1 | #!/usr/bin/ruby -I /usr/local/lib/utt -I $HOME/.local/lib/utt |
---|
2 | |
---|
3 | $: << "#{ENV['HOME']}/.local/lib/utt" |
---|
4 | $: << "/usr/local/lib/utt" |
---|
5 | |
---|
6 | require 'getoptlong' |
---|
7 | require 'seg.rb' |
---|
8 | |
---|
9 | opts = GetoptLong.new( |
---|
10 | [ '--help', '-h', GetoptLong::NO_ARGUMENT ], |
---|
11 | [ '--debug', '-d', GetoptLong::NO_ARGUMENT ], |
---|
12 | [ '--format', '-F', GetoptLong::REQUIRED_ARGUMENT ], |
---|
13 | [ '--info', '-I', GetoptLong::REQUIRED_ARGUMENT ], |
---|
14 | [ '--only-trees','-t', GetoptLong::NO_ARGUMENT ]) |
---|
15 | |
---|
16 | $helptext= |
---|
17 | "The program generates trees from the graph output by dgp. dgp must\n"+ |
---|
18 | "must be run with '-i ds' option.\n\n"+ |
---|
19 | "Command: tre [options]\n\n"+ |
---|
20 | "Options:\n"+ |
---|
21 | "--help -h Print help (this text) and exit.\n"+ |
---|
22 | "--debug -d Verbose output. For developers only.\n"+ |
---|
23 | "--format=s -F s Output format. Recognized values:\n"+ |
---|
24 | " a root + list of arcs\n"+ |
---|
25 | " p parenthesized notation\n"+ |
---|
26 | " h human readable indented tree format\n"+ |
---|
27 | " Multiple values are allowed. (default p)\n"+ |
---|
28 | "--info=s -I s Information printed. Recognized values:\n"+ |
---|
29 | " n node identifier\n"+ |
---|
30 | " f surface form\n"+ |
---|
31 | " m morphological information\n"+ |
---|
32 | " l arc labels\n"+ |
---|
33 | "--only-trees -t Do not copy input. Print trees only.\n" |
---|
34 | |
---|
35 | $DEBUG=false |
---|
36 | $FORMAT='p' |
---|
37 | $INFO='DEFAULT' |
---|
38 | $ONLYTREES=false |
---|
39 | |
---|
40 | opts.each do |opt, arg| |
---|
41 | case opt |
---|
42 | when '--help' |
---|
43 | print $helptext |
---|
44 | exit 0 |
---|
45 | when '--debug' |
---|
46 | $DEBUG=true |
---|
47 | when '--format' |
---|
48 | $FORMAT=arg |
---|
49 | when '--info' |
---|
50 | $INFO=arg |
---|
51 | when '--only-trees' |
---|
52 | $ONLYTREES=true |
---|
53 | else |
---|
54 | print "Unknown option #{opt}. Ignored.\n" |
---|
55 | end |
---|
56 | end |
---|
57 | |
---|
58 | if $INFO=='DEFAULT' |
---|
59 | case $FORMAT |
---|
60 | when 'p','a' |
---|
61 | $INFO='nl' |
---|
62 | when 'h' |
---|
63 | $INFO='fmnl' |
---|
64 | end |
---|
65 | end |
---|
66 | |
---|
67 | $dgpsep=';' |
---|
68 | |
---|
69 | def tre(input) |
---|
70 | $gphid=[] |
---|
71 | $form=[] |
---|
72 | $lem=[] |
---|
73 | nodes=[] |
---|
74 | count=0 |
---|
75 | seg=Seg.new |
---|
76 | for line in input |
---|
77 | print line unless $ONLYTREES |
---|
78 | seg.set(line) |
---|
79 | if dgp=seg['dgp'] |
---|
80 | if nodes==[] && seg[3]!='BOS' |
---|
81 | print "A sentence must start with BOS segment. Aborting.\n" |
---|
82 | return |
---|
83 | end |
---|
84 | |
---|
85 | id=dgp[/^\d+/].to_i |
---|
86 | |
---|
87 | if gph=seg['gph'] |
---|
88 | $gphid[id]=gph[/^\d+/].to_i |
---|
89 | else |
---|
90 | print "No gph field. Aborting.\n" |
---|
91 | return |
---|
92 | end |
---|
93 | |
---|
94 | $form[$gphid[id]]=seg[4] |
---|
95 | $lem[$gphid[id]]=seg['lem'] |
---|
96 | |
---|
97 | nodes[id] = [seg[1].to_i,dgp] |
---|
98 | |
---|
99 | if seg[3]=='EOS' |
---|
100 | $pref = "#{seg[1]} #{seg[2]} SYN *" |
---|
101 | parsegraph(nodes) |
---|
102 | printgraph if $DEBUG |
---|
103 | $thetrees=[] |
---|
104 | gentrees2 |
---|
105 | for t in $thetrees |
---|
106 | count += 1 |
---|
107 | t1=ground(t) |
---|
108 | case $FORMAT |
---|
109 | when /a/ |
---|
110 | print "#{$pref} tre:#{count} arc:" |
---|
111 | printarcs(t1[0],t1[1]) |
---|
112 | print "\n" |
---|
113 | when /p/ |
---|
114 | print "#{$pref} tre:#{count} par:" |
---|
115 | printpar(t1[0],t1[1]) |
---|
116 | print "\n" |
---|
117 | when /h/ |
---|
118 | print "#\n# tree #{count}\n# ------\n" |
---|
119 | printtree(t1[0],t1[1],0) |
---|
120 | end |
---|
121 | end |
---|
122 | nodes=[] |
---|
123 | end |
---|
124 | end |
---|
125 | end |
---|
126 | end |
---|
127 | |
---|
128 | |
---|
129 | def nodeinfo(id) |
---|
130 | info="" |
---|
131 | if $INFO =~ /n/ |
---|
132 | info += id.to_s |
---|
133 | info += '.' if $INFO =~ /[fm]/ |
---|
134 | end |
---|
135 | if $INFO =~ /f/ |
---|
136 | info += $form[id] |
---|
137 | info += ';' if $INFO =~ /m/ |
---|
138 | end |
---|
139 | if $INFO =~ /m/ |
---|
140 | info += $lem[id] |
---|
141 | end |
---|
142 | info |
---|
143 | end |
---|
144 | |
---|
145 | |
---|
146 | def printarcs(root,arcs) |
---|
147 | print nodeinfo(root) |
---|
148 | for a in arcs |
---|
149 | print ';' |
---|
150 | print "#{a[2]}:" if $INFO =~ /l/ |
---|
151 | print nodeinfo(a[0])+'-'+nodeinfo(a[1]) |
---|
152 | end |
---|
153 | end |
---|
154 | |
---|
155 | def printtree(root,arcs,o) |
---|
156 | if o==0 |
---|
157 | print "# %-16s" % "root: " |
---|
158 | end |
---|
159 | print nodeinfo(root),"\n" |
---|
160 | for arc in arcs.select{ |a| a[0]==root }.sort{|a,b| a[1]<=>b[1] } |
---|
161 | print '# '," "*(o+1) |
---|
162 | print "%-16s" % (arc[2]+": ") |
---|
163 | printtree(arc[1],arcs,o+1) |
---|
164 | end |
---|
165 | end |
---|
166 | |
---|
167 | def printpar(root,arcs) |
---|
168 | print nodeinfo(root) |
---|
169 | deps = arcs.select{ |a| a[0]==root }.sort{|a,b| a[1]<=>b[1] } |
---|
170 | unless deps == [] |
---|
171 | print '(' |
---|
172 | cont=false |
---|
173 | for arc in deps |
---|
174 | if cont then print ',' else cont=true end |
---|
175 | print arc[2],':' if $INFO =~ /l/ |
---|
176 | printpar(arc[1],arcs) |
---|
177 | end |
---|
178 | print ')' |
---|
179 | end |
---|
180 | end |
---|
181 | |
---|
182 | |
---|
183 | def parsegraph(nodes) |
---|
184 | |
---|
185 | $n =nodes.length |
---|
186 | $sat =[]; |
---|
187 | |
---|
188 | $vis =[]; |
---|
189 | $succ=[]; |
---|
190 | $lhs =[]; |
---|
191 | $arcs=[]; |
---|
192 | $pos=[] |
---|
193 | |
---|
194 | for dgp in nodes |
---|
195 | |
---|
196 | parts = dgp[1].split($dgpsep,6) |
---|
197 | |
---|
198 | if parts[3]==nil || parts[4]==nil || parts[5]==nil |
---|
199 | $stderr.print "ERR: tre requires dgp be called with '--info s' option. Aborting.\n" |
---|
200 | exit |
---|
201 | end |
---|
202 | |
---|
203 | i = parts[0].to_i |
---|
204 | $pos[i] = dgp[0].to_i |
---|
205 | $sat << i if parts[1]=="s" |
---|
206 | $arcs |= parts[2].split(',').map{ |a| case a |
---|
207 | when /\-\-(\w+)-(\d+)\/(\d+)/ |
---|
208 | [i, $2.to_i, $1, $3.to_i] |
---|
209 | when /\+\+(\d+)-(\w+)\/(\d+)/ |
---|
210 | [$1.to_i, i, $2, $3.to_i] |
---|
211 | end } |
---|
212 | $succ |= parts[3][1..-2].split(',').map{|x| [x.to_i,i]} |
---|
213 | $vis |= parts[4][1..-2].split(',').map{|x| [x.to_i,i]} |
---|
214 | $lhs |= parts[5][1..-2].split(',').map{|x| [x.to_i,i]} + [[i,i]] |
---|
215 | |
---|
216 | end |
---|
217 | end |
---|
218 | |
---|
219 | |
---|
220 | def ground(t) |
---|
221 | [ $gphid[t[0]] , t[1].map{|a| [$gphid[a[0]],$gphid[a[1]],a[2]]} ] |
---|
222 | end |
---|
223 | |
---|
224 | |
---|
225 | def gentrees2() |
---|
226 | $thetrees=[]; |
---|
227 | bos=0; eos=$n-1; |
---|
228 | roots = (1...eos).select{|i| $vis.include? [i,eos]}.select{|i| $vis.include? [bos,i]} |
---|
229 | if $DEBUG then print "ROOTS: #{roots.inspect}\n" end |
---|
230 | for i in roots |
---|
231 | $theroot=i |
---|
232 | for r in buildR(i , eos, []) |
---|
233 | (rmin,rmax,rtree) = r |
---|
234 | buildR(bos, rmin, rtree) |
---|
235 | end |
---|
236 | end |
---|
237 | end |
---|
238 | |
---|
239 | |
---|
240 | def buildR(min, max, tree) |
---|
241 | if $DEBUG then print "buildR--#{min}--#{max}--#{tree.inspect}\n" end |
---|
242 | trees=[] |
---|
243 | for a in $arcs.select{|a| a[0]==max && $vis.include?([min,a[1]]) } |
---|
244 | if $DEBUG then print "ARC: #{a.inspect}\n" end |
---|
245 | for r in buildR(a[1],a[3],tree+[a]) |
---|
246 | (rmin,rmax,rarcs) = r |
---|
247 | for l in buildR(min,rmin,rarcs) |
---|
248 | (lmin,lmax,larcs) = l |
---|
249 | trees << [lmin,rmax,larcs] |
---|
250 | end |
---|
251 | end |
---|
252 | end |
---|
253 | for i in (0...$n).select{|i| $succ.include?([i,max])}.select{|i| $lhs.include?([min,i])} |
---|
254 | for l in buildL(min,i,tree) |
---|
255 | (lmin,lmax,larcs) = l |
---|
256 | trees << [lmin,lmax,larcs] |
---|
257 | end |
---|
258 | end |
---|
259 | trees |
---|
260 | end |
---|
261 | |
---|
262 | |
---|
263 | def buildL(min,max,tree) |
---|
264 | if $DEBUG then print "buildL--#{min}--#{max}--#{tree.inspect}\n" end |
---|
265 | if $pos[min]==$pos[max] |
---|
266 | if min==0 && max==0 |
---|
267 | $thetrees.push [$theroot,tree] |
---|
268 | if $DEBUG then print "adding tree: #{tree.inspect}\n" end |
---|
269 | end |
---|
270 | return [[max,max,tree]] |
---|
271 | end |
---|
272 | trees=[] |
---|
273 | for arc in $arcs.select{|a| a[1]==max && $lhs.include?([min,a[0]]) } |
---|
274 | if $DEBUG then print "ARC: #{arc.inspect}\n" end |
---|
275 | for r in buildR(arc[3],max,tree+[arc]) |
---|
276 | (rmin,rmax,rarcs) = r |
---|
277 | for l in buildL(min,rmin,rarcs) |
---|
278 | (lmin,lmax,larcs) = l |
---|
279 | trees << [lmin,lmax,larcs] |
---|
280 | end |
---|
281 | end |
---|
282 | end |
---|
283 | trees |
---|
284 | end |
---|
285 | |
---|
286 | |
---|
287 | def printgraph() |
---|
288 | |
---|
289 | print "N: #{$n}\n" |
---|
290 | print "SAT: #{set_to_s($sat)}\n" |
---|
291 | print "SUCC: #{rel_to_s($succ)}\n" |
---|
292 | print "VIS: #{rel_to_s($vis)}\n" |
---|
293 | print "LHS: #{rel_to_s($lhs)}\n" |
---|
294 | print "ARCS: #{arcs_to_s($arcs)}\n" |
---|
295 | end |
---|
296 | |
---|
297 | def set_to_s(s) "{#{s.join(',')}}" end |
---|
298 | def rel_to_s(r) "{#{r.map{|p| "(#{p[0]},#{p[1]})"}.join(',')}}" end |
---|
299 | def arc_to_s(q) "-#{q[0]}-#{q[2]}-#{q[1]}/#{q[3]}" end |
---|
300 | def arcs_to_s(a) "{#{a.map{|q| arc_to_s(q)}.join(',')}}" end |
---|
301 | |
---|
302 | ###################################################################### |
---|
303 | |
---|
304 | tre($stdin) |
---|