Re: (no subject)

On 7/07/2013, at 8:32 AM, Josef Frydl wrote:

> I per may message:
> Hi, I am writing the small system in swi-prolog, which would search google and then bring down   individual sites and hrefs. which are in the query result.  Before I reinvent the wheel, let me ask you if there is already the utility which would would pick only 
> this    www.swi-prolog.org  from this  <cite>www.swi-<b>prolog</b>.org/</cite>
> 
> I decided to write DCG to do just that, which is simple, first it is very simplistic, since it is just first cut but got this error from  phrase_from_file

The important thing here is NOT to write your own stuff for parsing HTML
but to use the SGML support provided in SWI Prolog -- the sgml package.

Note that
1. <cite>www.swi-<b>prolog</b>.org/</cite>
2. <cite
   >www.swi-<!--
   --><b>prolog</b
   >.<!--
   -->org/</cite
   >
3. <cite><![CDATA[www.swi-]]><b><![CDATA[prolog]]></b><![CDATA[[.org/]]></cite>

are all legal, and in SGML, indistinguishable, ways to express the same
thing.  (Yes, I know that there are comment and CDATA objects in the XML
DOM.  I said *SGML*, where such things do *not* officially exist.)  Even
in XML,

4. <cite
   >www.swi-<b
   >prolog</b
   >.org/</cite
   >
is *identical* to 1.  And of course, let's never forget

5. <cite>www&#46;swi&#45;<b>prolog</b>&#46;org&#47;</cite>

which is also identical to 1 in XML and SGML.
(Claims verified by pushing all examples through an XML parser than
can be told to ignore comments and treat CDATA as plain data.)