如何在VTD-XML中按节点名称-值合并（合并）两个不同的xml文件？

Question

I am a newbie in Java, after evaluate some java libraries i choosed VTD-XML by its performance tests and the option to use Xpath, I tried StaX and i think is not for human beings, really hard to understand how the parsing works (almost for me XD).

So, my target is to "inject" the geo_code node from partial_geo_codes.xml into geo_code accommodations.xml matching the values on node ext_id from both

accommodation.xml

<accommodations>
 <accommodation>
  <ext_id>12345</ext_id>
  <type>A</type>
  <details>D</details>
  <geo_code />
  </accommodation>

and this is the file to be appended into accommodation.xml:

partial_geo_codes.xml

<geo_codes>
 <geo_code>
  <ext_id>12345</ext_id>
  <geo_idlocacion>77500</geo_idlocacion>
  <latitude>42.578114</latitude>
  <longitude>1.648293</longitude>
  </geo_code>
  <geo_code>
      ...
  <geo_code>
  <geo_code>
      ...
  <geo_code>
 <geo_codes>

this is the expected output:

accommodation_new.xml

<accommodations>
 <accommodation>
  <ext_id>12345</ext_id>
  <type>A</type>
  <details>D</details>
  <geo_code>
    <ext_id>12345</ext_id>
    <geo_idlocacion>77500</geo_idlocacion>
    <latitude>42.578114</latitude>
    <longitude>1.648293</longitude>
  <geo_code> 
  </accommodation>
  <accommodation>
   .....
  </accommodation>
  ...... 
</accommodations>

and this is my "wannabe-really-sucks" java class:

import com.ximpleware.extended.*;
import java.io.*;

public class MergeVtd  {

 public static void main(String args[]) throws Exception {

    String filesPath = new java.io.File("").getAbsolutePath() .concat("/main/src/");
    long start = System.currentTimeMillis();


    //init original xml
    VTDGenHuge vgh = new VTDGenHuge();
    //init tobemerged xml
    VTDGenHuge vgm = new VTDGenHuge();


    if (vgm.parseFile(filesPath.concat("partial_geo_code.xml"),true,VTDGenHuge.MEM_MAPPED)){

        VTDNavHuge vnm = vgm.getNav();
        AutoPilotHuge apm = new AutoPilotHuge(vnm);
        apm.selectElement("ext_id");


        int  count=0;
        while (apm.iterate()){
            int t = vnm.getText();
            if (t!=-1)    {
                System.out.println("Value vnm ==> "+vnm.toNormalizedString(t));

            //we have id to match....

            if (vgh.parseFile(filesPath.concat("accommodation.xml"),true,VTDGenHuge.MEM_MAPPED)){
                VTDNavHuge vnh = vgh.getNav();
                AutoPilotHuge aph = new AutoPilotHuge(vnh);
                aph.selectXPath("/accommodations/accommodation/ext_id[text()='" + vnm.toNormalizedString(t) + "']" );


                int result = -1;
                while ((result=aph.evalXPath())!=-1){
                    int g = vnh.getText();
                    if (g!=-1)  {
                        System.out.println("Value vnh ==> "+vnh.toNormalizedString(g));

                    }  else {
                        System.out.println("no match in vnh !======= ");
                    }
                }
            }

            }

            System.out.println("============================== " + count);
            count++;

        }

    }

    long end = System.currentTimeMillis();
    System.out.println("Execution time was "+ (end - start) +" ms.");
    System.exit(0);

 }

}

i really appreciate any clue helping me how to iterate into 2 xml files at once and merge by ext_id node value much faster, now really takes too much time.

Answer 1

How big is partial_geo_codes.xml? Can it fit in memory? If yes then I would recommend indexing it using hash-map. Just create simple HashMap, and put there references to geo_code nodes with values of ext_id as keys.

Having done that you'll need to pass accomodations.xml only once. Right now your algorithm complexity is O(n^2), what's worse is that involves n reads from disk! Version with HashMap will take O(n) time and will require only single pass through both xml files.