Export larger pages from Draw using PDF 1.6

Posted on: Fri 08 May 2020

Estimated read time: 2 minutes

Draw/Impress now has support for exporting larger page sizes into PDF. The previous limit was 200 " (508 cm), and now practically there is no such limit.

First, thanks Vector who made this work by Collabora possible.

Motivation

You can use Draw with a document which has a single page, which more or less acts as a canvas with unlimited size to handle vector graphics. The current limit of such a canvas in size is 600 x 600 cm. (And that can be increased further if there is demand without too large problems.)

Exporting such a document to PDF is a different matter, though. The specification (up to, and including version 1.5) says that the unit to specify sizes is points, and the maximum allowed value is 14 400. This means that there is no markup to describe that your page is 600 cm wide. PDF 1.6 (and newer versions) introduce a UserUnit markup to allow unlimited page size, and now Draw (and other apps) can use this to describe the increased size.

Another use-case can be a large sheet in Calc, exporting it to a single PDF page, so you can pan around easily on a touch device. If you have enough rows, then getting rid of this limit is helpful to deal with the large page height.

Results so far

Here is how a large page looks like now in Draw and then in Adobe Reader:

https://lh3.googleusercontent.com/wfVdQklYbX9mpfNG0hk_A0-I495Nsu4r8Twokq_ZZR3Z0IondP33BDE_XjGFsNfucDJ9kVpDps_J_tcQATKhwJxAFeZn6hS1NN7i5BVlS9UBroUIGrJU1TiwqDB88k6HpMbnK9x9gQ=w640

Figure 1. Export of 6m-wide.odg to PDF

You can see how both Draw and Adobe Reader show that the page width is larger than 200 ".

How is this implemented?

If you would like to know a bit more about how this works, continue reading… :-)

The PDF export already converts from an internal unit (e.g. Draw uses 100th millimeters, Writer uses twips) to PDF’s unit
The trick is that now PDF’s unit is no longer points all the time, but we can dynamically switch to a larger unit as needed.

Here is how the PDF markup looks like for a 600 cm wide page:

1 0 obj
<</Type/Page/Parent 4 0 R/Resources 11 0 R/MediaBox[0 0 8503.93700787402 396]
/UserUnit 2/Group<</S/Transparency/CS/DeviceRGB/I true>>/Contents 2 0 R>>
endobj

Notice how we still avoid values larger than 14 400, but now the UserUnit says that 1 unit means 2 points.

Want to start using this?

You can get a snapshot / demo of Collabora Office and try it out yourself right now: try unstable snapshot. Collabora is a major contributor to LibreOffice and all of this work will be available in TDF’s next release too (7.0).

Padded numbering in Writer, part 2

Posted on: Wed 08 April 2020

Estimated read time: 2 minutes

I already posted about the start of padded numbering support in Writer, there the focus was to insert 0 characters to pad up the result to 2 characters. Let’s see how that got extended in the recent past…

First, thanks Nicolas Christener who made this work by Collabora possible.

Motivation

Padded numbering is a style where you insert 0 characters in front of an otherwise normal (Arabic) numbering, making sure that the result always has at least N characters. Up to now, you had to number your content manually to have this effect, while Word supports this feature.

OOXML supports padding up to 2, 3, 4 and 5 characters. The news is now now it’s possible to not only pad up to 2 characters, but also to any number between 2 and 5.

Results so far

Here is how the current rendering of padded numbering looks like, with a custom prefix and suffix:

https://lh3.googleusercontent.com/JHqkVdHkLLkFV1Mrh5jR1FFqq8PvU3lmjuOrl6SwBnM-ygsbugL-FccMHIod9Uyj2-hAyADRX7VwozUHgzBTZTdo72FB_nuHzEH-iQngSl5ND0o6h1sZDTs1uv8H5cLNv0cHDgRv2A=w640

Figure 1. numbering-padded4.docx, current rendering

You can see how 0 is inserted before 999, but not before 1000 as this is the pad-to-4 case.

How is this implemented?

If you would like to know a bit more about how this works, continue reading… :-)

Padding to a custom number is not something that works in general, because both ODF and OOXML has a separate number format for each padding. So Writer supports the 4 cases Word supports, but (for now) not more.
Padding to 3 or more is more complicated than pad to 2, because OOXML has different markups for them.

Here is how the pad-to-2 markup looks like:

      <w:numFmt w:val="decimalZero"/>

And here is how you define pad-to-3:

      <mc:AlternateContent>
        <mc:Choice Requires="w14">
          <w:numFmt w:val="custom" w:format="001, 002, 003, ..."/>
        </mc:Choice>
        <mc:Fallback>
          <w:numFmt w:val="decimal"/>
        </mc:Fallback>
      </mc:AlternateContent>

This required taking the w14 branch when we hit such a conditional, we used to read the fallback branch previously.
This required mapping the data of the <w:numFmt> XML element not to an enumeration value, but to a pair of objects: the numbering format’s value and format.

The rest was reasonably straightforward, since the actual padding implementation just had to be generalized.

Want to start using this?

My hack week at Collabora: (start of) padded numbering in Writer

Posted on: Tue 10 March 2020

Estimated read time: 3 minutes

As mentioned in a previous such report, a hack week is when we are allowed to hack on anything we want in LibreOffice for a few days at Collabora. I used this time to implement core support for padded numbering in Writer.

Motivation

OOXML supports padding up to 2, 3, 4 and 5 characters. Padding up to 2 characters is the older feature, supported in DOC and RTF as well, so I focused on that piece.

Results so far

Here is how the current, the baseline and the reference rendering of padded numbering looks like:

https://lh3.googleusercontent.com/n-XvNw0xPkBK_u9eqVraa7meGxYPX8dpKtfmHkN54y60x5HvUhQBouGEnfLX3XjzQFDjG7oWhCGVvFLnINneGZwDIjB8FW1hgwYSEpWrt3kniLGeKNFAfpa_Y9NGHNadj50ecy3FVw=s300-p-k

Figure 1. numbering-padded2.docx, current

https://lh3.googleusercontent.com/BgSZhU4WNkwTvOejqbfaKSiEQUudnDCUeVaQ62xPETwzhP0-FUPsBZMrUrgwhfi3fSou3YIQ_Yb0tuDzJmeIxHk2LhOpS9ENvxwLCr3-aCn4rIS0e9vYOh2__cHAvP82-MUJzQS-Zg=s300-p-k

Figure 2. numbering-padded2.docx, baseline

https://lh3.googleusercontent.com/TXI-klcS5xzPUX0SaV_iqhweMcUX0aN1rc7rwwAbKdulmPYZ6wcYqcQTO94aHGZl_p4FuVSt_drCI1blKRHLupNjC6j08GjdppbkJ8o72xNmjpV_X2_LcPUgZmOfGBeUJRhDKptxqg=s300-p-k

Figure 3. numbering-padded2.docx, reference

You can see how 0 is inserted before 1..9, but not before 10.

How is this implemented?

If you would like to know a bit more about how this works, continue reading… :-)

sw padded numbering: add doc model and UNO API introduced ARABIC_ZERO externally and SVX_NUM_ARABIC_ZERO internally
sw padded numbering: add layout implements the actual padding in a new lcl_formatArabicZero() function
sw padded numbering: add ODF filter decides how this handled for ODF import/export; the spec needs no extension in this case
sw padded numbering: add DOCX filter implements OOXML paragraph numbering
sw padded numbering: add DOC filter implements [MS-DOC] paragraph numbering
sw padded numbering: add RTF export doesn’t touch RTF import, which worked out of the box, due to code sharing with DOCX import

Then I found that footnote numbering needs explicit handling, so added support for padding in that case as well:

sw padded numbering: add DOCX footnote export reuses the same markup, just in a different context
sw padded numbering: add DOC footnote filter fixed custom number formatting for DOC footnote export in general

Finally I had a little bit of remaining time, so I extended support for the recently added Chicago numbering:

sw chicago numbering: add DOCX footnote export builds on top of the existing paragraph numbering support
sw chicago numbering: add DOC footnote export only touches footnotes, as this type is not supported for paragraphs
sw chicago numbering: add RTF footnote export fixed custom number formatting for RTF footnote export in general

Future work

Padding up to 3, 4 and 5 characters would be possible to do, but it’s DOCX-only, and uses a different markup, planned to be done later.

All this is available in master (towards LibreOffice 7.0), so you can grab a daily build and try it out right now. :-)

Semi-transparent text in Writer

Posted on: Wed 26 February 2020

Estimated read time: 1 minutes

The problem

Last year I posted about work to have semi-transparent rendering of not only shape fill and shape borders, but also shape text in Draw. Now the same semi-transparent text feature is available in Writer as well. This improves compatibility with Word, which supports the same feature in its DOCX format. You can access the new option in the Format → Character dialog.

Result

Here is how the new, old and reference rendering looks like:

https://lh3.googleusercontent.com/oifzSVcdmKhomkQQKy4CNnPmamHtoo9CwRsbg9_ekbeK9Ov_4dFrzrJrYmDLHZNm-IvsoYNy8YfSs5kp4yEA42jybRO-8P-YKI8fYRu5ColRbzFHQBotP4MvkjanZG7JX9vP8Mcnnw=w640

Figure 1. Semi-transparent text in Writer, new rendering

https://lh3.googleusercontent.com/RxtyxxIdrW14TtN30cf37e7RNHwnQpxlX-zhjhpWrggjStEERXvUnH7eDv8acM_8DcySNpnvpz5yWdwO12_cmQ6ZyCo4-fpIOdd3x9Q8SAXxAwuyMhfmjzhvS9VDBps8tQX-xB70RA=w640

Figure 2. Semi-transparent text in Writer, old rendering

https://lh3.googleusercontent.com/6rkCyXnzfiVN96D1h8PiKSIva1ydRFQXzUbXbbLC-9plXOUJnMRHZT2ba4eKt7W1wqfL3vscVgE1-QB5ztbp2uqSmguOALV8wyZDnCF4LPlyHSbVEm5atwZvf29w3PzRmjy11aQahA=w640

Figure 3. Semi-transparent text in Writer, reference rendering

Want to start using this?

Improved rotated text handling in Writer's table rows with automatic height

Posted on: Fri 24 January 2020

Estimated read time: 2 minutes

Writer now has better support for rotated text in tables containing rows with automatic height. This post also presents two related fixes.

First, thanks Otevřená města who made this work by Collabora possible.

Before diving into improved rotated text handling, first a continuous section break import problem (tdf#128605) was fixed: this was a case when we created a new page style, but only a new section was intended. Here is how the fix looks:

https://lh3.googleusercontent.com/ycsIqjpdm5t2n9FQQhKBbWobpsX2cQ0s-ugLn970XwuGafzL79HzNVly6dYUpLMhTQmZq8Aa36kaGIfWfk-MzYmuSr92zhRUxFk1aNqwDOnTy1cyyrRzCguuvp-Wd3LD8XXzo_Aurg=w640

Figure 1. Importing continuous section break, fixed

https://lh3.googleusercontent.com/jLUVDTY3Z6-IWVCmSfia7ue_aSpXSG7YJK2WqhRCfJkdTgA3cVFRh8TiYZwKRut7vvC0yFkdGBzl9Wpv2auZpg6d84nmDI1gVHBZPmWsf_YA-5qD-SF8Uf2lBG7NOu9RcidR1bQ0ng=w640

Figure 2. Importing continuous section break, original

https://lh3.googleusercontent.com/izfhxSHtYg_QmeJdIY_NnX5uiaRsmDHQoPlAsofQSaLJERNzVq_gkOMShR-fzckvAd6E5D-ZEMbr41Db2PEMeUZA8gL-oqJsvf_3PytuxJhL95wz4Fsh-2JakWefx4Tes2MoHlFWWw=w640

Figure 3. Importing continuous section break, reference

What you can see is that the continuous section break used to result in an unwanted page break, and this is gone now.

Regarding the actual rotated text handling (tdf#128611), there were two problems there:

There was a missing flip handling for line shapes
Poor layout of rotated text in table cells with automatic height

Here is how the fix looks like:

https://lh3.googleusercontent.com/GFCNQDcvXsN2w1ADVo10H16FUOYB8i6jBTrPad0r6qa8nt8PmDBDD1oiDfG9Zs9_3eVqdyB-O-u6oWIBAujcyso2utvnDpdE4Bs2PykeJkbh0vSYMZDSIKsjv7vI06j9HUCru2Z19w=w640

Figure 4. Rotated text in table cells with auto height, fixed (both flip of line and text layout)

https://lh3.googleusercontent.com/IqsWjKbTD03_wnvUMUe_QK1pqdTfZKizoNEHRQDQuvpw_O8mrEQ8AgNk_2qbP49QUpezknqHDxnhn0eYDpfZL833dXaClRPD1e4_wxuTj5mTRTeEgioBCX0djcWa5vNSRdPmnalphA=w640

Figure 5. Rotated text in table cells with auto height, fixed (only flip of line so far)

https://lh3.googleusercontent.com/2Tkkj7oz9Bm9ckPs5lTGwAGBRokEJ0MWxfqrbqu6afp6Xm1I77zft8riW-kTxKo4kJsO6bwMtaZb0XNToFssEc1kkZuzm0qk0H4eLU-xk8DOd-e5eVMtNR6cYD_1FRDWbCIYok1oJg=w640

Figure 6. Rotated text in table cells with auto height, original

https://lh3.googleusercontent.com/m3X4dGiUTsFBnVwQhiEpcAQYGAJ_8vbKN93jivrcsKcCvjaornrk0upCvsTmck1e39XISyuYNc47BF1tuyrskoWLuMhpegtbyPh1KjBD7_wD-oLueaaLbjYnrgReSRkS5krvK1y4LQ=w640

Figure 7. Rotated text in table cells with auto height, reference

What you can see is that the line clearing the table cell had a vertical flip, which was lost on import. Also, the rotated text in the "row heading" cells was broken into multiple columns. Both are fixed now.

All this is available in LibreOffice master (towards 6.4), so you can try it out right now, if interested.

Semi-transparent text in Draw/Impress

Posted on: Fri 20 December 2019

Estimated read time: 1 minutes

Draw/Impress now has support for semi-transparent shape text, next to the existing transparency support in the context of shape fill colors and shape border colors.

First, thanks Vector who made this work by Collabora possible.

Here is how it looks:

https://lh3.googleusercontent.com/FJ55-BJ_Mc75TiyrPYuvtOscHOqFp81sEI4SfJvybPzaXG5Y2n4mIoLRzkXydEEmtEf--l9s_h-L4CyYjuGpTyOngOTi8YAzIFp8bQBEn0k2DWsRzJHLdfttf74wTbdPgZUzgSnOYQ=w640

Figure 1. Semi-transparent text in Draw

What you can see is that next to the existing character color, now you can also set a transparency percentage to decide if the text should be entirely transparent, entirely opaque or something between the two.

The primary focus was Draw in this case, but this also helps PPTX support, as the importer/exporter now handles this for Impress documents as well.

All this is available in LibreOffice master (towards 6.5), so you can try it out right now, if interested.

Better math import from PPTX into Impress

Posted on: Thu 21 November 2019

Estimated read time: 1 minutes

Impress now has a much improved math handling in its importer from PPTX, eliminating annoying duplicated objects you had to delete after import, manually.

First, thanks TU Dresden who made this work by Collabora possible.

Here is how it looks:

https://lh3.googleusercontent.com/zI1fDb7tfl0a-6e1JljnBuCz67cGBSpx3s_tMViW_xWpTjy0FrsLGws-VAXWcdQhmSwXddr9wjADs5UvhNu_sNjETT7VfMJB8pdQ4PxPpHliCVbuMOx67Wwrd6brfE-eJtWIb2K36A=w640

Figure 1. Impress, new math import from PPTX

https://lh3.googleusercontent.com/rpC9KOUaadEoYKekczLTZakB98SBwqbKGqmwf_VLY6D-8OpBLh9Lpxi_RDl6CMQHJKqsz0NnxgTN8kn63CB9fGmP6AWUKIqK0huvjoRCxvoh6dtd0BMRXclsWlaKr3jbdzb1TOJ1LQ=w640

Figure 2. Reference rendering

https://lh3.googleusercontent.com/hUBhfLLFPZFDa50EMPEQxl2kih91oKIfQFBxNK7F9uWGexGbUKnV1FMQZgvmj4EctmCwlxlYK56hPSX0RZS_MpOAKZZagmfOALZ3LRrvBgoOGAx9dsM1N3dp9bGvPU7c0WVvROu8iQ=w640

Figure 3. Impress, old math import from PPTX

What you can see is that the in the old case we managed to read both the actual math object and its replacement image from PPTX, which caused visible duplication due to working transparency handling. The new way just imports the math objects, and ignores the replacement graphic.

This new way (handling of the OOXML a14 feature flag) is enabled by default, except for Calc documents, where more works is needed before it can be enabled.

All this is available in LibreOffice master (towards 6.4), so you can try it out right now, if interested.

Continuous endnotes in Writer

Posted on: Mon 21 October 2019

Estimated read time: 1 minutes

Writer now has a new "continuous endnotes" compatibility setting in its layout, allowing rendering endnotes in a way which is closer to what Word users expect.

First, thanks TU Dresden who made this work by Collabora possible.

Here is how it looks:

https://lh3.googleusercontent.com/g4kmuKduOm-Xc5sHH5oBx9y-U8S6hK5pgamrLxG-5fM9XCRJwMJFEPKxWXrG6IzEQLVpHdxCHI0c45EjDax02SpDbIzyQjZvUd-gsvpTWlkgZ0p3Y5-FDXARIPggF0KyIwG1AFwoiQ=w640

Figure 1. Writer, new ContinuousEndnodes layout flag enabled

https://lh3.googleusercontent.com/dlUGc11qUKVvvkKXh5mhj9BlEw4bLxgKxW_t3pNAAu2rnuWuIrFtWdlKk-8nadbSur8CaDe-CIUNJiSJlQSTvA22Q89TkY-uRsdJH3EKX6RcyUJsr-W5YqJrVqaM4KgXAIDPxPXPvg=w640

Figure 2. Reference rendering

https://lh3.googleusercontent.com/odkN2eqFxYthY7qhxHwStIlodTZJmcUa5wzh6p74PWPT6jKtsAmAvZyNzTjwff_CrDm5AtTcdAq_Rw2QyMNQN1Xr_ypIYL_cEZ3g1DRJof9em6acQ1fxrNQWsx_LxUmBhmu4TkTmPw=w640

Figure 3. Writer, old separate endnote page layout

What you can see is that endnotes unconditionally start after the end of the document content in Word, while endnotes are unconditionally on separate endnote pages in Writer. The new ContinuousEndnotes layout compatibility flag in Writer allows rendering endnotes the Word way.

This new flag is enabled by default for DOC files, disabled otherwise.

All this is available in LibreOffice master (towards 6.4), so you can try it out right now, if interested.

Allow-overlap shape property in Writer

Posted on: Fri 04 October 2019

Estimated read time: 1 minutes

Writer now has a new "allow overlap" shape property for anchored objects, which can ensure that objects with overlapping positioning properties don’t actually overlap.

First, thanks TU Dresden who made this work by Collabora possible.

Here is a video showing how it works:

https://lh3.googleusercontent.com/fdBYhwk3jykbX7drs6Q3ehQsQyAOBk_GWukw3KzcJBlaMsmgVg0C7KHM1uyODgqIro7HSjOnHzV6zfy9RgdMjY1KJW5K_EmfhyB4d-HyadZszIDAhVGbxUe4DMQbemSiZJA2k0h7vw=w640

Figure 1. Allow-overlap shape property in Writer

Under the hood, this is a full new Writer feature, since previously shapes were always allowed to overlap with each other:

macros get a new AllowOverlap shape property (true by default)
layout takes this into account if you set the property on the UI
both DOCX and ODT filters serialize and load this property (ODT proposal)

All this is available in LibreOffice master (towards 6.4), so you can try it out right now, if interested.

Bottom to top, left to right writing direction in Writer conference talk

Posted on: Thu 12 September 2019

Estimated read time: 1 minutes

https://lh3.googleusercontent.com/7A-ATV-MLBkNG6T7cX7kmq2bd7cgsmUs81plgl1ZXlbQIhonNNWZDFMHB5iMPr3NhtT5S_-TO3HTPLh8mqKfUduGDGEZg0LRwRMZO5POBbqn5133-TDo0FipATD9_hlfFe7J19g9Mw=w640

Yesterday I gave a Bottom to top, left to right writing direction in Writer talk at the LibreOffice Conference 2019. The room was well-crowded — perhaps because it was on the first day and in the largest room. ;-)

It contains some details which are not available in previous btLr blog posts, like what natural languages use this direction, how to replace real-world clocks without breaking compatibility and more!

I expect quite some other slides from other Collaborans and the wider community will be available on Planet, don’t miss them.

You can get a snapshot / demo of Collabora Office 6.2 and try the presented feature out yourself right now: try unstable snapshot. Collabora is a major contributor to LibreOffice and all of this work will be available in TDF’s next release, too (6.4).