Restore Pandoc Citation Keys

restore-citekeys.js

/*
I use this to restore citeproc keys to my Pandoc MS's after de-compiling Word
documents that have been edited using track changes.
*/
const fs = require('fs')
const child_process = require('child_process')

// Tiny ARGV Parser:
const TinyArgv = function (argv) {
  this.__argv = argv
}
/**
 * Figure out the location of an argument in ARGV based on a key and an alias.
 * Returns -1 if neither is found.
 */
TinyArgv.prototype.__argIndex = function (key, alias) {
  const index = {
    key: this.__argv.indexOf(`--${key}`),
    alias: alias ? this.__argv.indexOf(`-${alias}`) : -1
  }
  return index.key >= 0 ? index.key : index.alias >= 0 ? index.alias : -1
}
/**
 * Extracts the value of a switch, which may be more than one word.
 */
TinyArgv.prototype.__extractValue = function (index) {
  const nextSwitch = this.__argv.slice(index + 1).reduce((acc, cur, idx) => acc >= 0 ? acc : cur.indexOf('-') === 0 ? idx : acc, -1)
  return this.__argv.slice(index + 1, nextSwitch < 0 ? undefined : index + 1 + nextSwitch).join(' ')
}
/**
 * Extract a flag (boolean) argument.
 * Can have a single-character alias and a default value (so can be true instead
 * of false).
 */
TinyArgv.prototype.flag = function ({key, defaultValue = false, alias = false}) {
  const indexOf = this.__argIndex(key, alias)
  this[key] = indexOf >= 0 ? !defaultValue : defaultValue
  return this;
}
/*
 * Extract a string-based argument.
 * Can have a single-character alias or a default value.
 */
TinyArgv.prototype.value = function ({key, defaultValue = null, alias = false}) {
  const indexOf = this.__argIndex(key, alias)
  this[key] = indexOf >= 0 ? this.__extractValue(indexOf) : defaultValue
  return this;
}
/**
 * Convenient chain-initiating function. Call this to start the processing.
 * Optional argument: can process a different array than process.argv
 */
const argv = function (argv) {
  return new TinyArgv(argv ? argv : process.argv.slice(2))
}

const arguments = argv()
  .value({key: 'original', defaultValue: './chapter.md', alias: 'r'})
  .value({key: 'formatted', defaultValue: './chapter-revised.md', alias: 'f'})
  .value({key: 'output', defaultValue: './chapter-corrected.md', alias: 'o'})
  .value({key: 'csl', defaultValue: fs.readFileSync('./Rakefile').toString().match(/\$cite_style = \"(.+?)\"/)[1]})
  .value({key: 'bibliography', defaultValue: fs.readFileSync('./Rakefile').toString().match(/bib_file = \"(.+?)\"/)[1].replace('#{ENV[\'HOME\']}', process.env.HOME)})

// Thanks, Stack Overflow:
// https://stackoverflow.com/questions/5069464/replace-multiple-strings-at-once/34560648#34560648
function str_replace($f, $r, $s) {
  return $s.replace(new RegExp("(" + (typeof ($f) === "string" ? $f.replace(/[.?*+^$[\]\\(){}|-]/g, "\\$&") : $f.map(function (i) {return i.replace(/[.?*+^$[\]\\(){}|-]/g, "\\$&")}).join("|")) + ")", "g"), typeof ($r) === "string" ? $r : typeof ($f) === "string" ? $r[0] : function (i) {return $r[$f.indexOf(i)]});
}

const rawCitations = fs.readFileSync(arguments.original).toString().match(/\[\@[^\]]+\]/g)
const substitutions = {}

// Process the list of all the raw citations from the original file to find the cite keys:
process.stdout.write('Processing citations in Pandoc (this is slow)…')
rawCitations.forEach(cite => {
  substitutions[cite] = child_process.execSync(`echo "${cite}" | pandoc -f markdown+smart -t plain+smart --bibliography="${arguments.bibliography}" --csl="${arguments.csl}.csl"`).toString().split(/\n/g)[0]
})
console.log('done')

fs.readFile(arguments.formatted, (_err, text) => {
  fs.writeFile(arguments.output, str_replace(Object.values(substitutions), Object.keys(substitutions), text.toString()), () => {})
})