From tim at tenkan.org  Sun Oct 23 12:54:10 2005
From: tim at tenkan.org (Tim Daly Jr.)
Date: 23 Oct 2005 14:54:10 +0200
Subject: [cl-pdf-devel] patch, pdf templates
Message-ID: <87ek6cfl71.fsf@beer.intern>


Hi everybody,

I'd like to contribute a small patch, and some code for creating
"templates" from pdf pages.

The patch mostly changes pdf-parser.lisp.  It adds location
information to a couple of error messages, a couple of comments, names
a constant, and fixes a couple of small parsing bugs.  There is one
change in pdf.lisp -- when printing numbers, we have to make sure that
they are printed in a form that's legal PDF syntax.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: cl-pdf.patch
Type: text/x-patch
Size: 3994 bytes
Desc: miscellaneous patch
URL: <https://mailman.common-lisp.net/pipermail/cl-pdf-devel/attachments/20051023/6809bdfe/attachment.bin>
-------------- next part --------------


The template code allows you to create a "Form XObject" from a PDF
page which can then be drawn onto new PDF pages.  The file stands
alone, but it might be nicer to integrate it into the cl-pdf system
and package definitions.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: pdf-template.lisp
Type: application/octet-stream
Size: 12318 bytes
Desc: pdf templates
URL: <https://mailman.common-lisp.net/pipermail/cl-pdf-devel/attachments/20051023/6809bdfe/attachment.obj>
-------------- next part --------------


The source itself can be converted into a PDF using pbook.el, which
might be a nicer way to read it:

  http://tenkan.org/~tim/pdf-template.pdf

This is an example output file, created with
(pdf:test-template "/tmp/pdf-template.pdf" 0 "/tmp/example.pdf"):

  http://tenkan.org/~tim/example.pdf

Thanks for cl-pdf.  It's beautiful code, and a lot of fun to work
with! :)

-- 
-Tim


From marc.battyani at fractalconcept.com  Sun Oct 23 20:36:54 2005
From: marc.battyani at fractalconcept.com (Marc Battyani)
Date: Sun, 23 Oct 2005 22:36:54 +0200
Subject: [cl-pdf-devel] patch, pdf templates
References: <87ek6cfl71.fsf@beer.intern>
Message-ID: <021301c5d811$94bd3a40$1401a8c0@marcxp>

"Tim Daly Jr." <tim at tenkan.org> wrote:
>
> Hi everybody,
>
> I'd like to contribute a small patch, and some code for creating
> "templates" from pdf pages.

Good idea!

> The patch mostly changes pdf-parser.lisp.  It adds location
> information to a couple of error messages, a couple of comments, names
> a constant, and fixes a couple of small parsing bugs.  There is one
> change in pdf.lisp -- when printing numbers, we have to make sure that
> they are printed in a form that's legal PDF syntax.

OK

> The template code allows you to create a "Form XObject" from a PDF
> page which can then be drawn onto new PDF pages.  The file stands
> alone, but it might be nicer to integrate it into the cl-pdf system
> and package definitions.

This is great. It was on my to-do list, so I'm glad you did it before me ;-)
I will look at this.

> The source itself can be converted into a PDF using pbook.el, which
> might be a nicer way to read it:
>
>   http://tenkan.org/~tim/pdf-template.pdf
>
> This is an example output file, created with
> (pdf:test-template "/tmp/pdf-template.pdf" 0 "/tmp/example.pdf"):
>
>   http://tenkan.org/~tim/example.pdf
>
> Thanks for cl-pdf.  It's beautiful code, and a lot of fun to work
> with! :)

You're welcome! Thanks for your contribution. ;-)

Marc



From a_bakic at yahoo.com  Sat Oct 29 23:38:12 2005
From: a_bakic at yahoo.com (Aleksandar Bakic)
Date: Sat, 29 Oct 2005 16:38:12 -0700 (PDT)
Subject: [cl-pdf-devel] unicode support (ttf + sb-unicode)
Message-ID: <20051029233813.68666.qmail@web40627.mail.yahoo.com>

Hi,

Attached is my attempt at adding Unicode support to cl-pdf. (I noticed that it
does not work correctly when pdf:*compress-streams* is true; I am using salza
with sbcl compiled for x86 on AMD64...)

One needs a Unicode TrueType font, such as MS Windows Arial and Times New
Roman. Then, one needs ttf2ufm (included in
http://www.acko.net/files/ufpdf.zip) to generate an ufm file. For example,

$ ttf2ufm -a -F arial.ttf arial

The rest should be relatively easy to figure out from the example (and a
comment in ttu-font.lisp at the end of the attachment).

So, without compression, it worked for me. I haven't tested it much, and the
patch is probably incomplete. BTW, does anyone know how I can make a smaller
font out of a TrueType one, containing only desired characters?

Regards,
Alex


	
		
__________________________________ 
Yahoo! Mail - PC Magazine Editors' Choice 2005 
http://mail.yahoo.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: cl-pdf-diffs
Type: application/octet-stream
Size: 16139 bytes
Desc: 1381307520-cl-pdf-diffs
URL: <https://mailman.common-lisp.net/pipermail/cl-pdf-devel/attachments/20051029/cc4adcef/attachment.obj>

From a_bakic at yahoo.com  Sat Oct 29 23:47:39 2005
From: a_bakic at yahoo.com (Aleksandar Bakic)
Date: Sat, 29 Oct 2005 16:47:39 -0700 (PDT)
Subject: [cl-pdf-devel] unicode support (ttf + sb-unicode)
In-Reply-To: <20051029233813.68666.qmail@web40627.mail.yahoo.com>
Message-ID: <20051029234739.42323.qmail@web40624.mail.yahoo.com>

> patch is probably incomplete

I forgot to mention that the patch includes someone else's patch that I picked
from this list's archive a few days ago.

Alex


		
__________________________________ 
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com


From a_bakic at yahoo.com  Sun Oct 30 13:24:49 2005
From: a_bakic at yahoo.com (Aleksandar Bakic)
Date: Sun, 30 Oct 2005 05:24:49 -0800 (PST)
Subject: [cl-pdf-devel] unicode support (ttf + sb-unicode)
In-Reply-To: <20051029233813.68666.qmail@web40627.mail.yahoo.com>
Message-ID: <20051030132449.45801.qmail@web34604.mail.mud.yahoo.com>

> BTW, does anyone know how I can make a smaller
> font out of a TrueType one, containing only desired characters?

This can be done using fontforge, although it is tedious...

Alex


	
		
__________________________________ 
Yahoo! Mail - PC Magazine Editors' Choice 2005 
http://mail.yahoo.com


From marc.battyani at fractalconcept.com  Sun Oct 30 19:15:22 2005
From: marc.battyani at fractalconcept.com (Marc Battyani)
Date: Sun, 30 Oct 2005 20:15:22 +0100
Subject: [cl-pdf-devel] unicode support (ttf + sb-unicode)
References: <20051029233813.68666.qmail@web40627.mail.yahoo.com>
Message-ID: <010301c5dd86$473f63e0$0a02a8c0@marcxp>

Aleksandar Bakic <a_bakic at yahoo.com> wrote:

>Attached is my attempt at adding Unicode support to cl-pdf.

Great!

>(I noticed that it
>does not work correctly when pdf:*compress-streams* is true; I am using
salza
>with sbcl compiled for x86 on AMD64...)

I fixed that.

>One needs a Unicode TrueType font, such as MS Windows Arial and Times New
>Roman. Then, one needs ttf2ufm (included in
>http://www.acko.net/files/ufpdf.zip) to generate an ufm file. For example,
>
>$ ttf2ufm -a -F arial.ttf arial
>
>The rest should be relatively easy to figure out from the example (and a
>comment in ttu-font.lisp at the end of the attachment).
>
>So, without compression, it worked for me. I haven't tested it much, and
the
>patch is probably incomplete.

It's OK for the compression now, I also made some changes for the different
functions to show strings, characters, etc.
So I think the base Unicode stuff is OK. I will look at making a full
Unicode support (Encoding stuff mostly) and also add Unicode support to
cl-typesetting.

>BTW, does anyone know how I can make a smaller
>font out of a TrueType one, containing only desired characters?

Sorry no.

Marc



From marc.battyani at fractalconcept.com  Sun Oct 30 19:15:47 2005
From: marc.battyani at fractalconcept.com (Marc Battyani)
Date: Sun, 30 Oct 2005 20:15:47 +0100
Subject: [cl-pdf-devel] cl-pdf version 2.90
References: <074c01c503f3$13b432c0$0a02a8c0@marcxp>
	<08b301c5198a$52a9fb20$0a02a8c0@marcxp>
Message-ID: <010401c5dd86$555ecec0$0a02a8c0@marcxp>

I've commited the latest version of cl-pdf to the cl-pdf repository.

In addition to the usual bunch of fixes and small improvements, this release
contains 2 new functionnalities:

-Support for basic Unicode TrueType fonts and handling [Aleksandar Bakic]
-The transformation of pages of existing pdf files as graphic templates [Tim
Daly Jr]

As usual, patches welcome ;-)

Rev 103:
cl-pdf: Version 2.9

-Added base Unicode support [Aleksandar Bakic]

-Added Unicode TrueType font support [Aleksandar Bakic]

-Extraction and transformation of pages from existing pdf files as templates
which can be drawn as graphic objects (in pdf-parser). [Tim Daly Jr]

-Change in the floting point formats for a better to improve precision

-Nil values are now allowed in the xy-charts

-Modifications for SBCL

-New examples (l-system.lisp and example9)

-Added all the small patches and fixes sent to me by a lot of cool guys ;-)


The full repository is here:
http://www.fractalconcept.com:8000/public/open-source/
(It's a subversion 1.0 repository.)

The cl-pdf part:
http://www.fractalconcept.com:8000/public/open-source/cl-pdf/

The cl-typesetting part:
http://www.fractalconcept.com:8000/public/open-source/cl-typesetting/

The mod_lisp part:
http://www.fractalconcept.com:8000/public/open-source/mod_lisp/

There are also tarballs for cl-pdf and cl-typesetting:

http://www.fractalconcept.com/download/cl-pdf-current.tgz
http://www.fractalconcept.com/download/cl-typesetting-current.tgz

BTW the (somewhat outdated) home pages for these projects are here:
http://www.fractalconcept.com/asp/html/cl-pdf.html
http://www.fractalconcept.com/asp/html/cl-typesetting.html
http://www.fractalconcept.com/asp/html/mod_lisp.html

Marc



From a_bakic at yahoo.com  Sun Oct 30 20:08:40 2005
From: a_bakic at yahoo.com (Aleksandar Bakic)
Date: Sun, 30 Oct 2005 12:08:40 -0800 (PST)
Subject: [cl-pdf-devel] unicode support (ttf + sb-unicode)
In-Reply-To: <010301c5dd86$473f63e0$0a02a8c0@marcxp>
Message-ID: <20051030200841.90278.qmail@web34615.mail.mud.yahoo.com>

> >(I noticed that it
> >does not work correctly when pdf:*compress-streams* is true; I am using
> salza
> >with sbcl compiled for x86 on AMD64...)
> 
> I fixed that.

I fixed it, too: first made byte arrays be strings of simple characters (code <
256), then modified zlib.lisp a bit (not using salza), and finally modified
sbcl-naturalize-cstring from uffi a bit. I did not like the last part because I
am not very familiar with cl-pdf and uffi, so I will probably use your
solution. Or, anyway, attached is a patch to my previous solution.

> >BTW, does anyone know how I can make a smaller
> >font out of a TrueType one, containing only desired characters?
> 
> Sorry no.

Using fontforge and Times New Roman to make a TTF font with three alphabets,
together with the fixed compression, I got ex1-ttu.pdf drop from around 470kB
to around 86kB.

Alex


		
__________________________________ 
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: new-diffs
Type: application/octet-stream
Size: 11989 bytes
Desc: 3194234562-new-diffs
URL: <https://mailman.common-lisp.net/pipermail/cl-pdf-devel/attachments/20051030/49d8ccee/attachment.obj>

From marc.battyani at fractalconcept.com  Sun Oct 30 21:32:07 2005
From: marc.battyani at fractalconcept.com (Marc Battyani)
Date: Sun, 30 Oct 2005 22:32:07 +0100
Subject: [cl-pdf-devel] unicode support (ttf + sb-unicode)
References: <20051030200841.90278.qmail@web34615.mail.mud.yahoo.com>
Message-ID: <015701c5dd99$60d8c180$0a02a8c0@marcxp>

Aleksandar Bakic <a_bakic at yahoo.com> wrote:

>I fixed it, too: first made byte arrays be strings of simple characters
(code <
>256), then modified zlib.lisp a bit (not using salza), and finally modified
>sbcl-naturalize-cstring from uffi a bit. I did not like the last part
because I
>am not very familiar with cl-pdf and uffi, so I will probably use your
>solution. Or, anyway, attached is a patch to my previous solution.

Seems a little bit too complex. ;-)
And doing this would need to be done for the other versions of zlib...

There are more important things to look at IMO. The encoding stuff and the
various functions to get the sizes, kerning information, etc. don't work
with the ttu-font-metrics. In a second time it would probably be a good idea
to try to optimize somewhat the in-memory representation of the font data.
(If it is possible of course.) I will look at that.

>Using fontforge and Times New Roman to make a TTF font with three
alphabets,
>together with the fixed compression, I got ex1-ttu.pdf drop from around
470kB
>to around 86kB.

Good to know.

Marc



From a_bakic at yahoo.com  Mon Oct 31 11:43:09 2005
From: a_bakic at yahoo.com (Aleksandar Bakic)
Date: Mon, 31 Oct 2005 03:43:09 -0800 (PST)
Subject: [cl-pdf-devel] a small patch, perhaps
Message-ID: <20051031114309.71966.qmail@web34614.mail.mud.yahoo.com>

Hi,

I had to make this patch to make cl-typesetting test.lisp work.

Index: pdf-base.lisp
===================================================================
--- pdf-base.lisp       (revision 103)
+++ pdf-base.lisp       (working copy)
@@ -35,7 +35,7 @@

 (defun write-cid-char (char)
   (write-char #\( *page-stream*)
-  (if (typep (font-metrics *font*) 'ttu-font-metrics)
+  (if (and *font* (typep (font-metrics *font*) 'ttu-font-metrics))
       (let* ((code (char-code char))
              (hi (ldb (byte 8 8) code))
              (lo (ldb (byte 8 0) code)))

Alex


		
__________________________________ 
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com


From marc.battyani at fractalconcept.com  Mon Oct 31 14:13:41 2005
From: marc.battyani at fractalconcept.com (Marc Battyani)
Date: Mon, 31 Oct 2005 15:13:41 +0100
Subject: [cl-pdf-devel] a small patch, perhaps
References: <20051031114309.71966.qmail@web34614.mail.mud.yahoo.com>
Message-ID: <032f01c5de25$4b9bb8e0$0a02a8c0@marcxp>

Aleksandar Bakic wrote:

>I had to make this patch to make cl-typesetting test.lisp work.

> (defun write-cid-char (char)
>   (write-char #\( *page-stream*)
>-  (if (typep (font-metrics *font*) 'ttu-font-metrics)
>+  (if (and *font* (typep (font-metrics *font*) 'ttu-font-metrics))

OK. I will do the same for write-cid-string too.

Marc



From a_bakic at yahoo.com  Mon Oct 31 17:55:07 2005
From: a_bakic at yahoo.com (Aleksandar Bakic)
Date: Mon, 31 Oct 2005 09:55:07 -0800 (PST)
Subject: [cl-pdf-devel] a small patch, perhaps
In-Reply-To: <032f01c5de25$4b9bb8e0$0a02a8c0@marcxp>
Message-ID: <20051031175507.33012.qmail@web34615.mail.mud.yahoo.com>

> OK. I will do the same for write-cid-string too.

BTW, I have been trying to write-document into a string instead of a file (I
need not save the PDf into a file), using sbcl with sb-unicode. There is a
problem with it because a string cannot have :external-format set to an 8-bit
type. When compression is turned on, the resulting PDF string is for some
reason truncated (I guess, some sequences of bytes are interpreted as Unicode
characters; it's strange that I haven't hit an invalid one yet).

On the other hand, I think I've read that PDF is essentially a binary format
(it allows compression, and whatever is human-readable is "by chance"). How
about using write-byte instead of write-char all over the code? There would
probably still be a problem with strings: I can (MAKE-STRING-OUTPUT-STREAM
:ELEMENT-TYPE '(UNSIGNED-BYTE 8)) but GET-OUTPUT-STREAM-STRING expects a string
of characters... Perhaps there is way around the problem?

Alex


		
__________________________________ 
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com


From marc.battyani at fractalconcept.com  Mon Oct 31 20:24:17 2005
From: marc.battyani at fractalconcept.com (Marc Battyani)
Date: Mon, 31 Oct 2005 21:24:17 +0100
Subject: [cl-pdf-devel] a small patch, perhaps
References: <20051031175507.33012.qmail@web34615.mail.mud.yahoo.com>
Message-ID: <040801c5de59$11e71660$0a02a8c0@marcxp>

Aleksandar Bakic wrote:

> > OK. I will do the same for write-cid-string too.
>
> BTW, I have been trying to write-document into a string instead of a file
(I
> need not save the PDf into a file), using sbcl with sb-unicode. There is a
> problem with it because a string cannot have :external-format set to an
8-bit
> type. When compression is turned on, the resulting PDF string is for some
> reason truncated (I guess, some sequences of bytes are interpreted as
Unicode
> characters; it's strange that I haven't hit an invalid one yet).

On what Lisp ?

> On the other hand, I think I've read that PDF is essentially a binary
format
> (it allows compression, and whatever is human-readable is "by chance").
How
> about using write-byte instead of write-char all over the code? There
would
> probably still be a problem with strings: I can (MAKE-STRING-OUTPUT-STREAM
> :ELEMENT-TYPE '(UNSIGNED-BYTE 8)) but GET-OUTPUT-STREAM-STRING expects a
string
> of characters... Perhaps there is way around the problem?

Dmitri is a fan of turning cl-pdf in binary mode. ;-)
I'm convinced too that it's the way to go but I fear it would break on a lot
of implementations. So why not, but this needs to have enough people to test
on all the impementations.

Anyway let's finish the Unicode stuff first. I'm working on the
unicode-encoding now.

Marc



From a_bakic at yahoo.com  Mon Oct 31 22:56:30 2005
From: a_bakic at yahoo.com (Aleksandar Bakic)
Date: Mon, 31 Oct 2005 14:56:30 -0800 (PST)
Subject: [cl-pdf-devel] a small patch, perhaps
In-Reply-To: <040801c5de59$11e71660$0a02a8c0@marcxp>
Message-ID: <20051031225630.67271.qmail@web34612.mail.mud.yahoo.com>

> On what Lisp ?

SBCL. I figured it out. It is always due to mixing characters and bytes...

> Dmitri is a fan of turning cl-pdf in binary mode. ;-)
> I'm convinced too that it's the way to go but I fear it would break on a lot
> of implementations. So why not, but this needs to have enough people to test
> on all the impementations.

On SBCL, I have found a way: using arrays of (unsigned-byte 8) instead of
strings of (unsigned-byte 8). That would require using vector-push-extend
instead of write-byte. Once you have a binary array, write-sequence dumps it
nicely into a file; whereas, if it were a binary string, that would not work in
SBCL for some reason (maybe it will in the future).

> Anyway let's finish the Unicode stuff first. I'm working on the
> unicode-encoding now.

Great, I am looking forward to it (and am not sure what is all there to
finish). I am also trying to generate PDFs dynamically using cl-typesetting in
the meantime. Are all those macros involving *document*, *page*, etc.,
necessary? Could some of them be turned into functions? I would also like if it
were possible to link objects (even) later (I am happy to notice that the PDF
macros do not generate PDF result immediately, unlike many HTML-generating
libraries), so that I can build pieces of a documents independently and then
compose them. Now, I cannot just build a page, because whenever I make an
instance of indirect-object (sub)class, *document* is used for linking... (or
something along those lines).

Alex


		
__________________________________ 
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com


From marc.battyani at fractalconcept.com  Mon Oct 31 23:32:32 2005
From: marc.battyani at fractalconcept.com (Marc Battyani)
Date: Tue, 1 Nov 2005 00:32:32 +0100
Subject: [cl-pdf-devel] a small patch, perhaps
References: <20051031225630.67271.qmail@web34612.mail.mud.yahoo.com>
Message-ID: <04aa01c5de73$61230ee0$0a02a8c0@marcxp>

Aleksandar Bakic wrote:

> On SBCL, I have found a way: using arrays of (unsigned-byte 8) instead of
> strings of (unsigned-byte 8). That would require using vector-push-extend
> instead of write-byte. Once you have a binary array, write-sequence dumps
it
> nicely into a file; whereas, if it were a binary string, that would not
work in
> SBCL for some reason (maybe it will in the future).
>
> > Anyway let's finish the Unicode stuff first. I'm working on the
> > unicode-encoding now.
>
> Great, I am looking forward to it (and am not sure what is all there to
> finish).

I've almost finished. In fact there is a problem with the ufm file. This
file does not contains the bounding boxes for the characters. So both the
afm and ufm files must be parsed.

> I am also trying to generate PDFs dynamically using cl-typesetting in
> the meantime.

Good idea, it's made for that. ;-)

> Are all those macros involving *document*, *page*, etc.,
> necessary? Could some of them be turned into functions?

It's always possible but then you will have to create lambda functions to
pass to them.
Maybe a functional version in addition to the macros would be good.

> I would also like if it
> were possible to link objects (even) later (I am happy to notice that the
PDF
> macros do not generate PDF result immediately, unlike many HTML-generating
> libraries), so that I can build pieces of a documents independently and
then
> compose them. Now, I cannot just build a page, because whenever I make an
> instance of indirect-object (sub)class, *document* is used for linking...
(or
> something along those lines).

I don't see where is the problem, do you have an example.

Marc



From a_bakic at yahoo.com  Mon Oct 31 23:58:25 2005
From: a_bakic at yahoo.com (Aleksandar Bakic)
Date: Mon, 31 Oct 2005 15:58:25 -0800 (PST)
Subject: [cl-pdf-devel] a small patch, perhaps
In-Reply-To: <04aa01c5de73$61230ee0$0a02a8c0@marcxp>
Message-ID: <20051031235825.55646.qmail@web34607.mail.mud.yahoo.com>

> Maybe a functional version in addition to the macros would be good.

OK, I'll assume there are no fundamental issues with this and give it a try
soon.

> I don't see where is the problem, do you have an example.

Well, take the first example in test.lisp and remove the wrapping
with-document. Just try to evaluate the with-page form (with its contents or
not).

Alex


		
__________________________________ 
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com