Can this XPath expression be minified - xpath

Question

I'm not an XPath expert and I couldn't find a tool to do this. Is there a way to minify this XPath statement? 80% is redundant.
XPATH 2
(//CustomField | //fields)[
(fullName/text[not(matches(#Image, "chk_[a-zA-Z0-9]+__c"))] and type/text[#Image="Checkbox"]) or
(fullName/text[not(matches(#Image, "txt_[a-zA-Z0-9]+__c"))] and type/text[#Image="Text"]) or
(fullName/text[not(matches(#Image, "txa_[a-zA-Z0-9]+__c"))] and type/text[#Image="Textarea"]) or
(fullName/text[not(matches(#Image, "txr_[a-zA-Z0-9]+__c"))] and type/text[#Image="Richtext"]) or
(fullName/text[not(matches(#Image, "txl_[a-zA-Z0-9]+__c"))] and type/text[#Image="LongTextArea"]) or
(fullName/text[not(matches(#Image, "num_[a-zA-Z0-9]+__c"))] and type/text[#Image="Number"]) or
(fullName/text[not(matches(#Image, "dat_[a-zA-Z0-9]+__c"))] and type/text[#Image="Date"]) or
(fullName/text[not(matches(#Image, "lkp_[a-zA-Z0-9]+__c"))] and type/text[#Image="Lookup"]) or
(fullName/text[not(matches(#Image, "mdr_[a-zA-Z0-9]+__c"))] and type/text[#Image="MasterDetail"]) or
(fullName/text[not(matches(#Image, "dtm_[a-zA-Z0-9]+__c"))] and type/text[#Image="DateTime"]) or
(fullName/text[not(matches(#Image, "url_[a-zA-Z0-9]+__c"))] and type/text[#Image="Url"]) or
(fullName/text[not(matches(#Image, "pkl_[a-zA-Z0-9]+__c"))] and type/text[#Image="Picklist"]) or
(fullName/text[not(matches(#Image, "pkm_[a-zA-Z0-9]+__c"))] and type/text[#Image="MultiselectPicklist"]) or
(fullName/text[not(matches(#Image, "cur_[a-zA-Z0-9]+__c"))] and type/text[#Image="Currency"]) or
(fullName/text[not(matches(#Image, "pct_[a-zA-Z0-9]+__c"))] and type/text[#Image="Percent"]) or
(fullName/text[not(matches(#Image, "fcur_[a-zA-Z0-9]+__c"))] and type/text[#Image="Currency"] and formula) or
(fullName/text[not(matches(#Image, "rcnt_[a-zA-Z0-9]+__c"))] and type/text[#Image="Summary"] and summaryOperation/text[#Image="count"])
]
This is a stripped down version of an XML document with one match
<?xml version="1.0" encoding="UTF-8"?>
<CustomObject xmlns="http://soap.sforce.com/2006/04/metadata">
<fields>
<fullName>txl_Description__c</fullName>
<deprecated>false</deprecated>
<externalId>false</externalId>
<label>Price Schedule Description</label>
<length>4000</length>
<trackHistory>false</trackHistory>
<trackTrending>false</trackTrending>
<type>LongTextArea</type>
<visibleLines>5</visibleLines>
</fields>
<fields>
<fullName>Reason_For_Override__c</fullName>
<deprecated>false</deprecated>
<externalId>false</externalId>
<label>Reason for Override</label>
<length>1000</length>
<trackHistory>false</trackHistory>
<trackTrending>false</trackTrending>
<type>LongTextArea</type>
<visibleLines>2</visibleLines>
</fields>
</CustomObject>

Answer 1

You can replace all those occurrences of text() with ., which is better practice anyhow.
You could pass in a parameter $data containing a lookup table: <data><key a="chk" b="Checkbox"/><key a="..." b="..."/>...</data> and then do
some $key in $data/data/key
satisfies fullName[not(matches(., concat($key/#a, "_[a-zA-Z0-9]+__c"))]
and type[.=$key/#b])

Answer 2

I thoroughly checked your expression and it looks like that there is near zero redundancy. The part _[a-zA-Z0-9]+__c is of low relevance in optimizing the expression.
Creating tuples like ("chk","Checkbox") and ("txt","TextBox") will neither improve readability nor performance, even if it seems like that. So you're good to go, AFAIK.