src/compdic/compdic-fst-to-bin
@
d2f119e
| Last change on this file since d2f119e was 555c7f8, checked in by Tomasz Obrebski <to@…>, 13 years ago | |
|---|---|
|
|
| File size: 1.1 KB | |
| Rev | Line | |
|---|---|---|
| [555c7f8] | 1 | #!/bin/bash |
| [f4bf33e] | 2 | |
| 3 | if [ $# -lt 2 ] | |
| 4 | then | |
| 5 | echo "Usage:" | |
| 6 | echo " compdic-fst-to-bin <fstautomaton> <binautomaton>" | |
| 7 | echo "where" | |
| 8 | echo " <fstautomaton> - file containig automaton in openfst format" | |
| 9 | echo " <binautomaton> - a file to which the compiled binary automaton (lem/cor/kor format) shoul be written" | |
| 10 | exit 0 | |
| 11 | fi | |
| 12 | ||
| 13 | tempdir=`mktemp -d /tmp/compdic.XXXXXX` | |
| 14 | ||
| 15 | alphabet=`tempfile -d $tempdir` | |
| 16 | ||
| 17 | cat <<EOF > $alphabet | |
| 18 | <eps> 0 | |
| 19 | a 1 | |
| 20 | A 2 | |
| 21 | ä 3 | |
| 22 | ± 4 | |
| 23 | ¡ 5 | |
| 24 | b 6 | |
| 25 | B 7 | |
| 26 | c 8 | |
| 27 | C 9 | |
| 28 | æ 10 | |
| 29 | Æ 11 | |
| 30 | d 12 | |
| 31 | D 13 | |
| 32 | e 14 | |
| 33 | E 15 | |
| 34 | é 16 | |
| 35 | ê 17 | |
| 36 | Ê 18 | |
| 37 | f 19 | |
| 38 | F 20 | |
| 39 | g 21 | |
| 40 | G 22 | |
| 41 | h 23 | |
| 42 | H 24 | |
| 43 | i 25 | |
| 44 | I 26 | |
| 45 | j 27 | |
| 46 | J 28 | |
| 47 | k 29 | |
| 48 | K 30 | |
| 49 | l 31 | |
| 50 | L 32 | |
| 51 | ³ 33 | |
| 52 | £ 34 | |
| 53 | m 35 | |
| 54 | M 36 | |
| 55 | n 37 | |
| 56 | N 38 | |
| 57 | ñ 39 | |
| 58 | Ñ 40 | |
| 59 | o 41 | |
| 60 | O 42 | |
| 61 | ö 43 | |
| 62 | ó 44 | |
| 63 | Ó 45 | |
| 64 | p 46 | |
| 65 | P 47 | |
| 66 | q 48 | |
| 67 | Q 49 | |
| 68 | r 50 | |
| 69 | R 51 | |
| 70 | s 52 | |
| 71 | S 53 | |
| 72 | ¶ 54 | |
| 73 | Š 55 | |
| 74 | t 56 | |
| 75 | T 57 | |
| 76 | u 58 | |
| 77 | U 59 | |
| 78 | ü 60 | |
| 79 | v 61 | |
| 80 | V 62 | |
| 81 | w 63 | |
| 82 | W 64 | |
| 83 | x 65 | |
| 84 | X 66 | |
| 85 | y 67 | |
| 86 | Y 68 | |
| 87 | z 69 | |
| 88 | Z 70 | |
| 89 | Œ 71 | |
| 90 | ¬ 72 | |
| 91 | ¿ 73 | |
| 92 | ¯ 74 | |
| 93 | 0 75 | |
| 94 | 1 76 | |
| 95 | 2 77 | |
| 96 | 3 78 | |
| 97 | 4 79 | |
| 98 | 5 80 | |
| 99 | 6 81 | |
| 100 | 7 82 | |
| 101 | 8 83 | |
| 102 | 9 84 | |
| 103 | _ 85 | |
| 104 | - 86 | |
| 105 | ? 87 | |
| 106 | ! 88 | |
| 107 | ~ 89 | |
| 108 | ; 90 | |
| 109 | , 91 | |
| 110 | / 92 | |
| 111 | * 93 | |
| 112 | + 94 | |
| 113 | EOF | |
| 114 | ||
| 115 | cat $1 | fstrmepsilon | fstdeterminize | fstminimize | fstprint --acceptor --isymbols=$alphabet | fsm2aut | aut2fsa > $2 | |
| 116 | rm -r $tempdir | |
| 117 | ||
| 118 | #echo generating cats file ... | |
| 119 | ||
| 120 | #cat $1 | cut -d ',' -f 2 | sort -u $2.cats |
Note: See TracBrowser
for help on using the repository browser.