handling-malformed-xml-in-pptx.txt

pptx file contains this malformed xml in member presentation.xml:

<p:presentation>
  <p:sldMasterIdLst>
    <p:sldMasterId id="2147483648" r:id="rId2"/>
  </p:sldMasterIdLst>
  <p:sldIdLst>
    <p:sldId id="256" r:id="rId3"/>
    <p:sldId id="257" r:id="rId4"/>
    <p:sldId id="258" r:id="rId5"/>
    <p:sldId id="259" r:id="rId6"/>
    <p:sldId id="260" r:id="rId7"/>
    <p:sldId id="261" r:id="rId8"/>
  </p:sldIdLst>
  <p:sldSz cx="12192000" cy="6858000"/>
  <p:notesSz cx="7772400" cy="10058400"/>
</p:presentation>

It's malformed because there are no namespace declarations. Nokogiri can't handle that.

Notes for dealing with it:

0. Extract the presentation.xml entry from the pptx package using rubyzip

package.extract('ppt/presentation.xml', some-file-name)

1. Get the raw text from the extracted file. Remove newlines as these will be converted into useless XML text nodes by Nokogiri.

raw = IO.read(some-file-name).gsub(/\n/,'')

2. Replace the namespace prefix delimiters with plain text so Nokogiri can process the XML. This will change 'p:presentation' to 'p__presentation' and so forth. This is a workaround for the fact the namespaces aren't declared in the presentation.xml file.

mod = raw.gsub(/:/,'__')

3. Make a Nokogiri document out of the munged XML text.

xml = Nokogiri::XML(mod)

4. Get the last sldId element in the document. Get the last node from the Nodeset and then the last Element in that node. You end up with a Nokogiri Element. 

last_sldId_element = xml.xpath("//p__presentation/p__sldIdLst").last.last_element_child

At this point, 'puts last_sldId_element' should return something like this:

<p__sldId id="261" r__id="rId8"/>

5. Now save the id value as an integer.   

last_sldId_id_value = last_sldId_element['id'].to_i

6. Save the number at the end of the r:id value as an integer. The value looks like 'rId8'.

last_sldId_rid_value = last_sldId_element['r__id']
last_sldId_rid_value = last_sldId_rid_value[3..last_sldId_rid_value.length].to_i

6. For each additional slide that was added to the deck:

6.1. Increment the id value

last_sldId_idvalue += 1

6.2. Increment the r:id value and save it as a string

last_sldId_rid_value += 1
sldId_rid_value_str = 'rId' + last_sldId_rid_value.to_s

6.2. Make a new child element under <p__presentation><p__sldIdLst>

sldId_node = Nokogiri::XML::Node.new('p__sldId',xml)
sldId_node['id'] = last_sldId_id_value.to_s
sldId_node['r__id'] = sldId_rid_value_str
xml.xpath("//p__presentation/p__sldIdLst").add_next_child(sldId_node)

----

Now we need to replace the presentation.xml entry in the pptx with the modified one.

7. Reverse the string replacement to restore the namespace prefixes

updated_xml = xml.to_s.gsub(/__/,':')