Re: fo:external-graphic and xml:base

The source of the problem is that FOP is mistakenly calculating the 
base-URI for the images. The purpose of xml:base is to rectify 
situations like this, but it seems that FOP does not know what to do 
with xml:base. For you, it's reporting xml:base as an invalid attribute 
of the fo:block element. For me, it just ignores xml:base and continues 
(of course still incorrectly resolving the image URIs).

First thing I would verify is that your XSLT and/or XSL-FO files are 
aware of the "xml" namespace. I tried this first on my setup, but it did 
not help. Maybe it will help you?

If that doesn't work, there are certainly ways to work around the problem:

1) Download the images to a corresponding local directory in a 
pre-processing step. Or you could program a Xalan extension that 
basically does the same thing "inline" in the XSLT.

2) Explicitly compute the base URI for the particular HTML document 
you're processing and use that value to create an absolute URI in 
<fo:external-graphic>. Something like "concat($base-URI,@src)". You 
could either pass in $base-URI as a global XSLT parameter (if you know 
this beforehand), or you can calculate it dynamically for each document 
you process by finding the directory of the a/@href attribute (the 
portion of the URI leading up to the file, but not the file name 
itself). Then you'd send that value along to the matching templates with 
<xsl:apply-templates><xsl:with-param name="base-URI" select="blah"/>, 
and you'd need to make sure your matching templates include that 
parameter with <xsl:param name="base-URI"/>. Then, when you're 
outputting the external graphic, you'd do something like 
<fo:external-graphic src="url('{concat($base-URI,@src)}')"/>.

3) Tell FOP to use a given base-URI via the configuration file. See 
http://xmlgraphics.apache.org/fop/0.94/configuration.html. It seems you 
can only do this if you already know the base-URI value beforehand, and 
you can only have one value per FOP run. I actually used this option 
with your example files and it worked for me. No change was needed to 
your files, and I didn't need to use xml:base, but of course you need to 
know the base-URI beforehand, or otherwise somehow loop through calls to 
FOP, updating the config file appropriately on each call.

Cheers,
Nathan Nadeau


Jack Bates wrote:
> I originally posted this question to the FOP users list,
> http://thread.gmane.org/gmane.text.xml.fop.user/29778
>
> I'm writing a stylesheet that includes this template,
>
>   <xsl:template match="html:img">
>     <fo:external-graphic src="url('{@src}')"/>
>   </xsl:template>
>
> - but when I process it I get errors like,
>
>  $ fop -xml index.html -xsl index.xsl index.pdf
>  [...]
>  16-Nov-2009 11:06:32 AM org.apache.fop.events.LoggingEventListener processEvent
>  SEVERE: Image not found. URI: /docs/images/thumb/b/bb/UM-2.1.png/500px-UM-2.1.png. (No context info available)
>  [...]
>
> - and the generated PDF is missing some images : (
>
> I think the problem is that the stylesheet uses the document() function
> to get HTML pages from the web,
>
>   <xsl:template match="html:a" mode="foo">
>     <fo:block>
>       <xsl:apply-templates select="document(@href)//html:body"/>
>     </fo:block>
>   </xsl:template>
>
> - and these pages, e.g. http://ica-atom.org/docs/index.php?title=UM-2.1
>
> - have some relative images,
> e.g. /docs/images/thumb/b/bb/UM-2.1.png/500px-UM-2.1.png
>
> The <img> elements are part of the HTML document, and so relative to
> http://ica-atom.org/docs/index.php?title=UM-2.1, but I guess the
> <external-graphic> elements aren't?
>
> I tried adding an xml:base="" attribute,
>
>   <xsl:template match="html:a" mode="foo">
>     <fo:block xml:base="{@href}">
>       <xsl:apply-templates select="document(@href)//html:body"/>
>     </fo:block>
>   </xsl:template>
>
> Here's a simplified, complete example,
>
> http://www.sfu.ca/~jdbates/tmp/fop/200911160/index.html
>
> http://www.sfu.ca/~jdbates/tmp/fop/200911160/index.xsl
>
> - but now when I process this example I get this error,
>
>  $ fop -xml index.html -xsl index.xsl index.pdf
>  16-Nov-2009 11:29:57 AM org.apache.fop.cli.Main startFOP
>  SEVERE: Exception
>  javax.xml.transform.TransformerException: org.apache.fop.fo.ValidationException: Invalid property encountered on "fo:block": xml:base (No context info available)
>          at org.apache.fop.cli.InputHandler.transformTo(InputHandler.java:314)
>          at org.apache.fop.cli.InputHandler.renderTo(InputHandler.java:146)
>          at org.apache.fop.cli.Main.startFOP(Main.java:174)
>          at org.apache.fop.cli.Main.main(Main.java:205)
>  [...]
>
> How can I automatically convert these HTML pages to a PDF with XSL-FO?
>