CODEWITHSUNDEEP
docxpandocodt
Tilo Wiklund
Tilo Wiklund

Reputation: 821

Add styling rules in pandoc tables for odt/docx output (table borders)

I'm generating some odt/docx reports via markdown using knitr and pandoc and am now wondering how you'd go about formating tables. Primarily I'm interested in adding rules (at least top, bottom and one below the header, but being able to add arbitrary ones inside the table would be nice too).

Running the following example from the pandoc documentation through pandoc (without any special parameters) just yields a "plain" table without any kind of rules/colours/guides (in either -t odt or -t docx).

+---------------+---------------+--------------------+
| Fruit         | Price         | Advantages         |
+===============+===============+====================+
| Bananas       | $1.34         | - built-in wrapper |
|               |               | - bright color     |
+---------------+---------------+--------------------+
| Oranges       | $2.10         | - cures scurvy     |
|               |               | - tasty            |
+---------------+---------------+--------------------+

I've looked through the "styles" for the possibility of specifying table formating in a reference .docx/.odt but found nothing obvious beyond "table header" and "table contents" styles, both of which seem to concern only the formatting of text within the table.

Being rather unfamiliar with WYSIWYG-style document processors I'm lost as to how to continue.

Upvotes: 33

Views: 21804

Answers (9)

Mike Loomis
Mike Loomis

Reputation: 21

This is possible to do with lua filters since pandoc version 3.4 because of this PR: https://github.com/jgm/pandoc/pull/10009

You need to update the element attribute "custom-style" for the table, and pandoc will correctly set the w:tblStyle value in the resulting docx xml structure to have a non-default table.

If you are using pandoc markdown as your source, and you are building a docx file, you can't set that element attribute using default pandoc markdown syntax and that is why you'd need to approach this with something like a lua filter.

Here is an example lua filter that turns all tables in your docx output into the custom style of your choice, even ones that you create in a reference doc:

-- table-filter.lua
local tablestyle = "TableCustom"

function Table(elem)
    -- https://pandoc.org/lua-filters.html#type-table                                                                                                                                                   
    if elem.attr.attributes["custom-style"] == nil then
        io.stderr:write("Table style set to: " .. tablestyle .. "\n")
        elem.attr.attributes["custom-style"] = tablestyle
    end

    return elem
end

You would use this lua filter by running pandoc with the lua filter, and reference doc that defines a TableCustom table style that you made:

pandoc --reference-doc custom-reference-doc.docx -L table-filter.lua input.md -o output.docx

For me, this is superior to creating an additional stage of generation where you decompress and edit the docx XML structure directly.

Upvotes: 2

Loïc
Loïc

Reputation: 11942

Using a reference docx file and then python-docx does the job pretty easily :

https://python-docx.readthedocs.io/

First convert your document to docx :

Bash :

pandoc --standalone --data-dir=/path/to/reference/ --output=/tmp/xxx.docx input_file.md

Notes :

  • /path/to/reference/ points to the folder containing reference.docx
  • reference.docx is a file containing the styles you need for docx elements

Then give the tables of your document the style you want to use :

Python :

import docx
document = docx.Document('/tmp/xxx.docx')
for table in document.tables:
    table.style = document.styles['custom_style'] # custom_style must exist in your reference.docx file
document.save("target.docx") # thank you Anish

Upvotes: 4

junjun zhou
junjun zhou

Reputation: 1

add filter and custom your own Table style, see lua filter: https://github.com/ZhouJunjun/TyporaLuaFilter

Upvotes: -1

codeulike
codeulike

Reputation: 23064

I really liked gbjbaanb's answer - here's a powershell version:

Background: Set up a PanDoc --reference-doc template as described in the pandoc documentation for the --reference-doc parameter

Open up Word and create a new custom table style in the template doc. In our example that custom table style is called 'MyCustomTable'

Generate your word doc using the --reference-doc parameter - the custom table style will be included in the doc, you just have to insert its name in the right place. This bit of powershell will do that for you:

$outFile = "C:\Path\To\Your\Doc.docx"
$workFolder = "C:\Some\Temp\Folder\Somewhere\"
# then this replaces table style in $outFile:
$zipFile = $outFile.Replace(".docx",".zip")
Rename-Item $outFile $zipFile 
Expand-Archive $zipFile -DestinationPath $workFolder -Force
$wordXml = Get-Content "${workFolder}Word\Document.xml"
$updatedXml = $wordXml.Replace('<w:tblStyle w:val="Table" />','<w:tblStyle w:val="MyCustomTable" />')
Set-Content -Path "${workFolder}Word\Document.xml" -Value $updatedXml
Compress-Archive -Path "${workFolder}*" -DestinationPath $zipFile -Force
Rename-Item $zipFile $outFile

... where $outFile is the docx, and $workFolder is a temp folder somewhere.

In some earlier versions of PanDoc, instead of seaching for <w:tblStyle w:val="Table" /> you'll need to search for <w:tblStyle w:val="TableNormal" />

Upvotes: 2

henry
henry

Reputation: 49

Add a table style named "TableNormal" in reference.docx.

Upvotes: 4

ZHUOQI LI
ZHUOQI LI

Reputation: 51

Just add a table style what every you want called "Table" in the reference-doc file。And update pandoc to latest.

Upvotes: 2

gbjbaanb
gbjbaanb

Reputation: 52689

edi9999 has the best answer but here's what I do:

When creating the docx, use a reference docx to get styles. That reference will contain a heap of other styles that just aren't used by Pandoc to create, but they are still in there. Typically you'll get the default sets, but you can add a new table style too.

Then, you only need to update the word\document.xml file to reference the new table style, and you can do that programmatically (by unzipping, running sed, and updating the docx archive), eg:

7z.exe x mydoc.docx word\document.xml
sed "s/<w:tblStyle w:val=\"TableNormal\"/<w:tblStyle w:val=\"NewTableStyle\"/g" word\document.xml > word\document2.xml
copy word\document2.xml word\document.xml /y
7z.exe u mydoc.docx word\document.xml

Upvotes: 8

lcn
lcn

Reputation: 2289

Same suggestion as edi9999: hack the xml content of converted docx. And the following is my R code for doing that.

The tblPr variable contains the definition of style to be added to the tables in docx. You could modify the string to satisfy your own need.

require(XML)

docx.file <- "report.docx"
tblPr <- '<w:tblPr xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main"><w:tblStyle w:val="a8"/><w:tblW w:w="0" w:type="auto"/><w:tblBorders><w:top w:val="single" w:sz="4" w:space="0" w:color="000000" w:themeColor="text1"/><w:left w:val="single" w:sz="4" w:space="0" w:color="000000" w:themeColor="text1"/><w:bottom w:val="single" w:sz="4" w:space="0" w:color="000000" w:themeColor="text1"/><w:right w:val="single" w:sz="4" w:space="0" w:color="000000" w:themeColor="text1"/><w:insideH w:val="single" w:sz="4" w:space="0" w:color="000000" w:themeColor="text1"/><w:insideV w:val="single" w:sz="4" w:space="0" w:color="000000" w:themeColor="text1"/></w:tblBorders><w:jc w:val="center"/></w:tblPr>'

## unzip the docx converted by Pandoc
system(paste("unzip", docx.file, "-d temp_dir"))
document.xml <- "temp_dir/word/document.xml"

doc <- xmlParse(document.xml)
tbl <- getNodeSet(xmlRoot(doc), "//w:tbl")
tblPr.node <- lapply(1:length(tbl), function (i)
                   xmlRoot(xmlParse(tblPr)))
added.Pr <- names(xmlChildren(tblPr.node[[1]]))
for (i in 1:length(tbl)) {
    tbl.node <- tbl[[i]]
    if ('tblPr' %in% names(xmlChildren(tbl.node))) {
        children.Pr <- xmlChildren(xmlChildren(tbl.node)$tblPr)
        for (j in length(added.Pr):1) {
            if (added.Pr[j] %in% names(children.Pr)) {
                replaceNodes(children.Pr[[added.Pr[j]]],
                             xmlChildren(tblPr.node[[i]])[[added.Pr[j]]])
            } else {
                ## first.child <- children.Pr[[1]]
                addSibling(children.Pr[['tblStyle']],
                           xmlChildren(tblPr.node[[i]])[[added.Pr[j]]],
                           after=TRUE)
            }
        }
    } else {
        addSibling(xmlChildren(tbl.node)[[1]], tblPr.node[[i]], after=FALSE)
    }
}

## save hacked xml back to docx
saveXML(doc, document.xml, indent = F)
setwd("temp_dir")
system(paste("zip -r ../", docx.file, " *", sep=""))
setwd("..")
system("rm -fr temp_dir")

Upvotes: 11

edi9999
edi9999

Reputation: 20574

Here's how I searched how to do this:

The way to add a table in Docx is to use the <w:tbl> tag. So I searched for this in the github repository, and found it in this file (called Writers/Docx.hs, so it's not a big surprise)

blockToOpenXML opts (Table caption aligns widths headers rows) = do
  let captionStr = stringify caption
  caption' <- if null caption
                 then return []
                 else withParaProp (pStyle "TableCaption")
                      $ blockToOpenXML opts (Para caption)
  let alignmentFor al = mknode "w:jc" [("w:val",alignmentToString al)] ()
  let cellToOpenXML (al, cell) = withParaProp (alignmentFor al)
                                    $ blocksToOpenXML opts cell
  headers' <- mapM cellToOpenXML $ zip aligns headers
  rows' <- mapM (\cells -> mapM cellToOpenXML $ zip aligns cells)
           $ rows
  let borderProps = mknode "w:tcPr" []
                    [ mknode "w:tcBorders" []
                      $ mknode "w:bottom" [("w:val","single")] ()
                    , mknode "w:vAlign" [("w:val","bottom")] () ]
  let mkcell border contents = mknode "w:tc" []
                            $ [ borderProps | border ] ++
                            if null contents
                               then [mknode "w:p" [] ()]
                               else contents
  let mkrow border cells = mknode "w:tr" [] $ map (mkcell border) cells
  let textwidth = 7920  -- 5.5 in in twips, 1/20 pt
  let mkgridcol w = mknode "w:gridCol"
                       [("w:w", show $ (floor (textwidth * w) :: Integer))] ()
  return $
    [ mknode "w:tbl" []
      ( mknode "w:tblPr" []
        ( [ mknode "w:tblStyle" [("w:val","TableNormal")] () ] ++
          [ mknode "w:tblCaption" [("w:val", captionStr)] ()
          | not (null caption) ] )
      : mknode "w:tblGrid" []
        (if all (==0) widths
            then []
            else map mkgridcol widths)
      : [ mkrow True headers' | not (all null headers) ] ++
      map (mkrow False) rows'
      )
    ] ++ caption'

I'm not familiar at all with Haskell, but I can see that the border-style is hardcoded, since there is no variable in it:

let borderProps = mknode "w:tcPr" []
                    [ mknode "w:tcBorders" []
                      $ mknode "w:bottom" [("w:val","single")] ()
                    , mknode "w:vAlign" [("w:val","bottom")] () ]

What does that mean ?

That means that you can't change the style of the docx tables with the current version of PanDoc. Howewer, there's a way to get your own style.

How to get your own style ?

  1. Create a Docx Document with the style you want on your table (by creating that table)
  2. Change the extension of that file and unzip it
  3. Open word/document.xml and search for the <w:tbl>
  4. Try to find out how your style translates in XML and change the borderProps according to what you see.

Here's a test with a border-style I created: Custom border style

And here is the corresponding XML: 

<w:tblBorders>
  <w:top w:val="dotted" w:sz="18" w:space="0" w:color="C0504D" w:themeColor="accent2"/>
  <w:left w:val="dotted" w:sz="18" w:space="0" w:color="C0504D" w:themeColor="accent2"/>
  <w:bottom w:val="dotted" w:sz="18" w:space="0" w:color="C0504D" w:themeColor="accent2"/>
  <w:right w:val="dotted" w:sz="18" w:space="0" w:color="C0504D" w:themeColor="accent2"/>
  <w:insideH w:val="dotted" w:sz="18" w:space="0" w:color="C0504D" w:themeColor="accent2"/>
  <w:insideV w:val="dotted" w:sz="18" w:space="0" w:color="C0504D" w:themeColor="accent2"/>
</w:tblBorders>

What about odt ?

I didn't have a look at it yet, ask if you don't find by yourself using a similar method.

Hope this helps and don't hesitate to ask something more

Upvotes: 25

Related Questions