# Last edited on 2000-10-11 07:02:52 by stolfi
Obtaining component statistics (simple words only), without braces:
set statargs = ( \
tc-y/prefix.pm \
tc-y/core.c \
tc-y/suffix.ns \
tc-y/crustmantle.pmns \
\
tf-z/precrust.p \
tf-z/premantle.m \
tf-z/sufmantle.n \
tf-z/sufcrust.s \
tf-z/mantle.mn \
tf-z/crust.ps \
)
foreach sec ( ${secs} txt.n lab.n )
echo "${sec}"
foreach dpt ( ${statargs} )
set tagpair = "${dpt:h}"
set pt = "${dpt:t}"
set part = "${pt:r}"
set seltype = "${pt:e}"
gather-stats ${sec} ${tagpair} ${seltype} ${part}
end
end
Listing it all:
foreach sec ( "hea.1" "bio.1" "str.2" "txt.n" "lab.n" )
if ( ${sec} == "txt.n" ) then
set wdtype = "text"; set xsec = "tot";
set wspfile = ( )
else if ( ${sec} == "lab.n" ) then
set wdtype = "labs"; set xsec = "tot"
set wspfile = ( )
else
set wdtype = "text"; set xsec = "${sec}";
set wspfile = ( "data/gud/text/${xsec}.wsp" )
endif
dicio-wc \
${wspfile} \
stats/${wdtype}/{tc-y,tf-z,tw-w}/${xsec}.frq \
prob/obs/${sec}/*.frq
end
dicio-wc prob/obs/*/word.frq
lines words bytes file
------- ------- --------- ------------
6703 6703 103030 data/gud/text/hea.1.wsp
1039 3117 27061 stats/text/tc-y/hea.1.frq
608 1824 14230 stats/text/tf-z/hea.1.frq
1833 5499 54265 stats/text/tw-w/hea.1.frq
22 66 430 prob/obs/hea.1/core.frq
305 915 6511 prob/obs/hea.1/crust.frq
695 2085 15369 prob/obs/hea.1/crustmantle.frq
20 60 421 prob/obs/hea.1/mantle.frq
60 180 1167 prob/obs/hea.1/precrust.frq
68 204 1354 prob/obs/hea.1/prefix.frq
15 45 305 prob/obs/hea.1/premantle.frq
187 561 3888 prob/obs/hea.1/sufcrust.frq
254 762 5426 prob/obs/hea.1/suffix.frq
13 39 262 prob/obs/hea.1/sufmantle.frq
1980 5940 45165 prob/obs/hea.1/word.frq
lines words bytes file
------- ------- --------- ------------
6555 6555 104974 data/gud/text/bio.1.wsp
744 2232 19346 stats/text/tc-y/bio.1.frq
421 1263 9719 stats/text/tf-z/bio.1.frq
1240 3720 36567 stats/text/tw-w/bio.1.frq
14 42 271 prob/obs/bio.1/core.frq
218 654 4594 prob/obs/bio.1/crust.frq
529 1587 11690 prob/obs/bio.1/crustmantle.frq
12 36 246 prob/obs/bio.1/mantle.frq
50 150 968 prob/obs/bio.1/precrust.frq
67 201 1335 prob/obs/bio.1/prefix.frq
13 39 261 prob/obs/bio.1/premantle.frq
106 318 2153 prob/obs/bio.1/sufcrust.frq
134 402 2777 prob/obs/bio.1/suffix.frq
10 30 198 prob/obs/bio.1/sufmantle.frq
1325 3975 30134 prob/obs/bio.1/word.frq
lines words bytes file
------- ------- --------- ------------
10097 10097 170197 data/gud/text/str.2.wsp
1451 4353 38269 stats/text/tc-y/str.2.frq
786 2358 18476 stats/text/tf-z/str.2.frq
2533 7599 75981 stats/text/tw-w/str.2.frq
22 66 430 prob/obs/str.2/core.frq
394 1182 8428 prob/obs/str.2/crust.frq
974 2922 21800 prob/obs/str.2/crustmantle.frq
24 72 510 prob/obs/str.2/mantle.frq
66 198 1282 prob/obs/str.2/precrust.frq
76 228 1515 prob/obs/str.2/prefix.frq
19 57 389 prob/obs/str.2/premantle.frq
253 759 5297 prob/obs/str.2/sufcrust.frq
379 1137 8248 prob/obs/str.2/suffix.frq
20 60 417 prob/obs/str.2/sufmantle.frq
2779 8337 64286 prob/obs/str.2/word.frq
lines words bytes file
------- ------- --------- ------------
2996 8988 79819 stats/text/tc-y/tot.frq
1561 4683 37317 stats/text/tf-z/tot.frq
5820 17460 175716 stats/text/tw-w/tot.frq
35 105 691 prob/obs/txt.n/core.frq
881 2643 19124 prob/obs/txt.n/crust.frq
2146 6438 48366 prob/obs/txt.n/crustmantle.frq
36 108 773 prob/obs/txt.n/mantle.frq
136 408 2700 prob/obs/txt.n/precrust.frq
178 534 3645 prob/obs/txt.n/prefix.frq
27 81 564 prob/obs/txt.n/premantle.frq
475 1425 10030 prob/obs/txt.n/sufcrust.frq
698 2094 15203 prob/obs/txt.n/suffix.frq
26 78 546 prob/obs/txt.n/sufmantle.frq
6604 19812 153703 prob/obs/txt.n/word.frq
lines words bytes file
------- ------- --------- ------------
519 1557 13325 stats/labs/tc-y/tot.frq
420 1260 9893 stats/labs/tf-z/tot.frq
671 2013 19655 stats/labs/tw-w/tot.frq
22 66 430 prob/obs/lab.n/core.frq
181 543 3949 prob/obs/lab.n/crust.frq
281 843 6253 prob/obs/lab.n/crustmantle.frq
12 36 246 prob/obs/lab.n/mantle.frq
44 132 854 prob/obs/lab.n/precrust.frq
40 120 785 prob/obs/lab.n/prefix.frq
9 27 179 prob/obs/lab.n/premantle.frq
158 474 3335 prob/obs/lab.n/sufcrust.frq
193 579 4152 prob/obs/lab.n/suffix.frq
13 39 263 prob/obs/lab.n/sufmantle.frq
721 2163 16460 prob/obs/lab.n/word.frq
TRIVIAL GRAMMARS
Creating "trivial" (straight enumeration) grammars:
set parts = ( core premantle sufmantle mantle precrust sufcrust crust prefix suffix word )
foreach sec ( hea.1 bio.1 str.2 lab.n txt.n )
foreach part ( ${parts} )
set gram = "gram/trivial/${sec}/${part}.grx"
echo ${gram}
mkdir -p ${gram:h}
echo '# Last edited on DATE TIME by USER' > ${gram}
echo '# Trivial grammar for '"${part}" >> ${gram}
echo "${part}"':' | tr 'a-z' 'A-Z' >> ${gram}
cat prob/obs/${sec}/${part}.frq >> ${gram}
end
end
Trivial check of the enumeration procedure:
set sec = "bio.1"
set class = "trivial"
mkdir prob/gen/${class}
mkdir prob/cmp/${class}
foreach part ( ${parts} )
echo ${class}/${sec}/${part}
set grxfile = "gram/${class}/${sec}/${part}.grx"
set genfile = "prob/gen/${class}/${sec}/${part}.prb"
cat ${grxfile} \
| enum-language \
> ${genfile}
compare-probs ${class} ${sec} ${part} ${part}
end