| 1 |
dpavlin |
90 |
#!/usr/bin/perl -w |
| 2 |
ulpfr |
19 |
# -*- Mode: Perl -*- |
| 3 |
|
|
# $Basename: sman $ |
| 4 |
|
|
# $Revision: 1.14 $ |
| 5 |
|
|
# Author : Ulrich Pfeifer |
| 6 |
|
|
# Created On : Fri Aug 30 15:52:25 1996 |
| 7 |
|
|
# Last Modified By: Ulrich Pfeifer |
| 8 |
|
|
# Last Modified On: Mon May 8 11:03:46 2000 |
| 9 |
|
|
# Language : CPerl |
| 10 |
|
|
# |
| 11 |
|
|
# (C) Copyright 1996-2000, Ulrich Pfeifer |
| 12 |
|
|
# |
| 13 |
ulpfr |
10 |
|
| 14 |
|
|
use strict; |
| 15 |
|
|
|
| 16 |
|
|
use Term::ReadLine; |
| 17 |
|
|
use Getopt::Long; |
| 18 |
laperla |
65 |
use Fcntl qw(O_RDONLY); |
| 19 |
ulpfr |
10 |
use Config; |
| 20 |
|
|
|
| 21 |
dpavlin |
90 |
use lib '/data/wait/lib'; |
| 22 |
dpavlin |
86 |
|
| 23 |
ulpfr |
10 |
require WAIT::Config; |
| 24 |
|
|
require WAIT::Database; |
| 25 |
|
|
require WAIT::Query::Base; |
| 26 |
|
|
require WAIT::Query::Wais; |
| 27 |
|
|
|
| 28 |
|
|
|
| 29 |
|
|
$SIG{PIPE} = 'IGNORE'; |
| 30 |
|
|
my %OPT = (database => 'DB', |
| 31 |
|
|
dir => $WAIT::Config->{WAIT_home} || '/tmp', |
| 32 |
|
|
table => 'man', |
| 33 |
|
|
pager => $WAIT::Config->{'pager'} || 'more', |
| 34 |
|
|
filter => 0, |
| 35 |
|
|
max => 15, |
| 36 |
|
|
); |
| 37 |
|
|
|
| 38 |
|
|
GetOptions(\%OPT, |
| 39 |
|
|
'database=s', |
| 40 |
|
|
'dir=s', |
| 41 |
|
|
'table=s', |
| 42 |
|
|
'filter=i', |
| 43 |
|
|
'max=i', |
| 44 |
laperla |
65 |
'pager:s') || die " |
| 45 |
|
|
Usage: $0 |
| 46 |
|
|
[--database database] |
| 47 |
|
|
[--dir dir ] |
| 48 |
|
|
[--table table ] |
| 49 |
|
|
[--filter integer ] |
| 50 |
|
|
[--max integer ] |
| 51 |
|
|
[--pager pager ] |
| 52 |
|
|
"; |
| 53 |
ulpfr |
10 |
|
| 54 |
|
|
my $db = WAIT::Database->open(name => $OPT{database}, |
| 55 |
|
|
mode => O_RDONLY, |
| 56 |
|
|
directory => $OPT{dir}) |
| 57 |
|
|
or die "Could not open database $OPT{database}: $@"; |
| 58 |
|
|
|
| 59 |
|
|
my $tb = $db->table(name => $OPT{table}) |
| 60 |
|
|
or die "Could not open table $OPT{table}: $@"; |
| 61 |
|
|
|
| 62 |
|
|
# not used: my $layout = $tb->layout; # a WAIT::Parse::Nroff object |
| 63 |
|
|
|
| 64 |
|
|
my $term = new Term::ReadLine 'Simple Query Interface'; |
| 65 |
|
|
|
| 66 |
|
|
require WAIT::Format::Term; |
| 67 |
|
|
my $format; |
| 68 |
|
|
if ($Config::Config{'archname'} eq 'i586-linux') { |
| 69 |
|
|
# for color xterm |
| 70 |
|
|
$format = new WAIT::Format::Term query_s => "[01;31m", query_e => "[00m"; |
| 71 |
|
|
} else { |
| 72 |
|
|
$format = new WAIT::Format::Term; |
| 73 |
|
|
} |
| 74 |
|
|
|
| 75 |
|
|
my $pager = ($OPT{pager}) ? \&pager : \&less; |
| 76 |
dpavlin |
91 |
$pager = \&less; # force pager usage |
| 77 |
ulpfr |
10 |
my $OUT = $term->OUT; |
| 78 |
|
|
|
| 79 |
|
|
my $st = 1; |
| 80 |
|
|
print $OUT "Enter 'h' for help.\n"; |
| 81 |
|
|
|
| 82 |
|
|
# sman is basically offering three services: find the hits and show |
| 83 |
|
|
# them (a query), show metadata for a hit (a view), show a hot (display) |
| 84 |
|
|
|
| 85 |
|
|
my($query, @did); |
| 86 |
|
|
|
| 87 |
|
|
while (defined ($_ = &myreadline("$st> "))) { |
| 88 |
|
|
chomp; $st++; |
| 89 |
|
|
|
| 90 |
|
|
my(%hits, $query_text); |
| 91 |
|
|
if (/^$/) { |
| 92 |
|
|
next; |
| 93 |
|
|
} elsif (/^m (\d+)$/) { |
| 94 |
|
|
$OPT{max} = $1; |
| 95 |
|
|
} elsif (/^f\s*(\d+)?$/) { |
| 96 |
|
|
$OPT{filter} = $1; |
| 97 |
|
|
next; |
| 98 |
|
|
} elsif (/^t$/i) { |
| 99 |
|
|
if ($pager eq \&less) { |
| 100 |
|
|
$pager = \&pager; |
| 101 |
|
|
} else { |
| 102 |
|
|
$pager = \&less; |
| 103 |
|
|
} |
| 104 |
|
|
next; |
| 105 |
|
|
} elsif (/^(\d+)$/) { |
| 106 |
|
|
if (defined $did[$1]) { |
| 107 |
|
|
display($did[$1]); # <----------- display (full doc) |
| 108 |
|
|
next; |
| 109 |
|
|
} |
| 110 |
|
|
} elsif (/^d\s*(\d+)/) { |
| 111 |
|
|
if (defined $did[$1]) { |
| 112 |
|
|
view($did[$1]); # <----------- view (metadata from WAIT) |
| 113 |
|
|
next; |
| 114 |
|
|
} |
| 115 |
|
|
} elsif (/^q$/i) { |
| 116 |
|
|
last; |
| 117 |
|
|
} elsif (/^l$/i) { |
| 118 |
|
|
# fall through |
| 119 |
|
|
} elsif (/^[h?]$/i) { |
| 120 |
|
|
help(); |
| 121 |
|
|
next; |
| 122 |
|
|
} elsif (/^hh$/i) { |
| 123 |
|
|
extended_help(); |
| 124 |
|
|
next; |
| 125 |
|
|
} else { # <----------- A query (Display a list) |
| 126 |
|
|
$query_text = $_; |
| 127 |
|
|
eval {$query = WAIT::Query::Wais::query($tb, $_)}; |
| 128 |
|
|
if ($@ ne '') { |
| 129 |
|
|
print $OUT "$_ => $query\n\$\@='$@'\n"; |
| 130 |
|
|
} elsif (ref($query)) { |
| 131 |
ulpfr |
19 |
%hits = $query->execute(top => $OPT{max}, picky => 1); |
| 132 |
ulpfr |
10 |
# the hash %hits has as keys document numbers and as values |
| 133 |
|
|
# quality figures. The doc numbers are not what we have as docid |
| 134 |
|
|
# to find the item in the access class, they are WAIT's private |
| 135 |
|
|
# numbers. |
| 136 |
|
|
} else { |
| 137 |
|
|
next; |
| 138 |
|
|
} |
| 139 |
|
|
} |
| 140 |
|
|
|
| 141 |
|
|
next unless %hits; |
| 142 |
|
|
my $no = 1; # numbering the hits for the result table that is |
| 143 |
|
|
# presented to the user |
| 144 |
|
|
|
| 145 |
|
|
@did = (); # store the internal numbers (keys of %hits). The user |
| 146 |
|
|
# will use $no in sman's interface to select a hit. |
| 147 |
|
|
|
| 148 |
|
|
# the following loop uses the values of %hits to sort the results |
| 149 |
|
|
# according to the quality and cut after a number of rows. After |
| 150 |
|
|
# that %hits isn't needed anymore. |
| 151 |
|
|
print "Query: $query_text\n"; |
| 152 |
|
|
for my $did (sort {$hits{$b} <=> $hits{$a}} keys %hits) { |
| 153 |
|
|
|
| 154 |
|
|
my %tattr = $tb->fetch($did); |
| 155 |
|
|
# the hash %tattr contains several attributes of the item we are |
| 156 |
|
|
# referring to, namely the attributes that we named in the "attr" |
| 157 |
|
|
# argument of the create_table statement in smakewhatis |
| 158 |
|
|
|
| 159 |
|
|
printf $OUT "%2d %6.3f %s\n", $no, $hits{$did}, |
| 160 |
|
|
substr($tattr{headline} ||'',0,68); |
| 161 |
|
|
$did[$no] = $did; |
| 162 |
|
|
last if $no++ >= $OPT{max}; |
| 163 |
|
|
|
| 164 |
|
|
} |
| 165 |
|
|
|
| 166 |
|
|
} continue { |
| 167 |
ulpfr |
19 |
# we don't do this since Andreas Koenig does not think of it as feature |
| 168 |
ulpfr |
10 |
# $term->SetHistory(grep length($_)>4, $term->GetHistory) |
| 169 |
|
|
} |
| 170 |
ulpfr |
19 |
warn "Thank you for using sman\n"; |
| 171 |
ulpfr |
10 |
|
| 172 |
ulpfr |
19 |
$tb->close; |
| 173 |
|
|
$db->close; |
| 174 |
|
|
|
| 175 |
ulpfr |
10 |
sub myreadline { |
| 176 |
|
|
if (@ARGV) { |
| 177 |
|
|
return shift @ARGV; |
| 178 |
|
|
} else { |
| 179 |
|
|
$term->readline(@_); |
| 180 |
|
|
} |
| 181 |
|
|
} |
| 182 |
|
|
sub help { |
| 183 |
|
|
my $idb = "\n\t'". join(q[', '], $tb->fields()) . "'"; |
| 184 |
|
|
print $OUT qq[Available commands: |
| 185 |
|
|
|
| 186 |
|
|
<num> Show the document <num> |
| 187 |
|
|
d <num> Show the db entry of document <num> |
| 188 |
|
|
f <num> Display only <num> lines context |
| 189 |
|
|
h,? Display this help message |
| 190 |
|
|
hh Display query examples |
| 191 |
|
|
m <num> Set maxhits to <num> |
| 192 |
|
|
t Toggle display mode (term/less) |
| 193 |
|
|
q Exit from $0 |
| 194 |
|
|
l redisplay last ranking |
| 195 |
|
|
Other input is tried as wais query. |
| 196 |
dpavlin |
91 |
The following fields for table '$OPT{table}' are known: $idb |
| 197 |
ulpfr |
10 |
] ; |
| 198 |
|
|
} |
| 199 |
|
|
|
| 200 |
|
|
sub extended_help { |
| 201 |
|
|
print q{ |
| 202 |
|
|
Here are some query examples: |
| 203 |
|
|
|
| 204 |
|
|
information retrieval free text query |
| 205 |
|
|
information or retrieval same as above |
| 206 |
|
|
des=information retrieval `information' must be in the description |
| 207 |
|
|
des=(information retrieval) one of them in description |
| 208 |
|
|
des=(information or retrieval) same as above |
| 209 |
|
|
des=(information and retrieval) both of them in description |
| 210 |
|
|
des=(information not retrieval) `information' in description and |
| 211 |
|
|
`retrieval' not in description |
| 212 |
|
|
des=(information system*) wild-card search |
| 213 |
|
|
au=ilia author names may be misspelled |
| 214 |
|
|
|
| 215 |
|
|
You can build arbitary boolean combination of the above examples. |
| 216 |
|
|
Field names may be abbreviated. |
| 217 |
|
|
} |
| 218 |
|
|
} |
| 219 |
|
|
|
| 220 |
|
|
sub view { |
| 221 |
|
|
my $did = shift; |
| 222 |
|
|
my %tattr = $tb->fetch($did); |
| 223 |
|
|
for (keys %tattr) { |
| 224 |
|
|
print $OUT "$_ $tattr{$_}\n"; |
| 225 |
|
|
} |
| 226 |
|
|
} |
| 227 |
|
|
|
| 228 |
|
|
sub display { |
| 229 |
|
|
my $did = shift; |
| 230 |
|
|
|
| 231 |
|
|
return unless defined $query and defined $did; |
| 232 |
|
|
|
| 233 |
|
|
print $OUT "Wais display document $did\n"; |
| 234 |
|
|
my %tattr = $tb->fetch($did); |
| 235 |
|
|
my $tdid = $tattr{docid}; |
| 236 |
|
|
# WHAT DOES HE DO HERE? ULI??? |
| 237 |
|
|
# Re: some indexing scripts did use pathnames relative to the table directory |
| 238 |
|
|
# especially the cpanwait script does this. uli |
| 239 |
dpavlin |
86 |
if ($tdid !~ m(^/)) { |
| 240 |
|
|
$tdid = $tb->dir . '/' . $tdid; |
| 241 |
|
|
} |
| 242 |
ulpfr |
10 |
|
| 243 |
|
|
# The main task of all that follows from here is highlighting. WAIT |
| 244 |
|
|
# is designed to make it possible to show the user why a certain |
| 245 |
|
|
# document was chosen by the indexer. |
| 246 |
|
|
|
| 247 |
|
|
my $buf = $tb->fetch_extern($tdid); |
| 248 |
|
|
# This $buf can be an object that can have enough information to do |
| 249 |
|
|
# highlighting without WAIT's help. If you prefer to implement your |
| 250 |
|
|
# own highlighting, you can do so now with e.g. print |
| 251 |
|
|
# $buf->highlight(query => $query) |
| 252 |
|
|
|
| 253 |
|
|
# All you need to know to implement highlighting is how a |
| 254 |
|
|
# WAIT::Query::Base object looks like (left as an exercise for the |
| 255 |
|
|
# reader). |
| 256 |
|
|
|
| 257 |
|
|
# The impatient reader may want to implement something without |
| 258 |
|
|
# highlighting, in which case he does not need any info about the |
| 259 |
|
|
# query object and can rightaway run e.g. |
| 260 |
|
|
# print $buf->as_string |
| 261 |
|
|
|
| 262 |
|
|
# Thus the impatient reader does not necessarily need the following |
| 263 |
|
|
# heavy wizardry. Just to give you an idea what's going on: every |
| 264 |
|
|
# word in the text must be compared to every word in the query if it |
| 265 |
|
|
# is worth highlighting, and which part of the word is worth |
| 266 |
|
|
# highlighting. This must be done differently for every field in the |
| 267 |
|
|
# table and for every index defined for that field. Try to run a |
| 268 |
|
|
# query with 100 words and you'll be amazed to see it really works. |
| 269 |
|
|
# Or maybe it doesn't. You should be aware that the hilighting code |
| 270 |
|
|
# is to be regarded as alpha. It is certainly the least tested part |
| 271 |
|
|
# of WAIT so far. |
| 272 |
|
|
|
| 273 |
|
|
if ($buf) { |
| 274 |
|
|
my @txt = $query->hilight($buf); |
| 275 |
|
|
# In this operation the following things melt into one piece: |
| 276 |
|
|
# $query: The query entered by the user (Class isa WAIT::Query::Base) |
| 277 |
|
|
# $tb: The table we queried (Class WAIT::Table) |
| 278 |
|
|
# $buf: The document to display (User defined class or string) |
| 279 |
|
|
# The steps taken are: |
| 280 |
|
|
# 1.) $query calls "hilight" on $tb and passes |
| 281 |
|
|
# filtered and raw search terms ($query->{Plain} and $query->{Raw}). |
| 282 |
|
|
# 2.) $tb asks the layout object to tag the object which results |
| 283 |
|
|
# in an array with alternating elements of tags (anon HASHes) and |
| 284 |
|
|
# strings. |
| 285 |
|
|
# 3.) $tb adds some markup on its own: {qt=>1} or some such |
| 286 |
|
|
|
| 287 |
|
|
# The result of that process can optionally be sent through a |
| 288 |
|
|
# filter, just to impress your friends with yet more heavy |
| 289 |
|
|
# wizardry |
| 290 |
|
|
if ($OPT{filter}) { |
| 291 |
|
|
@txt = &filter(@txt); |
| 292 |
|
|
} |
| 293 |
|
|
|
| 294 |
|
|
# And then a formatter (in our case a terminal formatter) turns |
| 295 |
|
|
# all the markup into escape sequences and strings that can in |
| 296 |
|
|
# turn be sent through a pager for instance |
| 297 |
|
|
&$pager($format->as_string(\@txt)); |
| 298 |
|
|
} |
| 299 |
|
|
|
| 300 |
|
|
# Hey, that's it. The user out there is deeply impressed now. You |
| 301 |
|
|
# can lean back again:-) He got a document that has some words |
| 302 |
|
|
# hilighted and will probably read and enjoy it. Maybe he'll send |
| 303 |
|
|
# you an email. |
| 304 |
|
|
} |
| 305 |
|
|
|
| 306 |
|
|
sub filter { |
| 307 |
|
|
my @result; |
| 308 |
|
|
my @context; |
| 309 |
|
|
my $lines = 0; |
| 310 |
|
|
my $clines = 0; |
| 311 |
|
|
my $elipsis = 0; |
| 312 |
|
|
|
| 313 |
|
|
print STDERR "Filter ..."; |
| 314 |
|
|
while (@_) { |
| 315 |
|
|
my %tag = %{shift @_}; |
| 316 |
|
|
my $txt = shift @_; |
| 317 |
|
|
|
| 318 |
|
|
for (split /(\n)/, $txt) { |
| 319 |
|
|
if ($_ eq "\n") { |
| 320 |
|
|
if (exists $tag{_qt}) { |
| 321 |
|
|
#die "Weird!"; |
| 322 |
|
|
push @result, {_i=>1}, "[WEIRD]"; |
| 323 |
|
|
} elsif ($lines) { |
| 324 |
|
|
push @result, {}, $_; |
| 325 |
|
|
$lines--; |
| 326 |
|
|
} else { |
| 327 |
|
|
push @context, {}, $_; |
| 328 |
|
|
$clines++; |
| 329 |
|
|
} |
| 330 |
|
|
} else { |
| 331 |
|
|
if (exists $tag{_qt}) { |
| 332 |
|
|
push @result, {_i=>1}, "\n[ $elipsis lines ]\n" if $elipsis; |
| 333 |
|
|
push @result, @context, {%tag}, $_; |
| 334 |
|
|
delete $tag{_qt}; |
| 335 |
|
|
@context = (); $clines = 0; $elipsis=0; |
| 336 |
|
|
$lines = $OPT{filter}+1; |
| 337 |
|
|
} elsif ($lines) { |
| 338 |
|
|
push @result, \%tag, $_; |
| 339 |
|
|
} else { |
| 340 |
|
|
push @context, \%tag, $_; |
| 341 |
|
|
} |
| 342 |
|
|
} |
| 343 |
|
|
if ($clines>$OPT{filter}) { |
| 344 |
|
|
my (%tag, $txt); |
| 345 |
|
|
while ($clines>$OPT{filter}) { |
| 346 |
|
|
%tag = %{shift @context}; |
| 347 |
|
|
$txt = shift @context; |
| 348 |
|
|
if ($txt =~ /\n/) { |
| 349 |
|
|
$clines--; |
| 350 |
|
|
$elipsis++; |
| 351 |
|
|
} |
| 352 |
|
|
} |
| 353 |
|
|
} |
| 354 |
|
|
} |
| 355 |
|
|
} |
| 356 |
|
|
print STDERR " done\n"; |
| 357 |
|
|
@result; |
| 358 |
|
|
} |
| 359 |
|
|
|
| 360 |
|
|
sub less { |
| 361 |
|
|
my $flags; |
| 362 |
|
|
if ($WAIT::Config->{pager} =~ /less/) { |
| 363 |
|
|
$flags = '-r'; |
| 364 |
|
|
} elsif ($WAIT::Config->{pager} =~ /more/) { |
| 365 |
|
|
$flags = '-c'; |
| 366 |
|
|
} |
| 367 |
|
|
open(PAGER, "|$WAIT::Config->{pager} $flags") or die; |
| 368 |
|
|
print PAGER @_; |
| 369 |
|
|
close PAGER; |
| 370 |
|
|
} |
| 371 |
|
|
|
| 372 |
|
|
sub pager { |
| 373 |
|
|
my @lines = split /\n/, $_[0]; |
| 374 |
|
|
my $line = 0; |
| 375 |
|
|
for (@lines) { |
| 376 |
|
|
print "$_\n"; $line++; |
| 377 |
|
|
if ($line % 24 == 0) { |
| 378 |
|
|
my $key = $term->readline("[return]"); |
| 379 |
|
|
return if $key =~ /^q/i; |
| 380 |
|
|
} |
| 381 |
|
|
} |
| 382 |
|
|
} |
| 383 |
|
|
|
| 384 |
|
|
|
| 385 |
|
|
__END__ |
| 386 |
|
|
## ################################################################### |
| 387 |
|
|
## pod |
| 388 |
|
|
## ################################################################### |
| 389 |
|
|
|
| 390 |
|
|
=head1 NAME |
| 391 |
|
|
|
| 392 |
|
|
sman - Search and disply manuals interactive |
| 393 |
|
|
|
| 394 |
|
|
=head1 SYNOPSIS |
| 395 |
|
|
|
| 396 |
|
|
B<sman> |
| 397 |
|
|
[B<-database> I<database name>] |
| 398 |
|
|
[B<-dir> I<database directory>] |
| 399 |
|
|
[B<-table> I<name>] |
| 400 |
|
|
[B<-less>] |
| 401 |
|
|
[B<-filter> I<num>] |
| 402 |
|
|
[B<-max> I<num>] |
| 403 |
|
|
|
| 404 |
|
|
=head1 DESCRIPTION |
| 405 |
|
|
|
| 406 |
|
|
B<Sman> is an interactive search interface to your systems manual pages. |
| 407 |
|
|
|
| 408 |
|
|
=head2 OPTIONS |
| 409 |
|
|
|
| 410 |
|
|
=over 10 |
| 411 |
|
|
|
| 412 |
|
|
=item B<-database> I<database name> |
| 413 |
|
|
|
| 414 |
|
|
Change the default database name to I<database name>. |
| 415 |
|
|
|
| 416 |
|
|
=item B<-dir> I<database directory> |
| 417 |
|
|
|
| 418 |
|
|
Change the default database directory to I<database directory>. |
| 419 |
|
|
|
| 420 |
|
|
=item B<-table> I<name> |
| 421 |
|
|
|
| 422 |
|
|
Use I<name> instead of C<man> as table name. |
| 423 |
|
|
|
| 424 |
|
|
=item B<-pager> I<name> |
| 425 |
|
|
|
| 426 |
|
|
Use I<name> instead of the default pager. If no I<name> is supplied a |
| 427 |
|
|
buildin pager is used. |
| 428 |
|
|
|
| 429 |
|
|
=item B<-filter> I<num> |
| 430 |
|
|
|
| 431 |
|
|
Display only I<num> lines above and below an occurance of a search |
| 432 |
|
|
term in the manual. |
| 433 |
|
|
|
| 434 |
|
|
=item B<-max> I<num> |
| 435 |
|
|
|
| 436 |
|
|
Display only I<num> hits. Default is to 10. |
| 437 |
|
|
|
| 438 |
|
|
=head1 SEE ALSO |
| 439 |
|
|
|
| 440 |
|
|
L<smakewhatis>. |
| 441 |
|
|
|
| 442 |
|
|
=head1 AUTHOR |
| 443 |
|
|
|
| 444 |
|
|
Ulrich Pfeifer E<lt>F<pfeifer@ls6.informatik.uni-dortmund.de>E<gt> |
Parent Directory
| 
Revision Log