attempting to solve intersect([pacs,keywords,affiliation,fax,...], <this journal's tex idiosyncracy>) with yaml, R, knitr/brew, mustache, and pandoc

_metadata.yaml

title: On physics and chemistry
authors: 
  - name: Lise Meitner
    affiliation: [Kaiser Wilhelm Institute,
                University of Berlin,
                Manne Siegbahn Institute]
    email: [email protected]
  - name: Pierre Curie
    affiliation: École Normale Supérieure
    email: [email protected]
  - name: Marie Curie
    affiliation: [University of Paris,
                 Institut du Radium,
                 École Normale Supérieure]
    email: [email protected]
    fax: 123
pacs: [123, 456, 789]
keywords: [physics, science, everything]
abbreviations: [UV,IR]

brew_template.R

brew_template <- function(in, out){
 ## pre-process input file to insert bits of glue extracted from the metadata and formatted according to journal syntax
 ## output a minimal pandoc tex template file with mostly hard-coded preamble
}

format_journal1.tex

\author{Andrew N. Other}
\altaffiliation{A shared footnote}
\author{Fred T. Secondauthor}
\altaffiliation{Current address: Some other place, Othert\"own,
Germany}
\author{I. Ken Groupleader}
\altaffiliation{A shared footnote}
\email{[email protected]}
\phone{+123 (0)123 4445556}
\fax{+123 (0)123 4445557}
\affiliation[Unknown University]
{Department of Chemistry, Unknown University, Unknown Town}
\alsoaffiliation[Second University]
{Department of Chemistry, Second University, Nearby Town}
\author{Susanne K. Laborator}
\email{[email protected]}
\affiliation[BigPharma]
{Lead Discovery, BigPharma, Big Town, USA}
\author{Kay T. Finally}
\affiliation[Unknown University]
{Department of Chemistry, Unknown University, Unknown Town}
\alsoaffiliation[Second University]
{Department of Chemistry, Second University, Nearby Town}

format_journal3.tex

\author[1,2,3]{Author One}
\author[2,*]{Author Two}
\author[1]{Author Three}

\affil[1]{Publications Department, The Optical Society (OSA), 2010 Massachusetts Avenue NW, Washington D.C., 20036}
\affil[2]{School of Science, University of Technology, 2000 J St. NW, Washington DC, 20036}
\affil[3]{School of Optics, University of Technology, 2000 J St. NW, Washington DC, 20036}

\affil[*]{Corresponding author: [email protected]}

manuscript.Rmd

---
date: "5/7/2017"
output: custom_process_journal:
 journal: _journal1.tex
 metadata: _metadata.yaml
 pre_knit: brew_template
---

## First section

Lorem Ipsum

test.R

library(yaml)

meta <- yaml.load_file("_metadata.yaml")

unique(unlist(lapply(meta$authors, "[[", "affiliation")))
# [1] "Kaiser Wilhelm Institute" "University of Berlin"     "Manne Siegbahn Institute"
# [4] "École Normale Supérieure" "University of Paris"      "Institut du Radium" 

library(glue)
tpl_author <- '{name}\\n{paste(affiliation, collapse=", ")}'

lapply(meta[["authors"]], glue_data, tpl_author)
# [[1]]
# Lise Meitner\nKaiser Wilhelm Institute, University of Berlin, Manne Siegbahn Institute
# 
# [[2]]
# Pierre Curie\nÉcole Normale Supérieure
# 
# [[3]]
# Marie Curie\nUniversity of Paris, Institut du Radium, École Normale Supérieure


## journal-dependent partial functions
fun_title <- function(meta) glue_data(meta, "\\title{{ {title} }}") # pretty standard
# this varies wildly: some require $^\dagger$ and to manually group unique addresses, etc.
# achemso example is pretty simple as it requires the full list and takes care of the groups at latex level
fun_authors <- function(meta) lapply(meta[["authors"]], 
                                     glue_data, 
                                     '\\author{{ {name} }}
                                     \\address{{ {paste(affiliation, collapse=", ")} }}')
# combine pieces to be passed to the template skeleton to create the template for pandoc
preamble <- list(title = fun_title, authors = fun_authors)

lapply(preamble, do.call, list(meta=meta))
# $title
# \title{ On physics and chemistry }
# 
# $authors
# $authors[[1]]
# \author{ Lise Meitner }
# \address{ Kaiser Wilhelm Institute, University of Berlin, Manne Siegbahn Institute }
# 
# $authors[[2]]
# \author{ Pierre Curie }
# \address{ École Normale Supérieure }
# 
# $authors[[3]]
# \author{ Marie Curie }
# \address{ University of Paris, Institut du Radium, École Normale Supérieure }



fun_title <- function(meta) glue_data(meta, "\\title{{ {title} }}") 
fun_authors <- function(meta) lapply(meta[["authors"]], 
                                     glue_data, 
                                     '\\author{{ {name} }}
                                     { paste(sprintf("\\\\affiliation{%s}", affiliation), collapse="\\n") }')
preamble_journal2 <- list(title = fun_title, authors = fun_authors)

lapply(preamble_journal2, do.call, list(meta=meta))
# $title
# \title{ On physics and chemistry }
# 
# $authors
# $authors[[1]]
# \author{ Lise Meitner }
# \affiliation{Kaiser Wilhelm Institute}
# \affiliation{University of Berlin}
# \affiliation{Manne Siegbahn Institute}
# 
# $authors[[2]]
# \author{ Pierre Curie }
# \affiliation{École Normale Supérieure}
# 
# $authors[[3]]
# \author{ Marie Curie }
# \affiliation{University of Paris}
# \affiliation{Institut du Radium}
# \affiliation{École Normale Supérieure}

tpl_journal1.tex

\documentclass[journal=jacsat,manuscript=article]{achemso}

{title}
{authors}
{keywords}

\begin{document}

{body}

\end{document}

tpl_journal2.tex

\documentclass[%
 reprint,
 amsmath,amssymb,
 aps,
]{revtex4-1}


\begin{document}

{title}
{authors}
{keywords}

\begin{abstract}
{abstract}
\end{abstract}

{pacs}

\maketitle

{body}

\end{document}

The goal is to provide a unified interface to multiple journal templates (which typically use different syntax for affiliation, keywords, etc.). My idea is to bypass pandoc templates, the syntax of which I find extremely cumbersome (\author{$for(author)$$author$$sep$ \and $endfor$}), and generate on-the-fly simpler, partially hard-coded templates, based on the available metadata and journal-specific formatting rules (lapply(meta[["authors"]], glue_data, '\\author{{ {name} }})

The envisaged structure is as follows:

• _metadata.yaml stores all metadata about the article in a generic human-readable form
• a minimal Rmd header specifies a custom pandoc template with "mustache" {{}} syntax where those values are to be inserted via a pre_knit hook (using brew and/or glue/whisker)
• knit + rmarkdown + pandoc the resulting document as usual using this auto-generated template for pandoc

you can use glue::collapse(affiliation, ", ") in place of the paste calls, there are also some purrr functions that might make the list processing little more concise...

for reference the standard strategy with pandoc templates looks like this,

$if(author)$
\author{$for(author)$$author$$sep$ \and $endfor$}
$endif$
$if(institute)$
\providecommand{\institute}[1]{}
\institute{$for(institute)$$institute$$sep$ \and $endfor$}
$endif$
\date{$date$}

\begin{document}
$if(title)$
\maketitle
$endif$
$if(abstract)$
\begin{abstract}
$abstract$
\end{abstract}
$endif$