indexMarkdownFiles.js

/**
 * find all markdown files in given directory, (here just current directory)
 * and then fetch their title to generate index markdown file
 *
 * use -v to show verbose detail
 * use -o to specify output file, default to ./index.md
 * use -d to specify scanner directory, such as -d the_directory, default to ./
 */
var fs = require('fs'),
  path = require('path')

var mdExts = ['.md', '.markdown']
var excludeDir = ['.git']

function isMarkdownFile(filePath) {
  return mdExts.indexOf(path.extname(filePath).toLowerCase()) !== -1
}

function isNotExlude(dir) {
  return excludeDir.indexOf(dir) == -1
}

function getNamedCommandLineArgs(list) {
  var ret = {},
    p = /^\-([^\-].*)/

  for (var i = 0, l = list.length; i < l; ) {
    var m = list[i].match(p)
    if (m && m[1]) {
      var k = m[1],
        v = ''

      if (i + 1 < l) {
        if (!p.test(list[i+1])) {
          v = list[i+1]
          i += 2
        } else {
          i++
        }
      } else {
        i++
      }

      ret[k] = v
    } else {
      i++
    }
  }

  return ret
}

function findMarkdownFiles(dir) {
  if (!dir || !fs.existsSync(dir)) return []

  var mdFiles = []
  var files = fs.readdirSync(dir)
  files.forEach(function(file) {
    filePath = path.join(dir, file)

    var st = fs.statSync(filePath)
    if (st && st.isFile() && isMarkdownFile(filePath)) {
      mdFiles.push(filePath)
    } else if (st && st.isDirectory() && isNotExlude(file)) {
      mdFiles = mdFiles.concat(findMarkdownFiles(filePath))
    }
  })

  return mdFiles
}

function isTitleDash(str) {
  var p = /^(?:\-+|=+)[\s\t]*$/
  return p.test(str)
}

function getMarkDownFileTitle(filePath) {
  /*
    1. readFile
    2. trim
    3. first line: index of #, and limit in (1, 6) yes to 4, no to 5
    4. get string from the last continues #
    5. if the second line is = or - , return the first line, else return empty string
  */
  var title = ''

  // try to use file name as default
  try {
    title = path.basename(filePath, path.extname(filePath))
  } catch(e) {}

  try {
    var lines = fs.readFileSync(filePath).toString().split('\n')
    var i = 0, l = lines.length
    while (i < l && lines[i].trim() === '') {
      i++
    }

    // starts with #
    var startPattern = /^#{1,6}(?:[\s\t]*)(.*)/,
      startMatches = lines[i].match(startPattern)

    if (startMatches) {
      if (startMatches[1]) {
        title = startMatches[1].trim()
      }
    } else {
      var nextLine = lines[i + 1]
      if (i + 1 < l && nextLine) {
        if (isTitleDash(nextLine)) {
          title = lines[i].trim()
        }
      }
    }
  } catch(e) {
    typeof console === 'object' && console.warn && console.warn(e)
  }

  return title
}

function createIndexPageWithData(targetFile, data) {
  var lines = ['# wiki 索引']

  data.forEach(function(item) {
    lines.push('- [' + item.title + '](' + item.file + ')')
  })

  try {
    fs.writeFile(targetFile, lines.join('\n'))
    console.log('success generate', targetFile)
  } catch(e) {
    console.error('failed to generate', targetFile, e)
  }
}


// run
var argsObj = getNamedCommandLineArgs(process.argv.slice(2)),
  scanDir = argsObj.d || './',
  outputFile = argsObj.o || './index.md',
  verbose = argsObj.v !== undefined

if (verbose) {
  console.log('name args', argsObj, '\n')
  console.log('title from md files')
}

var ret = findMarkdownFiles(scanDir).map(function(mdFilePath) {
  var title = getMarkDownFileTitle(mdFilePath)

  if (verbose) {
    console.log('  ', title, '<=', mdFilePath)
  }

  // remove .md extension (actually remove all extension)
  // so gitlab will render it rather than show source code
  return {
    file: path.join(path.dirname(mdFilePath), path.basename(mdFilePath, path.extname(mdFilePath))),
    title: title
  }
})

if (verbose) {
  console.log('\n')
}

createIndexPageWithData(outputFile, ret)