PDF created from wiki - Pending tasks

narayan · 2010-09-25 18:37

The pdf conversion has some problems. Can they be resolved?
The following list also includes some pending TODO tasks.

1. Hotspots still point at internet, not within the book.
2. Each new web-page should start on a new pdf page. There is no page break.
3. The pdf sections are named like "ReNamer:Rules:CleanUp". Should be just "Cleanup".
4. The navigation strip needs to be more attractive and consistent across all pages.
5. Many pages are yet to be added to the book.
6. The inner table in Delete rule is not converted well in the pdf version. There may be more.
7. The screenshots are a little bloated up in size.
8. The external links are in [n] format, where n is a running number. They can be mistaken for footnotes/end notes.
9. The pdf file does not have bookmarks. They would be nice to have!
10. How to add a fantastic-looking cover page? (I can design it)
11. Adding other pages for the book version (preface, introduction, about the author, etc)

Also, many Pascal Script functions are not defined adequately yet.

den4b · 2010-09-26 14:45

1. Hotspots still point at internet, not within the book.
If the link points to the article which is included in the book, the link will be changed to point to the page within the book. I tested it several times and it works as expected.

4. The navigation strip needs to be more attractive and consistent across all pages.
Do you mean the navigation bar which shows "up", "back" and "next" links? I was thinking that we should exclude it from printing all together.

5. Many pages are yet to be added to the book.
First, we need to add all pages about ReNamer into Category:ReNamer. Then the whole category can be added to the book. Adding a category to a page is easy, just insert [[Category:ReNamer]] at the bottom of the page. @ wiki/Talk:ReNamer

10. How to add a fantastic-looking cover page?
This can be done simply by creating a dedicated article (e.g. ReNamer:Book cover) and including it into the Book.

11. Adding other pages for the book version (preface, introduction, about the author, etc)
Same as above.

Pascal Script functions are not defined adequately yet.
I fixed that this morning.

2. Each new web-page should start on a new pdf page. There is no page break.
3. The pdf sections are named like "ReNamer:Rules:CleanUp". Should be just "Cleanup".
6. The inner table in Delete rule is not converted well in the pdf version. There may be more.
7. The screenshots are a little bloated up in size.
8. The external links are in [n] format, where n is a running number. They can be mistaken for footnotes/end notes.
9. The pdf file does not have bookmarks. They would be nice to have!

There is nothing I can do about them. This is just the way PDF printing service works.

narayan · 2010-09-27 03:09

well, 4 and 9 are interlinked: If the pdf does not have bookmarks, then at least the navigation strip will help the reader go around in the book.

10-> will do.

11-> You can write those. After all, these are the author's prerogative- Not to be ghost-written

I have raised a ticket at pediapress.com for the rest of the issues. Hopefully some of these features will get implemented in near future.

BTW I added all the remaining sections. The pdf is complete now.
Check out the download link!

The hotspots in pdf do not work for me even in the latest book (which has all the pages included).
But you confirmed that they do work. So what's wrong? :?

Last edited by narayan (2010-09-27 03:27)

narayan · 2010-09-27 05:47

I discovered a major bug with the pdf created with prediapress: All the internal hyperlinks are not even visible to the reader! (The text looks perfectly normal). If he does not know where a hyperlink is, where is he supposed to click??

To sum up, the pdf file has major problems: No hyperlinks, no bookmarks, no hotspots.
But then the extension is designed to create a PRINTED book, so these are not really bugs for that purpose.

But it certainly makes the pdf file useless for most users who'd read it on screen.

There are other issues as well, such as tables are broken up across pages at odd places.

Although I have raised these issues to printpedia as tickets, the solutions will take time.
(The designers may refuse these requests, because none of them are meant for a printed book.)

So, we can have our own workaround: export the wiki as odt, edit that odt file and then create a pdf out of it.

Can you install the extension to export the wiki as odt?
Then I edit the odt file and create the pdf (just like the User Manual for xplorer2).

Although this would be a manual process, it is NOT tedious, and it overcomes all inherent problems in the printing service.

Note that the "export to odt" extension may be related to pediapress or it may be an independent extension for mediawiki.

I found this. There may be others.

Last edited by narayan (2010-09-27 06:15)

narayan · 2010-09-27 15:22

I must say the PediaPress guys are superfast! All my tickets are already answered!

OK here is the response:

1. To start each new web-page on a new page:
page_break_after_article = False (customconfig.py)

2. The pdf sections are named like "ReNamer:Rules:CleanUp". The naming convention should be just "Cleanup".
Manually set a display title on the saved collection page by using named mediawiki links.
So change [[ReNamer:Rules:CleanUp]] to [[ReNamer:Rules:CleanUp|CleanUp]]

3. Some inner tables get converted to plain text. (e.g. the inner table here).
The problem is that the border is not rendered for the nested table.
That can be achieved by adding something like "border=1" to the inner table

4. The images are bloated up in size in pdf (they look ok in wiki version).
Sizing images is not easy - maybe something went wrong. Could you please provide a minimal example.
(I have to send the sample now. Hopefully we will receive a solution soon.)

5. The external links are shown with [n] symbol. They can be mistaken for footnotes/end notes.
That is done on purpose: If the PDF is printed out, the information about the link is lost.
Therefore the links are added to the references.

6. The pdf file does not have bookmarks.
This is a known problem which can hopefully be solved soon.

7. Show hyperlinks with underlines.
if you are using a custom render server you can customize the render server to render underlined links.
This is done by adding the following two lines to your customconfig.py:
import reportlab
reportlab.rl_config.platypus_link_underline = 1
Refer to the readme.txt of mwlib.rl and our public mailing list for further info.

************
So some of the problems are resolved. But I still think we should have that odt extension installed to have a much better manual.
(I am still thinking of the bundled offline manual, which would be updated very infrequently.)

Last edited by narayan (2010-09-27 15:34)

narayan · 2010-09-28 09:05

One more issue resolved by PediaPress guys:

8. The pdf is created with file name "Collection", not "ReNamer User Manual".

You can configure the filename in your LocalSettings.php by tweaking the array $wgCollectionContentTypeToFilename to something like this $wgCollectionContentTypeToFilename = array( 'application/pdf' => 'ReNamer User Manual.pdf').

I've just documented this in the README.txt of the Collection extension in r73874:
http://www.mediawiki.org/wiki/Special:C … Wiki/73874

Edit: One more solution arrived:

9. In pdf, the flowchart images on the PascalScript page show gray border (the wiki is OK)

Although the original image has no borders, the problem is that the PDF renderer inserts one. If you don't want the border you need to suppress it by making it white. --> img_border_color = (1.0, 1.0, 1.0) in your customconfig.py

****
Denis, someone will have to configure the server as recommended by the PediaPress guys (see the previous posts also).

Last edited by narayan (2010-09-28 12:50)

Andrew · 2010-09-28 21:36

Once the PDF is created and I see the link, how do I save it? I couldn't directly right-click the link and choose to save as PDF. Instead I had to wait for the entire thing to get loaded in the browser, after which I could save the file using the toolbar buttons (I use the Foxit plugin).

Last edited by Andrew (2010-09-28 21:36)

narayan · 2010-09-29 05:19

Two more responses received from PediaPress:

10. The text in RegEx chapter loses all color; making it meaningless.

Unfortunately the rendering software does not evaluate css files. This is why the orange/blue/red colors get lost in your example. But there is a workaround: All css classes you want evaluated need to be defined in your customconfig.py. In your example that would be: css_map = {'hl-orange': 'text-color:rgb(200, 100, 0);', } essentially the key of the css_map dictionary is the css class name and the value is a list of css attributes (like in a regular css file. just make sure not to include pagebreaks). Some more info can be found in the file customnodetransformer.py

11. The imagemap hotspots do not work in pdf

Unfortunately hotspots do not work in pdf.

****

To sum up, even after you modify all those Python files, we will still not have hotspots and bookmarks.

****
Andrew, I have installed FreeDownloadManager, and I can drag-n-drop the link into FDM's dropbox. It works.

However, if I click on that link, it loads the pdf into Firefox.
Curiously, my PDFdownload addon is not able to detect the incoming pdf file.
(It is supposed to ask me whether I want to open the pdf in Firefox or download it directly.)

I think no one would want to just load it in browser as default, and then decide whether he wants to save it or not.
So the link should do what it says: Directly download the pdf.

Denis, can you help here?

SafetyCar · 2010-09-29 20:14

narayan wrote:

Andrew, I have installed FreeDownloadManager, and I can drag-n-drop the link into FDM's dropbox. It works.
However, if I click on that link, it loads the pdf into Firefox.
Curiously, my PDFdownload addon is not able to detect the incoming pdf file.
(It is supposed to ask me whether I want to open the pdf in Firefox or download it directly.)
I think no one would want to just load it in browser as default, and then decide whether he wants to save it or not.
So the link should do what it says: Directly download the pdf.

I think this is a browser behavior, but I think with Right-Click in the link you can get the option to "Save as".

Maybe this hint could be added in that page?

Last edited by SafetyCar (2010-09-29 20:15)

Andrew · 2010-09-30 14:48

Hmm... Apparently on right-clicking the generated file link and choosing Save As, Firefox chooses to save the PDF as index.php, whereas IE correctly names it collection.pdf.

1) The file is a PDF and the exact same as the one downloaded with IE, but something needs to be done about FF (and possibly other browsers that might behave similarly). After all, can't very well expect the common user to rename a PHP file to a PDF to use it, right? That's just not intuitive and not the correct way to go about it.

2) Even with IE, the generated PDF really needs a better name. collection.pdf just doesn't indicate what the PDF is all about. Something like ReNamer Manual.pdf etc. is what is needed.

den4b Forum

#1 2010-09-25 18:37

PDF created from wiki - Pending tasks

#2 2010-09-26 14:45

Re: PDF created from wiki - Pending tasks

#3 2010-09-27 03:09

Re: PDF created from wiki - Pending tasks

#4 2010-09-27 05:47

Re: PDF created from wiki - Pending tasks

#5 2010-09-27 15:22

Re: PDF created from wiki - Pending tasks

#6 2010-09-28 09:05

Re: PDF created from wiki - Pending tasks

#7 2010-09-28 21:36

Re: PDF created from wiki - Pending tasks

#8 2010-09-29 05:19

Re: PDF created from wiki - Pending tasks

#9 2010-09-29 20:14

Re: PDF created from wiki - Pending tasks

#10 2010-09-30 14:48

Re: PDF created from wiki - Pending tasks

Board footer