Re: [Exist-development] org.xml.sax.SAXParseException - Prefixed namespace bindings may not be empty.

[Adam R. <ad...@ex...>]

Perhaps one option would be to switch namespace processing off in Neko
for this and then later add the namespaces back in using a custom
filter (if that is desirable).

http://nekohtml.sourceforge.net/settings.html#namespaces
http://nekohtml.sourceforge.net/settings.html#filters

On 1 September 2014 06:36, Alister Pillow <gsp...@gm...> wrote:
> The error in more detail… Only happening when storing the document, not when parsing.
>
> 2014-08-31 17:19:02,585 [eXistThread-31] DEBUG (ModuleUtils.java [htmlToXHtml]:251) - Converting HTML to XML using NekoHTML parser for: alternative
> 2014-08-31 17:19:03,724 [eXistThread-31] DEBUG (TransactionManager.java [execute]:159) - Starting new transaction: 5
> 2014-08-31 17:19:03,730 [eXistThread-31] DEBUG (Collection.java [validateXMLResourceInternal]:1631) - Scanning document /db/test/message-6.xml
> 2014-08-31 17:19:03,731 [eXistThread-31] DEBUG (GrammarPool.java [retrieveInitialGrammarSet]:81) - Retrieve initial grammarset (http://www.w3.org/TR/REC-xml).
> 2014-08-31 17:19:03,731 [eXistThread-31] DEBUG (GrammarPool.java [retrieveInitialGrammarSet]:85) - Found 0 grammars.
> 2014-08-31 17:19:03,740 [eXistThread-31] DEBUG (Indexer.java [fatalError]:419) - fatal error at (58,2064) : The value of the attribute "prefix="xmlns",localpart="o",rawname="xmlns:o"" is invalid. Prefixed namespace bindings may not be empty.
> 2014-08-31 17:19:03,740 [eXistThread-31] DEBUG (GrammarPool.java [retrieveInitialGrammarSet]:81) - Retrieve initial grammarset (http://www.w3.org/TR/REC-xml).
> 2014-08-31 17:19:03,740 [eXistThread-31] DEBUG (GrammarPool.java [retrieveInitialGrammarSet]:85) - Found 0 grammars.
> 2014-08-31 17:19:03,741 [eXistThread-31] ERROR (XMLDBStore.java [evalWithCollection]:220) - The XML parser reported a problem: fatal error at (58,2064) : The value of the attribute "prefix="xmlns",localpart="o",rawname="xmlns:o"" is invalid. Prefixed namespace bindings may not be empty.
> org.xmldb.api.base.XMLDBException: The XML parser reported a problem: fatal error at (58,2064) : The value of the attribute "prefix="xmlns",localpart="o",rawname="xmlns:o"" is invalid. Prefixed namespace bindings may not be empty.
>         at org.exist.xmldb.LocalCollection.storeXMLResource(LocalCollection.java:893)
>         at org.exist.xmldb.LocalCollection.storeResource(LocalCollection.java:766)
>         at org.exist.xmldb.LocalCollection.storeResource(LocalCollection.java:754)
>
> I’ve tried adding a RemoveElement filter to the NekoHTML parser - but can’t make that work - no content is returned. The documentation seems to suggest that I have to add an acceptElement for every element I want to keep - which means all of HTML!!!
>
> Now I’m trying a FilterInputStream approach - but that seems such a blunt instrument - any other suggestions on how to remove <o:p></o:p> from the result of htmlTXHtml?
>
> This is how I modified ModuleUtils.htmlToXHtml
>
>   LOG.debug("Converting HTML to XML using NekoHTML parser for: " + url);
>             reader = (XMLReader) Class.forName("org.cyberneko.html.parsers.SAXParser").newInstance();
>
>             ElementRemover remover = new ElementRemover();
>             remover.removeElement("o:p");
>             // setup filter chain
>             XMLDocumentFilter[] filters = {
>                     remover
>             };
>
>             reader.setProperty("http://cyberneko.org/html/properties/names/elems","match");
>             reader.setProperty("http://cyberneko.org/html/properties/names/attrs","no-change");
>             reader.setProperty("http://cyberneko.org/html/properties/filters", filters); // ADDED FILTER here
>
>
> On 31 Aug 2014, at 6:11 pm, Alister Pillow <gsp...@gm...> wrote:
>
>> Hi,
>> Trying to finish off the mail:get-messages function. The error appears when trying to store an email as xml in the /db.
>>
>> I’ve hit a nasty little bug - nothing to do with eXist - and (surprise) related to mail from Microsoft Outlook - via Apple Mail. I now suspect that Apple Mail is the culprit.
>>
>> I forwarded an email from my Inbox to another account and then retrieved it using mail:get-messages.
>> The original html section is full of <o:p></o:p> tags. These are end-of-paragraph markers inserted by MS Word when creating HTML.
>>
>> In the original email, the prefix and namespace is declared - but in the forwarded message, it is missing - consequently I get an error from SAXParser when trying to store this content in the DB.
>>
>> Is there some way to tell the parser to skip these (empty) elements? Or will I have to write a filter for the text before parsing it?
>>
>> I’m using Wolfgang’s suggestion:
>> DocumentImpl html = ModuleUtils.htmlToXHtml(context, "alternative", new StreamSource(part.getInputStream()), null, null);
>> ElementImpl rootElem = (ElementImpl)html.getDocumentElement();
>>
>> (Otherwise, mail:get-messages is working quite nicely.)
>>
>> Thanks,
>> Alister.
>
>
> ------------------------------------------------------------------------------
> Slashdot TV.
> Video for Nerds.  Stuff that matters.
> http://tv.slashdot.org/
> _______________________________________________
> Exist-development mailing list
> Exi...@li...
> https://lists.sourceforge.net/lists/listinfo/exist-development



-- 
Adam Retter

eXist Developer
{ United Kingdom }
ad...@ex...
irc://irc.freenode.net/existdb