readHistogram.js

// This function reads only the Histogram files generated by eBird.
// This is useful for seeing what birds have already been seen, but does not
// cover seeing all birds which might be seen or which have been seen outside
// of the area of the Histogram file.

const fs = require('fs').promises
const removeEmptyLines = require("remove-blank-lines");

async function getData (input) {
  let results = {}
  if (fs) {
    // Yes, this is brittle. I don't expect this format to change soon.
    input = removeEmptyLines(await fs.readFile(input, 'utf8')).split('\n')
    results.taxa = input.filter(s => s.includes('Number of taxa'))[0].split('\t')[1]
    results.sampleSize = input.filter(s => s.includes('Sample Size'))[0].split('\t').slice(1).filter(x => x !== '')
    input = input.slice(4,-1).map(species => {
      let dict = {}
      let speciesMatch = species.split('\t')[0].match(/(?<species>.*) \(.*>(?<latin>.*)</)
      let speciesName = speciesMatch.groups.species
      let scientificName = speciesMatch.groups.latin
      let frequency = species.split('\t').slice(1)
      // Basically, remove spuh
      let paraspecial = ['sp.', '/']
      if (!paraspecial.some(s => speciesName.includes(s))) {
        console.log(speciesName)
        dict[speciesName] = {
          species: speciesName,
          'Scientific Name': scientificName,
          frequency
        }
        return dict
      }
      return null
    })
    results.species = input.filter(s => s !== null)
  }

  console.log(results.species.length)
  return results
}

// TODO Is there a way to automatically generate these? As in, download them each morning from eBird?
async function washingtonCounty2020 () {
  return await getData('data/ebird_US-VT-023__2020_2020_1_12_barchart.txt')
}

module.exports = {
  washingtonCounty2020
}