Allow for comments in pecan.xml by stripping them in `read.settings()`

[Aariq]

Currently comments in pecan.xml (see example below) don't get stripped and can cause problems. In this example, the comment in the <ed2in_tags> ends up as a length 0 element in the settings list named comment which causes write.settings.ED2 to error with Error in vapply(custom_tags, function(x) grepl(numvec_rxp, x), logical(1)) : values must be length 1,

<ed2in_tags>
      <PFT_1ST_CHECK>0</PFT_1ST_CHECK>
      <IED_INIT_MODE>0</IED_INIT_MODE>
      <!-- turn off cohort and patch fusion. Not necessary for short runs -->
      <MAXCOHORT>0</MAXCOHORT>
      <MAXPATCH>0</MAXPATCH>
    </ed2in_tags>

Here's a possible pattern to strip comments using the xml2 package:

tt <-
  '<x>
     <a>text</a>
     <b foo="1"/>
     <!-- a comment -->
     <c bar="me">
        <d>a phrase</d>
        <!-- deeper comment -->
     </c>
  </x>'
xml <- xml2::read_xml(tt)
xml
#> {xml_document}
#> <x>
#> [1] <a>text</a>
#> [2] <b foo="1"/>
#> [3] <c bar="me">\n  <d>a phrase</d>\n  <!-- deeper comment -->\n</c>
# https://stackoverflow.com/questions/784745/accessing-comments-in-xml-using-xpath
comments <- xml2::xml_find_all(xml, "//comment()")
xml2::xml_remove(comments)
xml
#> {xml_document}
#> <x>
#> [1] <a>text</a>
#> [2] <b foo="1"/>
#> [3] <c bar="me">\n  <d>a phrase</d>\n</c>

I know pecan currently uses XML, but it might be a good opportunity to start the switch to xml2 since XML is no longer maintained.

full pecan.xml

<?xml version="1.0"?>
<pecan>
  <info>
    <!--Edit this-->
    <notes>site name</notes>
  </info>
  <!-- Edit the outdir name to replace SN with the two-letter abreviation for this site -->  
  <outdir>/data/tests/ed2_transect_SN</outdir>
  <database>
    <bety>
      <driver>PostgreSQL</driver>
      <user>bety</user>
      <password>bety</password>
      <host>postgres</host>
      <dbname>bety</dbname>
      <write>FALSE</write>
    </bety>
    <dbfiles>/data/dbfiles</dbfiles>
  </database>
  <pfts>
    <pft>
      <name>SetariaWT</name>
      <ed2_pft_number>1</ed2_pft_number>
    </pft>
    <!-- Add additional PFTs here -->
    <pft>
      <name>ebifarm.c3grass</name>
      <ed2_pft_number>5</ed2_pft_number>
    </pft>
  </pfts>

  <meta.analysis>
    <iter>3000</iter>
    <random.effects>TRUE</random.effects>
    <threshold>1.2</threshold>
    <update>TRUE</update>
  </meta.analysis>
<!--
  <sensitivity.analysis>
      <quantiles>
          <sigma>-1</sigma>
          <sigma>1</sigma>
      </quantiles>
    <variable>NPP</variable>
    <start.year>2019</start.year>
    <end.year>2019</end.year>
  </sensitivity.analysis>
-->
  <ensemble>
    <size>50</size>
    <variable>NPP</variable>
    <samplingspace>
      <parameters>
        <method>lhc</method>
      </parameters>
  </samplingspace>
  </ensemble>

  <model>
    <type>ED2</type>
    <binary>/groups/dlebauer/ed2_results/global_inputs/ed2_2.2.0_singularity.sh</binary>
    <id>14</id>
    <edin>/pecan/models/ed/inst/ED2IN.r2.2.0</edin>
    <config.header>
      <radiation>
        <lai_min>0.01</lai_min>
      </radiation>
      <ed_misc>
        <output_month>12</output_month>
      </ed_misc>
    </config.header>
    <phenol.scheme>0</phenol.scheme>
    <ed2in_tags>
      <PFT_1ST_CHECK>0</PFT_1ST_CHECK>
      <IED_INIT_MODE>0</IED_INIT_MODE>
      <!-- turn off cohort and patch fusion. Not necessary for short runs -->
      <MAXCOHORT>0</MAXCOHORT>
      <MAXPATCH>0</MAXPATCH>
    </ed2in_tags>
  </model>

  <run>
    <site>
      <!-- edit the site ID to match the site on BETYdb -->
      <id>678</id>
      <met.start>2019-01-01 00:00:00</met.start>
      <met.end>2019-12-31 23:59:59</met.end>
    </site>

    <inputs>
      <!-- edit the site ID in this path also -->
      <met>/data/sites/MERRA_ED2_site_0-678/ED_MET_DRIVER_HEADER</met>
      <veg>/data/oge2OLD/OGE2_</veg>
      <soil>/data/faoOLD/FAO_</soil>
      <lu>/data/ed_inputs/glu/</lu>
      <thsum>/data/ed_inputs/</thsum>
    </inputs>

    <start.date>2019-01-01</start.date>
    <end.date>2019-12-31</end.date>
  </run>
  
  <host>
      <!-- edit to whatever is in your .ssh/config -->
      <name>puma</name>
      <!-- don't edit the rest unless you know what you're doing -->
      <folder>/groups/dlebauer/ed2_results/pecan_remote</folder>
      <qsub>sbatch --job-name=@NAME@ --account=dlebauer --ntasks=28 --nodes=3 --time=25:00:00 -o @STDOUT@ -e @STDERR@</qsub>
      <qsub.jobid>\\D*([0-9]+)\\D*</qsub.jobid>
      <qstat>'squeue --job @JOBID@ &amp;> /dev/null || echo DONE'</qstat>
      <prerun>module load openmpi3</prerun>
      <modellauncher>
        <binary>/groups/dlebauer/ed2_results/pecan/contrib/modellauncher/modellauncher</binary>
        <qsub.extra>--partition=standard</qsub.extra>
        <mpirun>module load openmpi3; mpirun</mpirun>
      </modellauncher>
  </host>

</pecan>

[dlebauer]

Other options:

• use this xml2 solution without abandoning XML (adds a dependency, but easiest)
• find a way of doing this with the XML package
• find a way of doing this on the settings object
• do a find and replace on the text prior to parsing as xml

(note that XML is once again maintained by CRAN and has many reverse depends and reverse imports)

[Aariq]

Possibly helpful for doing this on the settings object: https://stackoverflow.com/questions/37853679/removing-elements-in-a-nested-r-list-by-name