Hugo Koopmans Hugo Koopmans - 5 months ago 12
JSON Question

rest extension will not transform json

I have a MarkLogic Rest extension which needs to transform JSON into OBI objects. For starters I created an XML input and created transforms for it, therse work.

Now the real data is slightly different and in JSON, so I need to transform the JSON to XML. I have had an example xsl transform that I cannot seem to get to work...

The JSON doc that I want to load via the rest extension:

{
"wifi_raw": [
{
"id": "4354279",
"hostname": "rb-0046",
"mac": "00:0C:43:00:08:F4",
"firstseen": "2015-08-12 13:54:50",
"lastseen": "2015-08-12 13:54:50",
"rssi": "-1",
"packets": "1",
"bssid": "24:A4:3C:53:19:62",
"probes": "",
"processed": "0"
},
{
"id": "4354257",
"hostname": "rb-0046",
"mac": "00:0E:58:BC:E9:03",
"firstseen": "2015-08-12 13:48:45",
"lastseen": "2015-08-12 13:52:10",
"rssi": "-58",
"packets": "3",
"bssid": "B8:E9:37:17:DA:EF",
"probes": "sonos_hbbzdjrspta2htbcqeb0gcjouc",
"processed": "0"
},
{
"id": "4354273",
"hostname": "rb-0046",
"mac": "00:0E:58:BC:E9:03",
"firstseen": "2015-08-12 13:48:45",
"lastseen": "2015-08-12 13:54:32",
"rssi": "-61",
"packets": "4",
"bssid": "B8:E9:37:17:DA:EF",
"probes": "sonos_hbbzdjrspta2htbcqeb0gcjouc",
"processed": "0"
}
]
}


On ingest I want to transform this to OBI objects which are all defined and work for the equivalent XML input doc...

If I do


json:transform-from-json($source)


In the ingest extension the JSON transforms to this XML:

<?xml version="1.0"?>
<json xmlns="http://marklogic.com/xdmp/json/basic" type="object">
<wifi__raw type="array">
<json type="object">
<id type="string">4354279</id>
<hostname type="string">rb-0046</hostname>
<mac type="string">00:0C:43:00:08:F4</mac>
<firstseen type="string">2015-08-12 13:54:50</firstseen>
<lastseen type="string">2015-08-12 13:54:50</lastseen>
<rssi type="string">-1</rssi>
<packets type="string">1</packets>
<bssid type="string">24:A4:3C:53:19:62</bssid>
<probes type="string"/>
<processed type="string">0</processed>
</json>
<json type="object">
<id type="string">4354257</id>
<hostname type="string">rb-0046</hostname>
<mac type="string">00:0E:58:BC:E9:03</mac>
<firstseen type="string">2015-08-12 13:48:45</firstseen>
<lastseen type="string">2015-08-12 13:52:10</lastseen>
<rssi type="string">-58</rssi>
<packets type="string">3</packets>
<bssid type="string">B8:E9:37:17:DA:EF</bssid>
<probes type="string">sonos_hbbzdjrspta2htbcqeb0gcjouc</probes>
<processed type="string">0</processed>
</json>
<json type="object">
<id type="string">4354273</id>
<hostname type="string">rb-0046</hostname>
<mac type="string">00:0E:58:BC:E9:03</mac>
<firstseen type="string">2015-08-12 13:48:45</firstseen>
<lastseen type="string">2015-08-12 13:54:32</lastseen>
<rssi type="string">-61</rssi>
<packets type="string">4</packets>
<bssid type="string">B8:E9:37:17:DA:EF</bssid>
<probes type="string">sonos_hbbzdjrspta2htbcqeb0gcjouc</probes>
<processed type="string">0</processed>
</json>
</wifi__raw>
</json>


My minimum xsl transform now is this

<?xml version="1.0" encoding="UTF-8"?>
<xsl:stylesheet version="2.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform"

xmlns:xdmp="http://marklogic.com/xdmp"
xmlns:json="http://marklogic.com/xdmp/json"
xmlns:basic="http://marklogic.com/xdmp/json/basic"
xmlns:map="http://marklogic.com/xdmp/map"

xmlns:obj="http://marklogic.com/solutions/obi/object"

extension-element-prefixes="xdmp">

<xdmp:import-module namespace="http://marklogic.com/xdmp/json" href="/MarkLogic/json/json.xqy"/>

<xsl:param name="params" as="map:map"/>
<xsl:variable name="ingest-time" select="(map:get($params, 'ingest-time'), 'false')[1] = 'true'"/>

<xsl:template match="/text()[empty(../*)]">
<xsl:apply-templates select="json:transform-from-json(.)"/>
</xsl:template>

<xsl:template match="basic:json">
<results>
<objects>
<!-- test -->
<xsl:text>example</xsl:text>
</objects>

<links>
<!-- links go here-->
<xsl:text>example</xsl:text>
</links>
</results>
</xsl:template>
</xsl:stylesheet>


I get NO results back from the

let $result := eput:apply-document-transform(fn:concat($dataset, '-transform'), $params, $context, $source)/element()


In the ingest extension.

As I understood this the match "match="/text()[empty(../*)]">" would be true for the JSON document node, so the same transform-from-json would generate the XML version of the original JSON doc also posted here above so I could at least find the root node with "match="basic:json""

What am I missing here?

EDIT WORKING SOLUTION

This is basically what worked. First we made sure the new JSON object-node() is matched, then you have to be really carefull with the matching XSL templates. I had two templates matched for "basic:json" witch resulted in only the last to be executed...

Now the root is matched with

<xsl:template match="/basic:json">


And the one inside the array is matched with ANY child of wifi__raw like

<xsl:template match="basic:wifi__raw/*">


The complete XSL now looks like:

<?xml version="1.0" encoding="UTF-8"?>
<xsl:stylesheet version="2.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform"

xmlns:xdmp="http://marklogic.com/xdmp"
xmlns:json="http://marklogic.com/xdmp/json"
xmlns:basic="http://marklogic.com/xdmp/json/basic"
xmlns:map="http://marklogic.com/xdmp/map"

xmlns:obj="http://marklogic.com/solutions/obi/object"

extension-element-prefixes="xdmp">

<xdmp:import-module namespace="http://marklogic.com/xdmp/json" href="/MarkLogic/json/json.xqy"/>

<xsl:param name="params" as="map:map"/>
<xsl:variable name="ingest-time" select="(map:get($params, 'ingest-time'), 'false')[1] = 'true'"/>

<!-- voor MarkLogic 7 -->
<xsl:template match="/text()[empty(../*)]">
<xsl:apply-templates select="json:transform-from-json(.)"/>
</xsl:template>

<!-- voor MarkLogic 8 -->
<xsl:template match="/node()[not(self::*)]">
<xsl:apply-templates select="json:transform-from-json(.)"/>
</xsl:template>

<xsl:template match="/basic:json">
<xsl:apply-templates select="basic:wifi__raw"/>
</xsl:template>

<xsl:template match="basic:wifi__raw">
<xsl:variable name="objects" as="element()*">
<xsl:apply-templates select="*" />
</xsl:variable>

<results>
<objects>
<xsl:sequence select="$objects" />
</objects>

</results>
</xsl:template>

<xsl:template match="basic:wifi__raw/*">
<foundId>
<xsl:value-of select="basic:id" />
</foundId>
</xsl:template>

</xsl:stylesheet>

Answer

This is caused by MarkLogic 8 handling JSON datatype differently from MarkLogic 7 and before. In MarkLogic 8, JSON data gets processed into object-node() data, not into text(). Unfortunately, you cannot explicitly test for object-node() in XSLT, but there are a few ways to work around that. Change the match for text() like this:

<xsl:template match="/node()[not(self::*)]">
  <xsl:apply-templates select="json:transform-from-json(.)"/>
</xsl:template>

This should work in both MarkLogic 7 and 8.

HTH!

Comments