When passing a character variable to cattable, the label on the variable is lost when it is converted to a factor. The results is unlabeled variables in the output and is not the desired result. Labels should be preserved.

Adding a line as simple as as.data.frame(data) in these cases should be enough.

This isn't an exported function, so back compatibility shouldn't be too big an issue.

Still to do.

Completed

• change default for 'lazyReportFormat' to 'markdown'
• Add option 'lazyWeave_cat = TRUE' This will cause output of table and figure generating functions to wrap the output in 'cat'
• 'lazy.matrix' should print column names in bold text
• Add pvalString as an exported function
• Add 'cat=getOption(lazyWeave_cat)' argument to lazy.table, lazy.matrix, write.ctable, write.univ, lazy.figure
• Configure lazy.table, lazy.matrix, write.ctable, write.univ, lazy.figure to act on the cat argument.
• Remove pvalue.QHS
• Replace functions calls to pvalue.QHS with pvalString

The only purpose for the survival dependency is in the examples. With the addition of the Delivery data, this isn't necessary anymore. Adjustments to the examples would allow for removal of the survival dependency.

Unfortunately, the labeling functions in Hmisc are pretty essential to cattable and conttable. This dependency can't be avoided because not all of the labelling functions are exported in the Hmisc namespace. I'm contemplating copying those functions into lazyWeave so that Hmisc won't be a dependency any more. In my ideal world, I'd like to see the labeling functions have their own package, but this might require too many changes to Hmisc.

latexTranslate is the other function I rely on heavily, but this is exported in Hmisc's namespace, so we can at least access that one.

lazy.footnote, for HTML files, requires the user to provide two tag names to generate the links between the footnote and the in-document reference. This could probably be simplified by having the user provide a tag name and appending it with [tag]_doc and [tag]_end.

The documentation claims it does midranks, but it doesn't.

write.ctable doesnt' play nicely with hooks for numbering and referencing tables

title: "Auto-numbering and cross-referencing of figures and tables in rmarkdown"

output: html_document

# This is a knitr hook to enable figure and table captions
# number figures
# from http://stackoverflow.com/a/18672268/1036500
library("knitr")
tn = local({
  i = 0
  function(x) {
    i <<- i + 1
    paste("<table><caption>", 'Figure ', i, ': ', x, "</caption><colgroup><col width='100'></col></colgroup><thead><tr class='header'></tr></thead><tbody></tbody></table><p>", sep = '')
  }
})
knit_hooks$set(fig.cap = function(before, options, envir) {
  if(!before)
    tn(options$fig.cap)
})
default_output_hook = knit_hooks$get("output")
knit_hooks$set(output = function(x, options) {
  if (is.null(options$fig.cap) == F)  
    x
  else
    default_output_hook(x,options)
})

# number tables
# from http://stackoverflow.com/a/18672268/1036500
to = local({
  i = 0
  function(x) {
    i <<- i + 1
    paste("<table><caption>", 'Table ', i, ': ', x, "</caption><colgroup><col width='100'></col></colgroup><thead><tr class='header'></tr></thead><tbody></tbody></table><p>", sep = '')
  }
})
knit_hooks$set(tab.cap = function(before, options, envir) {
  if(!before)
    to(options$tab.cap)
})
default_output_hook = knit_hooks$get("output")
knit_hooks$set(output = function(x, options) {
  if (is.null(options$tab.cap) == F)  
    x
  else
    default_output_hook(x,options)
})

library(knitr)
library(kfigr) # devtools::install_github("github mkoohafkan/kfigr")
opts_chunk$set(message=FALSE, warning=FALSE)
library(lazyWeave)
library(stringr)
options(lazyReportFormat = "markdown")

summaries

Using the knitr option "echo=F" seems to cause the problem.

data(mtcars)

mtcars$group = sample(c("Bad", "Good"), size=nrow(mtcars), replace=T)

tab1 = catconttable(data = mtcars,
                    vars = c("mpg", "cyl", "disp", "hp", "drat", "wt", "qsec", "vs", "am", "gear", "carb"),
                    normal = c("mpg", "cyl", "disp", "hp", "drat", "wt", "qsec", "vs", "am", "gear", "carb"),
                    byVar = "group")
tab = write.ctable(tab1)
tab = str_split(tab, "\n")[[1]]
tab = tab[c(6,4,5,7:length(tab))]
tab = paste(tab, collapse="\n")
cat(tab)

See r figr("ATable", T,F) for summaries of the variables in the cars dataset.

data(mtcars)

mtcars$group = sample(c("Bad", "Good"), size=nrow(mtcars), replace=T)

tab1 = catconttable(data = mtcars,
                    vars = c("mpg", "cyl", "disp", "hp", "drat", "wt", "qsec", "vs", "am", "gear", "carb"),
                    normal = c("mpg", "cyl", "disp", "hp", "drat", "wt", "qsec", "vs", "am", "gear", "carb"),
                    byVar = "group")
tab = write.ctable(tab1)
tab = str_split(tab, "\n")[[1]]
tab = tab[c(6,4,5,7:length(tab))]
tab = paste(tab, collapse="\n")
cat(tab)

See r figr("ATable2", T,F) for summaries of the variables in the cars dataset.

Currently, using the text.size argument produces a warning that the argument was scheduled for deletion in 2013. It's almost 2015. Maybe I should take this out finally.

When building the zip folder for an HTML document, lazy.build copies everything into a top level directory. This is because I hadn't yet figured out how to avoid copying the whole directory tree into the file. I can avoid some of this overhead work by using the "-j" flag in the flags argument of zip

Would allow tables to break over multiple pages more naturally.

The big question is if this will give me the same control over tables. Can I have columns spanning multiple cells? Can I control the HTML fonts? Can I add shading?