sandal / prawn

Currently, it is not possible to combine the :align option with absolute-positioned text (i.e. when using :at option). The text starts at the :at point, making it always left-aligned. It would be very useful if this point could be used as reference for centered or right aligns as well. So, when :align == :center, the text would be positioned just in middle of :at's x; when :align == :right, the text would be positioned before :at's x; and when :align == :left, the text would be positioned after :at's x, as it currently does. In fact, it seems very simple to implement this using width_of method.

would be useful for go_to_page() and other things. Returns the current page number

On Acrobat Reader 9.2 on Gentoo linux, the following code illustrates an issue with the color refactoring commit at:

http://github.com/sandal/prawn/commit/28393eb96734ebd017a6fc7b9c794c57a66c3f1e

What I'm seeing is that it displays the stroked rectangle correctly on the first page, but all subsequent pages are left blank. I am reverting for now. If we can fix this, feel free to pull the change back in.

Prawn supports jpg and png images to get embbeded on pdf it would be nice if it supports gif images too.

Perhaps this is a prawn/format issue, and if so I apologize for posting this in the wrong place. In 0.5.x, you could put HTML line breaks (
) inside a cell and have the table render correctly. Since 0.6, it now occasionally adds a page break instead (and then, not always).

For example:

Prawn::Document.generate('test1.pdf') do
  table [['A', 'B<br/>C'], ['D', 'E<br/>F']]
end

The result is a 3-page pdf that looks very bad.

From 126f494 Mon Sep 17 00:00:00 2001
From: JeanLuc HAVY 
Date: Sun, 15 Nov 2009 01:45:18 +0100
Subject: [PATCH] Fixed links creation in annotations.rb

In method annotations.rb:annotate a reference identifier was used instead
of the actual reference. The links in generated pdf weren't working.

Results in pdf were for example :
/Annots [7]

instead of :
/Annots [7 0 R]
---
 lib/prawn/document/annotations.rb |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/lib/prawn/document/annotations.rb b/lib/prawn/document/annotations.rb
index 61df1ab..e3925f9 100644
--- a/lib/prawn/document/annotations.rb
+++ b/lib/prawn/document/annotations.rb
@@ -23,7 +23,7 @@ module Prawn
       def annotate(options)
         current_page.data[:Annots] ||= []
         options = sanitize_annotation_hash(options)
-        current_page.data[:Annots] << ref(options)
+        current_page.data[:Annots] << ref!(options)
         return options
       end
 
-- 
1.6.5.2

Jean-Luc

Hi,
I just upgrade from 0.5.1 to 0.6.2 and discover that my code break
with 0.6.2.
I narrow the problem down to the new 'extensions' class instance
variable (or whatever it is called in ruby terminology).
The following code works with 0.5.1 and doesn't with 0.6.2 :

require 'prawn'
require 'prawn/format'

class TestPdf < Prawn::Document
end

t1 = Prawn::Document.new
puts Prawn::Document.extensions.inspect
puts t1.tags.inspect

t2 = TestPdf.new
puts TestPdf.extensions.inspect
puts t2.tags.inspect

And here is the problem, 'extensions' is empty for the TestPdf class
as it isn't inherited.

So if I want to subclass a Prawn::Document class, I have to add
TestPdf.extensions << Prawn::Format.

So the question is : Is this the desired behavior ??
If not, maybe 'extensions' could be a class variable (@@ extensions)
which should solve this issue.

Jean-Luc

I am trying to use transactions. I am creating a complex document (ballot), and I die when trying to start a transaction every time. Here is a 10 liner reproducible case:

require 'rubygems'
require 'prawn' 

pdf = Prawn::Document.new(:skip_page_creation => true)
pdf.start_new_page
pdf.font_families.update({
    "Helvetica" => { :normal => "/Library/Fonts/Arial Unicode.ttf",
                     :bold => "/Library/Fonts/Arial Bold.ttf" }}
   )
pdf.font("Helvetica", :size => 13)
pdf.text "How are you"
pdf.transaction do 
    pdf.text "Nice, thank you"
end
puts pdf.render

The code dies HERE:
prawn-core-0.6.1/lib/prawn/document/snapshot.rb

def take_snapshot
  {:page_content    => Marshal.load(Marshal.dump(page_content)),
   :current_page    => Marshal.load(Marshal.dump(current_page)), <<< HERE

The problem is that
current_page #
    {:Font=>
      {"F1.0"=>
        #:CourierNewPSMT, :Type=>:Font},
         @gen=0,
         @identifier=7,
         @on_encode=
          #,
         @stream=nil>,

You can see that font reference has on_encode object of type Proc, which cannot be marshaled.
http://corelib.rubyonrails.org/classes/Marshal.html

For now, my workaround will be to not declare custom fonts.

The origin of BBox is at the lower-left corner of the page. Therefore, it must be as wide and as tall as the page, not the bounds (which already account for margin), or else we get cut off at the right and top.

Fix available through http://github.com/Bluejade/prawn

-Daniel

The general/page_numbering example shows this problem. When pages 1 and 2 are reopened, their content stream dictionary writes the wrong length (94 for page 1; should be around 170). The length does not match the actual length of the stream.

Preview on OS X reads until endstream without regard for the length; Adobe Reader 9.2 on Windows stops reading the stream at the dictionary-specified length.

Line 33 of Font.rb included direct references to @current_page, instead of current_page. I've added spec coverage and the fix and pushed them to http://github.com/Bluejade/prawn/tree/

-Daniel

Here is a list of features I've submitted to be considered for the 0.6 release, all available through the master branch at http://github.com/Bluejade/prawn:>

stroke dash
stroke cap style (a late addition)
stroke join style (another late addition)
transparency
stamps

Add option to position an image on the page, but not move the text cursor afterwards. See shammond42/prawn (http://github.com/shammond42/prawn/tree/) for implementation, documentation and examples.

Greg, Brad, I've incorporated the changes you recommended and made a few other improvements (such as a couple new specs). Please note that everything for both transparency and stamp is now on the master of my Prawn fork: http://github.com/bluejade/prawn. I hope that works okay for you.

Please let me know if anything else should be done on these.

Best,

Daniel

I just found what i think is a bug in the nametree << method :
That code taken from name_tree_spec.rb works ok:

  it "should keep tree balanced when subtree split cascades to root" do
    node = Prawn::NameTree::Node.new(@pdf, 3)
    tree_add(node, ["one", 1], ["two", 2], ["three", 3], ["four", 4])
    tree_add(node, ["five", 5], ["six", 6], ["seven", 7], ["eight", 8])
    tree_dump(node).should == "[[[eight=8,five=5],[four=4,one=1]],[[seven=7,six=6],[three=3,two=2]]]"
  end

but the same code with the values pre-ordered doesn't :

  it "should keep tree balanced when subtree split cascades to root" do
    node = Prawn::NameTree::Node.new(@pdf, 3)
    tree_add(node, ["eight", 8], ["five", 5], ["four", 4,],  ["one",1]
    tree_add(node, ['seven', 7], ['six', 6], ['three', 3], ['two', 2])
    tree_dump(node).should == "[[[eight=8,five=5],[four=4,one=1]],[[seven=7,six=6],[three=3,two=2]]]"
  end

Here is the output :

/usr/lib/ruby/gems/1.9.1/gems/prawn-core-0.5.1/lib/prawn/name_tree.rb:83:in `<<': undefined method `<<' for nil:NilClass (NoMethodError)
        from /usr/lib/ruby/gems/1.9.1/gems/prawn-core-0.5.1/lib/prawn/
name_tree.rb:42:in `add'
        from name_tree_spec.rb:13:in `block in tree_add'
        from name_tree_spec.rb:12:in `each'
        from name_tree_spec.rb:12:in `tree_add'

nametree.rb:83 is "fit << value"
but fit is nil because in the previous line : fit = children.detect { |
child| child >= value }
"child >= value" is always false if the values inserted are already ordered.

This patch fixes it for me :

diff --git a/lib/prawn/name_tree.rb b/lib/prawn/name_tree.rb
index 267a6e9..c7d20c8 100644
--- a/lib/prawn/name_tree.rb
+++ b/lib/prawn/name_tree.rb
@@ -79,6 +79,7 @@ module Prawn
           split! if children.length > limit
         else
           fit = children.detect { |child| child >= value }
+          fit = children.last unless fit
           fit << value
         end

-- jlh

As detailed here: http://groups.google.com/group/prawn-ruby/browse_thread/thread/5109563b0d766de

As detailed here: http://groups.google.com/group/prawn-ruby/browse_thread/thread/5109563b0d766de

See title :)

Hello

For example for german umlauts compute_width_of() reports up to the double width of the same character without diaeresis (e.g. Ü has twice the width of U).

Here some sample code:

#!/usr/local/bin/ruby
  
require 'rubygems'
require 'prawn'

pdf = Prawn::Document.new

%w(wertbestätigung wertbestatigung Ä A Ü U Ö O Â A).each do |heading|
  str_width = pdf.font.compute_width_of(heading)
  pdf.text "#{heading} width: '#{str_width}' "
end

def print_stretched pdf, string, gap, position_y
  char_h = {}
  string_width = 0.0
  offset = 0
  char_a = string.split('')
  char_a.each { |c| string_width += char_h[c] = pdf.font.compute_width_of(c).to_f }
  string_width += ( char_a.length - 1 ) * gap
  position_x = pdf.bounds.width / 2 - string_width / 2
  char_a.each do |c|
    pdf.text c, :at => [position_x+offset, position_y]
    offset += char_h[c] + gap
  end
end

pos = 400
%w(rechnung kostenvoranschlag wertbestätigung angebot).each do |heading|
  print_stretched pdf, heading, 2.0, pos
  pos+= 30
end

pdf.render_file('stretched.pdf')

Font#normalize_encoding is described in the docs as modifiying the argument, which it does not anymore. Font#normalize_encoding! is the destructive version of the method.

Would it be possible (or even a good idea) to change document.width_of so that it always does a normalize_encoding automagically?

Having both :at and :align options on the text method, makes you think they would work together, which they don't. This should be fixed, or reflected in the documentation.

Hi,

I got an error
  /usr/lib64/ruby/gems/1.8/gems/prawn-format-0.2.0.1/lib/prawn/

format/parser.rb:120:in `start_parse': closing b, but no tags are open
(Prawn::Format::Parser::TagError)

when displaying text like that

pdf.text "<b>this is simple plain text where plain = true</b>

", :plain => true

This is working fine in earlier version.

If a table cell is wide, the next table header wraps, or all subsequent table headers overlap into the final. This can occur in the middle of the table.

you wrote that i should create an issue to remind you. so please apply my fixes:
http://github.com/corny/prawn-layout/commit/7524535bd421e36594ae1b64c8160fca0cb4fb36
http://github.com/corny/prawn-layout/commit/3883f8b9c6ac3b0dff017c28b3e01b8715b31d7f
http://github.com/corny/prawn-layout/commit/8738c3d51cacbf57b2452e47323a5a9d27721270

In ruby 1.8.7, enumerator.rb doesn't exist. It's not a problem if a script using prawn is executed by ruby interpreter, however, a Windows binary generated by exerb from the script cannot execute correctly because the binary tries to load enumerator.so but cannot load it.

Please fix this issue. Thanks!

The four binary characters recommended by the spec (as a sentinel for file-transfer programs) should be in a PDF comment: "it is recommended that the header line be immediately followed by a
comment line containing at least four binary characters". Confirmed that at least Apple's PDF framework does this correctly.

Fixed in bradediger/prawn@a14f21e4. Sending pull request.

using center layout in prawn is not correct as the length of lines is not correctly determined (incloding '\n' character)

This was published and discussed on in google group:
http://groups.google.com/group/prawn-ruby/browse_thread/thread/6754ada443dd22cd
workaround is to overwrite width_of method by manually removing tailing '\n' characters. (sorry the layout is broken as githup is always reformating this as coding)

module Prawn
class Document

alias_method :orig_width_of, :width_of
def width_of(string, options={})
  orig_width_of(string.gsub(/\n\z/,""),options)
end

end end

Including a PNG with RGB+alpha (color type 6) in a PDF is not feasible because the method "unfilter_image_data" is incredibly slow. A quick benchmark of spec/png_spec.rb with and without the tests that hit this method:

(2.4GHz Intel Core 2 Duo) with: "Finished in 13.779863 seconds."
w/out: "Finished in 0.391898 seconds."

That's on a 320x240 pixel PNG. It gets a lot worse the bigger they get. I have a use case where a 474x719 PNG is included in a PDF and it takes 24 seconds to create.

The bummer here is that many graphical apps (Pixelmator, Illustrator, etc) always flip the alpha transparency bit so this affects many images.

I had used prawn to generate some chinese pdf documents successfully
when set the :wrap option to :character, but the wrap function failed for document which include tags
after I required 'prawn/format'.

Allow printer's marks (bleed, crop, etc) to be drawn on the pages through use of something like :prepress => [bleed, crop] in Document.generate.

Examples:
http://www.jukeboxprint.com/images/crop_marks.gif
http://www.ewbi.com/ewbi.develop/graphics/word.crop.marks.jpg

A bit pie in the sky at the moment, but....

grouped do
   # ...
end

Would attempt to keep contents on the same page

Hi

 I am creating a request here according to the discussions at

http://groups.google.com/group/prawn-ruby/browse_thread/thread/aef186542b250187

    Could anybody give a solution to this?

Thanks
sk

Trying to keep my images in a vector format so that printing and scaling of the prawn generated document stays crisp and lightweight. Would love to be able to embed either EPS or PDF vector art into my prawn generated document.

(Note: I've replaced this ticket with a different description --greg)

See http://github.com/rogergl/prawn/commit/5f11d7701e9b63c62d6b7ad3922a3029740f2344 for a reference implementation, but we really should centralize our alignment hackery so we don't repeat this same basic procedure again and again

It seems that font.width_of was changed to font.compute_width_of

Hello

As I already reported in a ticket on prawn.lighthouseapp.com/ (URL: http://is.gd/1bGe3 ) for 0.4 in 0.5.1 remains a bug with table cells and a :leading greater than 0. In particular the bug is in prawn/document/text/wrapping.rb and does probably not only concern tables.

A sample script that produces the misbehavior is here: https://prawn.lighthouseapp.com/attachments/128185/prawntest.rb

I fixed it for me in the following way (diff for wrapping.rb):

22c22,23
<           string.lines.to_a.length * font.height_at(size)
---
>           lcount = string.lines.to_a.length
>           lcount * font.height_at(size) + ( lcount - 1 ) * text_options[:spacing].to_f

So long, T.

Allow :margins => [a,b,c,d] or even :margins => k

having to do pdf.y = dy + bounds.absolute_bottom sucks.

Currently, Prawn does not support displaying the total page count of the generated PDF. From this post I implemented the suggested hack in my forked version of prawn located here

ImageMagick (and reportedly Ghostscript) doesn't like Prawn's EOF syntax. I don't believe the particular file in question is relevant -- any PDF from Prawn spits out this error:

$ convert --version
Version: ImageMagick 6.5.1-0 2009-05-23 Q16 http://www.imagemagick.org
Copyright: Copyright (C) 1999-2009 ImageMagick Studio LLC

$ convert -quality 100 out/0000.pdf out/0000.jpg

**** This file has a corrupted %%EOF marker, or garbage after the %%EOF.
**** The file was produced by Prawn:
**** please notify the author of this software
**** that the file does not conform to Adobe's published PDF
**** specification.  Processing of the file will continue normally.

Of course, this is a lie; the file is in compliance with the Adobe spec, but ImageMagick expected a newline after the %%EOF marker. Adding the newline silences that warning. According to the PDF spec (reference 1.3, appendix H, implementation note 15), Adobe readers accept %%EOF anywhere in the last 1024 bytes of the file.

Patched at bradediger/prawn@f8f5257. Hopefully patches are accepted to make my life easier on busted unsupported viewers.

A clone of :row_color but for the columns of a table.

While trying to work with column_box I found out that plain text works well but tables don't flow properly.

This works fine:

    column_box [0,cursor], :width => bounds.width, :columns => 2 do
      text " This is a test " * 100
    end

This just causes the tables to continue in the first column over multiple pages, never flowing into the second column:

    column_box [0,cursor], :width => bounds.width, :columns => 2 do
      table data, :headers => headers,
        :header_color => 'C0C0C0',
        :font_size    => 7,
        :border_style => :grid,
        :border_width => 0.5,
        :width        => bounds.width-10
    end 

rename prawn-core/lib/prawn.rb to prawn-core/lib/prawn/core.rb and have the Prawn gem require it.

This fixes weird rubygems issues.

The min_version() method in my fork will be useful for extensions that use features that require PDF versions > 1.3

Most readers seem to ignore the version anyway, but we probably should at least try to be good PDF citizens.

Need to add documentation for cases where non-standard page size is required (ie not included in Document::PageGeometry::SIZES):

Use [width, height] instead of a symbol. (In PDF points 1/72 in)

As per Greg's request in the following mailing list post, here's a feature request: http://groups.google.com/group/prawn-ruby/browse_thread/thread/229740129eb9a506

The behaviour of bounding_box should be changed to fully enclose nested elements, such that Prawn elements that follow a bounding box containing multiple nested elements with varying heights will align with the bottom edge of the bounding_box, and not the bottom edge of the last rendered nested element. This should obviously only be the case if the following element uses 'cursor' to denote it's y-axis alignment.

I hope I've elucidated adequately.

Samå

https://prawn.lighthouseapp.com/attachments/61121/interlaced.png

Here's a sample image Prawn chokes on. This is your dice.png. I just reopened it in Photoshop and resaved it with the interlaced setting on. Prawn now chokes on parsing it, reporting the wrong error mid-parse.

In addition, see the discussion at:
https://prawn.lighthouseapp.com/projects/9398-prawn/tickets/82-adding-a-png-to-a-pdf-appears-corrupted

Improves consistency with typesetting vocabulary.

It would be very convenient to a project I'm doing if it were possibly to do, e.g.

pdf.cell blahblah, :vertical_align => :center

and have that work as easily as :horizontal_align => :center does. You can "fake it" with vertical padding fairly easily... right until you end up word wrapping.

I'm currently banging my head against this for my project but if I get a generalizable solution working I'll try getting you guys the code.

In PDF::Writer I could set the document properties "author", "title", "keywords", "subject", "creator", "producer".

(It's been my experience that these help with SEO.)

This is a request for several enhancements to Prawn#text_box so that it would support the features of #add_text_wrap in PDF::Writer.

1. When Prawn#text_box is called with the :truncate or :ellipses option the text which is not displayed in the text box is returned. (Implemented by geeksam and merged for 0.6)
   

2. When Prawn#text_box is called with the :angle => (int in range 0..259) option, the text is rotated as it would be using PDF::Writer #add_text_wrap

3. When Prawn#text_box is called with the :justification => :left | :center option, the text is justified as it would be using PDF::Writer #add_text_wrap

Testing this new issues hotness