| 1 | C0XMAIN ; GPL - Fileman Triples entry point routine ;10/13/11 17:05
|
|---|
| 2 | ;;0.1;C0X;nopatch;noreleasedate;Build 1
|
|---|
| 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 | ;
|
|---|
| 32 | INITFARY(ZFARY) ; INITIALIZE FILE NUMBERS AND OTHER USEFUL THINGS
|
|---|
| 33 | ; FOR THE DEFAULT TRIPLE STORE. USE OTHER VALUES FOR SUPPORTING ADDITIONAL
|
|---|
| 34 | ; TRIPLE STORES
|
|---|
| 35 | S @ZFARY@("C0XTFN")=172.101 ; TRIPLES FILE NUMBER
|
|---|
| 36 | S @ZFARY@("C0XSFN")=172.201 ; TRIPLES STRINGS FILE NUMBER
|
|---|
| 37 | S @ZFARY@("C0XTN")=$NA(^C0X(101)) ; TRIPLES GLOBAL NAME
|
|---|
| 38 | S @ZFARY@("C0XSN")=$NA(^C0X(201)) ; STRING FILE GLOBAL NAME
|
|---|
| 39 | S @ZFARY@("C0XDIR")="/home/glilly/rdf/"
|
|---|
| 40 | D USEFARY(ZFARY)
|
|---|
| 41 | Q
|
|---|
| 42 | ;
|
|---|
| 43 | USEFARY(ZFARY) ; INITIALIZES VARIABLES SAVED IN ARRAY ZFARY
|
|---|
| 44 | N ZI S ZI=""
|
|---|
| 45 | F S ZI=$O(@ZFARY@(ZI)) Q:ZI="" D
|
|---|
| 46 | . ;N ZX
|
|---|
| 47 | . S ZX="S "_ZI_"="""_@ZFARY@(ZI)_""""
|
|---|
| 48 | . ;W !,ZX
|
|---|
| 49 | . X ZX
|
|---|
| 50 | Q
|
|---|
| 51 | ;
|
|---|
| 52 | FILEIN ; INTERACTIVE ENTRY POINT FOR OPTION TO READ IN A FILE
|
|---|
| 53 | I '$D(C0XFARY) D INITFARY("C0XFARY")
|
|---|
| 54 | D USEFARY("C0XFARY")
|
|---|
| 55 | S DIR(0)="F^3:240"
|
|---|
| 56 | S DIR("A")="File Directory"
|
|---|
| 57 | S DIR("B")=C0XDIR
|
|---|
| 58 | D ^DIR
|
|---|
| 59 | I Y="^" Q ;
|
|---|
| 60 | S C0XDIR=Y
|
|---|
| 61 | S C0XFARY("C0XDIR")=Y
|
|---|
| 62 | S DIR(0)="F^3:240"
|
|---|
| 63 | S DIR("A")="File Name"
|
|---|
| 64 | I '$D(C0XFN) S DIR("B")="qds.rdf"
|
|---|
| 65 | E S DIR("B")=C0XFN
|
|---|
| 66 | D ^DIR
|
|---|
| 67 | I Y="" Q ;
|
|---|
| 68 | I Y="^" Q ;
|
|---|
| 69 | S C0XFN=Y
|
|---|
| 70 | D IMPORT(C0XFN,C0XDIR,,"C0XFARY")
|
|---|
| 71 | Q
|
|---|
| 72 | ;
|
|---|
| 73 | IMPORT(FNAME,FDIR,FURL,FARY) ; EXTRINSIC THAT READS A FILE FROM THE STANDARD
|
|---|
| 74 | ; DIRECTORY, LOADS IT INTO THE TRIPLESTORE AS TEXT, AND RETURNS THE
|
|---|
| 75 | ; NODE NAME OF THE TEXT TRIPLE
|
|---|
| 76 | ; FDIR IS THE OPTIONAL DIRECTORY (DEFAUTS TO STANDARD DIR)
|
|---|
| 77 | ; FURL IS THE OPTIONAL URI FOR ACCESSING THE FILE FROM THE TRIPLE STORE
|
|---|
| 78 | ; FARY IS THE OPTIONAL FILE ARRAY OF THE TRIPLE STORE TO USE
|
|---|
| 79 | I '$D(FARY) D ;
|
|---|
| 80 | . D INITFARY("C0XFARY")
|
|---|
| 81 | . S FARY="C0XFARY"
|
|---|
| 82 | D USEFARY(FARY)
|
|---|
| 83 | N ZD,ZTMP
|
|---|
| 84 | I '$D(FDIR) S FDIR=C0XDIR ; DIRECTORY OF THE RDF FILE
|
|---|
| 85 | I '$D(FURL) D ;
|
|---|
| 86 | . N ZN2 S ZN2=$TR(FNAME,".","_") ; REMOVE THE DOT FROM THE NAME
|
|---|
| 87 | . S FURL=FDIR_ZN2
|
|---|
| 88 | N ZTMP
|
|---|
| 89 | S ZTMP=$NA(^TMP("C0X",$J,"FILEIN",1)) ; WHERE TO PUT THE INCOMING FILE
|
|---|
| 90 | K @ZTMP ; MAKE SURE IT'S CLEAR
|
|---|
| 91 | S C0XSTART=$$NOW^XLFDT
|
|---|
| 92 | W !,"STARTED: ",C0XSTART
|
|---|
| 93 | W !,"READING IN: ",FNAME
|
|---|
| 94 | I '$$FILEREAD(ZTMP,FDIR,FNAME,4) D Q ; QUIT IF NO SUCCESS
|
|---|
| 95 | . W !,"ERROR READING FILE: ",FDIR,FNAME
|
|---|
| 96 | S ZRDF=$NA(^TMP("C0X",$J,"FILEIN")) ; WITHOUT THE SUBSCRIPT
|
|---|
| 97 | W !,$O(@ZRDF@(""),-1)," LINES READ"
|
|---|
| 98 | D INSRDF(ZRDF,FURL,FARY) ; IMPORT AND PROCESS THE RDF
|
|---|
| 99 | Q
|
|---|
| 100 | ;
|
|---|
| 101 | WGET(ZURL,FARY) ; GET FROM THE INTERNET AN RDF FILE AND INSERT IT
|
|---|
| 102 | ;
|
|---|
| 103 | I '$D(FARY) D ;
|
|---|
| 104 | . D INITFARY("C0XFARY")
|
|---|
| 105 | . S FARY="C0XFARY"
|
|---|
| 106 | D USEFARY(FARY)
|
|---|
| 107 | ;N ZLOC,ZTMP
|
|---|
| 108 | K ZTMP
|
|---|
| 109 | S ZLOC=$NA(^TMP("C0X","WGET",$J))
|
|---|
| 110 | K @ZLOC
|
|---|
| 111 | S C0XSTART=$$NOW^XLFDT
|
|---|
| 112 | W !,"STARTED: ",C0XSTART
|
|---|
| 113 | W !,"DOWNLOADING: ",ZURL
|
|---|
| 114 | S OK=$$httpGET^%zewdGTM(ZURL,.ZTMP)
|
|---|
| 115 | M @ZLOC=ZTMP
|
|---|
| 116 | S C0XLINES=$O(@ZLOC@(""),-1)
|
|---|
| 117 | W !,C0XLINES," LINES READ"
|
|---|
| 118 | S C0XDLC=$$NOW^XLFDT ; DOWNLOAD COMPLETE
|
|---|
| 119 | W !,"DOWNLOAD COMPLETE AT ",C0XDLC
|
|---|
| 120 | S C0XDIFF=$$FMDIFF^XLFDT(C0XDLC,C0XSTART,2)
|
|---|
| 121 | W !," ELAPSED TIME: ",C0XDIFF," SECONDS"
|
|---|
| 122 | I C0XDIFF'=0 W !," APPROXIMATELY ",$P(C0XLINES/C0XDIFF,".")," LINES PER SEC"
|
|---|
| 123 | D INSRDF(ZLOC,ZURL,FARY)
|
|---|
| 124 | Q
|
|---|
| 125 | ;
|
|---|
| 126 | INSRDF(ZRDF,ZNAME,FARY) ; INSERT AN RDF FILE INTO THE STORE AND PROCESS
|
|---|
| 127 | ; ZRDF IS PASSED BY NAME
|
|---|
| 128 | I '$D(FARY) D ;
|
|---|
| 129 | . D INITFARY("C0XFARY")
|
|---|
| 130 | . S FARY="C0XFARY"
|
|---|
| 131 | D USEFARY(FARY)
|
|---|
| 132 | N ZGRAPH,ZSUBJECT
|
|---|
| 133 | S ZGRAPH="_:G"_$$LKY9 ; RANDOM GRAPH NAME
|
|---|
| 134 | S ZSUBJECT=$$ANONS ; RANDOM ANOYMOUS SUBJECT
|
|---|
| 135 | D ADD(ZGRAPH,ZSUBJECT,"fmts:url",ZNAME,FARY)
|
|---|
| 136 | N ZTXTNM
|
|---|
| 137 | S ZTXTNM="_TXT_INCOMING_RDF_FILE_"_ZNAME_"_"_$$LKY9 ; NAME FOR TEXT NODE
|
|---|
| 138 | D ADD(ZGRAPH,ZSUBJECT,"fmts:rdfSource",ZTXTNM,FARY)
|
|---|
| 139 | D UPDIE(.C0XFDA) ; TRY IT OUT
|
|---|
| 140 | K C0XCNT ;RESET FOR NEXT TIME
|
|---|
| 141 | D STORETXT(ZRDF,ZTXTNM,FARY)
|
|---|
| 142 | W !,"ADDED: ",ZGRAPH," ",ZSUBJECT," fmts:rdfSource ",ZTXTNM
|
|---|
| 143 | D PROCESS2(.G,ZRDF,ZNAME,ZGRAPH,FARY) ; PARSE AND INSERT THE RDF
|
|---|
| 144 | Q
|
|---|
| 145 | ;
|
|---|
| 146 | STORETXT(ZTXT,ZNAME,FARY) ; STORE TEXT IN THE TRIPLESTORE AT ZNAME
|
|---|
| 147 | ;
|
|---|
| 148 | I '$D(FARY) D ;
|
|---|
| 149 | . D INITFARY("C0XFARY")
|
|---|
| 150 | . S FARY="C0XFARY"
|
|---|
| 151 | D USEFARY(FARY)
|
|---|
| 152 | N ZIEN
|
|---|
| 153 | S ZIEN=$$IENOF(ZNAME,FARY) ; GET THE IEN
|
|---|
| 154 | D CLEAN^DILF
|
|---|
| 155 | K ZERR
|
|---|
| 156 | D WP^DIE(C0XSFN,ZIEN_",",1,,ZTXT,"ZERR")
|
|---|
| 157 | I $D(ZERR) D ;
|
|---|
| 158 | . ZWR ZERR
|
|---|
| 159 | Q
|
|---|
| 160 | ;
|
|---|
| 161 | GETTXT(ZRTN,ZNAME,FARY) ; RETURNS RDF SOURCE OR OTHER TEXT
|
|---|
| 162 | ; ZRTN IS PASSED BY REFERENCE
|
|---|
| 163 | I '$D(FARY) D ;
|
|---|
| 164 | . D INITFARY("C0XFARY")
|
|---|
| 165 | . S FARY="C0XFARY"
|
|---|
| 166 | D USEFARY(FARY)
|
|---|
| 167 | N ZIEN
|
|---|
| 168 | S ZIEN=$$IENOF(ZNAME)
|
|---|
| 169 | S OK=$$GET1^DIQ(C0XSFN,ZIEN_",",1,,"ZRTN")
|
|---|
| 170 | Q
|
|---|
| 171 | ;
|
|---|
| 172 | WHERETXT(ZNAME,FARY) ; EXTRINSIC WHICH RETURNS THE NAME OF THE GLOBAL
|
|---|
| 173 | ; WHERE THE TEXT IS LOCATED. NAME IS THE NAME OF THE STRING
|
|---|
| 174 | I '$D(FARY) D ;
|
|---|
| 175 | . D INITFARY("C0XFARY")
|
|---|
| 176 | . S FARY="C0XFARY"
|
|---|
| 177 | D USEFARY(FARY)
|
|---|
| 178 | N ZIEN
|
|---|
| 179 | S ZIEN=$$IENOF(ZNAME)
|
|---|
| 180 | Q $NA(@C0XSN@(ZIEN,1))
|
|---|
| 181 | ;
|
|---|
| 182 | FILEREAD(ZINTMP,ZDIR,ZFNAME,ZLVL) ; READS A FILE INTO ZINTMP USING FTG^%ZISH
|
|---|
| 183 | ; ZINTMP IS PASSED BY NAME AND INCLUDES THE NEW SUBSCRIPT
|
|---|
| 184 | ; IE ^TMP("C0X","FILEIN",1)
|
|---|
| 185 | ; ZLVL IN THIS CASE WOULD BE 3 INCREMENTING THE 1
|
|---|
| 186 | ; EXTRINSIC WHICH RETURNS THE RESULT OF FTG^%ZISH
|
|---|
| 187 | S OK=$$FTG^%ZISH(ZDIR,FNAME,ZINTMP,ZLVL)
|
|---|
| 188 | Q OK
|
|---|
| 189 | ;
|
|---|
| 190 | TESTPROC ; TEST PROCESS WITH EXISTING SMALL RDF FILE
|
|---|
| 191 | S ZIN=$NA(^TMP("C0X",12226,"FILEIN"))
|
|---|
| 192 | S ZGRAPH="/test/rdfFile"
|
|---|
| 193 | S ZM="/test/rdfFile/meta"
|
|---|
| 194 | D PROCESS(.G,ZIN,ZGRAPH,ZM)
|
|---|
| 195 | Q
|
|---|
| 196 | ;
|
|---|
| 197 | PROCESS(ZRTN,ZRDF,ZGRF,ZMETA,FARY) ; PROCESS AN INCOMING RDF FILE
|
|---|
| 198 | ; ZRTN IS PASS BY REFERENCE AND RETURNS MESSAGES ABOUT THE PROCESSING
|
|---|
| 199 | ; ZRDF IS PASSED BY NAME AND IS THE GLOBAL CONTAINING THE RDF FILE
|
|---|
| 200 | ; ZGRF IS THE NAME OF THE GRAPH TO USE IN THE TRIPLE STORE FOR RESULTS
|
|---|
| 201 | ; ZMETA IS OPTIONAL AND IS THE NAME OF THE GRAPH TO STORE METADATA
|
|---|
| 202 | ;
|
|---|
| 203 | I '$D(FARY) D ;
|
|---|
| 204 | . D INITFARY("C0XFARY")
|
|---|
| 205 | . S FARY="C0XFARY"
|
|---|
| 206 | D USEFARY(FARY)
|
|---|
| 207 | ; -- first parse the rdf file with the MXML parser
|
|---|
| 208 | ;S C0XDOCID=$$PARSE^C0CNHIN(ZRDF,"C0XARRAY") ; PARSE WITH MXML
|
|---|
| 209 | S C0XDOCID=$$EN^MXMLDOM(ZRDF,"W")
|
|---|
| 210 | ; -- assign the MXLM dom global name to ZDOM
|
|---|
| 211 | S ZDOM=$NA(^TMP("MXMLDOM",$J,C0XDOCID))
|
|---|
| 212 | W !,$O(@ZDOM@(""),-1)," XML NODES PARSED"
|
|---|
| 213 | ; -- populate the metagraph to point to the graph with status unfinished
|
|---|
| 214 | S METAS=$$ANONS ; GET AN ANONOMOUS RANDOM SUBJECT
|
|---|
| 215 | I '$D(ZMETA) S ZMETA="_:G"_$$LKY9 ; RANDOM GRAPH NAME FOR METAGRAPH
|
|---|
| 216 | D ADD(ZMETA,METAS,"fmts:about",ZGRF,FARY) ; POINT THE META TO THE GRAPH
|
|---|
| 217 | D ADD(ZMETA,METAS,"fmts:status","unfinished",FARY) ; mark as unfinished
|
|---|
| 218 | ;S C0XDATE=$$FMDTOUTC^C0CUTIL($$NOW^XLFDT,"DT")
|
|---|
| 219 | S C0XDATE=$$NOW^XLFDT
|
|---|
| 220 | D ADD(ZMETA,METAS,"fmts:dateTime",C0XDATE,FARY)
|
|---|
| 221 | D UPDIE(.C0XFDA) ; commit the metagraph changes to the triple store
|
|---|
| 222 | ; --
|
|---|
| 223 | ; -- pull out the vocabularies in the RDF statement. marked with xmlns:
|
|---|
| 224 | ; -- put them in a local variable for quick reference
|
|---|
| 225 | ; -- TODO: create a graph for vocabularies and validate incoming against it
|
|---|
| 226 | ;
|
|---|
| 227 | S C0XVOC=""
|
|---|
| 228 | N ZI,ZJ,ZK S ZI=""
|
|---|
| 229 | F S ZI=$O(@ZDOM@(1,"A",ZI)) Q:ZI="" D ; FOR EACH xmlns
|
|---|
| 230 | . S ZVOC=$P(ZI,"xmlns:",2)
|
|---|
| 231 | . I ZVOC'="" S C0XVOC(ZVOC)=$G(@ZDOM@(1,"A",ZI))
|
|---|
| 232 | ;W !,"VOCABS:" ZWR C0XVOC
|
|---|
| 233 | ;
|
|---|
| 234 | ; -- look for children called rdf:Description. quit if none. not an rdf file
|
|---|
| 235 | ;
|
|---|
| 236 | S ZI=$O(@ZDOM@(1,"C",""))
|
|---|
| 237 | I $G(@ZDOM@(1,"C",ZI))'="rdf:Description" D Q ; not an rdf file
|
|---|
| 238 | . W !,"Error. Not an RDF file. Cannot process."
|
|---|
| 239 | ;
|
|---|
| 240 | ; -- now process the rdf description children
|
|---|
| 241 | ;
|
|---|
| 242 | S ZI=""
|
|---|
| 243 | S (C0XSUB,C0XPRE,C0XOBJ)="" ; INITIALIZE subject, object and predicate
|
|---|
| 244 | F S ZI=$O(@ZDOM@(1,"C",ZI)) Q:ZI="" D ;
|
|---|
| 245 | . ; -- we are skipping any child that is not rdf:Description
|
|---|
| 246 | . ; -- TODO: check to see if this is right in general
|
|---|
| 247 | . ;
|
|---|
| 248 | . IF $G(@ZDOM@(1,"C",ZI))'="rdf:Description" D Q ;
|
|---|
| 249 | . . W !,"SKIPPING NODE: ",ZI
|
|---|
| 250 | . ; -- now looking for the subject for the triples
|
|---|
| 251 | . S ZX=$G(@ZDOM@(ZI,"A","rdf:about"))
|
|---|
| 252 | . I ZX'="" D ; we have the subject
|
|---|
| 253 | . . ;W " about: ",ZX
|
|---|
| 254 | . . S C0XSUB=ZX
|
|---|
| 255 | . E D ;
|
|---|
| 256 | . . S ZX=$G(@ZDOM@(ZI,"A","rdf:nodeID")) ; node id is another style of subject
|
|---|
| 257 | . . I ZX'="" D ;
|
|---|
| 258 | . . . S C0XSUB=ZX
|
|---|
| 259 | . I C0XSUB="" S C0XSUB=$$ANONS ; DEFAULT TO BLANK SUBJECT
|
|---|
| 260 | . ;
|
|---|
| 261 | . ; -- we now have the subject. the children of this node have the rest
|
|---|
| 262 | . ;
|
|---|
| 263 | . S ZJ="" ; for the children of the rdf:Description nodes
|
|---|
| 264 | . F S ZJ=$O(@ZDOM@(ZI,"C",ZJ)) Q:ZJ="" D ; for each child
|
|---|
| 265 | . . S C0XPRE=@ZDOM@(ZJ) ; the predicate without a prefix
|
|---|
| 266 | . . S ZX=$G(@ZDOM@(ZJ,"A","xmlns")) ; name space
|
|---|
| 267 | . . I ZX'="" S C0XPRE=ZX_C0XPRE ; add the namespace prefix
|
|---|
| 268 | . . I C0XPRE[":" D ; expand using vocabulary
|
|---|
| 269 | . . . N ZB,ZA
|
|---|
| 270 | . . . S ZB=$P(C0XPRE,":",1)
|
|---|
| 271 | . . . S ZA=$P(C0XPRE,":",2)
|
|---|
| 272 | . . . I $G(C0XVOC(ZB))'="" D ;
|
|---|
| 273 | . . . . S C0XPRE=C0XVOC(ZB)_ZA ; expanded
|
|---|
| 274 | . . S ZY=$G(@ZDOM@(ZJ,"A","rdf:resource")) ; potential object
|
|---|
| 275 | . . I ZY'="" D Q ;
|
|---|
| 276 | . . . S C0XOBJ=ZY ; object
|
|---|
| 277 | . . . D ADD(ZGRF,C0XSUB,C0XPRE,C0XOBJ) ; finally. our first real triple
|
|---|
| 278 | . . ; -- this is an else because of the quit above
|
|---|
| 279 | . . S ZX=$G(@ZDOM@(ZJ,"A","rdf:nodeID")) ; fishing for nodeId object
|
|---|
| 280 | . . I ZX'="" D Q ; got one
|
|---|
| 281 | . . . S C0XOBJ=ZX ; we are using the incoming nodeIDs as object/subject
|
|---|
| 282 | . . . ; without change... this could be foolish .. look at it again later
|
|---|
| 283 | . . . D ADD(ZGRF,C0XSUB,C0XPRE,C0XOBJ) ; go for it and add a node
|
|---|
| 284 | . . S C0XOBJ=$G(@ZDOM@(ZJ,"T",1)) ; hopefully an object is here
|
|---|
| 285 | . . I C0XOBJ="" D Q ; not a happy situation
|
|---|
| 286 | . . . W !,"ERROR, NO OBJECT FOUND FOR NODE: ",ZJ
|
|---|
| 287 | . . D ADD(ZGRF,C0XSUB,C0XPRE,C0XOBJ) ; go for it and add a node
|
|---|
| 288 | W !,"INSERTING ",C0XCNT," TRIPLES"
|
|---|
| 289 | D UPDIE(.C0XFDA) ; commit the updates to the file
|
|---|
| 290 | ; next, mark the graph as finished
|
|---|
| 291 | S C0XEND=$$NOW^XLFDT
|
|---|
| 292 | W !," ENDED AT: ",C0XEND
|
|---|
| 293 | S C0XDIFF=$$FMDIFF^XLFDT(C0XEND,C0XSTART,2)
|
|---|
| 294 | W !," ELAPSED TIME: ",C0XDIFF," SECONDS"
|
|---|
| 295 | I C0XDIFF'=0 D ;
|
|---|
| 296 | . W !," APPROXIMATELY ",$P(C0XCNT/C0XDIFF,".")," TRIPLES PER SECOND"
|
|---|
| 297 | Q
|
|---|
| 298 | ;
|
|---|
| 299 | PROCESS2(ZRTN,ZRDF,ZGRF,ZMETA,FARY) ; PROCESS AN INCOMING RDF FILE
|
|---|
| 300 | ; ZRTN IS PASS BY REFERENCE AND RETURNS MESSAGES ABOUT THE PROCESSING
|
|---|
| 301 | ; ZRDF IS PASSED BY NAME AND IS THE GLOBAL CONTAINING THE RDF FILE
|
|---|
| 302 | ; ZGRF IS THE NAME OF THE GRAPH TO USE IN THE TRIPLE STORE FOR RESULTS
|
|---|
| 303 | ; ZMETA IS OPTIONAL AND IS THE NAME OF THE GRAPH TO STORE METADATA
|
|---|
| 304 | ;
|
|---|
| 305 | I '$D(FARY) D ;
|
|---|
| 306 | . D INITFARY("C0XFARY")
|
|---|
| 307 | . S FARY="C0XFARY"
|
|---|
| 308 | D USEFARY(FARY)
|
|---|
| 309 | ;N BATCNT
|
|---|
| 310 | ;N BATMAX
|
|---|
| 311 | S BATCNT=0 ; BATCH COUNTER
|
|---|
| 312 | S BATMAX=10000 ; TRY BATCHES OF THIS SIZE
|
|---|
| 313 | ; -- first parse the rdf file with the MXML parser
|
|---|
| 314 | ;S C0XDOCID=$$PARSE^C0CNHIN(ZRDF,"C0XARRAY") ; PARSE WITH MXML
|
|---|
| 315 | S C0XDLC2=$$NOW^XLFDT ; START OF PARSE
|
|---|
| 316 | S C0XDOCID=$$EN^MXMLDOM(ZRDF,"W")
|
|---|
| 317 | ; -- assign the MXLM dom global name to ZDOM
|
|---|
| 318 | S ZDOM=$NA(^TMP("MXMLDOM",$J,C0XDOCID))
|
|---|
| 319 | S C0XNODE=$O(@ZDOM@(""),-1)
|
|---|
| 320 | W !,C0XNODE," XML NODES PARSED"
|
|---|
| 321 | S C0XPRS=$$NOW^XLFDT ; PARSE COMPLETE
|
|---|
| 322 | W !,"PARSE COMPLETE AT ",C0XPRS
|
|---|
| 323 | S C0XDIFF=$$FMDIFF^XLFDT(C0XPRS,C0XDLC2,2)
|
|---|
| 324 | W !," ELAPSED TIME: ",C0XDIFF," SECONDS"
|
|---|
| 325 | I C0XDIFF'=0 D ;
|
|---|
| 326 | . W !," APPROXIMATELY ",$P(C0XNODE/C0XDIFF,".")," NODES PER SECOND"
|
|---|
| 327 | ; -- populate the metagraph to point to the graph with status unfinished
|
|---|
| 328 | S METAS=$$ANONS ; GET AN ANONOMOUS RANDOM SUBJECT
|
|---|
| 329 | I '$D(ZMETA) S ZMETA="_:G"_$$LKY9 ; RANDOM GRAPH NAME FOR METAGRAPH
|
|---|
| 330 | D ADD(ZMETA,METAS,"fmts:about",ZGRF,FARY) ; POINT THE META TO THE GRAPH
|
|---|
| 331 | D ADD(ZMETA,METAS,"fmts:status","unfinished",FARY) ; mark as unfinished
|
|---|
| 332 | ;S C0XDATE=$$FMDTOUTC^C0CUTIL($$NOW^XLFDT,"DT")
|
|---|
| 333 | S C0XDATE=$$NOW^XLFDT
|
|---|
| 334 | D ADD(ZMETA,METAS,"fmts:dateTime",C0XDATE,FARY)
|
|---|
| 335 | D UPDIE(.C0XFDA) ; commit the metagraph changes to the triple store
|
|---|
| 336 | ; --
|
|---|
| 337 | ; -- pull out the vocabularies in the RDF statement. marked with xmlns:
|
|---|
| 338 | ; -- put them in a local variable for quick reference
|
|---|
| 339 | ; -- TODO: create a graph for vocabularies and validate incoming against it
|
|---|
| 340 | ;
|
|---|
| 341 | S C0XVOC=""
|
|---|
| 342 | N ZI,ZJ,ZK S ZI=""
|
|---|
| 343 | F S ZI=$O(@ZDOM@(1,"A",ZI)) Q:ZI="" D ; FOR EACH xmlns
|
|---|
| 344 | . S ZVOC=$P(ZI,"xmlns:",2)
|
|---|
| 345 | . I ZVOC'="" S C0XVOC(ZVOC)=$G(@ZDOM@(1,"A",ZI))
|
|---|
| 346 | ;W !,"VOCABS:" ZWR C0XVOC
|
|---|
| 347 | ;
|
|---|
| 348 | ; -- look for children called rdf:Description. quit if none. not an rdf file
|
|---|
| 349 | ;
|
|---|
| 350 | S ZI=$O(@ZDOM@(1,"C",""))
|
|---|
| 351 | I $G(@ZDOM@(1,"C",ZI))'="rdf:Description" D Q ; not an rdf file
|
|---|
| 352 | . W !,"Error. Not an RDF file. Cannot process."
|
|---|
| 353 | ;
|
|---|
| 354 | ; -- now process the rdf description children
|
|---|
| 355 | ;
|
|---|
| 356 | S ZI=""
|
|---|
| 357 | S (C0XSUB,C0XPRE,C0XOBJ)="" ; INITIALIZE subject, object and predicate
|
|---|
| 358 | F S ZI=$O(@ZDOM@(1,"C",ZI)) Q:ZI="" D ;
|
|---|
| 359 | . ; -- we are skipping any child that is not rdf:Description
|
|---|
| 360 | . ; -- TODO: check to see if this is right in general
|
|---|
| 361 | . ;
|
|---|
| 362 | . IF $G(@ZDOM@(1,"C",ZI))'="rdf:Description" D Q ;
|
|---|
| 363 | . . W !,"SKIPPING NODE: ",ZI
|
|---|
| 364 | . ; -- now looking for the subject for the triples
|
|---|
| 365 | . S ZX=$G(@ZDOM@(ZI,"A","rdf:about"))
|
|---|
| 366 | . I ZX'="" D ; we have the subject
|
|---|
| 367 | . . ;W " about: ",ZX
|
|---|
| 368 | . . S C0XSUB=ZX
|
|---|
| 369 | . E D ;
|
|---|
| 370 | . . S ZX=$G(@ZDOM@(ZI,"A","rdf:nodeID")) ; node id is another style of subject
|
|---|
| 371 | . . I ZX'="" D ;
|
|---|
| 372 | . . . S C0XSUB=ZX
|
|---|
| 373 | . I C0XSUB="" S C0XSUB=$$ANONS ; DEFAULT TO BLANK SUBJECT
|
|---|
| 374 | . ;
|
|---|
| 375 | . ; -- we now have the subject. the children of this node have the rest
|
|---|
| 376 | . ;
|
|---|
| 377 | . S ZJ="" ; for the children of the rdf:Description nodes
|
|---|
| 378 | . F S ZJ=$O(@ZDOM@(ZI,"C",ZJ)) Q:ZJ="" D ; for each child
|
|---|
| 379 | . . S C0XPRE=@ZDOM@(ZJ) ; the predicate without a prefix
|
|---|
| 380 | . . S ZX=$G(@ZDOM@(ZJ,"A","xmlns")) ; name space
|
|---|
| 381 | . . I ZX'="" S C0XPRE=ZX_C0XPRE ; add the namespace prefix
|
|---|
| 382 | . . I C0XPRE[":" D ; expand using vocabulary
|
|---|
| 383 | . . . N ZB,ZA
|
|---|
| 384 | . . . S ZB=$P(C0XPRE,":",1)
|
|---|
| 385 | . . . S ZA=$P(C0XPRE,":",2)
|
|---|
| 386 | . . . I $G(C0XVOC(ZB))'="" D ;
|
|---|
| 387 | . . . . S C0XPRE=C0XVOC(ZB)_ZA ; expanded
|
|---|
| 388 | . . S ZY=$G(@ZDOM@(ZJ,"A","rdf:resource")) ; potential object
|
|---|
| 389 | . . I ZY'="" D Q ;
|
|---|
| 390 | . . . S C0XOBJ=ZY ; object
|
|---|
| 391 | . . . D ADD2(ZGRF,C0XSUB,C0XPRE,C0XOBJ) ; finally. our first real triple
|
|---|
| 392 | . . ; -- this is an else because of the quit above
|
|---|
| 393 | . . S ZX=$G(@ZDOM@(ZJ,"A","rdf:nodeID")) ; fishing for nodeId object
|
|---|
| 394 | . . I ZX'="" D Q ; got one
|
|---|
| 395 | . . . S C0XOBJ=ZX ; we are using the incoming nodeIDs as object/subject
|
|---|
| 396 | . . . ; without change... this could be foolish .. look at it again later
|
|---|
| 397 | . . . D ADD2(ZGRF,C0XSUB,C0XPRE,C0XOBJ) ; go for it and add a node
|
|---|
| 398 | . . S C0XOBJ=$G(@ZDOM@(ZJ,"T",1)) ; hopefully an object is here
|
|---|
| 399 | . . I C0XOBJ="" D Q ; not a happy situation
|
|---|
| 400 | . . . W !,"ERROR, NO OBJECT FOUND FOR NODE: ",ZJ
|
|---|
| 401 | . . D ADD2(ZGRF,C0XSUB,C0XPRE,C0XOBJ) ; go for it and add a node
|
|---|
| 402 | S C0XTRP=$$NOW^XLFDT ; PARSE COMPLETE
|
|---|
| 403 | W !,"TRIPLES COMPLETE AT ",C0XTRP
|
|---|
| 404 | S C0XDIFF=$$FMDIFF^XLFDT(C0XTRP,C0XPRS,2)
|
|---|
| 405 | W !," ELAPSED TIME: ",C0XDIFF," SECONDS"
|
|---|
| 406 | I C0XDIFF'=0 D ;
|
|---|
| 407 | . W !," APPROXIMATELY ",$P(C0XCNT/C0XDIFF,".")," TRIPLES PER SECOND"
|
|---|
| 408 | W !,"INSERTING ",C0XCNT," TRIPLES"
|
|---|
| 409 | I $D(C0XFDA) D UPDIE(.C0XFDA) ; commit the updates to the file
|
|---|
| 410 | ; next, mark the graph as finished
|
|---|
| 411 | S C0XINS=$$NOW^XLFDT ; PARSE COMPLETE
|
|---|
| 412 | W !,"INSERTION COMPLETE AT ",C0XPRS
|
|---|
| 413 | S C0XDIFF=$$FMDIFF^XLFDT(C0XINS,C0XTRP,2)
|
|---|
| 414 | W !," ELAPSED TIME: ",C0XDIFF," SECONDS"
|
|---|
| 415 | W !," APPROXIMATELY ",$P(C0XCNT/C0XDIFF,".")," NODES PER SECOND"
|
|---|
| 416 | S C0XEND=$$NOW^XLFDT
|
|---|
| 417 | W !," ENDED AT: ",C0XEND
|
|---|
| 418 | S C0XDIFF=$$FMDIFF^XLFDT(C0XEND,C0XSTART,2)
|
|---|
| 419 | W !," ELAPSED TIME: ",C0XDIFF," SECONDS"
|
|---|
| 420 | W !," APPROXIMATELY ",$P(C0XCNT/C0XDIFF,".")," TRIPLES PER SECOND"
|
|---|
| 421 | Q
|
|---|
| 422 | ;
|
|---|
| 423 | SHOW(ZN) ;
|
|---|
| 424 | ZWR ^TMP("MXMLDOM",$J,1,ZN,*)
|
|---|
| 425 | Q
|
|---|
| 426 | ;
|
|---|
| 427 | ANONS() ; RETURNS AN ANONOMOUS SUBJECT
|
|---|
| 428 | Q "_S:"_$$LKY9
|
|---|
| 429 | ;
|
|---|
| 430 | NEWG(NGRAPH,NMETA) ; CREATES A NEW META GRAPH, MARKS IT AS UNFINISHED
|
|---|
| 431 | ; THEN CREATES A NEW GRAPH AND POINTS THE METAGRAPH TO IT
|
|---|
| 432 | ; NGRAPH AND NMETA ARE PASSED BY REFERENCE AND ARE THE RETURN
|
|---|
| 433 | S NGRAPH="G"_$$LKY9
|
|---|
| 434 | S NMETA=NGRAPH_"A"
|
|---|
| 435 | Q
|
|---|
| 436 | ;
|
|---|
| 437 | ADD(ZG,ZS,ZP,ZO,FARY) ; ADD A TRIPLE TO THE TRIPLESTORE. ALL VALUES ARE TEXT
|
|---|
| 438 | ; THE FDA IS SET UP BUT THE FILES ARE NOT UPDATED. CALL UPDIE TO COMPLETE
|
|---|
| 439 | I '$D(FARY) D ;
|
|---|
| 440 | . D INITFARY("C0XFARY")
|
|---|
| 441 | . S FARY="C0XFARY"
|
|---|
| 442 | D USEFARY(FARY)
|
|---|
| 443 | I '$D(C0XCNT) S C0XCNT=0
|
|---|
| 444 | N ZNODE
|
|---|
| 445 | S ZNODE="N"_$$LKY17
|
|---|
| 446 | N ZNARY ; GET READY TO CALL IENOFA
|
|---|
| 447 | S ZNARY("ZG",ZG)=""
|
|---|
| 448 | S ZNARY("ZS",ZS)=""
|
|---|
| 449 | S ZNARY("ZP",ZP)=""
|
|---|
| 450 | S ZNARY("ZO",ZO)=""
|
|---|
| 451 | D IENOFA(.ZIENS,.ZNARY,FARY) ; RESOLVE/ADD STRINGS
|
|---|
| 452 | ;S ZGIEN=$$IENOF(ZG) ; LAYGO TO GET IEN
|
|---|
| 453 | ;S ZSIEN=$$IENOF(ZS)
|
|---|
| 454 | ;S ZPIEN=$$IENOF(ZP)
|
|---|
| 455 | ;S ZOIEN=$$IENOF(ZO)
|
|---|
| 456 | ;I $D(C0XFDA) D UPDIE ; ADD THE STRINGS IF NEEDED
|
|---|
| 457 | S C0XCNT=C0XCNT+1
|
|---|
| 458 | S C0XFDA(C0XTFN,"?+"_C0XCNT_",",.01)=ZNODE
|
|---|
| 459 | S C0XFDA(C0XTFN,"?+"_C0XCNT_",",.02)=$O(ZIENS("IEN","ZG",""))
|
|---|
| 460 | S C0XFDA(C0XTFN,"?+"_C0XCNT_",",.03)=$O(ZIENS("IEN","ZS",""))
|
|---|
| 461 | S C0XFDA(C0XTFN,"?+"_C0XCNT_",",.04)=$O(ZIENS("IEN","ZP",""))
|
|---|
| 462 | S C0XFDA(C0XTFN,"?+"_C0XCNT_",",.05)=$O(ZIENS("IEN","ZO",""))
|
|---|
| 463 | ; REMEMBER TO CALL UPDIE WHEN YOU'RE DONE
|
|---|
| 464 | Q
|
|---|
| 465 | ;
|
|---|
| 466 | ADD2(ZG,ZS,ZP,ZO,FARY) ; ADD A TRIPLE TO THE TRIPLESTORE. ALL VALUES ARE TEXT
|
|---|
| 467 | ; THE FDA IS SET UP BUT THE FILES ARE NOT UPDATED. CALL UPDIE TO COMPLETE
|
|---|
| 468 | I '$D(FARY) D ;
|
|---|
| 469 | . D INITFARY("C0XFARY")
|
|---|
| 470 | . S FARY="C0XFARY"
|
|---|
| 471 | D USEFARY(FARY)
|
|---|
| 472 | I '$D(C0XCNT) S C0XCNT=0
|
|---|
| 473 | N ZNODE
|
|---|
| 474 | S ZNODE="N"_$$LKY17
|
|---|
| 475 | N ZNARY ; GET READY TO CALL IENOFA
|
|---|
| 476 | S ZNARY("ZG",ZG)=""
|
|---|
| 477 | S ZNARY("ZS",ZS)=""
|
|---|
| 478 | S ZNARY("ZP",ZP)=""
|
|---|
| 479 | S ZNARY("ZO",ZO)=""
|
|---|
| 480 | D IENOFA(.ZIENS,.ZNARY,FARY) ; RESOLVE/ADD STRINGS
|
|---|
| 481 | ;S ZGIEN=$$IENOF(ZG) ; LAYGO TO GET IEN
|
|---|
| 482 | ;S ZSIEN=$$IENOF(ZS)
|
|---|
| 483 | ;S ZPIEN=$$IENOF(ZP)
|
|---|
| 484 | ;S ZOIEN=$$IENOF(ZO)
|
|---|
| 485 | ;I $D(C0XFDA) D UPDIE ; ADD THE STRINGS IF NEEDED
|
|---|
| 486 | S BATCNT=BATCNT+1
|
|---|
| 487 | S C0XCNT=C0XCNT+1
|
|---|
| 488 | S C0XFDA(C0XTFN,"?+"_BATCNT_",",.01)=ZNODE
|
|---|
| 489 | S C0XFDA(C0XTFN,"?+"_BATCNT_",",.02)=$O(ZIENS("IEN","ZG",""))
|
|---|
| 490 | S C0XFDA(C0XTFN,"?+"_BATCNT_",",.03)=$O(ZIENS("IEN","ZS",""))
|
|---|
| 491 | S C0XFDA(C0XTFN,"?+"_BATCNT_",",.04)=$O(ZIENS("IEN","ZP",""))
|
|---|
| 492 | S C0XFDA(C0XTFN,"?+"_BATCNT_",",.05)=$O(ZIENS("IEN","ZO",""))
|
|---|
| 493 | I BATCNT=BATMAX D ; BATCH IS DONE
|
|---|
| 494 | . D UPDIE(.C0XFDA)
|
|---|
| 495 | . K C0XFDA
|
|---|
| 496 | . S BATCNT=0 ; RESET COUNTER
|
|---|
| 497 | ; REMEMBER TO CALL UPDIE WHEN YOU'RE DONE
|
|---|
| 498 | Q
|
|---|
| 499 | ;
|
|---|
| 500 | LKY9() ;EXTRINIC THAT RETURNS A RANDOM 9 DIGIT NUMBER. USED FOR GENERATING
|
|---|
| 501 | ; UNIQUE NODE AND GRAPH NAMES
|
|---|
| 502 | N ZN,ZI
|
|---|
| 503 | S ZN=""
|
|---|
| 504 | F ZI=1:1:9 D ;
|
|---|
| 505 | . S ZN=ZN_$R(10)
|
|---|
| 506 | Q ZN
|
|---|
| 507 | ;
|
|---|
| 508 | LKY17() ;EXTRINIC THAT RETURNS A RANDOM 9 DIGIT NUMBER. USED FOR GENERATING
|
|---|
| 509 | ; UNIQUE NODE AND GRAPH NAMES
|
|---|
| 510 | N ZN,ZI
|
|---|
| 511 | S ZN=""
|
|---|
| 512 | F ZI=1:1:17 D ;
|
|---|
| 513 | . S ZN=ZN_$R(10)
|
|---|
| 514 | Q ZN
|
|---|
| 515 | ;
|
|---|
| 516 | IENOF(ZSTRING,FARY) ; EXTRINSIC WHICH RETURNS THE IEN OF ZS IN THE STRINGS FILE
|
|---|
| 517 | I '$D(FARY) D ;
|
|---|
| 518 | . D INITFARY("C0XFARY")
|
|---|
| 519 | . S FARY="C0XFARY"
|
|---|
| 520 | N ZIEN
|
|---|
| 521 | S ZIEN=$O(@C0XSN@("B",ZSTRING,""))
|
|---|
| 522 | I ZIEN="" D ;
|
|---|
| 523 | . S C0XFDA2(C0XSFN,"+1,",.01)=ZSTRING
|
|---|
| 524 | . D UPDIE(.C0XFDA2)
|
|---|
| 525 | . S ZIEN=$O(@C0XSN@("B",ZSTRING,""))
|
|---|
| 526 | . K C0XFDA2
|
|---|
| 527 | Q ZIEN
|
|---|
| 528 | ;
|
|---|
| 529 | IENOFA(ZOUTARY,ZINARY,FARY) ; RESOLVE STRINGS TO IEN IN STRINGS FILE
|
|---|
| 530 | ; OR ADD THEM IF
|
|---|
| 531 | ; MISSING. ZINARY AND ZOUTARY ARE PASSED BY REFERENCE
|
|---|
| 532 | ; ZINARY LOOKS LIKE ZINARY("VAR","VAL")=""
|
|---|
| 533 | ; RETURNS IN ZOUTARY OF THE FORM ZOUTARY("IEN","VAR",IEN)=""
|
|---|
| 534 | I '$D(FARY) D ;
|
|---|
| 535 | . D INITFARY("C0XFARY")
|
|---|
| 536 | . S FARY="C0XFARY"
|
|---|
| 537 | K ZOUTARY ; START WITH CLEAN RESULTS
|
|---|
| 538 | K C0XFDA2 ; USE A SEPARATE FDA FOR THIS
|
|---|
| 539 | N ZI S ZI=""
|
|---|
| 540 | N ZV,ZIEN
|
|---|
| 541 | N ZCNT S ZCNT=0
|
|---|
| 542 | F S ZI=$O(ZINARY(ZI)) Q:ZI="" D ; LOOK FOR MISSING STRINGS
|
|---|
| 543 | . S ZV=$O(ZINARY(ZI,""))
|
|---|
| 544 | . I $O(@C0XSN@("B",ZV,""))="" D ;
|
|---|
| 545 | . . S ZCNT=ZCNT+1
|
|---|
| 546 | . . S C0XFDA2(C0XSFN,"+"_ZCNT_",",.01)=ZV
|
|---|
| 547 | I $D(C0XFDA2) D ;
|
|---|
| 548 | . D UPDIE(.C0XFDA2) ; ADD MISSING STRINGS
|
|---|
| 549 | . K C0XFDA2 ; CLEAN UP
|
|---|
| 550 | F S ZI=$O(ZINARY(ZI)) Q:ZI="" D ; NOT GET ALL IENS
|
|---|
| 551 | . S ZV=$O(ZINARY(ZI,""))
|
|---|
| 552 | . S ZIEN=$O(@C0XSN@("B",ZV,"")) ; THEY SHOULD BE THERE NOW
|
|---|
| 553 | . I ZIEN="" D ;
|
|---|
| 554 | . . W !,"ERROR ADDING STRING: ",ZV
|
|---|
| 555 | . . B
|
|---|
| 556 | . S ZOUTARY("IEN",ZI,ZIEN)=""
|
|---|
| 557 | Q
|
|---|
| 558 | ;
|
|---|
| 559 | UPDIE(ZFDA) ; INTERNAL ROUTINE TO CALL UPDATE^DIE AND CHECK FOR ERRORS
|
|---|
| 560 | ; ZFDA IS PASSED BY REFERENCE
|
|---|
| 561 | ;ZWR ZFDA
|
|---|
| 562 | ;B
|
|---|
| 563 | K ZERR
|
|---|
| 564 | D CLEAN^DILF
|
|---|
| 565 | D UPDATE^DIE("","ZFDA","","ZERR")
|
|---|
| 566 | I $D(ZERR) S ZZERR=ZZERR ; ZZERR DOESN'T EXIST,
|
|---|
| 567 | ; INVOKE THE ERROR TRAP IF TASKED
|
|---|
| 568 | ;. W "ERROR",!
|
|---|
| 569 | ;. ZWR ZERR
|
|---|
| 570 | ;. B
|
|---|
| 571 | K ZFDA
|
|---|
| 572 | Q
|
|---|
| 573 | ;
|
|---|
