README

The game ain't in me no more. None of it.

xmlcutty is a simple tool for carving out elements from large XML files, fast. Since it works in a streaming fashion, it uses almost no memory and can process around 1G of XML per minute.

Why? Background.

Install

Use a deb or rpm release. It's in AUR, too.

Or install with the go tool:

$ go get github.com/miku/xmlcutty/cmd/xmlcutty

Usage

$ cat fixtures/sample.xml
<a>
    <b>
        <c></c>
    </b>
    <b>
        <c></c>
    </b>
</a>

Options:

$ xmlcutty -h
Usage of xmlcutty:
  -path string
        select path (default "/")
  -rename string
        rename wrapper element to this name
  -root string
        synthetic root element
  -v    show version

It looks a bit like XPath, but it really is only a simple matcher.

$ xmlcutty -path /a fixtures/sample.xml
<a>
    <b>
        <c></c>
    </b>
    <b>
        <c></c>
    </b>
</a>

You specify a path, e.g. /a/b and all elements matching this path are printed:

$ xmlcutty -path /a/b fixtures/sample.xml
<b>
    <c></c>
</b>
<b>
    <c></c>
</b>

You can end up with an XML document without a root. To make tools like xmllint happy, you can add a synthetic root element on the fly:

$ xmlcutty -root hello -path /a/b fixtures/sample.xml | xmllint --format -
<?xml version="1.0"?>
<hello>
    <b>
        <c></c>
    </b>
    <b>
        <c></c>
    </b>
</hello>

Rename wrapper element - that is the last element of the matching path:

$ xmlcutty -rename beee -path /a/b fixtures/sample.xml
<beee>
    <c></c>
</beee>
<beee>
    <c></c>
</beee>

All options, synthetic root element and a renamed path element:

$ xmlcutty -root hi -rename ceee -path /a/b/c fixtures/sample.xml | xmllint --format -
<?xml version="1.0"?>
<hi>
    <ceee/>
    <ceee/>
</hi>

It will parse XML files without a root element just fine.

$ head fixtures/oai.xml
<record>
    <header>
        <identifier>oai:arXiv.org:0704.0004</identifier>
        <datestamp>2007-05-23</datestamp>
        <setSpec>math</setSpec>
    </header>
    <metadata>
        <oai_dc:dc xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/"... >
            <dc:title>A determinant of Stirling cycle numbers counts ...
            <dc:type>text</dc:type>
            <dc:identifier>http://arxiv.org/abs/0704.0004</dc:identifier>
...

This is an example XML response from a web service. We can slice out the identifier elements. Note that any namespace - here oai_dc - is completely ignored for the sake of simplicity:

$ cat fixtures/oai.xml | xmlcutty -root x -path /record/metadata/dc/identifier \
                       | xmllint --format -
<?xml version="1.0"?>
<x>
    <identifier>http://arxiv.org/abs/0704.0004</identifier>
    <identifier>http://arxiv.org/abs/0704.0010</identifier>
    <identifier>http://arxiv.org/abs/0704.0012</identifier>
</x>

We can go a bit further and extract the text element, which is like a poor man text() in XPath terms. By using the a newline as argument to rename, we effectively get rid of the enclosing XML tag:

$ cat fixtures/oai.xml | xmlcutty -rename '\n' -path /record/metadata/dc/identifier \
                       | grep -v "^$"
http://arxiv.org/abs/0704.0004
http://arxiv.org/abs/0704.0010
http://arxiv.org/abs/0704.0012

This last feature is nice to quickly extract text from large XML files.

eddieHRS/xmlcutty

README

Install

Usage