Home   Package List   Routine Alphabetical List   Global Alphabetical List   FileMan Files List   FileMan Sub-Files List   Package Component Lists   Package-Namespace Mapping  
Routine: MXMLPRSE

MXMLPRSE.m

Go to the documentation of this file.
  1. MXMLPRSE ;SAIC/DKM - XML Parser ;09/08/08 11:50
  1. ;;7.3;TOOLKIT;**58,67,89,116,136**;Apr 25, 1995;Build 5
  1. ;Per VHA Directive 6402, this routine should not be modified
  1. ;=================================================================
  1. ; Main entry point.
  1. ; DOC = Closed reference to global array containing document
  1. ; CBK = Local array containing entry points for callback interface
  1. ; OPTION = Option flags; expected values are:
  1. ; D = Debug mode
  1. ; W = Do not report warnings
  1. ; V = Validate (checks only well-formedness by default)
  1. ; 0,1 = Terminate on encountering error at specified level
  1. EN(DOC,CBK,OPTION) ;
  1. N WS,ID,QT,EDC,DTD,LVL,CS,DOCSTK,LLEN,LPOS,CPOS,LCUR,ERR,XML,PFX,SFX,EOD,EOG,ST,PATH,OFX
  1. S ID=$T(+0),WS=$C(9,10,13,32),QT="""",(DOCSTK,EOG,EOD,LVL,CS,ST,LPOS,LLEN,LCUR)=0,(CPOS,LVL(0,"N"))=1,OPTION=$G(OPTION),(XML,PFX,SFX)="",PATH=$$PATH(DOC)
  1. K ^TMP(ID,$J)
  1. I $L($T(TOUCH^XUSCLEAN)) D TOUCH^XUSCLEAN ;Set the keepalive node
  1. D INIT^MXMLPRS1,EPOS,CBK("STARTDOCUMENT"),OPNDOC(DOC)
  1. F Q:EOD D READ,EPOS,@ST^MXMLPRS0:'EOD
  1. D UNRESLV^MXMLPRS1,ERROR(17):ST'=2,CBK("ENDDOCUMENT")
  1. K ^TMP(ID,$J)
  1. Q
  1. ; Open a document
  1. ; Saves state of current document on stack.
  1. ; DOCREF=Closed reference to array containing document
  1. ; PREFIX=Optional prefix to prepend to document
  1. ; SUFFIX=Optional suffix to append to document
  1. OPNDOC(DOCREF,PREFIX,SUFFIX) ;
  1. S:$E(DOCREF)'="^" DOCREF=$$EXTRNL(DOCREF)
  1. Q:'$L(DOCREF)
  1. D SAVRES(1)
  1. S DOC=$NA(@DOCREF)
  1. I '$D(^TMP(ID,$J,"DOC",DOC)) S ^(DOC)=""
  1. E D ERROR(43)
  1. S (LPOS,LLEN,LCUR)=0,CPOS=1,(OFX,XML)="",PFX=$G(PREFIX),SFX=$G(SUFFIX)
  1. S LCUR=DOC,DOC=$E(DOC,1,$L(DOC)-1) ;*rwf
  1. Q
  1. ; Close current document
  1. ; Restores state of previous document from stack.
  1. CLSDOC K ^TMP(ID,$J,"DOC",DOC_")") ;*rwf
  1. D SAVRES(0)
  1. Q
  1. ; Extract path from filespec
  1. PATH(DOC) ;
  1. N X
  1. Q:U[$E(DOC) ""
  1. F X="\","/","]",":","" Q:DOC[X
  1. Q $P(DOC,X,1,$L(DOC,X)-1)_X
  1. ; Save or restore document state
  1. SAVRES(SAVE) ;
  1. N X
  1. S:'SAVE DOCSTK=DOCSTK-1,EOD=DOCSTK=0
  1. I DOCSTK F X="LLEN","LPOS","CPOS","LCUR","XML","PFX","SFX","OFX","DOC" D
  1. .I SAVE S DOCSTK(DOCSTK,X)=@X
  1. .E S @X=DOCSTK(DOCSTK,X)
  1. I SAVE S DOCSTK=DOCSTK+1
  1. E K DOCSTK(DOCSTK)
  1. Q
  1. ; Retrieve text from document
  1. READ Q:((LLEN-CPOS)>50)!EOD ;Quit if still have 50 char or EOD
  1. I (CPOS'<LLEN)&EOG D CLSDOC S EOG=0 Q ;At end of text in file
  1. N TMP,X
  1. D SHIFT Q:$L(XML)>50
  1. I EOG!EOD Q ;Quit at end of document
  1. S LPOS=LPOS+1,LCUR=$Q(@LCUR) ;Get next node
  1. I LCUR'[DOC S EOG=1 Q ;At end of global
  1. S TMP=@LCUR ;Get next data chunk
  1. W:OPTION["D" !,$J(LPOS,3)_":",TMP,!
  1. S OFX=OFX_TMP
  1. D SHIFT
  1. I LLEN<50 G READ
  1. Q
  1. ;Shift OFX to XML
  1. SHIFT ;
  1. S XML=$E(XML,CPOS,9999),CPOS=1 ;Drop old
  1. I $L(PFX) S OFX=XML_OFX,XML=PFX,PFX=""
  1. I $L(OFX) S X=511-$L(XML),XML=XML_$E(OFX,1,X),OFX=$E(OFX,X+1,99999)
  1. S LLEN=$L(XML)
  1. Q
  1. ; Parse name
  1. ; ERN=Error to signal if invalid (optional)
  1. NAME(ERN) ;
  1. N X
  1. D EPOS
  1. S X=$E(XML,CPOS)
  1. I X'?1A,"_:"'[X D:$G(ERN) ERROR(ERN,X) Q ""
  1. Q $$NAMETKN(.ERN)
  1. ; Parse name token
  1. ; ERN=Error to signal if invalid (optional)
  1. NAMETKN(ERN) ;
  1. N X,Y
  1. D EPOS
  1. F X=CPOS:1:LLEN+1 S Y=$E(XML,X) I Y'?1AN,".-_:"'[Y Q
  1. S Y=$E(XML,CPOS,X-1),CPOS=X
  1. I '$L(Y),$G(ERN) D ERROR(ERN,Y)
  1. Q Y
  1. ; Parse quote-enclosed value
  1. ; ERF=If set, signal error if not found
  1. ; FLG=Special flag: 0=attribute literal, 1=general entity literal
  1. ; 2=parameter entity literal
  1. ; Returns value less quotes with normalized whitespace
  1. VALUE(ERF,FLG) ;
  1. N DLM,CHR,RTN,EXC
  1. D WS()
  1. S DLM=$S($$NEXT(QT):QT,$$NEXT("'"):"'",1:""),RTN="",FLG=+$G(FLG),EXC=$S(FLG=2:"",1:"<")
  1. I DLM="" D:$G(ERF) EPOS,ERROR(11) Q RTN
  1. F S CHR=$E(XML,CPOS) Q:DLM=CHR!(EXC[CHR)!EOD D
  1. .I $$NEXT("&#") S RTN=RTN_$$CHENTITY
  1. .E I 'FLG,$$NEXT("&") S RTN=RTN_$$ENTITY
  1. .E S RTN=RTN_CHR,CPOS=CPOS+1
  1. .;D:CPOS>LLEN READ
  1. .D:(LLEN-CPOS)<50 READ ;P136
  1. I DLM=CHR S CPOS=CPOS+1
  1. E D EPOS,ERROR($S('$L(CHR):12,EXC[CHR:13,1:12)) Q ""
  1. Q $$NMLWS(RTN)
  1. ; Normalize whitespace
  1. ; Note: used as input transform for Entity Catalog, so can't depend
  1. ; on any environment variables.
  1. ; TXT=Text to normalize
  1. ; Returns text stripped of leading and trailing whitespace and with
  1. ; imbedded contiguous whitespace reduced to single space.
  1. NMLWS(TXT,FG) ;
  1. N Z,CRLF
  1. S CRLF=$C(13,10)
  1. ;Normalize CRLF to one SP first
  1. F S Z=$F(TXT,CRLF) Q:'Z S TXT=$P(TXT,CRLF,1)_" "_$P(TXT,CRLF,2,999)
  1. S TXT=$TR(TXT,$C(9,10,13,32)," ")
  1. ;For CDATA or unk, this is where we should stop
  1. Q:'$G(FG) TXT
  1. F Z=1:1 Q:$E(TXT,Z)'=" "
  1. S TXT=$E(TXT,Z,9999)
  1. F Z=$L(TXT):-1 Q:$E(TXT,Z)'=" "
  1. S TXT=$E(TXT,1,Z)
  1. F Z=1:1:$L(TXT) D:$E(TXT,Z)=" "
  1. .F Q:$E(TXT,Z+1)'=" " S $E(TXT,Z+1)=""
  1. Q TXT
  1. ; Process parameter entity if found
  1. DOPARAM F D WS() Q:EOD!'$$NEXT("%") I $$ENTITY(1)
  1. Q
  1. ; Resolve general/parameter/character entity
  1. ; PARAM=1: parameter; PARAM=0: general or character (default)
  1. ENTITY(PARAM) ;
  1. N NAME,APND
  1. S PARAM=+$G(PARAM)
  1. I 'PARAM,$$NEXT("#") Q $$CHENTITY
  1. S NAME=$S(PARAM:"%",1:"")_$$NAME(2)
  1. Q:'$$NEXT(";",3) ""
  1. ;Handle the common ones inline
  1. S APND=$S(NAME="amp":"&",NAME="lt":"<",NAME="gt":">",NAME="quot":$C(34),NAME="apos":"'",1:"")
  1. Q:$L(APND) APND
  1. I $D(^TMP(ID,$J,"UNP",NAME)) D ERROR(40,NAME) Q ""
  1. I '$D(^TMP(ID,$J,"ENT",NAME)) D ERROR(14,NAME) Q ""
  1. S APND=$S(PARAM:" ",1:"")
  1. D OPNDOC(^TMP(ID,$J,"ENT",NAME),APND,APND)
  1. Q ""
  1. ; Parse character entity reference
  1. ; Returns character equivalent
  1. CHENTITY() ;
  1. N DIGIT,BASE,DIGITS,VAL
  1. S BASE=$S($$NEXT("x"):16,1:10),DIGITS="0123456789"_$S(BASE=16:"ABCDEF",1:""),VAL=0
  1. F CPOS=CPOS:1:LLEN+1 Q:$$NEXT(";")!EOD D
  1. .S DIGIT=$F(DIGITS,$$UP^XLFSTR($E(XML,CPOS)))-2,VAL=VAL*BASE+DIGIT
  1. .D:DIGIT<0 ERROR(19)
  1. I VAL<32,WS'[$C(VAL) D ERROR(19)
  1. Q $C(VAL)
  1. ; Set an entity value
  1. SETENT(NAME,VAL) ;
  1. K ^TMP(ID,$J,"ENT",NAME)
  1. S ^(NAME)=$NA(^(NAME)),^(NAME,1)=VAL
  1. Q
  1. ; Process all attributes
  1. ATTRIBS(ENAME,ATTR) ;
  1. N TYP,MOD,DEF,ANAME
  1. K ATTR
  1. F Q:'$$ATTRIB(ENAME,.ATTR)
  1. I OPTION["V" D
  1. .S ANAME="$"
  1. .F S ANAME=$O(^TMP(ID,$J,"ATT",ENAME,ANAME)) Q:'$L(ANAME) D
  1. ..S TYP=^(ANAME),MOD=$P(TYP,"^",2),DEF=$P(TYP,"^",3,9999),TYP=+TYP
  1. ..I MOD=1!(MOD=3),'$D(ATTR(ANAME)) D ERROR(36,ANAME) Q
  1. ..I MOD=3,ATTR(ANAME)'=DEF D ERROR(37,ATTR(ANAME)) Q
  1. ..I MOD=2,'$D(ATTR(ANAME)) Q
  1. ..S:'$D(ATTR(ANAME)) ATTR(ANAME)=DEF
  1. Q
  1. ; Parse attribute=value sequence
  1. ; ENAME=Element name to which attribute belongs
  1. ; ATTR=Local array (by reference) to receive attribute value.
  1. ; Format is ATTR("<attribute name>")="<attribute value>"
  1. ; Returns 1 if successful, 0 if not.
  1. ATTRIB(ENAME,ATTR) ;
  1. N ANAME
  1. D READ,WS() ;p116
  1. S ANAME=$$NAME
  1. Q:ANAME="" 0
  1. I $D(ATTR(ANAME)) D ERROR(4,ANAME) Q 0
  1. D:'$D(^TMP(ID,$J,"ATT",ENAME,ANAME)) ERROR(29,ANAME)
  1. D READ,WS() ;p116
  1. Q:'$$NEXT("=",3) 0
  1. D WS()
  1. S ATTR(ANAME)=$$VALUE(1)
  1. D CHKVAL^MXMLPRS1(ENAME,ANAME,ATTR(ANAME))
  1. Q 1
  1. ; Parse a processing instruction
  1. ; Returns 1 if PI found, 0 if not.
  1. PI() N PNAME,ARGS,DONE
  1. Q:'$$NEXT("<?") 0
  1. S PNAME=$$NAME(2),ARGS=0
  1. I $$UP^XLFSTR(PNAME)="XML" D ERROR(9) Q 0
  1. D WS(1)
  1. F S DONE=$F(XML,"?>",CPOS) D Q:DONE!EOD
  1. .S ARGS=ARGS+1,ARGS(ARGS)=$E(XML,CPOS,$S(DONE:DONE-3,1:LLEN))
  1. .S CPOS=$S(DONE:DONE,1:LLEN+1)
  1. .D READ
  1. I EOD D ERROR(7) Q 0
  1. D CBK("PI",PNAME,.ARGS)
  1. Q 1
  1. ; Parse a comment
  1. ; Returns 1 if comment found, 0 if not.
  1. COMMENT() Q $$PARSCT("<!--","--",">","COMMENT")
  1. ; Parse a CDATA section
  1. ; Returns 1 if found, 0 if not.
  1. CDATA() Q $$PARSCT("<![CDATA[","]]>","","CHARACTERS")
  1. ; Parse a section (for CDATA and COMMENT)
  1. ; BGN=Beginning delimiter
  1. ; END=Ending delimiter
  1. ; TRL=Trailing delimiter
  1. ; TYP=Event type
  1. PARSCT(BGN,END,TRL,TYP) ;
  1. N X
  1. Q:'$$NEXT(BGN) 0
  1. D EPOS
  1. I 'LVL,TYP'="COMMENT" D ERROR(6) Q 0
  1. F S X=$F(XML,END,CPOS) D Q:X!EOD
  1. .D CBK(TYP,$E(XML,CPOS,$S(X:X-$L(END)-1,1:LLEN)))
  1. .S CPOS=$S(X:X,1:LLEN+1)
  1. .D READ,EPOS
  1. I EOD D ERROR(7) Q 0
  1. I $L(TRL),$$NEXT(TRL,3)
  1. Q 1
  1. ; Fetch an external entity from file or entity catalog
  1. ; SYS=System identifier (i.e., a URL)
  1. ; PUB=Public identifier (i.e., Entity Catalog ID) - optional
  1. ; GBL=Optional global root to receive entity content
  1. ; Returns global reference or null if error
  1. EXTRNL(SYS,PUB,GBL) ;
  1. N X,Y
  1. S PUB=$$NMLWS($G(PUB)),GBL=$G(GBL)
  1. I '$L(GBL) D CBK("EXTERNAL",.SYS,.PUB,.GBL) Q:$L(GBL) GBL
  1. I $L(PUB) D Q:X $NA(^MXML(950,X,1))
  1. .S Y=$E(PUB,1,30),X=0
  1. .F S X=$O(^MXML(950,"B",Y,X)) Q:'X Q:$G(^MXML(950,X,0))=PUB
  1. S:'$L(GBL) GBL=$$TMPGBL
  1. S:$$PATH(SYS)="" SYS=PATH_SYS
  1. S X=$S($$FTG^%ZISH(SYS,"",$NA(@GBL@(1)),$QL(GBL)+1):GBL,1:"")
  1. D:'$L(X) ERROR(30,$S($L(SYS):SYS,1:PUB))
  1. Q X
  1. ; Return a unique scratch global reference
  1. TMPGBL() N SUB
  1. S SUB=$O(^TMP(ID,$J,$C(1)),-1)+1,^(SUB)=""
  1. Q $NA(^(SUB))
  1. ; Returns a SYSTEM and/or PUBLIC id
  1. ; SYS=Returned SYSTEM id
  1. ; PUB=Returned PUBLIC id
  1. ; FLG=If set, SYSTEM id is optional after PUBLIC id
  1. ; Optional return value: 0=neither, 1=PUBLIC, 2=SYSTEM
  1. SYSPUB(SYS,PUB,FLG) ;
  1. N RTN
  1. I $$NEXT("PUBLIC") D
  1. .D WS(1)
  1. .S PUB=$$VALUE(1),SYS=$$VALUE('$G(FLG)),RTN=1
  1. E I $$NEXT("SYSTEM") D
  1. .D WS(1)
  1. .S PUB="",SYS=$$VALUE(1),RTN=2
  1. E S (SYS,PUB)="",RTN=0
  1. Q:$Q RTN
  1. Q
  1. ; Save current document location for error reporting
  1. ; See EPOS^MXMLPRS0
  1. EPOS S ERR("XML")=XML,ERR("POS")=CPOS,ERR("LIN")=LPOS
  1. Q
  1. ; Setup error information
  1. ERROR(ERN,ARG) ;
  1. N DIHELP,DIMSG,DIERR,MSG
  1. D BLD^DIALOG(9500000+ERN,"","","MSG","")
  1. S ERR("NUM")=ERN
  1. S ERR("SEV")=$S($G(DIHELP):0,$G(DIMSG):1,1:2)
  1. S ERR("MSG")=$G(MSG(1))
  1. S ERR("ARG")=$G(ARG)
  1. I OPTION'["W"!ERR("SEV"),OPTION["V"!(ERR("SEV")'=1) D CBK("ERROR",.ERR)
  1. S:ERR("SEV")=2!(OPTION[ERR("SEV")) EOD=-1 ; Stop parsing on severe error
  1. Q
  1. ; Shortcuts to functions/procedures defined elsewhere
  1. WS(X) Q:$Q $$WS^MXMLPRS0(.X)
  1. D WS^MXMLPRS0(.X) Q
  1. CBK(X,Y1,Y2,Y3,Y4) D CBK^MXMLPRS0(.X,.Y1,.Y2,.Y3,.Y4) Q
  1. NEXT(X,Y) Q $$NEXT^MXMLPRS0(.X,.Y)