source: fmts/trunk/p/C0XF2N.m@ 1380

Last change on this file since 1380 was 1380, checked in by George Lilly, 12 years ago

change for non-xml incoming files

File size: 20.6 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_"_"_$$LKY9 ; NAME FOR TEXT NODE
149 D ADD(ZGRAPH,ZSUBJECT,"fmts:rdfSource",ZTXTNM,FARY)
150 D SWUPDIE(.C0XFDA) ; TRY IT OUT
151 K C0XCNT ;RESET FOR NEXT TIME
152 D STORETXT(ZRDF,ZTXTNM,FARY)
153 W !,"ADDED: ",ZGRAPH," ",ZSUBJECT," fmts:rdfSource ",ZTXTNM
154 D PROCESS(.G,ZRDF,ZNAME,ZGRAPH,FARY) ; PARSE AND INSERT THE RDF
155 Q
156 ;
157STORETXT(ZTXT,ZNAME,FARY) ; STORE TEXT IN THE TRIPLESTORE AT ZNAME
158 ;
159 I '$D(FARY) D ;
160 . D INITFARY("C0XFARY")
161 . S FARY="C0XFARY"
162 D USEFARY(FARY)
163 N ZIEN
164 S ZIEN=$$IENOF(ZNAME,FARY) ; GET THE IEN
165 D CLEAN^DILF
166 K ZERR
167 D WP^DIE(C0XSFN,ZIEN_",",1,,ZTXT,"ZERR")
168 I $D(ZERR) D Q ;
169 . W !,"ERROR CREATING WORD PROCESSING FIELD"
170 . S C0XERR="ERROR CREATING WORD PROCESSING FIELD"
171 . D ^%ZTER ; error trap
172 Q
173 ;
174GETTXT(ZRTN,ZNAME,FARY) ; RETURNS RDF SOURCE OR OTHER TEXT
175 ; ZRTN IS PASSED BY REFERENCE
176 I '$D(FARY) D ;
177 . D INITFARY("C0XFARY")
178 . S FARY="C0XFARY"
179 D USEFARY(FARY)
180 N ZIEN
181 S ZIEN=$$IENOF(ZNAME)
182 S OK=$$GET1^DIQ(C0XSFN,ZIEN_",",1,,"ZRTN")
183 Q
184 ;
185WHERETXT(ZNAME,FARY) ; EXTRINSIC WHICH RETURNS THE NAME OF THE GLOBAL
186 ; WHERE THE TEXT IS LOCATED. NAME IS THE NAME OF THE STRING
187 I '$D(FARY) D ;
188 . D INITFARY("C0XFARY")
189 . S FARY="C0XFARY"
190 D USEFARY(FARY)
191 N ZIEN
192 S ZIEN=$$IENOF(ZNAME)
193 Q $NA(@C0XSN@(ZIEN,1))
194 ;
195FILEREAD(ZINTMP,ZDIR,ZFNAME,ZLVL) ; READS A FILE INTO ZINTMP USING FTG^%ZISH
196 ; ZINTMP IS PASSED BY NAME AND INCLUDES THE NEW SUBSCRIPT
197 ; IE ^TMP("C0X","FILEIN",1)
198 ; ZLVL IN THIS CASE WOULD BE 3 INCREMENTING THE 1
199 ; EXTRINSIC WHICH RETURNS THE RESULT OF FTG^%ZISH
200 S OK=$$FTG^%ZISH(ZDIR,FNAME,ZINTMP,ZLVL)
201 Q OK
202 ;
203TESTPROC ; TEST PROCESS WITH EXISTING SMALL RDF FILE
204 S ZIN=$NA(^TMP("C0X",12226,"FILEIN"))
205 S ZGRAPH="/test/rdfFile"
206 S ZM="/test/rdfFile/meta"
207 D PROCESS(.G,ZIN,ZGRAPH,ZM)
208 Q
209 ;
210VISTAOWL ;
211 S ZRDF=$NA(^TMP("C0X",542,"FILEIN"))
212 S ZNAME="/home/glilly/vistaowl/VistAOWL.owl"
213 S ZGRAPH="_:G431590209"
214 S FARY="C0XFARY"
215 D INITFARY(FARY)
216 S C0XDOCID=1
217 S BATCNT=0
218 S BATMAX=10000
219 D PROCESS(.G,ZRDF,ZGRAPH,ZNAME,FARY)
220 Q
221 ;
222PROCESS(ZRTN,ZRDF,ZGRF,ZMETA,FARY) ; PROCESS AN INCOMING RDF FILE
223 ; ZRTN IS PASS BY REFERENCE AND RETURNS MESSAGES ABOUT THE PROCESSING
224 ; ZRDF IS PASSED BY NAME AND IS THE GLOBAL CONTAINING THE RDF FILE
225 ; ZGRF IS THE NAME OF THE GRAPH TO USE IN THE TRIPLE STORE FOR RESULTS
226 ; ZMETA IS OPTIONAL AND IS THE NAME OF THE GRAPH TO STORE METADATA
227 ;
228 I '$D(FARY) D ;
229 . D INITFARY("C0XFARY")
230 . S FARY="C0XFARY"
231 D USEFARY(FARY)
232 ;N BATCNT
233 ;N BATMAX
234 ; -- first parse the rdf file with the MXML parser
235 ;S C0XDOCID=$$PARSE^C0CNHIN(ZRDF,"C0XARRAY") ; PARSE WITH MXML
236 S C0XDLC2=$$NOW^XLFDT ; START OF PARSE
237 I @ZRDF@(1)'["<?xml" D Q ;
238 . W !,"Not an XML file"
239 S C0XDOCID=$$EN^MXMLDOM(ZRDF,"W") ;
240 ;B
241 K @ZRDF ; DON'T NEED INPUT BUFFER ANYMORE
242 ; -- assign the MXLM dom global name to ZDOM
243 S ZDOM=$NA(^TMP("MXMLDOM",$J,C0XDOCID))
244 ;S ZDOM=$NA(^TMP("MXMLDOM",16850,C0XDOCID)) ;VISTAOWL DOM
245 S C0XNODE=$O(@ZDOM@(""),-1)
246 W !,C0XNODE," XML NODES PARSED"
247 S C0XPRS=$$NOW^XLFDT ; PARSE COMPLETE
248 W !,"PARSE COMPLETE AT ",C0XPRS
249 S C0XDIFF=$$FMDIFF^XLFDT(C0XPRS,C0XDLC2,2)
250 W !," ELAPSED TIME: ",C0XDIFF," SECONDS"
251 I C0XDIFF'=0 D ;
252 . W !," APPROXIMATELY ",$P(C0XNODE/C0XDIFF,".")," NODES PER SECOND"
253 ; -- populate the metagraph to point to the graph with status unfinished
254 S METAS=$$ANONS ; GET AN ANONOMOUS RANDOM SUBJECT
255 I '$D(ZMETA) S ZMETA="_:G"_$$LKY9 ; RANDOM GRAPH NAME FOR METAGRAPH
256 D ADD(ZMETA,METAS,"fmts:about",ZGRF,FARY) ; POINT THE META TO THE GRAPH
257 D ADD(ZMETA,METAS,"fmts:status","unfinished",FARY) ; mark as unfinished
258 W !,"INSERTING GRAPH: ",ZGRF
259 ;S C0XDATE=$$FMDTOUTC^C0CUTIL($$NOW^XLFDT,"DT")
260 S C0XDATE=$$NOW^XLFDT
261 D ADD(ZMETA,METAS,"fmts:dateTime",C0XDATE,FARY)
262 D SWUPDIE(.C0XFDA) ; commit the metagraph changes to the triple store
263 ; --
264 ; -- pull out the vocabularies in the RDF statement. marked with xmlns:
265 ; -- put them in a local variable for quick reference
266 ; -- TODO: create a graph for vocabularies and validate incoming against it
267 ;
268 S C0XVOC=""
269 N ZI,ZJ,ZK S ZI=""
270 F S ZI=$O(@ZDOM@(1,"A",ZI)) Q:ZI="" D ; FOR EACH xmlns
271 . S ZVOC=$P(ZI,"xmlns:",2)
272 . I ZVOC'="" S C0XVOC(ZVOC)=$G(@ZDOM@(1,"A",ZI))
273 W !,"VOCABS:"
274 N ZZ S ZZ=""
275 F S ZZ=$O(C0XVOC(ZZ)) Q:ZZ="" W !,ZZ,":",C0XVOC(ZZ)
276 ;
277 ; -- look for children called rdf:Description. quit if none. not an rdf file
278 ;
279 S C0XTYPE("rdf:Description")=1
280 S C0XTYPE("owl:ObjectProperty")=1
281 S C0XTYPE("owl:Ontology")=1
282 S C0XTYPE("owl:Class")=1
283 S C0XTYPE("rdfs:subClassOf")=1
284 S ZI=$O(@ZDOM@(1,"C",""))
285 I '$G(C0XTYPE(@ZDOM@(1,"C",ZI))) D Q ; not an rdf file
286 . W !,"Error. Not an RDF file. Cannot process."
287 . D SHOW(1)
288 ;
289 ; -- now process the rdf description children
290 ;
291 S ZI=""
292 S (C0XSUB,C0XPRE,C0XOBJ)="" ; INITIALIZE subject, object and predicate
293 F S ZI=$O(@ZDOM@(1,"C",ZI)) Q:ZI="" D ;
294 . ; -- we are skipping any child that is not rdf:Description
295 . ; -- TODO: check to see if this is right in general
296 . ;
297 . IF '$G(C0XTYPE(@ZDOM@(1,"C",ZI))) D Q ;
298 . . W !,"SKIPPING NODE: ",ZI
299 . ; -- now looking for the subject for the triples
300 . S ZX=$G(@ZDOM@(ZI,"A","rdf:about"))
301 . I ZX'="" D ; we have the subject
302 . . ;W " about: ",ZX
303 . . S C0XSUB=ZX
304 . E D ;
305 . . S ZX=$G(@ZDOM@(ZI,"A","rdf:nodeID")) ; node id is another style of subject
306 . . I ZX'="" D ;
307 . . . S C0XSUB=ZX
308 . I C0XSUB="" S C0XSUB=$$ANONS ; DEFAULT TO BLANK SUBJECT
309 . ;
310 . ; -- we now have the subject. the children of this node have the rest
311 . ;
312 . S ZJ="" ; for the children of the rdf:Description nodes
313 . F S ZJ=$O(@ZDOM@(ZI,"C",ZJ)) Q:ZJ="" D ; for each child
314 . . S C0XPRE=@ZDOM@(ZJ) ; the predicate without a prefix
315 . . S ZX=$G(@ZDOM@(ZJ,"A","xmlns")) ; name space
316 . . I ZX'="" S C0XPRE=ZX_C0XPRE ; add the namespace prefix
317 . . I C0XPRE[":" D ; expand using vocabulary
318 . . . N ZB,ZA
319 . . . S ZB=$P(C0XPRE,":",1)
320 . . . S ZA=$P(C0XPRE,":",2)
321 . . . I $G(C0XVOC(ZB))'="" D ;
322 . . . . S C0XPRE=C0XVOC(ZB)_ZA ; expanded
323 . . S ZY=$G(@ZDOM@(ZJ,"A","rdf:resource")) ; potential object
324 . . I ZY'="" D Q ;
325 . . . S C0XOBJ=$$EXT^C0XUTIL(ZY) ; object
326 . . . D ADD(ZGRF,C0XSUB,C0XPRE,C0XOBJ) ; finally. our first real triple
327 . . ; -- this is an else because of the quit above
328 . . S ZX=$G(@ZDOM@(ZJ,"A","rdf:nodeID")) ; fishing for nodeId object
329 . . I ZX'="" D Q ; got one
330 . . . S C0XOBJ=ZX ; we are using the incoming nodeIDs as object/subject
331 . . . ; without change... this could be foolish .. look at it again later
332 . . . D ADD(ZGRF,C0XSUB,C0XPRE,C0XOBJ) ; go for it and add a node
333 . . S C0XOBJ=$G(@ZDOM@(ZJ,"T",1)) ; hopefully an object is here
334 . . I C0XOBJ="" D Q ; not a happy situation
335 . . . W !,"ERROR, NO OBJECT FOUND FOR NODE: ",ZJ
336 . . S C0XOBJ=$$EXT^C0XUTIL(C0XOBJ) ; might be namespaced
337 . . D ADD(ZGRF,C0XSUB,C0XPRE,C0XOBJ) ; go for it and add a node
338 S C0XTRP=$$NOW^XLFDT ; PARSE COMPLETE
339 W !,"TRIPLES COMPLETE AT ",C0XTRP
340 S C0XDIFF=$$FMDIFF^XLFDT(C0XTRP,C0XPRS,2)
341 W !," ELAPSED TIME: ",C0XDIFF," SECONDS"
342 I C0XDIFF'=0 D ;
343 . W !," APPROXIMATELY ",$P(C0XCNT/C0XDIFF,".")," TRIPLES PER SECOND"
344 W !,"INSERTING ",C0XCNT," TRIPLES"
345 I $D(C0XFDA) D ;
346 . I $G(BLKLOAD) D ;
347 . . D BULKLOAD(.C0XFDA)
348 . E D ;
349 . . D UPDIE(.C0XFDA) ; commit the updates to the file
350 ; next, mark the graph as finished
351 S C0XINS=$$NOW^XLFDT ; PARSE COMPLETE
352 W !,"INSERTION COMPLETE AT ",C0XPRS
353 S C0XDIFF=$$FMDIFF^XLFDT(C0XINS,C0XTRP,2)
354 W !," ELAPSED TIME: ",C0XDIFF," SECONDS"
355 I C0XDIFF'=0 W !," APPROXIMATELY ",$P(C0XCNT/C0XDIFF,".")," NODES PER SECOND"
356 S C0XEND=$$NOW^XLFDT
357 W !," ENDED AT: ",C0XEND
358 S C0XDIFF=$$FMDIFF^XLFDT(C0XEND,C0XSTART,2)
359 W !," ELAPSED TIME: ",C0XDIFF," SECONDS"
360 I C0XDIFF'=0 W !," APPROXIMATELY ",$P(C0XCNT/C0XDIFF,".")," TRIPLES PER SECOND"
361 Q
362 ;
363SHOW(ZN) ;
364 I '$D(C0XJOB) S C0XJOB=$J
365 N ZD
366 S ZD=$NA(^TMP("MXMLDOM",C0XJOB,1,ZN))
367 W ZD,"=",@ZD
368 F S ZD=$Q(@ZD) Q:$QS(ZD,4)'=ZN W !,ZD,"=",@ZD
369 ;ZWR ^TMP("MXMLDOM",C0XJOB,1,ZN,*)
370 Q
371 ;
372ANONS() ; RETURNS AN ANONOMOUS SUBJECT
373 Q "iDPsDPss"_$$LKY9
374 ;
375NEWG(NGRAPH,NMETA) ; CREATES A NEW META GRAPH, MARKS IT AS UNFINISHED
376 ; THEN CREATES A NEW GRAPH AND POINTS THE METAGRAPH TO IT
377 ; NGRAPH AND NMETA ARE PASSED BY REFERENCE AND ARE THE RETURN
378 S NGRAPH="G"_$$LKY9
379 S NMETA=NGRAPH_"A"
380 Q
381 ;
382ADD(ZG,ZS,ZP,ZO,FARY) ; ADD A TRIPLE TO THE TRIPLESTORE. ALL VALUES ARE TEXT
383 ; THE FDA IS SET UP BUT THE FILES ARE NOT UPDATED. CALL UPDIE TO COMPLETE
384 I '$D(FARY) D ;
385 . D INITFARY("C0XFARY")
386 . S FARY="C0XFARY"
387 D USEFARY(FARY)
388 I '$D(C0XCNT) S C0XCNT=0
389 N ZNODE
390 S ZNODE="N"_$$LKY17
391 N ZNARY ; GET READY TO CALL IENOFA
392 S ZNARY("ZG",ZG)=""
393 S ZNARY("ZS",ZS)=""
394 S ZNARY("ZP",ZP)=""
395 S ZNARY("ZO",ZO)=""
396 D IENOFA(.ZIENS,.ZNARY,FARY) ; RESOLVE/ADD STRINGS
397 ;S ZGIEN=$$IENOF(ZG) ; LAYGO TO GET IEN
398 ;S ZSIEN=$$IENOF(ZS)
399 ;S ZPIEN=$$IENOF(ZP)
400 ;S ZOIEN=$$IENOF(ZO)
401 ;I $D(C0XFDA) D UPDIE ; ADD THE STRINGS IF NEEDED
402 I '$D(BATCNT) S BATCNT=0
403 S BATCNT=BATCNT+1
404 S C0XCNT=C0XCNT+1
405 I $G(BLKLOAD)=1 D ; we are using bulk load
406 . S C0XFDA(C0XTFN,BATCNT,.01)=ZNODE
407 . S C0XFDA(C0XTFN,BATCNT,.02)=$O(ZIENS("IEN","ZG",""))
408 . S C0XFDA(C0XTFN,BATCNT,.03)=$O(ZIENS("IEN","ZS",""))
409 . S C0XFDA(C0XTFN,BATCNT,.04)=$O(ZIENS("IEN","ZP",""))
410 . S C0XFDA(C0XTFN,BATCNT,.05)=$O(ZIENS("IEN","ZO",""))
411 E D ;
412 . S C0XFDA(C0XTFN,"?+"_BATCNT_",",.01)=ZNODE
413 . S C0XFDA(C0XTFN,"?+"_BATCNT_",",.02)=$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)=$O(ZIENS("IEN","ZO",""))
417 I '$D(BATMAX) S BATMAX=10000
418 I BATCNT=BATMAX D ; BATCH IS DONE
419 . I $G(BLKLOAD) D ; bulk load
420 . . D BULKLOAD(.C0XFDA) ; bulk load the batch
421 . E D ; no bulk load
422 . . D UPDIE(.C0XFDA)
423 . K C0XFDA
424 . S BATCNT=0 ; RESET COUNTER
425 ; REMEMBER TO CALL UPDIE WHEN YOU'RE DONE
426 Q
427 ;
428LKY9() ;EXTRINIC THAT RETURNS A RANDOM 9 DIGIT NUMBER. USED FOR GENERATING
429 ; UNIQUE NODE AND GRAPH NAMES
430 N ZN,ZI
431 S ZN=""
432 F ZI=1:1:9 D ;
433 . S ZN=ZN_$R(10)
434 Q ZN
435 ;
436LKY17() ;EXTRINIC THAT RETURNS A RANDOM 9 DIGIT NUMBER. USED FOR GENERATING
437 ; UNIQUE NODE AND GRAPH NAMES
438 N ZN,ZI
439 S ZN=""
440 F ZI=1:1:17 D ;
441 . S ZN=ZN_$R(10)
442 Q ZN
443 ;
444 ; these routines add the string if it is not found
445 ;
446IENOF(ZSTRING,FARY) ; EXTRINSIC WHICH RETURNS THE IEN OF ZS IN THE STRINGS FILE
447 I '$D(FARY) D ;
448 . D INITFARY("C0XFARY")
449 . S FARY="C0XFARY"
450 N ZIEN
451 I $G(ZSTRING)="" Q "" ; NO STRING
452 S ZIEN=$O(@C0XSN@("B",ZSTRING,""))
453 I ZIEN="" D ;
454 . S C0XFDA2(C0XSFN,"+1,",.01)=ZSTRING
455 . D UPDIE(.C0XFDA2)
456 . S ZIEN=$O(@C0XSN@("B",ZSTRING,""))
457 . K C0XFDA2
458 Q ZIEN
459 ;
460IENOFA(ZOUTARY,INARY,FARY) ; RESOLVE STRINGS TO IEN IN STRINGS FILE
461 ; OR ADD THEM IF
462 ; MISSING. ZINARY AND ZOUTARY ARE PASSED BY REFERENCE
463 ; ZINARY LOOKS LIKE ZINARY("VAR","VAL")=""
464 ; RETURNS IN ZOUTARY OF THE FORM ZOUTARY("IEN","VAR",IEN)=""
465 I '$D(FARY) D ;
466 . D INITFARY("C0XFARY")
467 . S FARY="C0XFARY"
468 K ZOUTARY ; START WITH CLEAN RESULTS
469 K C0XFDA2 ; USE A SEPARATE FDA FOR THIS
470 I '$D(C0XVOC) D VOCINIT^C0XUTIL
471 N ZINARY
472 N ZI S ZI=""
473 F S ZI=$O(INARY(ZI)) Q:ZI="" D ;
474 . N ZK
475 . S ZK=$O(INARY(ZI,""))
476 . S ZINARY($$EXT^C0XUTIL(ZI),$$EXT^C0XUTIL(ZK))=""
477 N ZV,ZIEN
478 N ZCNT S ZCNT=0
479 F S ZI=$O(ZINARY(ZI)) Q:ZI="" D ; LOOK FOR MISSING STRINGS
480 . S ZV=$O(ZINARY(ZI,""))
481 . I ZV["^" S ZV=$TR(ZV,"^","|")
482 . I $O(@C0XSN@("B",ZV,""))="" D ;
483 . . S ZCNT=ZCNT+1
484 . . S C0XFDA2(C0XSFN,"+"_ZCNT_",",.01)=ZV
485 I $D(C0XFDA2) D ;
486 . D UPDIE(.C0XFDA2) ; ADD MISSING STRINGS
487 . K C0XFDA2 ; CLEAN UP
488 F S ZI=$O(ZINARY(ZI)) Q:ZI="" D ; NOT GET ALL IENS
489 . S ZV=$O(ZINARY(ZI,""))
490 . I ZV["^" S ZV=$TR(ZV,"^","|")
491 . S ZIEN=$O(@C0XSN@("B",ZV,"")) ; THEY SHOULD BE THERE NOW
492 . I ZIEN="" D ;
493 . . W !,"ERROR ADDING STRING: ",ZV
494 . . B
495 . S ZOUTARY("IEN",ZI,ZIEN)=""
496 Q
497 ;
498BULKLOAD(ZBFDA) ; BULK LOADER FOR LOADING TRIPLES INTO FILE 172.101
499 ; USING GLOBAL SETS INSTEAD OF UPDATE^DIE
500 ; QUITS IF FILE IS NOT 172.101
501 ; EXPECTS AN FDA WITHOUT STRINGS FOR THE IENS, STARTING AT 1
502 ; QUITS IF FIRST ENTRY IS NOT IENS 1
503 ; ASSUMES THAT THE LAST IENS IS THE COUNT OF ENTRIES
504 ; ZBFDA IS PASSED BY REFERENCE
505 ;
506 ; -- reserves a block of iens from file 172.101 by locking the zero node
507 ; -- ^C0X(101,0) and adding the count of entries to piece 2 and 3
508 ; -- then unlocking to minimize the duration of the lock
509 ;
510 W !,"USING BULKLOAD"
511 I '$D(ZBFDA) Q ; EMPTY FDA
512 I $O(ZBFDA(""))'=172.101 Q ; WRONG FILE
513 N ZCNT,ZP3,ZP4
514 ; -- find the number of nodes to insert
515 S ZCNT=$O(ZBFDA(172.101,""),-1)
516 I ZCNT="" D Q ;
517 . W !,"ERROR IN BULK LOAD - INVALID NODE COUNT"
518 . B
519 ; -- lock the zero node and reserve a block of iens to insert
520 W !,"LOCKING ZERO NODE"
521 LOCK +^C0X(101,0)
522 S ZP3=$P(^C0X(101,0),U,3)
523 S ZP4=$P(^C0X(101,0),U,4)
524 S $P(^C0X(101,0),U,3)=ZP3+ZCNT+1
525 S $P(^C0X(101,0),U,4)=ZP4+ZCNT+1
526 LOCK -^C0X(101,0)
527 N ZI,ZN,ZG,ZS,ZP,ZO,ZIEN,ZBASE
528 S ZBASE=ZP3 ; the last ien in the file
529 W !,"ZERO NODE UNLOCKED, IENS RESERVED=",ZCNT
530 W !,$$NOW^XLFDT
531 S ZI=""
532 F S ZI=$O(ZBFDA(172.101,ZI)) Q:ZI="" D ;
533 . S ZN=$G(ZBFDA(172.101,ZI,.01)) ; node name
534 . I ZN="" D BLKERR Q ;
535 . S ZG=$G(ZBFDA(172.101,ZI,.02)) ; graph pointer
536 . I ZG="" D BLKERR Q ;
537 . S ZS=$G(ZBFDA(172.101,ZI,.03)) ; subject pointer
538 . I ZS="" D BLKERR Q ;
539 . S ZP=$G(ZBFDA(172.101,ZI,.04)) ; predicate pointer
540 . I ZP="" D BLKERR Q ;
541 . S ZO=$G(ZBFDA(172.101,ZI,.05)) ; object pointer
542 . I ZO="" D BLKERR Q ;
543 . S ZIEN=ZI+ZBASE ; the new ien
544 . S ^C0X(101,ZIEN,0)=ZN_U_ZG_U_ZS_U_ZP_U_ZO ; set the zero node
545 . D INDEX(ZIEN,ZN,ZG,ZS,ZP,ZO)
546 Q
547 ;
548INDEX(ZIEN,ZN,ZG,ZS,ZP,ZO) ; HARD SET THE INDEX FOR ONE ENTRY
549 S ^C0X(101,"B",ZN,ZIEN)="" ; the B index
550 S ^C0X(101,"G",ZG,ZIEN)="" ; the G for Graph index
551 S ^C0X(101,"SPO",ZS,ZP,ZO,ZIEN)=""
552 S ^C0X(101,"SOP",ZS,ZO,ZP,ZIEN)=""
553 S ^C0X(101,"OPS",ZO,ZP,ZS,ZIEN)=""
554 S ^C0X(101,"OSP",ZO,ZS,ZP,ZIEN)=""
555 S ^C0X(101,"PSO",ZP,ZS,ZO,ZIEN)=""
556 S ^C0X(101,"POS",ZP,ZO,ZS,ZIEN)=""
557 S ^C0X(101,"GOPS",ZG,ZO,ZP,ZS,ZIEN)=""
558 S ^C0X(101,"GOSP",ZG,ZO,ZS,ZP,ZIEN)=""
559 S ^C0X(101,"GPSO",ZG,ZP,ZS,ZO,ZIEN)=""
560 S ^C0X(101,"GPOS",ZG,ZP,ZO,ZS,ZIEN)=""
561 S ^C0X(101,"GSPO",ZG,ZS,ZP,ZO,ZIEN)=""
562 S ^C0X(101,"GSOP",ZG,ZS,ZO,ZP,ZIEN)=""
563 Q
564 ;
565REINDEX ; REINDEX THE ^C0X(101, TRIPLE STORE
566 K ^C0X(101,"B")
567 K ^C0X(101,"G")
568 K ^C0X(101,"SPO")
569 K ^C0X(101,"SOP")
570 K ^C0X(101,"OPS")
571 K ^C0X(101,"OSP")
572 K ^C0X(101,"PSO")
573 K ^C0X(101,"POS")
574 K ^C0X(101,"GOPS")
575 K ^C0X(101,"GOSP")
576 K ^C0X(101,"GPSO")
577 K ^C0X(101,"GPOS")
578 K ^C0X(101,"GSPO")
579 K ^C0X(101,"GSOP")
580 N ZIEN,ZZ
581 S ZIEN=0
582 F S ZIEN=$O(^C0X(101,ZIEN)) Q:+ZIEN=0 D ; FOR EACH NODE
583 . S ZZ=$G(^C0X(101,ZIEN,0))
584 . I ZZ="" D Q ;
585 . . W !,"ERROR REINDEXING NODE ",ZI
586 . S ZN=$P(ZZ,"^",1)
587 . S ZG=$P(ZZ,"^",2)
588 . S ZS=$P(ZZ,"^",3)
589 . S ZP=$P(ZZ,"^",4)
590 . S ZO=$P(ZZ,"^",5)
591 . D INDEX(ZIEN,ZN,ZG,ZS,ZP,ZO)
592 Q
593 ;
594BLKERR ;
595 W !,"ERROR IN BULK LOAD"
596 S C0XERR="ERROR IN BULK LOAD"
597 S C0XLOC=ZBFDA(ZI)
598 D ^%ZTER ; report the error
599 B
600 Q
601 ;
602DELGRAPH(ZGRF,FARY) ; delete a graph from the triplestore
603 ; (doesn't delete strings)
604 ;
605 I '$D(FARY) D ;
606 . D INITFARY("C0XFARY")
607 . S FARY="C0XFARY"
608 D USEFARY(FARY)
609 N ZGRAPH
610 D TING(.ZGRAPH,ZGRF,FARY)
611 I '$D(ZGRAPH) D Q ;
612 . W !,"NO TRIPLES IN GRAPH"
613 K C0XFDA
614 N ZI S ZI=""
615 F S ZI=$O(ZGRAPH(ZI)) Q:ZI="" D ;
616 . S C0XFDA(C0XTFN,ZI_",",.01)="@"
617 D UPDIE(.C0XFDA)
618 Q
619 ;
620TING(ZRTN,ZGRF,FARY) ; return the iens for graph ZGRF
621 ; ZRTN is passed by reference
622 I '$D(FARY) D ;
623 . D INITFARY("C0XFARY")
624 . S FARY="C0XFARY"
625 D USEFARY(FARY)
626 K ZRTN
627 N ZI,ZG S ZI=""
628 S ZG=$$IENOF(ZGRF)
629 I ZG="" D Q ;
630 . W !,"ERROR GRAPH NOT FOUND"
631 I '$D(@C0XTN@("G",ZG)) Q ;
632 F S ZI=$O(@C0XTN@("G",ZG,ZI)) Q:ZI="" D ;
633 . S ZRTN(ZI)=""
634 Q
635 ;
636SWUPDIE(ZFDA) ; SWITCH BETWEEN UPDIE AND BULKLOAD
637 I $G(BLKLOAD)=1 D ; bulk load
638 . D BULKLOAD(.ZFDA) ; bulk load the batch
639 E D ; no bulk load
640 . D UPDIE(.ZFDA)
641 K ZFDA
642 Q
643 ;
644UPDIE(ZFDA) ; INTERNAL ROUTINE TO CALL UPDATE^DIE AND CHECK FOR ERRORS
645 ; ZFDA IS PASSED BY REFERENCE
646 ;ZWR ZFDA
647 ;B
648 K ZERR
649 D CLEAN^DILF
650 D UPDATE^DIE("","ZFDA","","ZERR")
651 I $D(ZERR) S ZZERR=ZZERR ; ZZERR DOESN'T EXIST,
652 ; INVOKE THE ERROR TRAP IF TASKED
653 ;. W "ERROR",!
654 ;. ZWR ZERR
655 ;. B
656 K ZFDA
657 Q
658 ;
Note: See TracBrowser for help on using the repository browser.