The Systems Group at ETH Zurich. XML and Databases Exercise Session 12. courtesy of Ghislain Fourny. Department of Computer Science ETH Zürich

Transcription

1 ETH Zürich XML and Databases Exercise Session 12 The Systems Group at ETH Zurich courtesy of Ghislain Fourny

2 Where we are: XQuery Implementation

3 The problem we are solving today Square peg Round hole

4 The problem we are solving today

5 The problem we are solving today: Storing XML in an RDBMS <root>some <b>xml</b> document</root> Relational Database

6 Good news: we will use node IDs (remember last week) Duck Donald Issued 11/25/09 Expires 11/24/19 Born June 9th, 1934 Duckburg, Calisota Duck County The mayor Scrooge McDuck

7 XML Documents: no nesting

8 XML Documents: with nesting

9 How to put XML in a relational DB? With a schema: Schema-based shredding Without schema (generic): Tree encoding Edge approach

10 Schema-based Shredding without nesting

11 Schema-based Shredding without nesting

12 Schema-based Shredding without nesting airid name tax LHR London Heathrow 100 ZRH Zurich 150 PAR Paris 100 Doc name passportnum ber address Santa Claus Somewhere Santa Claus Somewhere flightref passref LX LX Passenger Reservation flightid seats date departure arrival source destination LX :15:00 10:01:00 ZRH PAR LX :00:00 13:00:00 PAR ZRH Flight

13 Schema-based Shredding without nesting

14 Schema-based Shredding without nesting: the DB schema Passenger(passportNumber, name, address) Airport(airpId, name, tax) Flight(flightId, date, departure, arrival, seats, source, destination) Reservation(flightRef, passref)

15 Schema-based Shredding with nesting (general case)

16 Schema-based Shredding: The DTD <!ELEMENT Doc(Airport+, Passenger+, Flight+, Reservation+ > <!ELEMENT Passenger(passportNumber, name, address) > <!ELEMENT Airport(airpId, name, tax) > <!ELEMENT Flight(flightId, date, departure, arrival, seats, Airport, Airport) > <!ELEMENT Reservation(resId, Flight, Passenger)> <!ELEMENT passportnumber #PCDATA> <!ELEMENT name #PCDATA> <!ELEMENT address #PCDATA> <!ATTLIST Airport airpid CDATA> <!ELEMENT tax #PCDATA> <!ATTLIST Flight flightid CDATA> <!ELEMENT date #PCDATA> <!ELEMENT departure #PCDATA> <!ELEMENT arrival #PCDATA> <!ELEMENT seats #PCDATA> <!ELEMENT resid #PCDATA>

17 Schema-based Shredding: The DTD graph

18 Schema-based Shredding: The element (sub)graphs Since there are no cycles, these are all subgraphs.

19 Schema-based Shredding: DB Schema for the Doc subgraph We can not put the children inline because they are not unique... 19

20 Schema-based Shredding: DB Schema for the Doc subgraph So we create special tables for them 20

21 Schema-based Shredding: DB Schema for the Doc subgraph Here, there are attributes we can put inline. 21

22 Schema-based Shredding: DB Schema for the Doc subgraph Those need a special table as well, because they can occur more than once.

23 Schema-based Shredding: The other subgraphs

24 Schema-based Shredding: Filling in the tables Doc.Flight. AirportID Doc.Flight. Airport.pare ntid Doc.Flight. Airport.airpI d Doc.Flight. Airport.nam e Doc.Flight. Airport.tax 1 1 ZRH Zurich PAR Paris PAR Paris position 4 2 LHR London Heathrow Doc.Flight.Airport

25 Edge

26 Edge Ordinal Source Label Target 1 1 Doc Airport airid v1 2 3 name v2 3 3 tax v3 Id v1 v2 Value ZRH Zurich Id Value v3 150

27 Tree encoding

28 Tree encoding

29 Tree encoding: one single table

30 Tree encoding pre size level kind prop frag 1 (many) 1 element Doc element Airport attribute airid attvalue ZRH element name text Zurich

31 Tree encoding pre size level kind prop frag 1 (many) 1 element Doc element Airport attribute airid attvalue ZRH element name text Zurich Ancestor relationship?

32 Tree encoding pre size level kind prop frag 1 (many) 1 element Doc element Airport attribute airid attvalue ZRH element name text Zurich Ancestor relationship? descendant.pre > ancestor.pre & descendant.pre <= ancestor.pre+ancestor.size

33 Tree encoding pre size level kind prop frag 1 (many) 1 element Doc element Airport attribute airid attvalue ZRH element name text Zurich 1 Parent relationship?

34 Tree encoding pre size level kind prop frag 1 (many) 1 element Doc element Airport attribute airid attvalue ZRH element name text Zurich 1 Parent relationship? child.pre > parent.pre & child.pre <= parent.pre+parent.size & child.level = parent.level + 1

35 The problem we are solving today: Storing XML in an RDBMS <root>some <b>xml</b> document</root> Relational Database

36 The problem we are solving today: Storing XML in an RDBMS <root>some <b>xml</b> document</root> XQuery Relational Database

37 The problem we are solving today: Storing XML in an RDBMS <root>some <b>xml</b> document</root> XQuery SQL Relational Database

38 Our Query "Determine all destinations of passenger Santa Claus"

39 XML Documents: Query for Schemabased Shredding (no nesting) Passenger(passportNumber, name, address) Airport(airpId, name, tax) Flight(flightId, date, departure, arrival, seats, source, destination) Reservation(flightRef, passportnumber)

40 XML Documents: Query for Schemabased Shredding (no nesting)

41 XML Documents: Schema-based Shredding (nesting)

42 XML Documents: Query for Schemabased Shredding (nesting)

43 Edge Ordinal Source Label Target 1 1 Doc Airport airid v1 2 3 name v2 3 3 tax v3 Id v1 v2 Value ZRH Zurich Id Value v3 150

44 XML Documents: Query for Edge Passenger Name Santa Claus

45 XML Documents: Query for Edge Reservation Passenger Name Santa Claus

46 XML Documents: Query for Edge Reservation Passenger Flight Airport[2] Name Santa Claus

47 XML Documents: Query for Edge

48 Tree encoding pre size level kind prop frag 1 (many) 1 element Doc element Airport attribute airid attvalue LHR element name text London Heathrow 1

49 XML Documents: Query for Tree Encoding Reservation Passenger Airport[2] Flight Name Santa Claus

50 XML Documents: Query for Tree Encoding pn offspring of p

51 Good luck with the exam!