source: fmts/trunk/p/C0XF2N.m@ 1384

Last change on this file since 1384 was 1384, checked in by George Lilly, 13 years ago

fixed nodeID rdf processing

File size: 21.0 KB
Line 
1C0XF2N ; GPL - Fileman Triples entry point routine ;10/13/11 17:05
2 ;;0.1;C0X;nopatch;noreleasedate;Build 7
3 ;Copyright 2011 George Lilly. Licensed under the terms of the GNU
4 ;General Public License See attached copy of the License.
5 ;
6 ;This program is free software; you can redistribute it and/or modify
7 ;it under the terms of the GNU General Public License as published by
8 ;the Free Software Foundation; either version 2 of the License, or
9 ;(at your option) any later version.
10 ;
11 ;This program is distributed in the hope that it will be useful,
12 ;but WITHOUT ANY WARRANTY; without even the implied warranty of
13 ;MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 ;GNU General Public License for more details.
15 ;
16 ;You should have received a copy of the GNU General Public License along
17 ;with this program; if not, write to the Free Software Foundation, Inc.,
18 ;51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
19 ;
20 Q
21 ;
22 ; This is based on C0XMAIN but experiments with a fast load for triples
23 ; that will write directly to the fileman global
24 ; The file 172.101 is a F2N design style for triples, which means
25 ; that it is a Flat file with no subfiles, all fields at the root
26 ; ... it is a "2" file solution which means all strings are stored in
27 ; ... strings file and pointed to by the triples file
28 ; ... it is an N file because it has generated Node IDs instead of
29 ; ... DINUM which would use the IEN for the Node ID.
30 ; gpl 11/04/2011
31 ;
32INITFARY(ZFARY) ; INITIALIZE FILE NUMBERS AND OTHER USEFUL THINGS
33 ; FOR THE DEFAULT TRIPLE STORE. USE OTHER VALUES FOR SUPPORTING ADDITIONAL
34 ; TRIPLE STORES
35 I $D(@ZFARY) Q ; ALREADY INITIALIZED
36 S @ZFARY@("C0XTFN")=172.101 ; TRIPLES FILE NUMBER
37 S @ZFARY@("C0XSFN")=172.201 ; TRIPLES STRINGS FILE NUMBER
38 S @ZFARY@("C0XTN")=$NA(^C0X(101)) ; TRIPLES GLOBAL NAME
39 S @ZFARY@("C0XSN")=$NA(^C0X(201)) ; STRING FILE GLOBAL NAME
40 S @ZFARY@("C0XDIR")="/home/glilly/sage/test/"
41 S @ZFARY@("BLKLOAD")=1 ; this file supports block load
42 S @ZFARY@("FMTSSTYLE")="F2N" ; fileman style
43 S @ZFARY@("REPLYFMT")="JSON"
44 D USEFARY(ZFARY)
45 Q
46 ;
47USEFARY(ZFARY) ; INITIALIZES VARIABLES SAVED IN ARRAY ZFARY
48 N ZI S ZI=""
49 F S ZI=$O(@ZFARY@(ZI)) Q:ZI="" D
50 . ;N ZX
51 . S ZX="S "_ZI_"="""_@ZFARY@(ZI)_""""
52 . ;W !,ZX
53 . X ZX
54 Q
55 ;
56FILEIN ; INTERACTIVE ENTRY POINT FOR OPTION TO READ IN A FILE
57 I '$D(C0XFARY) D INITFARY("C0XFARY")
58 D USEFARY("C0XFARY")
59 S DIR(0)="F^3:240"
60 S DIR("A")="File Directory"
61 S DIR("B")=C0XDIR
62 D ^DIR
63 I Y="^" Q ;
64 S C0XDIR=Y
65 S C0XFARY("C0XDIR")=Y
66 S DIR(0)="F^3:240"
67 S DIR("A")="File Name"
68 I '$D(C0XFN) S DIR("B")="qds.rdf"
69 E S DIR("B")=C0XFN
70 D ^DIR
71 I Y="" Q ;
72 I Y="^" Q ;
73 S C0XFN=Y
74 D IMPORT(C0XFN,C0XDIR,,"C0XFARY")
75 K C0XFDA
76 Q
77 ;
78IMPORT(FNAME,INDIR,INURL,FARY) ; EXTRINSIC THAT READS A FILE FROM THE STANDARD
79 ; DIRECTORY, LOADS IT INTO THE TRIPLESTORE AS TEXT, AND RETURNS THE
80 ; NODE NAME OF THE TEXT TRIPLE
81 ; INDIR IS THE OPTIONAL DIRECTORY (DEFAUTS TO STANDARD DIR)
82 ; INURL IS THE OPTIONAL URI FOR ACCESSING THE FILE FROM THE TRIPLE STORE
83 ; FARY IS THE OPTIONAL FILE ARRAY OF THE TRIPLE STORE TO USE
84 I '$D(FARY) D ;
85 . D INITFARY("C0XFARY")
86 . S FARY="C0XFARY"
87 D USEFARY(FARY)
88 N ZD,ZTMP
89 I '$D(INDIR) S INDIR=C0XDIR ; DIRECTORY OF THE RDF FILE
90 I $G(INURL)="" D ;
91 . ;N ZN2 S ZN2=$P(FNAME,".",1)_"_"_$P(FNAME,".",2) ; REMOVE THE DOT
92 . ;S INURL=FDIR_ZN2
93 . S INURL=INDIR_FNAME
94 N ZTMP
95 S ZTMP=$NA(^TMP("C0X",$J,"FILEIN",1)) ; WHERE TO PUT THE INCOMING FILE
96 K @ZTMP ; MAKE SURE IT'S CLEAR
97 S C0XSTART=$$NOW^XLFDT
98 W !,"STARTED: ",C0XSTART
99 W !,"READING IN: ",FNAME
100 I '$$FILEREAD(ZTMP,INDIR,FNAME,4) D Q ; QUIT IF NO SUCCESS
101 . W !,"ERROR READING FILE: ",INDIR,FNAME
102 S ZRDF=$NA(^TMP("C0X",$J,"FILEIN")) ; WITHOUT THE SUBSCRIPT
103 W !,$O(@ZRDF@(""),-1)," LINES READ"
104 D INSRDF(ZRDF,INURL,FARY) ; IMPORT AND PROCESS THE RDF
105 K INURL
106 K C0XFDA
107 ;K ^TMP("MXMLDOM",$J)
108 Q
109 ;
110WGET(ZURL,FARY) ; GET FROM THE INTERNET AN RDF FILE AND INSERT IT
111 ;
112 I '$D(FARY) D ;
113 . D INITFARY("C0XFARY")
114 . S FARY="C0XFARY"
115 D USEFARY(FARY)
116 ;N ZLOC,ZTMP
117 K ZTMP
118 S ZLOC=$NA(^TMP("C0X","WGET",$J))
119 K @ZLOC
120 S C0XSTART=$$NOW^XLFDT
121 W !,"STARTED: ",C0XSTART
122 W !,"DOWNLOADING: ",ZURL
123 S OK=$$httpGET^%zewdGTM(ZURL,.ZTMP)
124 M @ZLOC=ZTMP
125 S C0XLINES=$O(@ZLOC@(""),-1)
126 W !,C0XLINES," LINES READ"
127 S C0XDLC=$$NOW^XLFDT ; DOWNLOAD COMPLETE
128 W !,"DOWNLOAD COMPLETE AT ",C0XDLC
129 S C0XDIFF=$$FMDIFF^XLFDT(C0XDLC,C0XSTART,2)
130 W !," ELAPSED TIME: ",C0XDIFF," SECONDS"
131 I C0XDIFF'=0 W !," APPROXIMATELY ",$P(C0XLINES/C0XDIFF,".")," LINES PER SEC"
132 D INSRDF(ZLOC,ZURL,FARY)
133 Q
134 ;
135INSRDF(ZRDF,ZNAME,FARY) ; INSERT AN RDF FILE INTO THE STORE AND PROCESS
136 ; ZRDF IS PASSED BY NAME
137 I '$D(FARY) D ;
138 . D INITFARY("C0XFARY")
139 . S FARY="C0XFARY"
140 D USEFARY(FARY)
141 S BATCNT=0 ; BATCH COUNTER
142 S BATMAX=10000 ; TRY BATCHES OF THIS SIZE
143 N ZGRAPH,ZSUBJECT
144 S ZGRAPH="_:G"_$$LKY9 ; RANDOM GRAPH NAME
145 S ZSUBJECT=$$ANONS() ; RANDOM ANOYMOUS SUBJECT
146 D ADD(ZGRAPH,ZSUBJECT,"fmts:url",ZNAME,FARY)
147 N ZTXTNM
148 S ZTXTNM="_TXT_INCOMING_RDF_FILE_"_ZNAME_"_"_$$LKY9 ; NAME FOR TEXT NODE
149 D ADD(ZGRAPH,ZSUBJECT,"fmts:fileSource",ZTXTNM,FARY)
150 D ADD(ZGRAPH,ZSUBJECT,"fmts:fileTag",$$name2tag(ZNAME),FARY)
151 D SWUPDIE(.C0XFDA) ; TRY IT OUT
152 K C0XCNT ;RESET FOR NEXT TIME
153 D STORETXT(ZRDF,ZTXTNM,FARY)
154 W !,"ADDED: ",ZGRAPH," ",ZSUBJECT," fmts:fileSource ",ZTXTNM
155 D PROCESS(.G,ZRDF,ZNAME,ZGRAPH,FARY) ; PARSE AND INSERT THE RDF
156 Q
157 ;
158name2tag(zname) ; extrinsic which returns a tag derived from a name
159 ; /home/vista/project.xml ==> project
160 q $p($re($p($re(zname),"/")),".")
161 ;
162STORETXT(ZTXT,ZNAME,FARY) ; STORE TEXT IN THE TRIPLESTORE AT ZNAME
163 ;
164 I '$D(FARY) D ;
165 . D INITFARY("C0XFARY")
166 . S FARY="C0XFARY"
167 D USEFARY(FARY)
168 N ZIEN
169 S ZIEN=$$IENOF(ZNAME,FARY) ; GET THE IEN
170 D CLEAN^DILF
171 K ZERR
172 D WP^DIE(C0XSFN,ZIEN_",",1,,ZTXT,"ZERR")
173 I $D(ZERR) D Q ;
174 . W !,"ERROR CREATING WORD PROCESSING FIELD"
175 . S C0XERR="ERROR CREATING WORD PROCESSING FIELD"
176 . D ^%ZTER ; error trap
177 Q
178 ;
179GETTXT(ZRTN,ZNAME,FARY) ; RETURNS RDF SOURCE OR OTHER TEXT
180 ; ZRTN IS PASSED BY REFERENCE
181 I '$D(FARY) D ;
182 . D INITFARY("C0XFARY")
183 . S FARY="C0XFARY"
184 D USEFARY(FARY)
185 N ZIEN
186 S ZIEN=$$IENOF(ZNAME)
187 S OK=$$GET1^DIQ(C0XSFN,ZIEN_",",1,,"ZRTN")
188 Q
189 ;
190WHERETXT(ZNAME,FARY) ; EXTRINSIC WHICH RETURNS THE NAME OF THE GLOBAL
191 ; WHERE THE TEXT IS LOCATED. NAME IS THE NAME OF THE STRING
192 I '$D(FARY) D ;
193 . D INITFARY("C0XFARY")
194 . S FARY="C0XFARY"
195 D USEFARY(FARY)
196 N ZIEN
197 S ZIEN=$$IENOF(ZNAME)
198 Q $NA(@C0XSN@(ZIEN,1))
199 ;
200FILEREAD(ZINTMP,ZDIR,ZFNAME,ZLVL) ; READS A FILE INTO ZINTMP USING FTG^%ZISH
201 ; ZINTMP IS PASSED BY NAME AND INCLUDES THE NEW SUBSCRIPT
202 ; IE ^TMP("C0X","FILEIN",1)
203 ; ZLVL IN THIS CASE WOULD BE 3 INCREMENTING THE 1
204 ; EXTRINSIC WHICH RETURNS THE RESULT OF FTG^%ZISH
205 S OK=$$FTG^%ZISH(ZDIR,FNAME,ZINTMP,ZLVL)
206 Q OK
207 ;
208TESTPROC ; TEST PROCESS WITH EXISTING SMALL RDF FILE
209 S ZIN=$NA(^TMP("C0X",12226,"FILEIN"))
210 S ZGRAPH="/test/rdfFile"
211 S ZM="/test/rdfFile/meta"
212 D PROCESS(.G,ZIN,ZGRAPH,ZM)
213 Q
214 ;
215VISTAOWL ;
216 S ZRDF=$NA(^TMP("C0X",542,"FILEIN"))
217 S ZNAME="/home/glilly/vistaowl/VistAOWL.owl"
218 S ZGRAPH="_:G431590209"
219 S FARY="C0XFARY"
220 D INITFARY(FARY)
221 S C0XDOCID=1
222 S BATCNT=0
223 S BATMAX=10000
224 D PROCESS(.G,ZRDF,ZGRAPH,ZNAME,FARY)
225 Q
226 ;
227PROCESS(ZRTN,ZRDF,ZGRF,ZMETA,FARY) ; PROCESS AN INCOMING RDF FILE
228 ; ZRTN IS PASS BY REFERENCE AND RETURNS MESSAGES ABOUT THE PROCESSING
229 ; ZRDF IS PASSED BY NAME AND IS THE GLOBAL CONTAINING THE RDF FILE
230 ; ZGRF IS THE NAME OF THE GRAPH TO USE IN THE TRIPLE STORE FOR RESULTS
231 ; ZMETA IS OPTIONAL AND IS THE NAME OF THE GRAPH TO STORE METADATA
232 ;
233 I '$D(FARY) D ;
234 . D INITFARY("C0XFARY")
235 . S FARY="C0XFARY"
236 D USEFARY(FARY)
237 ;N BATCNT
238 ;N BATMAX
239 ; -- first parse the rdf file with the MXML parser
240 ;S C0XDOCID=$$PARSE^C0CNHIN(ZRDF,"C0XARRAY") ; PARSE WITH MXML
241 S C0XDLC2=$$NOW^XLFDT ; START OF PARSE
242 I @ZRDF@(1)'["<?xml" D Q ;
243 . K @ZRDF ; don't need the input buffer
244 . W !,"Not an XML file"
245 S C0XDOCID=$$EN^MXMLDOM(ZRDF,"W") ;
246 ;B
247 K @ZRDF ; DON'T NEED INPUT BUFFER ANYMORE
248 ; -- assign the MXLM dom global name to ZDOM
249 S ZDOM=$NA(^TMP("MXMLDOM",$J,C0XDOCID))
250 ;S ZDOM=$NA(^TMP("MXMLDOM",16850,C0XDOCID)) ;VISTAOWL DOM
251 S C0XNODE=$O(@ZDOM@(""),-1)
252 W !,C0XNODE," XML NODES PARSED"
253 S C0XPRS=$$NOW^XLFDT ; PARSE COMPLETE
254 W !,"PARSE COMPLETE AT ",C0XPRS
255 S C0XDIFF=$$FMDIFF^XLFDT(C0XPRS,C0XDLC2,2)
256 W !," ELAPSED TIME: ",C0XDIFF," SECONDS"
257 I C0XDIFF'=0 D ;
258 . W !," APPROXIMATELY ",$P(C0XNODE/C0XDIFF,".")," NODES PER SECOND"
259 ; -- populate the metagraph to point to the graph with status unfinished
260 S METAS=$$ANONS ; GET AN ANONOMOUS RANDOM SUBJECT
261 I '$D(ZMETA) S ZMETA="_:G"_$$LKY9 ; RANDOM GRAPH NAME FOR METAGRAPH
262 D ADD(ZMETA,METAS,"fmts:about",ZGRF,FARY) ; POINT THE META TO THE GRAPH
263 D ADD(ZMETA,METAS,"fmts:status","unfinished",FARY) ; mark as unfinished
264 W !,"INSERTING GRAPH: ",ZGRF
265 ;S C0XDATE=$$FMDTOUTC^C0CUTIL($$NOW^XLFDT,"DT")
266 S C0XDATE=$$NOW^XLFDT
267 D ADD(ZMETA,METAS,"fmts:dateTime",C0XDATE,FARY)
268 D SWUPDIE(.C0XFDA) ; commit the metagraph changes to the triple store
269 ; --
270 ; -- pull out the vocabularies in the RDF statement. marked with xmlns:
271 ; -- put them in a local variable for quick reference
272 ; -- TODO: create a graph for vocabularies and validate incoming against it
273 ;
274 S C0XVOC=""
275 N ZI,ZJ,ZK S ZI=""
276 F S ZI=$O(@ZDOM@(1,"A",ZI)) Q:ZI="" D ; FOR EACH xmlns
277 . S ZVOC=$P(ZI,"xmlns:",2)
278 . I ZVOC'="" S C0XVOC(ZVOC)=$G(@ZDOM@(1,"A",ZI))
279 W !,"VOCABS:"
280 N ZZ S ZZ=""
281 F S ZZ=$O(C0XVOC(ZZ)) Q:ZZ="" W !,ZZ,":",C0XVOC(ZZ)
282 ;
283 ; -- look for children called rdf:Description. quit if none. not an rdf file
284 ;
285 S C0XTYPE("rdf:Description")=1
286 S C0XTYPE("owl:ObjectProperty")=1
287 S C0XTYPE("owl:Ontology")=1
288 S C0XTYPE("owl:Class")=1
289 S C0XTYPE("rdfs:subClassOf")=1
290 S ZI=$O(@ZDOM@(1,"C",""))
291 I '$G(C0XTYPE(@ZDOM@(1,"C",ZI))) D Q ; not an rdf file
292 . W !,"Error. Not an RDF file. Cannot process."
293 . D SHOW(1)
294 ;
295 ; -- now process the rdf description children
296 ;
297 S ZI=""
298 S (C0XSUB,C0XPRE,C0XOBJ)="" ; INITIALIZE subject, object and predicate
299 F S ZI=$O(@ZDOM@(1,"C",ZI)) Q:ZI="" D ;
300 . ; -- we are skipping any child that is not rdf:Description
301 . ; -- TODO: check to see if this is right in general
302 . ;
303 . IF '$G(C0XTYPE(@ZDOM@(1,"C",ZI))) D Q ;
304 . . W !,"SKIPPING NODE: ",ZI
305 . ; -- now looking for the subject for the triples
306 . S ZX=$G(@ZDOM@(ZI,"A","rdf:about"))
307 . I ZX'="" D ; we have the subject
308 . . ;W " about: ",ZX
309 . . S C0XSUB=ZX
310 . E D ;
311 . . S ZX=$G(@ZDOM@(ZI,"A","rdf:nodeID")) ; node id is another style of subject
312 . . I ZX'="" D ;
313 . . . S C0XSUB=ZX
314 . I C0XSUB="" S C0XSUB=$$ANONS ; DEFAULT TO BLANK SUBJECT
315 . ;
316 . ; -- we now have the subject. the children of this node have the rest
317 . ;
318 . S ZJ="" ; for the children of the rdf:Description nodes
319 . F S ZJ=$O(@ZDOM@(ZI,"C",ZJ)) Q:ZJ="" D ; for each child
320 . . S C0XPRE=@ZDOM@(ZJ) ; the predicate without a prefix
321 . . S ZX=$G(@ZDOM@(ZJ,"A","xmlns")) ; name space
322 . . I ZX'="" S C0XPRE=ZX_C0XPRE ; add the namespace prefix
323 . . I C0XPRE[":" D ; expand using vocabulary
324 . . . N ZB,ZA
325 . . . S ZB=$P(C0XPRE,":",1)
326 . . . S ZA=$P(C0XPRE,":",2)
327 . . . I $G(C0XVOC(ZB))'="" D ;
328 . . . . S C0XPRE=C0XVOC(ZB)_ZA ; expanded
329 . . S ZY=$G(@ZDOM@(ZJ,"A","rdf:resource")) ; potential object
330 . . I ZY'="" D Q ;
331 . . . S C0XOBJ=$$EXT^C0XUTIL(ZY) ; object
332 . . . D ADD(ZGRF,C0XSUB,C0XPRE,C0XOBJ) ; finally. our first real triple
333 . . ; -- this is an else because of the quit above
334 . . S ZX=$G(@ZDOM@(ZJ,"A","rdf:nodeID")) ; fishing for nodeId object
335 . . I ZX'="" D Q ; got one
336 . . . S C0XOBJ=ZX ; we are using the incoming nodeIDs as object/subject
337 . . . ; without change... this could be foolish .. look at it again later
338 . . . D ADD(ZGRF,C0XSUB,C0XPRE,C0XOBJ) ; go for it and add a node
339 . . S C0XOBJ=$G(@ZDOM@(ZJ,"T",1)) ; hopefully an object is here
340 . . I C0XOBJ="" D Q ; not a happy situation
341 . . . W !,"ERROR, NO OBJECT FOUND FOR NODE: ",ZJ
342 . . S C0XOBJ=$$EXT^C0XUTIL(C0XOBJ) ; might be namespaced
343 . . D ADD(ZGRF,C0XSUB,C0XPRE,C0XOBJ) ; go for it and add a node
344 S C0XTRP=$$NOW^XLFDT ; PARSE COMPLETE
345 W !,"TRIPLES COMPLETE AT ",C0XTRP
346 S C0XDIFF=$$FMDIFF^XLFDT(C0XTRP,C0XPRS,2)
347 W !," ELAPSED TIME: ",C0XDIFF," SECONDS"
348 I C0XDIFF'=0 D ;
349 . W !," APPROXIMATELY ",$P(C0XCNT/C0XDIFF,".")," TRIPLES PER SECOND"
350 W !,"INSERTING ",C0XCNT," TRIPLES"
351 I $D(C0XFDA) D ;
352 . I $G(BLKLOAD) D ;
353 . . D BULKLOAD(.C0XFDA)
354 . E D ;
355 . . D UPDIE(.C0XFDA) ; commit the updates to the file
356 ; next, mark the graph as finished
357 S C0XINS=$$NOW^XLFDT ; PARSE COMPLETE
358 W !,"INSERTION COMPLETE AT ",C0XPRS
359 S C0XDIFF=$$FMDIFF^XLFDT(C0XINS,C0XTRP,2)
360 W !," ELAPSED TIME: ",C0XDIFF," SECONDS"
361 I C0XDIFF'=0 W !," APPROXIMATELY ",$P(C0XCNT/C0XDIFF,".")," NODES PER SECOND"
362 S C0XEND=$$NOW^XLFDT
363 W !," ENDED AT: ",C0XEND
364 S C0XDIFF=$$FMDIFF^XLFDT(C0XEND,C0XSTART,2)
365 W !," ELAPSED TIME: ",C0XDIFF," SECONDS"
366 I C0XDIFF'=0 W !," APPROXIMATELY ",$P(C0XCNT/C0XDIFF,".")," TRIPLES PER SECOND"
367 Q
368 ;
369SHOW(ZN) ;
370 I '$D(C0XJOB) S C0XJOB=$J
371 N ZD
372 S ZD=$NA(^TMP("MXMLDOM",C0XJOB,1,ZN))
373 W ZD,"=",@ZD
374 F S ZD=$Q(@ZD) Q:$QS(ZD,4)'=ZN W !,ZD,"=",@ZD
375 ;ZWR ^TMP("MXMLDOM",C0XJOB,1,ZN,*)
376 Q
377 ;
378ANONS() ; RETURNS AN ANONOMOUS SUBJECT
379 Q "iDPsDPss"_$$LKY9
380 ;
381NEWG(NGRAPH,NMETA) ; CREATES A NEW META GRAPH, MARKS IT AS UNFINISHED
382 ; THEN CREATES A NEW GRAPH AND POINTS THE METAGRAPH TO IT
383 ; NGRAPH AND NMETA ARE PASSED BY REFERENCE AND ARE THE RETURN
384 S NGRAPH="G"_$$LKY9
385 S NMETA=NGRAPH_"A"
386 Q
387 ;
388ADD(ZG,ZS,ZP,ZO,FARY) ; ADD A TRIPLE TO THE TRIPLESTORE. ALL VALUES ARE TEXT
389 ; THE FDA IS SET UP BUT THE FILES ARE NOT UPDATED. CALL UPDIE TO COMPLETE
390 I '$D(FARY) D ;
391 . D INITFARY("C0XFARY")
392 . S FARY="C0XFARY"
393 D USEFARY(FARY)
394 I '$D(C0XCNT) S C0XCNT=0
395 N ZNODE
396 S ZNODE="N"_$$LKY17
397 N ZNARY ; GET READY TO CALL IENOFA
398 S ZNARY("ZG",ZG)=""
399 S ZNARY("ZS",ZS)=""
400 S ZNARY("ZP",ZP)=""
401 S ZNARY("ZO",ZO)=""
402 D IENOFA(.ZIENS,.ZNARY,FARY) ; RESOLVE/ADD STRINGS
403 ;S ZGIEN=$$IENOF(ZG) ; LAYGO TO GET IEN
404 ;S ZSIEN=$$IENOF(ZS)
405 ;S ZPIEN=$$IENOF(ZP)
406 ;S ZOIEN=$$IENOF(ZO)
407 ;I $D(C0XFDA) D UPDIE ; ADD THE STRINGS IF NEEDED
408 I '$D(BATCNT) S BATCNT=0
409 S BATCNT=BATCNT+1
410 S C0XCNT=C0XCNT+1
411 I $G(BLKLOAD)=1 D ; we are using bulk load
412 . S C0XFDA(C0XTFN,BATCNT,.01)=ZNODE
413 . S C0XFDA(C0XTFN,BATCNT,.02)=$O(ZIENS("IEN","ZG",""))
414 . S C0XFDA(C0XTFN,BATCNT,.03)=$O(ZIENS("IEN","ZS",""))
415 . S C0XFDA(C0XTFN,BATCNT,.04)=$O(ZIENS("IEN","ZP",""))
416 . S C0XFDA(C0XTFN,BATCNT,.05)=$O(ZIENS("IEN","ZO",""))
417 E D ;
418 . S C0XFDA(C0XTFN,"?+"_BATCNT_",",.01)=ZNODE
419 . S C0XFDA(C0XTFN,"?+"_BATCNT_",",.02)=$O(ZIENS("IEN","ZG",""))
420 . S C0XFDA(C0XTFN,"?+"_BATCNT_",",.03)=$O(ZIENS("IEN","ZS",""))
421 . S C0XFDA(C0XTFN,"?+"_BATCNT_",",.04)=$O(ZIENS("IEN","ZP",""))
422 . S C0XFDA(C0XTFN,"?+"_BATCNT_",",.05)=$O(ZIENS("IEN","ZO",""))
423 I '$D(BATMAX) S BATMAX=10000
424 I BATCNT=BATMAX D ; BATCH IS DONE
425 . I $G(BLKLOAD) D ; bulk load
426 . . D BULKLOAD(.C0XFDA) ; bulk load the batch
427 . E D ; no bulk load
428 . . D UPDIE(.C0XFDA)
429 . K C0XFDA
430 . S BATCNT=0 ; RESET COUNTER
431 ; REMEMBER TO CALL UPDIE WHEN YOU'RE DONE
432 Q
433 ;
434LKY5() ;EXTRINIC THAT RETURNS A RANDOM 5 DIGIT NUMBER. USED FOR GENERATING
435 ; UNIQUE NODE AND GRAPH NAMES
436 N ZN,ZI
437 S ZN=""
438 F ZI=1:1:5 D ;
439 . S ZN=ZN_$R(10)
440 Q ZN
441 ;
442LKY9() ;EXTRINIC THAT RETURNS A RANDOM 9 DIGIT NUMBER. USED FOR GENERATING
443 ; UNIQUE NODE AND GRAPH NAMES
444 N ZN,ZI
445 S ZN=""
446 F ZI=1:1:9 D ;
447 . S ZN=ZN_$R(10)
448 Q ZN
449 ;
450LKY17() ;EXTRINIC THAT RETURNS A RANDOM 9 DIGIT NUMBER. USED FOR GENERATING
451 ; UNIQUE NODE AND GRAPH NAMES
452 N ZN,ZI
453 S ZN=""
454 F ZI=1:1:17 D ;
455 . S ZN=ZN_$R(10)
456 Q ZN
457 ;
458 ; these routines add the string if it is not found
459 ;
460IENOF(ZSTRING,FARY) ; EXTRINSIC WHICH RETURNS THE IEN OF ZS IN THE STRINGS FILE
461 I '$D(FARY) D ;
462 . D INITFARY("C0XFARY")
463 . S FARY="C0XFARY"
464 N ZIEN
465 I $G(ZSTRING)="" Q "" ; NO STRING
466 S ZIEN=$O(@C0XSN@("B",ZSTRING,""))
467 I ZIEN="" D ;
468 . S C0XFDA2(C0XSFN,"+1,",.01)=ZSTRING
469 . D UPDIE(.C0XFDA2)
470 . S ZIEN=$O(@C0XSN@("B",ZSTRING,""))
471 . K C0XFDA2
472 Q ZIEN
473 ;
474IENOFA(ZOUTARY,INARY,FARY) ; RESOLVE STRINGS TO IEN IN STRINGS FILE
475 ; OR ADD THEM IF
476 ; MISSING. ZINARY AND ZOUTARY ARE PASSED BY REFERENCE
477 ; ZINARY LOOKS LIKE ZINARY("VAR","VAL")=""
478 ; RETURNS IN ZOUTARY OF THE FORM ZOUTARY("IEN","VAR",IEN)=""
479 I '$D(FARY) D ;
480 . D INITFARY("C0XFARY")
481 . S FARY="C0XFARY"
482 K ZOUTARY ; START WITH CLEAN RESULTS
483 K C0XFDA2 ; USE A SEPARATE FDA FOR THIS
484 I '$D(C0XVOC) D VOCINIT^C0XUTIL
485 N ZINARY
486 N ZI S ZI=""
487 F S ZI=$O(INARY(ZI)) Q:ZI="" D ;
488 . N ZK
489 . S ZK=$O(INARY(ZI,""))
490 . S ZINARY($$EXT^C0XUTIL(ZI),$$EXT^C0XUTIL(ZK))=""
491 N ZV,ZIEN
492 N ZCNT S ZCNT=0
493 F S ZI=$O(ZINARY(ZI)) Q:ZI="" D ; LOOK FOR MISSING STRINGS
494 . S ZV=$O(ZINARY(ZI,""))
495 . I ZV["^" S ZV=$TR(ZV,"^","|")
496 . I $O(@C0XSN@("B",ZV,""))="" D ;
497 . . S ZCNT=ZCNT+1
498 . . S C0XFDA2(C0XSFN,"+"_ZCNT_",",.01)=ZV
499 I $D(C0XFDA2) D ;
500 . D UPDIE(.C0XFDA2) ; ADD MISSING STRINGS
501 . K C0XFDA2 ; CLEAN UP
502 F S ZI=$O(ZINARY(ZI)) Q:ZI="" D ; NOT GET ALL IENS
503 . S ZV=$O(ZINARY(ZI,""))
504 . I ZV["^" S ZV=$TR(ZV,"^","|")
505 . S ZIEN=$O(@C0XSN@("B",ZV,"")) ; THEY SHOULD BE THERE NOW
506 . I ZIEN="" D ;
507 . . W !,"ERROR ADDING STRING: ",ZV
508 . . B
509 . S ZOUTARY("IEN",ZI,ZIEN)=""
510 Q
511 ;
512BULKLOAD(ZBFDA) ; BULK LOADER FOR LOADING TRIPLES INTO FILE 172.101
513 ; USING GLOBAL SETS INSTEAD OF UPDATE^DIE
514 ; QUITS IF FILE IS NOT 172.101
515 ; EXPECTS AN FDA WITHOUT STRINGS FOR THE IENS, STARTING AT 1
516 ; QUITS IF FIRST ENTRY IS NOT IENS 1
517 ; ASSUMES THAT THE LAST IENS IS THE COUNT OF ENTRIES
518 ; ZBFDA IS PASSED BY REFERENCE
519 ;
520 ; -- reserves a block of iens from file 172.101 by locking the zero node
521 ; -- ^C0X(101,0) and adding the count of entries to piece 2 and 3
522 ; -- then unlocking to minimize the duration of the lock
523 ;
524 W !,"USING BULKLOAD"
525 I '$D(ZBFDA) Q ; EMPTY FDA
526 I $O(ZBFDA(""))'=172.101 Q ; WRONG FILE
527 N ZCNT,ZP3,ZP4
528 ; -- find the number of nodes to insert
529 S ZCNT=$O(ZBFDA(172.101,""),-1)
530 I ZCNT="" D Q ;
531 . W !,"ERROR IN BULK LOAD - INVALID NODE COUNT"
532 . B
533 ; -- lock the zero node and reserve a block of iens to insert
534 W !,"LOCKING ZERO NODE"
535 LOCK +^C0X(101,0)
536 S ZP3=$P(^C0X(101,0),U,3)
537 S ZP4=$P(^C0X(101,0),U,4)
538 S $P(^C0X(101,0),U,3)=ZP3+ZCNT+1
539 S $P(^C0X(101,0),U,4)=ZP4+ZCNT+1
540 LOCK -^C0X(101,0)
541 N ZI,ZN,ZG,ZS,ZP,ZO,ZIEN,ZBASE
542 S ZBASE=ZP3 ; the last ien in the file
543 W !,"ZERO NODE UNLOCKED, IENS RESERVED=",ZCNT
544 W !,$$NOW^XLFDT
545 S ZI=""
546 F S ZI=$O(ZBFDA(172.101,ZI)) Q:ZI="" D ;
547 . S ZN=$G(ZBFDA(172.101,ZI,.01)) ; node name
548 . I ZN="" D BLKERR Q ;
549 . S ZG=$G(ZBFDA(172.101,ZI,.02)) ; graph pointer
550 . I ZG="" D BLKERR Q ;
551 . S ZS=$G(ZBFDA(172.101,ZI,.03)) ; subject pointer
552 . I ZS="" D BLKERR Q ;
553 . S ZP=$G(ZBFDA(172.101,ZI,.04)) ; predicate pointer
554 . I ZP="" D BLKERR Q ;
555 . S ZO=$G(ZBFDA(172.101,ZI,.05)) ; object pointer
556 . I ZO="" D BLKERR Q ;
557 . S ZIEN=ZI+ZBASE ; the new ien
558 . S ^C0X(101,ZIEN,0)=ZN_U_ZG_U_ZS_U_ZP_U_ZO ; set the zero node
559 . D INDEX(ZIEN,ZN,ZG,ZS,ZP,ZO)
560 Q
561 ;
562INDEX(ZIEN,ZN,ZG,ZS,ZP,ZO) ; HARD SET THE INDEX FOR ONE ENTRY
563 S ^C0X(101,"B",ZN,ZIEN)="" ; the B index
564 S ^C0X(101,"G",ZG,ZIEN)="" ; the G for Graph index
565 S ^C0X(101,"SPO",ZS,ZP,ZO,ZIEN)=""
566 S ^C0X(101,"SOP",ZS,ZO,ZP,ZIEN)=""
567 S ^C0X(101,"OPS",ZO,ZP,ZS,ZIEN)=""
568 S ^C0X(101,"OSP",ZO,ZS,ZP,ZIEN)=""
569 S ^C0X(101,"PSO",ZP,ZS,ZO,ZIEN)=""
570 S ^C0X(101,"POS",ZP,ZO,ZS,ZIEN)=""
571 S ^C0X(101,"GOPS",ZG,ZO,ZP,ZS,ZIEN)=""
572 S ^C0X(101,"GOSP",ZG,ZO,ZS,ZP,ZIEN)=""
573 S ^C0X(101,"GPSO",ZG,ZP,ZS,ZO,ZIEN)=""
574 S ^C0X(101,"GPOS",ZG,ZP,ZO,ZS,ZIEN)=""
575 S ^C0X(101,"GSPO",ZG,ZS,ZP,ZO,ZIEN)=""
576 S ^C0X(101,"GSOP",ZG,ZS,ZO,ZP,ZIEN)=""
577 Q
578 ;
579REINDEX ; REINDEX THE ^C0X(101, TRIPLE STORE
580 K ^C0X(101,"B")
581 K ^C0X(101,"G")
582 K ^C0X(101,"SPO")
583 K ^C0X(101,"SOP")
584 K ^C0X(101,"OPS")
585 K ^C0X(101,"OSP")
586 K ^C0X(101,"PSO")
587 K ^C0X(101,"POS")
588 K ^C0X(101,"GOPS")
589 K ^C0X(101,"GOSP")
590 K ^C0X(101,"GPSO")
591 K ^C0X(101,"GPOS")
592 K ^C0X(101,"GSPO")
593 K ^C0X(101,"GSOP")
594 N ZIEN,ZZ
595 S ZIEN=0
596 F S ZIEN=$O(^C0X(101,ZIEN)) Q:+ZIEN=0 D ; FOR EACH NODE
597 . S ZZ=$G(^C0X(101,ZIEN,0))
598 . I ZZ="" D Q ;
599 . . W !,"ERROR REINDEXING NODE ",ZI
600 . S ZN=$P(ZZ,"^",1)
601 . S ZG=$P(ZZ,"^",2)
602 . S ZS=$P(ZZ,"^",3)
603 . S ZP=$P(ZZ,"^",4)
604 . S ZO=$P(ZZ,"^",5)
605 . D INDEX(ZIEN,ZN,ZG,ZS,ZP,ZO)
606 Q
607 ;
608BLKERR ;
609 W !,"ERROR IN BULK LOAD"
610 S C0XERR="ERROR IN BULK LOAD"
611 S C0XLOC=ZBFDA(ZI)
612 D ^%ZTER ; report the error
613 B
614 Q
615 ;
616DELGRAPH(ZGRF,FARY) ; delete a graph from the triplestore
617 ; (doesn't delete strings)
618 ;
619 I '$D(FARY) D ;
620 . D INITFARY("C0XFARY")
621 . S FARY="C0XFARY"
622 D USEFARY(FARY)
623 N ZGRAPH
624 D TING(.ZGRAPH,ZGRF,FARY)
625 I '$D(ZGRAPH) D Q ;
626 . W !,"NO TRIPLES IN GRAPH"
627 K C0XFDA
628 N ZI S ZI=""
629 F S ZI=$O(ZGRAPH(ZI)) Q:ZI="" D ;
630 . S C0XFDA(C0XTFN,ZI_",",.01)="@"
631 D UPDIE(.C0XFDA)
632 Q
633 ;
634TING(ZRTN,ZGRF,FARY) ; return the iens for graph ZGRF
635 ; ZRTN is passed by reference
636 I '$D(FARY) D ;
637 . D INITFARY("C0XFARY")
638 . S FARY="C0XFARY"
639 D USEFARY(FARY)
640 K ZRTN
641 N ZI,ZG S ZI=""
642 S ZG=$$IENOF(ZGRF)
643 I ZG="" D Q ;
644 . W !,"ERROR GRAPH NOT FOUND"
645 I '$D(@C0XTN@("G",ZG)) Q ;
646 F S ZI=$O(@C0XTN@("G",ZG,ZI)) Q:ZI="" D ;
647 . S ZRTN(ZI)=""
648 Q
649 ;
650SWUPDIE(ZFDA) ; SWITCH BETWEEN UPDIE AND BULKLOAD
651 I $G(BLKLOAD)=1 D ; bulk load
652 . D BULKLOAD(.ZFDA) ; bulk load the batch
653 E D ; no bulk load
654 . D UPDIE(.ZFDA)
655 K ZFDA
656 Q
657 ;
658UPDIE(ZFDA) ; INTERNAL ROUTINE TO CALL UPDATE^DIE AND CHECK FOR ERRORS
659 ; ZFDA IS PASSED BY REFERENCE
660 ;ZWR ZFDA
661 ;B
662 K ZERR
663 D CLEAN^DILF
664 D UPDATE^DIE("","ZFDA","","ZERR")
665 I $D(ZERR) S ZZERR=ZZERR ; ZZERR DOESN'T EXIST,
666 ; INVOKE THE ERROR TRAP IF TASKED
667 ;. W "ERROR",!
668 ;. ZWR ZERR
669 ;. B
670 K ZFDA
671 Q
672 ;
Note: See TracBrowser for help on using the repository browser.