LinuxQuestions.org - Regular expression to select the field correctly which has a

- Programming (https://www.linuxquestions.org/questions/programming-9/)

- - Regular expression to select the field correctly which has a - in it (https://www.linuxquestions.org/questions/programming-9/regular-expression-to-select-the-field-correctly-which-has-a-in-it-4175626904/)

Regular expression to select the field correctly which has a - in it

Hello All,
I have an XML file like this and i need a regular expression to select the <todo-item> from below. I wrote the expression like this: ([\r\n]+)(?=\s*\<todo-item\>)

but i think due to the - which is between todo and item is causing it to not detect correctly. Can anyone shed some light on how to change the regex code?

PHP Code:




<todo-items type="array"> 
 <todo-item> 
 <project-id type="integer">353705</project-id> 
 <tasklist-istemplate type="boolean">false</tasklist-istemplate> 
 <hastickets type="boolean">false</hastickets> 
 <order type="integer">2003</order> 
 <comments-count type="integer">0</comments-count> 
 <created-on type="date">2018-02-21T06:26:43Z</created-on> 
 <canedit type="boolean">true</canedit> 
 <has-predecessors type="integer">0</has-predecessors> 
 <id type="integer">17223695</id> 
 <completed type="boolean">false</completed> 
 <position type="integer">2003</position> 
 <estimated-minutes type="integer">0</estimated-minutes> 
 <description/> 
 <progress type="integer">0</progress> 
 <harvest-enabled type="boolean">false</harvest-enabled> 
 <parenttaskid type="integer">17223687</parenttaskid> 
 <responsible-party-lastname>xxx</responsible-party-lastname> 
 <company-id type="integer">103131</company-id> 
 <creator-id type="integer">316954</creator-id> 
 <project-name>asdfasdfasdf</project-name> 
 <start-date type="integer">20180403</start-date> 
 <tasklist-private type="boolean">true</tasklist-private> 
 <lockdownid type="integer">806894</lockdownid> 
 <cancomplete type="boolean">true</cancomplete> 
 <responsible-party-id>317122,221525,316954</responsible-party-id> 
 <creator-lastname>asdfasdfsdf</creator-lastname> 
 <has-reminders type="boolean">false</has-reminders> 
 <has-unread-comments type="boolean">false</has-unread-comments> 
 <todo-list-name>Phase Two</todo-list-name> 
 <due-date-base type="integer">20180403</due-date-base> 
 <private type="integer">2</private> 
 <userfollowingcomments type="boolean">false</userfollowingcomments> 
 <responsible-party-summary>You + 2 others</responsible-party-summary> 
 <status>new</status> 
 <todo-list-id type="integer">1533948</todo-list-id> 
 <predecessors type="array"/> 
 <tags type="array"/> 
 <content>ffdddffdfdfdfdfdfdfdfdfdfd</content> 
 <responsible-party-type>Person</responsible-party-type> 
 <company-name>as dfcsdfsdfs</company-name> 
 <creator-firstname>asdfasdfasdfasdf</creator-firstname> 
 <last-changed-on type="date">2018-03-29T10:55:28Z</last-changed-on> 
 <due-date type="integer">20180403</due-date> 
 <has-dependencies type="integer">2</has-dependencies> 
 <attachments-count type="integer">0</attachments-count> 
 <userfollowingchanges type="boolean">false</userfollowingchanges> 
 <priority/> 
 <responsible-party-firstname>asdfasdfasdf</responsible-party-firstname> 
 <viewestimatedtime type="boolean">true</viewestimatedtime> 
 <responsible-party-ids>317122,221525,316954</responsible-party-ids> 
 <responsible-party-names>cdcdcdcdcdcdcdcd</responsible-party-names> 
 <tasklist-lockdownid type="integer">806894</tasklist-lockdownid> 
 <timeislogged type="integer">0</timeislogged> 
 </todo-item>

You haven't said which regex engine, or which tool you are using. Are we to guess php ?.

In general parsing XML youself is pointless - use one of the appropriate tools. No regex engine I use cares at all about a minus sign unless in a bracket selection.
Why do you require line feeds on input - normal stream tools strip the line-feed. I would remove the first subexpression completely - but I don't do php ...

Sorry for the confusion. I am actually importing this XML into splunk. So my props.conf is like
[project]
SHOULD_LINEMERGE = false
LINE_BREAKER = ([\r\n]+)(?=\s*\<todo-item\>)
DATETIME_CONFIG = CURRENT
KV_MODE =xml

With this i am able to injest other XML files without any issue. I was just thinking the data is not split because of the - in between. Thats why i posted the question. Again apologies for the confusion.
Thanks

Agreed. It is quite pointless to use pure regex to try to manage XML. A proper parser is needed for that. You have several easy-to-use, mature XML parsers in CPAN. See XML::TreeBuilder, XML::XPathEngine, or XML::Twig there on CPAN.

There are also several standalone XML parsers also based on XPath. xmllint is one.

Quote:

Originally Posted by Turbocapitalist (Post 5838742)

There are also several standalone XML parsers also based on XPath. xmllint is one.

... the nokogiri standalone executable is another.

I am responding, however, because I have made the same error a few times and tried to parse XML with Regexs. The way that the OP describes it and as commented by the other contributors, it is pointless, in deed.

An XML-parser, on the other hand, is called an XML-parser, because it pares XML. Think about it.

Quote:

Originally Posted by ranjitabraham (Post 5838741)

I am actually importing this XML into splunk.
... i am able to injest other XML files without any issue.

Sorry, I briefly looked at splunk when it first emerged years ago, but don't use it.
I just had a brief look at the regex doco - interesting; I've not seen \r and \n used like that as anchors before. Can't help - hopefully others with splunk experience can assist.

Whatever RE engine it is, a literal < and > should not be escaped.
\<todo-item\> should be <todo-item>

While you can backslash-escape a literal dash, Turbocapitalist's admonition to "use a real XML parser" is a very sound one.

There are two general approaches. One reads the XML and builds an in-memory data structure. The other parses the XML and, while doing so, makes subroutine-calls to back-end routines of your own devising. Both are "known good" tools for handling the vagaries of XML.