Hi,

Someone knows a fast html parser to use in Pascal code?

I need something like this:

HTML:
<select name="sel_x">
<option>1</option>
<option>2</option>
</select>

I need a function/object to give me only the values: 1
2

Something like:
S := GetHTMLValues('sel_x');

Regards,
Marcos Douglas

It's not that difficult to write yourself.

You're right. But I'm searching the faster HTML parser to use in huge HTML files... thousands of files.

Best regards,
Marcos Douglas

I disagree: it's damn difficult if one isn't working with tightly constrained input, and the original question says HTML without specifying it's a subset.

There's a couple of places where I parse HTML files that I've created myself, i.e. I know exactly what's in them, using- basically- a simple recursive-descent parser with some rather flexible ideas about comments (i.e. in the above example, name="sel_x" could be lost as a comment). However if I'm doing a brute-force job over a large number of files I usually use Lynx as a preprocessor, which allows me to use standard text-processing utilities to pull named rows out of tabulated reports.

I know the tokens to search, but the HTML could be very different each other. I can't use a external tool. Need to be a application (that already exists).

Thanks,
Marcos Douglas

Marcos Douglas wrote

I know the tokens to search, but the HTML could be very different each other. I can't use a external tool. Need to be a application (that already exists).

Take a look at POWtils (aka PWU or PSP or Pascal Server Pages) created by somebody known as Z505. There has been various locations for the source code, but I think the latest is at:

https://code.google.com/p/powtils/

It has (or at least had) a very simple to use HTML parser that was very fast. If you don't come write with the above URL, I have some release archives I know contains the code. Just let me know and I can make it available.

Regards,
- Graeme -

Marcos Douglas wrote

I know the tokens to search, but the HTML could be very different each other. I can't use a external tool. Need to be a application (that already exists).

It seems a copy of the Fast HTML Parser unit I spoke of has made its way into the FPC source code tree.

See <fpc_src>/packages/chm/src/fasthtmlparser.pas

Attached is the original one I got from powtils release. It includes the parser, a utility unit and a demo program showing the parser in action with some stats output.

Regards,
- Graeme -

There is the unit fasthtmlparser included with fpc in the packages/chm folder.

It is pretty basic and just has callbacks for tags and text. I don't think it's smart enough to tell you of the

name="sel_x" part of your tag. Maybe it can be improved.

Regards,

Andrew Haines

You're right, but I change my code to use fasthtmlparser and worked (at least for now). Thank you.

Maybe it can be improved.

I agree. If I change something, I'll send a patch.

Marcos Douglas

Rainer Stratmann wrote

It's not that difficult to write yourself.

In fact, my son once did write (using Delphi) a parser that creates a list of hierarchically linked objects from HTML code and also can write a HTML file from this structure.

So you can read a file, use straight forward programming to modify the content, and write it back.

As the HTML format is not very strict and is a moving target, the parser unit is far from perfect, but it is in daily use and does a rather nice job.

OTOH, I would not say it's fast, anyway :-( .

-Michael

But the fasthtmlparser, your tip before, is a powtils' source, don't? I have the code -- for many years -- but I did not know about fasthtmlparser. It's very simple. I did not found everything I want but it is a good start.

Best regards,
Marcos Douglas

PS: Like you I use FPC in real applications in production. So I have a deadline - always short - to fulfill. So finding good code to help in our projects is very good because it makes us save time. Thanks.

In our previous episode, Marcos Douglas said:

Yes it is. The CHM parser is also based on it, but there z505 is not listed as author but as contributor:

AUTHOR : James Azarja
http://www.jazarsoft.com/

CONTRIBUTORS : L505
http://z505.com

You can try http://www.benibela.de/sources_en.html#internettools

Luiz

2014-08-07 10:20 GMT-03:00 Marco van de Voort :

In our previous episode, luiz americo pereira camara said:

You can try http://www.benibela.de/sources_en.html#internettools

That seems more something like sax_html fromt the fcl-xml package.

2014-08-08 8:28 GMT-03:00 Marco van de Voort :

It's not a simple parser. It has the ability to extract part of html through templates. See http://videlibri.sourceforge.net/cgi-bin/xidelcgi

Luiz

In our previous episode, luiz americo pereira camara said:

> It's not a simple parser. It has the ability to extract part of html through templates. See http://videlibri.sourceforge.net/cgi-bin/xidelcgi

There is xpath support in fcl-xml?

luiz americo pereira camara wrote

You can try http://www.benibela.de/sources_en.html#internettools

I will see, thanks.

Regards,
Marcos Douglas

Marco van de Voort wrote

There is xpath support in fcl-xml?

Yes. But HTML files used to be very irregular XML. Some files can raise an error when trying to open.

Things like "<p>" without closing element were easy to find.