[Feature request] short option flags for --printed-node-format and --ignore-namespaces

[goekce]

I use xidel in a course where we parse XML files. We used xmllint for that, but xmllint does not support parsing without setting the namespaces, and I am grateful that I found xidel!

As my students are used to xmllint output, I have to use --printed-node-format=xml and --ignore-namespaces to get a similar output. xidel outputs the content of a node immediately without having to provide text(), which is convenient.

Example:

xidel https://www.uniprot.org/uniprot/P21817.xml \
  --printed-node-format=xml \
  --ignore-namespaces \
  -se '//protein'

<protein>
<recommendedName>
<fullName>Ryanodine receptor 1</fullName>
<shortName>RYR-1</shortName>
...
</protein>

It takes time for the users to write these long options. Is it possible to add shorter option flags for --printed-node-format and --ignore-namespaces? Or did I miss any alternative options?

[benibela]

There is --xml to set input and output format. Or use -se 'serialize(//protein)'

There is a nothing shorter for --ignore-namespaces. You could set it permanently though export XIDEL_OPTIONS=--ignore-namespaces. It is a hardly used option (I had even forgotten that it was there).

[goekce]

--xml is a convenient alternative with a minor inconvenience for my case (see below).

-se 'serialize(//protein)' sometimes messes up the output, and does not use coloring on my terminal:

$ xidel https://www.uniprot.org/uniprot/P21817.xml -se 'serialize(//fullname)'
<fullName xmlns="http://uniprot.org/uniprot" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">Ryanodine receptor 1</fullName><fullName xmlns="http://uniprot.org/uniprot" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">Skeletal muscle calcium release channel</fullName><fullName xmlns="http://uniprot.org/uniprot" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">Skeletal muscle ryanodine receptor</fullName><fullName xmlns="http://uniprot.org/uniprot" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">Skeletal muscle-type ryanodine receptor</fullName><fullName xmlns="http://uniprot.org/uniprot" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">Type 1 ryanodine receptor</fullName>

I use --ignore-namespaces for instructional and experimenting purposes to declutter the output, e.g.:

$ xidel https://www.uniprot.org/uniprot/P21817.xml --xml -se '//fullname'
<?xml version="1.0" encoding="UTF-8"?>
<xml>
<fullName xmlns="http://uniprot.org/uniprot" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">Ryanodine receptor 1</fullName>
<fullName xmlns="http://uniprot.org/uniprot" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">Skeletal muscle calcium release channel</fullName>
<fullName xmlns="http://uniprot.org/uniprot" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">Skeletal muscle ryanodine receptor</fullName>
<fullName xmlns="http://uniprot.org/uniprot" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">Skeletal muscle-type ryanodine receptor</fullName>
<fullName xmlns="http://uniprot.org/uniprot" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">Type 1 ryanodine receptor</fullName>
</xml>

with --ignore-namespaces:

$ xidel https://www.uniprot.org/uniprot/P21817.xml --xml -se '//fullname' --ignore-namespaces
<?xml version="1.0" encoding="UTF-8"?>
<xml>
<fullName>Ryanodine receptor 1</fullName>
<fullName>Skeletal muscle calcium release channel</fullName>
<fullName>Skeletal muscle ryanodine receptor</fullName>
<fullName>Skeletal muscle-type ryanodine receptor</fullName>
<fullName>Type 1 ryanodine receptor</fullName>
</xml>

For experimenting on the shell, I still find --printed-node-format=xml better than --xml, though, because xml nodes are completely gone:

$ xidel chebi_27732_P21817.xml --printed-node-format=xml -se '//fullname' --ignore-namespaces
<fullName>Ryanodine receptor 1</fullName>
<fullName>Skeletal muscle calcium release channel</fullName>
<fullName>Skeletal muscle ryanodine receptor</fullName>
<fullName>Skeletal muscle-type ryanodine receptor</fullName>
<fullName>Type 1 ryanodine receptor</fullName>

My cases may be not very common, I can understand that! Nevertheless XIDEL_OPTIONS="--printed-node-format=xml --ignore-namespaces" is also a good way to avoid the long options in subsequent iterations of the command. I hope that --ignore-namespaces and --printed-node-format stay as options in future. Thanks Benito!

[Reino17]

Found this (old) "issue" of yours.

-se 'serialize(//protein)' sometimes messes up the output, and does not use coloring on my terminal:

Only HTML/XML and JSON output gets coloring. The output of serialize() is always a string. And in addition, if //fullname is a sequence of nodes, then serialize(//fullname) serializes/"stringifies" and "minifies"/concatenates these strings, as you've seen already.

At the moment the only way to "prettify" XML (when the input is not a sequence) is to use serialize()'s "indent"-option:

$ xidel https://www.uniprot.org/uniprot/P21817.xml -se 'serialize(//protein,{"indent":true()})'

And if you really want coloring :D :

$ xidel https://www.uniprot.org/uniprot/P21817.xml -se 'serialize(//protein,{"indent":true()})' | \
  xidel -se '.' --printed-node-format=xml

$ xidel https://www.uniprot.org/uniprot/P21817.xml -se '
  parse-xml(
    serialize(//protein,{"indent":true()})
  )
' --printed-node-format=xml

Pretty good chance xidel well get a dedicated "prettify" command-line option in the near future. See https://sourceforge.net/p/videlibri/mailman/message/37373117/.

[goekce]

I used to use xidel during my classes to teach Xpath to students who are new to the Unix shell. My goal was to show students a fast way of extracting data from an XML file without configuring the tool. Using your serialize() is a good solution which could be interesting for shell users who know how to create wrappers around shell commands.

Thanks for your followup @Reino17 !

[benibela]

@Reino17

That reminds me that I wanted to rename --printed-node-format=xml to something like --output-node-format=xml and add a --output-indent or --output-node-indent option