Is it possible to read an element from an XML file without loading the entire file into memory?

Hi,

I have thousands of large XML files from which I need to extract only few elements. I am fairly new to XML files and I was wondering if it is possible to read a single element from an XML file based on its tag without loading the entire XML file into the memory?

Thanks,

Tuomas

[Sara]

You can try playing with textscan. Or read a line per time with fgetl and discard the lines that do not contain your tag. If you post an extract of your file, it's easier to try a solution.

[dpb]

doc memmapfile

as character, maybe.

Have you confirmed xmlread doesn't perform well enough, however?

[Tuomas Kenttä]

Thank you Sara for your comment.

Going through the file with fgetl was my initial thought as well. I was just wondering if there would be a more sophisticated way of doing this. If not, then I guess I have to go with fgetl .

The data contains EKG measurements at various sample points. My task is to extract the measured data at each sample point (which is stored in one location) and the signal from which those measurements were done (stored in another location within the file).

Here is an extract out of the file:

(Measurements)

<TrendEntry Idx="134">
    <EntryTime>
      <Minute>26</Minute>
      <Second>40</Second>
    </EntryTime>
    <HeartRate>95</HeartRate>
    <Mets>1.0</Mets>
    <RPP>202</RPP>
    <VECount>0</VECount>
    <PaceCount>0</PaceCount>
    <Artifact>0%</Artifact>
    <LeadMeasurements lead="I">
      <JPointAmplitude units="mV">-0.01</JPointAmplitude>
      <STAmplitude20ms units="mV">-0.01</STAmplitude20ms>
      <STAmplitude units="mV">-0.01</STAmplitude>
      <RAmplitude units="mV"> 0.35</RAmplitude>
      <R1Amplitude units="mV">-32.77</R1Amplitude>
      <STSlope units="mV/s">-0.06</STSlope>
      <STIntegral>0</STIntegral>
      <STIndex>abnormal</STIndex>
    </LeadMeasurements>
    <LeadMeasurements lead="II">
      <JPointAmplitude units="mV">-0.03</JPointAmplitude>
      <STAmplitude20ms units="mV">-0.03</STAmplitude20ms>
      <STAmplitude units="mV">-0.03</STAmplitude>
      <RAmplitude units="mV"> 0.49</RAmplitude>
      <R1Amplitude units="mV">-32.77</R1Amplitude>
      <STSlope units="mV/s">-0.10</STSlope>
      <STIntegral>-2</STIntegral>
      <STIndex>abnormal</STIndex>

(Signal)

<Median Idx="134">
      <Time>
        <Minute>26</Minute>
        <Second>40</Second>
      </Time>
      <I-Point>184</I-Point>
     <J-Point>231</J-Point>
      <JX-Point>261</JX-Point>
      <P-Point>104</P-Point>
      <PE-Point>158</PE-Point>
      <Q-Point>190</Q-Point>
      <T-Point>346</T-Point>
      <WaveformData lead="I" startIdx="134">1,1,1,1,1,1,1,1,....
                <WaveformData lead="II" startIdx="134">1,1,1,1,1,...

[Sara]

What are the tags you are interested in?

[dpb]

Again, why not use the XML file tools in Matlab first to prove they're not up to the task first???? Could probably have already done it in time have spent dreaming about fgetl and manual parsing plus reading the file line-by-line w/ fgetl is about as slow as it gets from an i/o process w/ only a fgetch being worse...

As for "big", what kind of size are we talking of?

[Sara]

Building on dpb comment, do you have any code you have written to read in the files and "search" them?