Delete specific fields from a document

I have

• Tested with the latest release
• Tested with the current master branch
• Searched for similar existing issues

Expected behaviour

Solr allows deleting a field by providing the null attribute to an add operation. In the example below, the field skills should be deleted from the document with “employeeId” 05991:

<add>
    <doc>
        <field name="employeeId">05991</field>
        <field name="skills" update="set" null="true" />
    </doc>
</add>

I would expect that the following call would be equivalent to the xml above:

docs = [{'empolyeeId':05991, 'skills':None}]
solr.add(docs, fieldUpdates={'skills':'set'})

Actual behaviour

As far as I understand, this operation is not yet supported in pysolr. I could not find a way to input the null attribute to the add method of Solr class.

Configuration

• Operating system version: Ubuntu 14.04, kernel 3.16.0-77-generic
• Search engine version: 4.7
• Python version: 2.7.6
• pysolr version: 3.5.0

[jogilder]

Hi, I updated the _build_doc method. At the moment it doesn’t do anything if there is a ‘null value’ passed. I have added in an extra check if the user passes in an empty string or list and they are doing a ‘set’ fieldUpdate then it will add in the document but will also add an extra attribute null=true. This works for me as I am passing in empty lists, it wouldn’t work if someone passed in field=None. `

def _build_field(self, value, key, _atomic_update_set, boost=None, fieldUpdates=None):
    attrs = {'name': key}
    if fieldUpdates and key in fieldUpdates:
        attrs['update'] = fieldUpdates[key]
        if self._is_null_value(value) and _atomic_update_set:
            attrs['null'] = u'true'
    if boost and key in boost:
        attrs['boost'] = force_unicode(boost[key])
    field = ElementTree.Element('field', **attrs)
    field.text = self._from_python(value)
    return field

def _is_atomic_update_set(self,field_update_text):
    _atomic_update_set = False
    if field_update_text.lower() == "set":
        _atomic_update_set = True
    return _atomic_update_set

def _build_doc(self, doc, boost=None, fieldUpdates=None):
    doc_elem = ElementTree.Element('doc')
    for key, value in doc.items():
        if key == NESTED_DOC_KEY:
            for child in value:
                doc_elem.append(self._build_doc(child, boost, fieldUpdates))
            continue

        if key == 'boost':
            doc_elem.set('boost', force_unicode(value))
            continue

        _atomic_update_set = self._is_atomic_update_set(fieldUpdates.get(key, u''))
        # To avoid multiple code-paths we'd like to treat all of our values as iterables:
        if isinstance(value, (list, tuple)):
            values = value
            #we want to be able to do the for loop if we have an empty list
            if _atomic_update_set and not values:
                values = (u'',)
        else:
            values = (value, )
        #if this is a list then we go through the list or tuple - doesn't work for an empty list
        for bit in values:
            #allow empty strings if this is a set update but create null attribute
            if self._is_null_value(bit) and not _atomic_update_set:
                continue
            field = self._build_field(bit, key, _atomic_update_set, boost, fieldUpdates)
            doc_elem.append(field)
    return doc_elem

`

[jogilder]

so is this fixed then?