Problem with nested fields with numeric field names

[maxguru]

I seem to have found a big issue with the newer version (0.10.3). I narrowed it down to issues with XFDF generator.

See files generated by the two versions:
xfdf.zip

Version 0.10.3 generate files that don't work fully with pdftk-java (some fields are not filled).

[mikehaertl]

I'm hesitant to download arbitrary ZIP files from unknown sources. Could you maybe better add the relevant information here? A code example would also help.

[maxguru]

GitHub didn't allow me to upload XFDF files but allowed ZIP. Go figure.

XFDF output from v0.9.2:

<?xml version="1.0" encoding="UTF-8"?>
<xfdf xmlns="http://ns.adobe.com/xfdf/" xml:space="preserve">
<fields><field name="Phone">
<value>+1 (168) 258-9206</value>
</field>
<field name="Relationship to You">
<value>Eius eligendi eius a</value>
</field>
<field name="Hospital">
<value>Veniam ex neque et</value>
</field>
<field name="Phone_2">
<value>+1 (452) 938-4562</value>
</field>
<field name="City">
<value>Do ut soluta distinc</value>
</field>
<field name="County">
<value>Unde est ut expedita</value>
</field>
<field name="Date">
<value>14-Feb-2007</value>
</field>
<field name="Client DOB">
<value>Rem eos aliqua Qui</value>
</field>
<field name="Date_2">
<value>16-Sep-2000</value>
</field>
<field name="Date_3">
<value>06-Sep-2003</value>
</field>
<field name="Please Initial that you have read this page">
<field name="0">
<value>Totam quia aut culpa</value>
</field>
<field name="1">
<field name="0">
<value>Aut sequi corrupti</value>
</field>
<field name="1">
<value>Dolorem duis aliquid</value>
</field>
<field name="2">
<field name="0">
<value>Quibusdam illo labor</value>
</field>
<field name="1">
<value>Reiciendis irure rec</value>
</field>
<field name="2">
<value>Quis duis sint rati</value>
</field>
</field>
</field>
</field>
<field name="Parents or Legal Guardians Name Please Print">
<field name="0">
<value>Irene Juarez</value>
</field>
<field name="1">
<value>Berk Rivers</value>
</field>
<field name="2">
<value>Jaime Carver</value>
</field>
</field>
<field name="Client Name Please Print">
<field name="0">
<value>Amal Harrington</value>
</field>
<field name="1">
<value>Sophia Decker</value>
</field>
</field>
<field name="Check Box2">
<field name="0">
<field name="0">
<value>Yes</value>
</field>
<field name="1">
<value>Yes</value>
</field>
</field>
<field name="1">
<field name="0">
<value>Yes</value>
</field>
<field name="1">
<value>Yes</value>
</field>
</field>
</field>
<field name="Name">
<value>Noelani Martin</value>
</field>
</fields>
</xfdf>

XFDF output from v0.10.3:

<?xml version="1.0" encoding="UTF-8"?>
<xfdf xmlns="http://ns.adobe.com/xfdf/" xml:space="preserve">
<fields>
<field name="Phone">
<value>+1 (168) 258-9206</value>
</field>
<field name="Relationship to You">
<value>Eius eligendi eius a</value>
</field>
<field name="Hospital">
<value>Veniam ex neque et</value>
</field>
<field name="Phone_2">
<value>+1 (452) 938-4562</value>
</field>
<field name="City">
<value>Do ut soluta distinc</value>
</field>
<field name="County">
<value>Unde est ut expedita</value>
</field>
<field name="Date">
<value>14-Feb-2007</value>
</field>
<field name="Client DOB">
<value>Rem eos aliqua Qui</value>
</field>
<field name="Date_2">
<value>16-Sep-2000</value>
</field>
<field name="Date_3">
<value>06-Sep-2003</value>
</field>
<field name="Please Initial that you have read this page">
<value>Totam quia aut culpa</value>
<value></value>
</field>
<field name="Parents or Legal Guardians Name Please Print">
<value>Irene Juarez</value>
<value>Berk Rivers</value>
<value>Jaime Carver</value>
</field>
<field name="Client Name Please Print">
<value>Amal Harrington</value>
<value>Sophia Decker</value>
</field>
<field name="Check Box2">
<value></value>
<value></value>
</field>
<field name="Name">
<value>Noelani Martin</value>
</field>
</fields>
</xfdf>

[mikehaertl]

Can you please also provide a simple code example that I can use to reproduce the issue? Especially including the format of the data array.

[maxguru]

$data = array(
  'Phone' => '+1 (168) 258-9206',
  'Relationship to You' => 'Eius eligendi eius a',
  'Hospital' => 'Veniam ex neque et',
  'Phone_2' => '+1 (452) 938-4562',
  'City' => 'Do ut soluta distinc',
  'County' => 'Unde est ut expedita',
  'Date' => '14-Feb-2007',
  'Client DOB' => 'Rem eos aliqua Qui',
  'Date_2' => '16-Sep-2000',
  'Date_3' => '06-Sep-2003',
  'Please Initial that you have read this page.0' => 'Totam quia aut culpa',
  'Please Initial that you have read this page.1.0' => 'Aut sequi corrupti',
  'Please Initial that you have read this page.1.1' => 'Dolorem duis aliquid',
  'Please Initial that you have read this page.1.2.0' => 'Quibusdam illo labor',
  'Please Initial that you have read this page.1.2.1' => 'Reiciendis irure rec',
  'Please Initial that you have read this page.1.2.2' => 'Quis duis sint rati',
  'Parents or Legal Guardians Name Please Print.0' => 'Irene Juarez',
  'Parents or Legal Guardians Name Please Print.1' => 'Berk Rivers',
  'Parents or Legal Guardians Name Please Print.2' => 'Jaime Carver',
  'Client Name Please Print.0' => 'Amal Harrington',
  'Client Name Please Print.1' => 'Sophia Decker',
  'Check Box2.0.0' => 'Yes',
  'Check Box2.0.1' => 'Yes',
  'Check Box2.1.0' => 'Yes',
  'Check Box2.1.1' => 'Yes',
  'Name' => 'Noelani Martin',
);
$xfdf = new \mikehaertl\pdftk\XfdfFile($data);
$xfdf->saveAs('data.xfdf');

By the way, even v0.9.2 output looks strange (nested field tags), although, I am not an expert on XFDF.

[mikehaertl]

Your field names contain dots . which are interpreted as nested fields. I don't remember if this is a XFDF thing or only in our library. I will look into this later but maybe it also helps you to further investigate.

[maxguru]

Version 0.9.2 output works with pdftk, so, I am not going to complain, but whatever version 0.10.3 outputs doesn't work (some fields are not filled).

[mikehaertl]

[DELETED]

[mikehaertl]

Sorry, forget what I wrote above. This will not work. The real problem is that you use numeric field names.

[mikehaertl]

Unfortunately I see no way to fix this without breaking things. The main problem is really that your field names are numbers. If you can somehow at least avoid a field with name 0 things should work again for you.

Background: If you have field names like this:

[
    'abc.0' => 'value1',
    'abc.1' => 'value2',
]

this gets converted internally into this:

[
    'abc' => [
        0 => 'value1',
        1 => 'value2',
    ],
]

But this is the same as the syntax that we use for fields with multiple values:

[
    'abc' => ['value1', 'value2'],
]

They must be handled differently. So we check if a key with value 0 exists and assume that its a multivalue field. That's why your fields are treated like multivalue fields and not like nested fields anymore.

There is no fix for this.

[maxguru]

I don't understand. The old version (0.9.2) is working fine...

[maxguru]

I can't control what my users use as a field names. Regardless, this sounds like a design issue in the php-pdftk library. I use another library which has no problems whatsoever with the same PDF file.

IMHO, dismissing this issue as a user problem is unreasonable. Would you consider a pull request if I was to fix it?

[mikehaertl]

The problem started with implemenation of #148 in 0.10.2 where we added support for multi-value fields.

But let me have another look, maybe we can still find a workaround somehow.

[mikehaertl]

@maxguru I've prepared a fix. Can you help testing this? See MR above for branch issue-260-nested-numeric-field-names.

[maxguru]

Thanks for the quick fix! It does produce the desired output for the case I described above. However, give me some time to test it with a large number of different PDF files.

[mikehaertl]

I've now created release 0.10.4 containing this fix. Let me know if there's still an issue.

[maxguru]

It seems to be working fine. Thanks!