Xml search

Discussion in 'XML' started by huamin_chen, Nov 1, 2012.

  1. huamin_chen

    huamin_chen Guest

    huamin_chen, Nov 1, 2012
    #1
    1. Advertisements

  2. huamin_chen

    Jongware Guest

    Did you generate these 1,000,002 lines of XML data, or is this from the
    real world?

    In case someone does not like downloading 57 megs of zipped file, or
    expanding it into 722 megs of rather pointless example lines: here is an
    abbreviated version:

    <?xml version="1.0" encoding="UTF-16"?>
    <Appdata>
    <Data Attr0="00" Attr1="01" Attr2="02" Attr3="03" Attr4="04" Attr5="05"
    Attr6="06" Attr7="07" Attr8="08" Attr9="09" Attr10="010" Attr11="011"
    Attr12="012" Attr13="013" Attr14="014" Attr15="015" Attr16="016"
    Attr17="017" Attr18="018" Attr19="019">Node_Number0</Data>
    <Data Attr0="10" Attr1="11" Attr2="12" Attr3="13" Attr4="14" Attr5="15"
    Attr6="16" Attr7="17" Attr8="18" Attr9="19" Attr10="110" Attr11="111"
    Attr12="112" Attr13="113" Attr14="114" Attr15="115" Attr16="116"
    Attr17="117" Attr18="118" Attr19="119">Node_Number1</Data>
    .... (999,998 similar lines omitted) ...
    <Data Attr0="9999980" Attr1="9999981" Attr2="9999982" Attr3="9999983"
    Attr4="9999984" Attr5="9999985" Attr6="9999986" Attr7="9999987"
    Attr8="9999988" Attr9="9999989" Attr10="99999810" Attr11="99999811"
    Attr12="99999812" Attr13="99999813" Attr14="99999814" Attr15="99999815"
    Attr16="99999816" Attr17="99999817" Attr18="99999818"
    Attr19="99999819">Node_Number999998</Data>
    <Data Attr0="9999990" Attr1="9999991" Attr2="9999992" Attr3="9999993"
    Attr4="9999994" Attr5="9999995" Attr6="9999996" Attr7="9999997"
    Attr8="9999998" Attr9="9999999" Attr10="99999910" Attr11="99999911"
    Attr12="99999912" Attr13="99999913" Attr14="99999914" Attr15="99999915"
    Attr16="99999916" Attr17="99999917" Attr18="99999918"
    Attr19="99999919">Node_Number999999</Data>
    </Appdata>

    I'm assuming you *generated* this file by way of example. If not, well,
    it's so extremely structured that you could throw it away and use a
    simple algorithm to generate the "data" for any line immediately. (And
    then it would not be "data", it would be a calculation.)

    Anyway, XML is a poor choice for this particular set of data. Write a
    program to convert it into a binary format, where each "line" uses 10
    integers and one string of a fixed length of 20 bytes. That takes up no
    more than 1,000,000 x (10 * sizeof(int) + 20) ~ 60 MB of memory. Small
    enough to be loaded into the RAM of today's computers.

    Search "quickly" depends on what you want to search for. If, for
    example, you may need to grab a single digit out of any attribute or
    content (say, a '9' that can occur in the middle of 'Attr2="4593252"'),
    you are better off storing everything as string. You could also sort the
    list on one or more of the Attr fields, and, if you prefer lookup speed
    over memory usage, you could even sort on *all* of the attribute fields
    plus the data field, and save pointers to the 'actual' data.

    [Jw]
     
    Jongware, Nov 1, 2012
    #2
    1. Advertisements

  3. huamin_chen

    wmedwardchan Guest

    Many thanks Jong. Can I have the details in Visual C++ codes? To search the binary format in the way you suggested.

    Many Thanks & Best Regards,
    HuaMin
     
    wmedwardchan, Nov 2, 2012
    #3
  4. huamin_chen

    Jongware Guest

    That would be

    qsort (...);
    result = bsearch (..);

    -- you can look up the correct syntax for both qsort and bsearch
    elsewhere. (It's beyond the scope of c.t.xml anyway.)

    [Jw]
     
    Jongware, Nov 2, 2012
    #4
  5. huamin_chen

    wmedwardchan Guest

    Thanks. But did you see my Xml file above? Qsort is to sort a list of items. How is it applicable to my Xml file?

    Many Thanks & Best Regards,
    Edward Chan
     
    wmedwardchan, Nov 3, 2012
    #5
  6. huamin_chen

    huamin_chen Guest

    JW,
    Furthermore, do you think it is feasible to load the very long list (shown above) into an array, like what you said

    Many Thanks & Best Regards,
    HuaMin
     
    huamin_chen, Nov 3, 2012
    #6
  7. huamin_chen

    Jongware Guest

    On 03-Nov-12 16:27 PM, wrote:> On Friday, November
    How >>is it applicable to my Xml file?

    bsearch is a function for very quickly looking up any item, but the
    items have to be sorted first.
    That's also the reason you have to pick a single key to sort on -- the
    key you want to look up 'quickly'. If you want to be able to look up
    *any* value of the 20 attributes, plus the content string, make 21
    sorted lists.
    To be able to give a less generic answer, we'd need to know much more of
    the data set and what data item(s) need to be looked up.

    Why would it not be feasible? It seems a very simple data array, with 20
    integers and a string content (possibly of a limited length).

    I advise you to ask on one of the comp.programming groups; preferably
    NOT on one dealing with 'Windows', because the requirement for Visual C
    is virtually unimportant here, but on one of the generic C/C++ groups.

    [Jw]
     
    Jongware, Nov 5, 2012
    #7
  8. huamin_chen

    wmedwardchan Guest

    Thanks a lot. What is the algorithm to sort my sample Xml file above? Which other group is better for me to have any other related question for my current issue?
     
    wmedwardchan, Nov 6, 2012
    #8
  9. huamin_chen

    huamin_chen Guest

    JW,
    Any advice to this?
     
    huamin_chen, Nov 7, 2012
    #9
  10. Manuel Collado, Nov 7, 2012
    #10
  11. Depends entirely on what kind of search you're doing and how often
    you're going to be searching the same document.

    A simple SAX parser feeding a SAX handler which discards everything but
    the data you're interested in would be one solution.

    Or XPath/XQuery if you need a serious search language. (XQuery or XSLT
    if your goal is to generate an XML report document.)

    Or use the SAX parser to load the XML into an in-memory data structure
    optimized for whatever kinds of searches you're performing, and run the
    search against that. Which is what most full implementations of
    XPath/XSLT/XQuery do under the covers.


    --
    Joe Kesselman,
    http://www.love-song-productions.com/people/keshlam/index.html

    {} ASCII Ribbon Campaign | "may'ron DaroQbe'chugh vaj bIrIQbej" --
    /\ Stamp out HTML mail! | "Put down the squeezebox & nobody gets hurt."
     
    Joe Kesselman, Nov 7, 2012
    #11
  12. huamin_chen

    huamin_chen Guest

    Thanks a lot. Can you please provide me with one c++ sample project using SAX parser?
     
    huamin_chen, Nov 13, 2012
    #12
  13. Most SAX parsers come with sample programs. Pick your favorite (I'm
    biased toward Apache Xerces) and look at those. And/or try checking the
    many tutorials and articles on http://www.ibm.com/DeveloperWorks/xml --
    those are mostly slanted toward Java, but the same principles apply.

    (I cite DeveloperWorks for several reasons. I admit to an association
    with IBM... but DeveloperWorks really is run almost as an independent
    web magazine, and the content is fairly extensive, better than average,
    and not noticeably biased. In fact, I've had arguments with the editors
    on occasion when they've included something that I thought conflicted
    with IBM's interests; their response was to keep the article I was
    complaining about up and invite a separate article disagreeing with it.)

    --
    Joe Kesselman,
    http://www.love-song-productions.com/people/keshlam/index.html

    {} ASCII Ribbon Campaign | "may'ron DaroQbe'chugh vaj bIrIQbej" --
    /\ Stamp out HTML mail! | "Put down the squeezebox & nobody gets hurt."
     
    Joe Kesselman, Nov 22, 2012
    #13
  14. huamin_chen

    wmedwardchan Guest

    Many thanks Joe. Is it possible to load the very big Xml fiile (I did originally showed in this thread) into Sax parser? Will it lead to bad processing speed? Have a great weekend!
     
    wmedwardchan, Nov 23, 2012
    #14
  15. Depends entirely on what you need to do with the document. Sax is just a
    parser; it produces events, and it's up to you to decide what to do in
    response to those events. One obvious thing you can do is build a
    complete in-memory model such as the DOM. On the other hand, for some
    tasks you may be able to note and discard most of the data as it goes
    by, keeping only the parts your task actually needs -- and possibly
    doing the computation as you go, to further minimize how much you need
    to keep.

    This is sometimes referred to as "streaming" processing. There are
    streaming subset implementations of XPath (Xerces-J comes with one; I'm
    not sure about Xerces-C), or of course you can hand-code the search logic.

    --
    Joe Kesselman,
    http://www.love-song-productions.com/people/keshlam/index.html

    {} ASCII Ribbon Campaign | "may'ron DaroQbe'chugh vaj bIrIQbej" --
    /\ Stamp out HTML mail! | "Put down the squeezebox & nobody gets hurt."
     
    Joe Kesselman, Nov 24, 2012
    #15
  16. huamin_chen

    huamin_chen Guest

    Thanks. Is there any sample to build a DOM from a given Xml file?
     
    huamin_chen, Nov 27, 2012
    #16
  17. Most parsers come with sample programs. Start with that. See also the
    many XML tutorials and articles on the web -- standard citation here for
    the resources at http://developerworks.ibm.com/xml, which I consider
    better than most.


    --
    Joe Kesselman,
    http://www.love-song-productions.com/people/keshlam/index.html

    {} ASCII Ribbon Campaign | "may'ron DaroQbe'chugh vaj bIrIQbej" --
    /\ Stamp out HTML mail! | "Put down the squeezebox & nobody gets hurt."
     
    Joe Kesselman, Nov 28, 2012
    #17
  18. (Reminder: if the document is "very big", you may find that the standard
    DOM is not the best answer. See my previous comments, and the resources
    mentioned.)
     
    Joe Kesselman, Nov 28, 2012
    #18
  19. huamin_chen

    wmedwardchan Guest

    Thanks Joe. Did you ever open my Xml file? Is it possible to work with it against DOM?
     
    wmedwardchan, Nov 29, 2012
    #19
  20. No, and yes assuming it's well-formed XML, respectively. Whether the DOM
    is the *best* way to work with it depends on what you're doing and on
    what kind of machine.

    --
    Joe Kesselman,
    http://www.love-song-productions.com/people/keshlam/index.html

    {} ASCII Ribbon Campaign | "may'ron DaroQbe'chugh vaj bIrIQbej" --
    /\ Stamp out HTML mail! | "Put down the squeezebox & nobody gets hurt."
     
    Joe Kesselman, Nov 29, 2012
    #20
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.