definma-api/data_import/import.js

const csv = require('csv-parser');
const fs = require('fs');
const axios = require('axios');
const {Builder} = require('selenium-webdriver');    // selenium and the chrome driver must be installed and configured separately
const chrome = require('selenium-webdriver/chrome');
const pdfReader = require('pdfreader');
const iconv = require('iconv-lite');

const metaDoc = 'C:\\Users\\vle2fe\\Documents\\Data\\Rng_200707\\metadata.csv';  // metadata files
const kfDoc = 'C:\\Users\\vle2fe\\Documents\\Data\\Rng_200707\\kf.csv';
const vzDoc = 'C:\\Users\\vle2fe\\Documents\\Data\\Rng_200707\\vz.csv';
const nmDocs = 'C:\\Users\\vle2fe\\Documents\\Data\\Rng_200707\\nmDocs';  // NormMaster Documents
const dptFiles = 'C:\\Users\\vle2fe\\Documents\\Data\\Rng_200707\\DPT';  // Spectrum files
// const host = 'http://localhost:3000';
const host = 'https://definma-api.apps.de1.bosch-iot-cloud.com';
let data = [];  // metadata contents
let materials = {};
let samples = [];
let normMaster = {};
let sampleDevices = {};

// TODO: BASF twice, BASF as color
// TODO: duplicate kf values
// TODO: conditions
// TODO: comment and reference handling


// TODO: check last color errors (filter out already taken) use location and device for user, upload to BIC

main();

async function main() {
  if (0) {  // materials
    await getNormMaster();
    await importCsv(metaDoc);
    await allMaterials();
    await saveMaterials();
    await importCsv(kfDoc);
    await allMaterials();
    await saveMaterials();
    await importCsv(vzDoc);
    await allMaterials();
    await saveMaterials();
  }
  if (0) {  // samples
    sampleDeviceMap();
    if (1) {
      console.log('-------- META ----------');
      await importCsv(metaDoc);
      await allSamples();
      await saveSamples();
    }
    if (1) {
      console.log('-------- KF ----------');
      await importCsv(kfDoc);
      await allSamples();
      await saveSamples();
      await allKfVz();
    }
    if (1) {
      console.log('-------- VZ ----------');
      await importCsv(vzDoc);
      await allSamples();
      await saveSamples();
      await allKfVz();
    }
  }
  if (1) {  // DPT
    await allDpts();
  }
  if (0) {  // pdf test
    console.log(await readPdf('N28_BN05-OX013_2016-03-11.pdf'));
  }
}

async function importCsv(doc) {
  data = [];
  await new Promise(resolve => {
    fs.createReadStream(doc)
      .pipe(iconv.decodeStream('win1252'))
      .pipe(csv())
      .on('data', (row) => {
        data.push(row);
      })
      .on('end', () => {
        console.info('CSV file successfully processed');
        if (data[0]['Farbe']) {  // fix German column names
          data.map(e => {e['Color'] = e['Farbe']; return e; });
        }
        resolve();
      });
  });
}

async function allDpts() {
  let res = await axios({
    method: 'get',
    url: host + '/template/measurements',
    auth: {
      username: 'admin',
      password: 'Abc123!#'
    }
  });
  const measurement_template = res.data.find(e => e.name === 'spectrum')._id;
  res = await axios({
    method: 'get',
    url: host + '/samples?status=all',
    auth: {
      username: 'admin',
      password: 'Abc123!#'
    }
  });
  const sampleIds = {};
  res.data.forEach(sample => {
    sampleIds[sample.number] = sample._id;
  });
  const dptRegex = /.*?_(.*?)_(\d+|\d+_\d+).DPT/;
  const dpts = fs.readdirSync(dptFiles);
  for (let i in dpts) {
    const regexRes = dptRegex.exec(dpts[i])
    if (regexRes && sampleIds[regexRes[1]]) {  // found matching sample
      console.log(dpts[i]);
      const f = fs.readFileSync(dptFiles + '\\' + dpts[i], 'utf-8');
      const data = {
        sample_id: sampleIds[regexRes[1]],
        values: {},
        measurement_template
      };
      data.values.dpt = f.split('\r\n').map(e => e.split(','));
      await axios({
        method: 'post',
        url: host + '/measurement/new',
        auth: {
          username: 'admin',
          password: 'Abc123!#'
        },
        data
      }).catch(err => {
        console.log(dpts[i]);
        console.error(err.response.data);
      });
    }
    else {
      console.log(`Could not find sample for ${dpts[i]} !!!!!!`);
    }
  }
}

async function allKfVz() {
  let res = await axios({
    method: 'get',
    url: host + '/template/measurements',
    auth: {
      username: 'admin',
      password: 'Abc123!#'
    }
  });
  const kf_template = res.data.find(e => e.name === 'kf')._id;
  const vz_template = res.data.find(e => e.name === 'vz')._id;
  res = await axios({
    method: 'get',
    url: host + '/samples?status=all',
    auth: {
      username: 'admin',
      password: 'Abc123!#'
    }
  });
  const sampleIds = {};
  res.data.forEach(sample => {
    sampleIds[sample.number] = sample._id;
  });
  for (let index in data) {
    console.info(`${index}/${data.length}`);
    let sample = data[index];
    if (sample['Sample number'] !== '') {
      let credentials = ['admin', 'Abc123!#'];
      if (sampleDevices[sample['Sample number']]) {
        credentials = [sampleDevices[sample['Sample number']], '2020DeFinMachen!']
      }
      if (sample['KF in Gew%']) {
        await axios({
          method: 'post',
          url: host + '/measurement/new',
          auth: {
            username: credentials[0],
            password: credentials[1]
          },
          data: {
            sample_id: sampleIds[sample['Sample number']],
            measurement_template: kf_template,
            values: {
              'weight %': sample['KF in Gew%'],
              'standard deviation': sample['Stabwn']
            }
          }
        }).catch(err => {
          console.log(sample['Sample number']);
          console.error(err.response.data);
        });
      }
      if (sample['VZ (ml/g)']) {
        await axios({
          method: 'post',
          url: host + '/measurement/new',
          auth: {
            username: credentials[0],
            password: credentials[1]
          },
          data: {
            sample_id: sampleIds[sample['Sample number']],
            measurement_template: vz_template,
            values: {
              vz: sample['VZ (ml/g)']
            }
          }
        }).catch(err => {
          console.log(sample['Sample number']);
          console.error(err.response.data);
        });
      }
    }
  }
}

async function allSamples() {
  samples = [];
  let res = await axios({
    method: 'get',
    url: host + '/materials?status=all',
    auth: {
      username: 'admin',
      password: 'Abc123!#'
    }
  });
  const dbMaterials = {}
  res.data.forEach(m => {
    dbMaterials[m.name] = m;
  })
  res = await axios({
    method: 'get',
    url: host + '/samples?status=all',
    auth: {
      username: 'admin',
      password: 'Abc123!#'
    }
  });
  const sampleColors = {};
  res.data.forEach(sample => {
    sampleColors[sample.number] = sample.color;
  });


  for (let index in data) {
    console.info(`${index}/${data.length}`);
    let sample = data[index];
    if (sample['Sample number'] !== '') {  // TODO: what about samples without color
      if (sample['Supplier'] === '') {  // empty supplier fields
        sample['Supplier'] = 'unknown';
      }
      if (sample['Granulate/Part'] === '') {  // empty supplier fields
        sample['Granulate/Part'] = 'unknown';
      }
      const material = dbMaterials[trim(sample['Material name'])];
      if (!material) { // could not find material, skipping sample
        continue;
      }
      console.log(sample['Material name']);
      console.log(material._id);
      samples.push({
        number: sample['Sample number'],
        type: sample['Granulate/Part'],
        batch: sample['Charge/batch granulate/part'] || '',
        material_id: material._id,
        notes: {
          comment: sample['Comments']
        }
      });
      const si = samples.length - 1;
      if (sample['Material number'] !== '' && material.numbers.find(e => e.number === sample['Material number'])) {  // TODO: fix because of false material/material number
        samples[si].color = material.numbers.find(e => e.number === sample['Material number']).color;
      }
      else if (sample['Color'] && sample['Color'] !== '') {
        let number = material.numbers.find(e => e.color.indexOf(trim(sample['Color'])) >= 0);
        if (!number && /black/.test(sample['Color'])) {  // special case bk for black
          number = material.numbers.find(e => e.color.toLowerCase().indexOf('bk') >= 0);
          if (!number) {  // try German word
            number = material.numbers.find(e => e.color.toLowerCase().indexOf('schwarz') >= 0);
          }
        }
        samples[si].color = number.color;
      }
      else if (sampleColors[sample['Sample number'].split('_')[0]]) {  // derive color from main sample for kf/vz
        samples[si].color = sampleColors[sample['Sample number'].split('_')[0]];
      }
      else {
        samples[si].color = '';
      }
    }
  }
}

async function saveSamples() {
  for (let i in samples) {
    console.info(`${i}/${samples.length}`);
    let credentials = ['admin', 'Abc123!#'];
    if (sampleDevices[samples[i].number]) {
      credentials = [sampleDevices[samples[i].number], '2020DeFinMachen!']
    }
    await axios({
      method: 'post',
      url: host + '/sample/new',
      auth: {
        username: credentials[0],
        password: credentials[1]
      },
      data: samples[i]
    }).catch(err => {
      if (err.response.data.status && err.response.data.status !== 'Sample number already taken') {
        console.log(samples[i]);
        console.error(err.response.data);
      }
    });
  }
  console.info('saved all samples');
}

async function allMaterials() {
  materials = {};
  for (let index in data) {
    let sample = data[index];
    if (sample['Sample number'] && sample['Sample number'] !== '') {
      if (sample['Supplier'] === '') {  // empty supplier fields
        sample['Supplier'] = 'unknown';
      }
      if (sample['Material name'] === '') {  // empty name fields
        sample['Material name'] = sample['Material'];
      }
      if (!sample['Material']) {  // column Material is named Plastic in VZ metadata
        sample['Material'] = sample['Plastic'];
      }
      sample['Material name'] = trim(sample['Material name']);
      if (materials.hasOwnProperty(sample['Material name'])) {  // material already found at least once
        if (sample['Material number'] && sample['Material number'] !== '') {
          if (materials[sample['Material name']].numbers.length === 0 || !materials[sample['Material name']].numbers.find(e => e.number === stripSpaces(sample['Material number']))) {  // new material number
            if (materials[sample['Material name']].numbers.find(e => e.color === sample['Color'] && e.number === '')) {  // color already in list, only number missing
              materials[sample['Material name']].numbers.find(e => e.color === sample['Color'] && e.number === '').number = stripSpaces(sample['Material number']);
            }
            else {
              materials[sample['Material name']].numbers.push({color: trim(sample['Color']), number: stripSpaces(sample['Material number'])});
            }
          }
        }
        else if (sample['Color'] && sample['Color'] !== '') {
          if (!materials[sample['Material name']].numbers.find(e => e.color === stripSpaces(sample['Color']))) {  // new material color
            materials[sample['Material name']].numbers.push({color: trim(sample['Color']), number: ''});
          }
        }
      }
      else {  // new material
        console.info(`${index}/${data.length}  ${sample['Material name']}`);
        materials[sample['Material name']] = {
          name: sample['Material name'],
          supplier: trim(sample['Supplier']),
          group: trim(sample['Material'])
        };
        let tmp = /M(\d+)/.exec(sample['Reinforcing material']);
        materials[sample['Material name']].mineral = tmp ? tmp[1] : 0;
        tmp = /GF(\d+)/.exec(sample['Reinforcing material']);
        materials[sample['Material name']].glass_fiber = tmp ? tmp[1] : 0;
        tmp = /CF(\d+)/.exec(sample['Reinforcing material']);
        materials[sample['Material name']].carbon_fiber = tmp ? tmp[1] : 0;
        materials[sample['Material name']].numbers = await numbersFetch(sample);
        console.log(materials[sample['Material name']]);
      }
    }
  }
}

async function saveMaterials() {
  const mKeys = Object.keys(materials)
  for (let i in mKeys) {
    console.info(`${i}/${mKeys.length}`);
    await axios({
      method: 'post',
      url: host + '/material/new',
      auth: {
        username: 'admin',
        password: 'Abc123!#'
      },
      data: materials[mKeys[i]]
    }).catch(err => {
      if (err.response.data.status && err.response.data.status !== 'Material name already taken') {
        console.info(materials[mKeys[i]]);
        console.error(err.response.data);
      }
    });
  }
  console.info('saved all materials');
}

async function numbersFetch(sample) {
  let nm = [];
  let res = [];
  if (sample['Material number']) {  // sample has a material number
    nm = normMaster[stripSpaces(sample['Material number'])]? [normMaster[stripSpaces(sample['Material number'])]] : [];
  }
  else {  // try finding via material name
    nm = Object.keys(normMaster).filter(e => normMaster[e].nameSpaceless === stripSpaces(sample['Material name'])).map(e => normMaster[e]);
  }
  if (nm.length > 0) {
    for (let i in nm) {
      // if (!fs.readdirSync(nmDocs).find(e => e.indexOf(nm[i].doc.replace(/ /g, '_')) >= 0)) {  // document not loaded
      //   await getNormMasterDoc(nm[i].url.replace(/ /g, '%20'));
      // }
      // if (!fs.readdirSync(nmDocs).find(e => e.indexOf(nm[i].doc.replace(/ /g, '_')) >= 0)) {  // document not loaded
      //   console.info('Retrying download...');
      //   await getNormMasterDoc(nm[i].url.replace(/ /g, '%20'), 2.2);
      // }
      // if (!fs.readdirSync(nmDocs).find(e => e.indexOf(nm[i].doc.replace(/ /g, '_')) >= 0)) {  // document not loaded
      //   console.info('Retrying download again...');
      //   await getNormMasterDoc(nm[i].url.replace(/ /g, '%20'), 5);
      // }
      if (fs.readdirSync(nmDocs).find(e => e.indexOf(nm[i].doc.replace(/ /g, '_')) >= 0)) {  // document loaded
        res = await readPdf(fs.readdirSync(nmDocs).find(e => e.indexOf(nm[i].doc.replace(/ /g, '_')) >= 0));
      }
      if (res.length > 0) {  // no results
        break;
      }
      else if (i + 1 >= nm.length) {
        console.error('Download failed!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!');
      }
    }
  }
  if (res.length === 0) {  // no results
    if ((sample['Color'] && sample['Color'] !== '') || (sample['Material number'] &&sample['Material number'] !== '')) {
      return [{color: trim(sample['Color']), number: sample['Material number']}];
    }
    else {
      return [];
    }
  }
  else {
    if (sample['Material number'] && !res.find(e => e.number === sample['Material number'])) {  // sometimes norm master does not include sample number even if listed
      res.push({color: trim(sample['Color']), number: sample['Material number']});
    }
    return res;
  }
}

async function getNormMaster(fetchAgain = false) {
  if (fetchAgain) {
    console.info('fetching norm master...');
    const res = await axios({
      method: 'get',
      url: 'http://rb-normen.bosch.com/cgi-bin/searchRBNorm4TradeName'
    });

    console.info('finding documents...');
    let match;
    // const regex = /<tr>.*?<td>.*?<\/span>(.*?)<\/td><td>(\d+)<\/td>.*?<a href="(.*?)"/gm;
    const regex = /<tr>.*?<td>.*?<\/span>(.*?)<\/td><td>(\d+)<\/td><td>40.*?<a href="(.*?)".*?<\/a>(.*?)<\/td>/gm;  // only valid materials
    do {
      match = regex.exec(res.data);
      if (match) {
        normMaster[match[2]] = {name: match[1], nameSpaceless: stripSpaces(match[1]), number: match[2], url: match[3], doc: match[4]};
      }
    } while (match);
    fs.writeFileSync('./data_import/normMaster.json', JSON.stringify(normMaster));
  }
  else {
    normMaster = JSON.parse(fs.readFileSync('./data_import/normMaster.json'), 'utf-8');
  }
}

function getNormMasterDoc(url, timing = 1) {
  console.info(url);
  return new Promise(async resolve => {
    const options = new chrome.Options();
    options.setUserPreferences({
      "download.default_directory": nmDocs,
      "download.prompt_for_download": false,
      "download.directory_upgrade": true,
      "plugins.always_open_pdf_externally": true
    });
    let driver = await new Builder().forBrowser('chrome').setChromeOptions(options).build();
    let timeout = 7000 * timing;
    try {
      await driver.get(url);
      if (await driver.getCurrentUrl() !== 'https://rb-wam-saml.bosch.com/tfim/sps/normmaster/saml20/login') {  // got document selection page
        timeout = 11000 * timing;
        await driver.executeScript('Array.prototype.slice.call(document.querySelectorAll(\'.functionlink\')).filter(e => e.innerText === \'English\')[0].click()').catch(() => {timeout = 0; });
      }
    }
    finally {
      setTimeout(async () => {  // wait until download is finished
        await driver.quit();
        resolve();
      }, timeout);
    }
  });
}

function readPdf(file) {
  return new Promise(async resolve => {
    const countdown = 100;  // value for text timeout
    let table = 0;      // > 0 when in correct table area
    let rows = [];      // found table rows
    let lastY = 0;      // y of last row
    let lastX = 0;      // right x of last item
    let lastText = '';  // text of last item
    let lastLastText = '';  // text of last last item
    await new pdfReader.PdfReader().parseFileItems(nmDocs + '\\' + file, (err, item) => {
      if (item && item.text) {
        if ((stripSpaces(lastLastText + lastText + item.text).toLowerCase().indexOf('colordesignationsupplier') >= 0) || (stripSpaces(lastLastText + lastText + item.text).toLowerCase().indexOf('colordesignatiomsupplier') >= 0)) {  // table area starts
          table = countdown;
        }
        if (table > 0) {
          // console.log(item);
          // console.log(item.y - lastY);
          // console.log(item.text);
          if (item.y - lastY > 0.8 && Math.abs(item.x - lastX) > 5) {  // new row
            lastY = item.y;
            rows.push(item.text);
          }
          else {  // still the same row row
            rows[rows.length - 1] += (item.x - lastX > 1.09 ? '$' : '') + item.text;  // push to row, detect if still same cell
          }
          lastX = (item.w * 0.055) + item.x;

          if (/\d \d\d\d \d\d\d \d\d\d/.test(item.text)) {
            table = countdown;
          }
          table --;
          if (table <= 0 || item.text.toLowerCase().indexOf('release document') >= 0 || item.text.toLowerCase().indexOf('normative references') >= 0) {  // table area ended
            table = -1;
            // console.log(rows);
            rows = rows.filter(e => /^\d{10}/m.test(stripSpaces(e)));  // filter non-table rows
            resolve(rows.map(e => {return {color: trim(e.split('$')[3]), number: stripSpaces(e.split('$')[0])}; }));
          }
        }
        lastLastText = lastText;
        lastText = item.text;
      }
      if (!item && table !== -1) {  // document ended
        rows = rows.filter(e => /^\d{10}/m.test(stripSpaces(e)));  // filter non-table rows
        resolve(rows.map(e => {return {color: trim(e.split('$')[3]), number: stripSpaces(e.split('$')[0])}; }));
      }
    });
  });
}

function sampleDeviceMap() {
  const dpts = fs.readdirSync(dptFiles);
  const regex = /(.*?)_(.*?)_(\d+|[^_]+_\d+).DPT/;
  for (let i in dpts) {
    const regexRes = regex.exec(dpts[i])
    if (regexRes) {  // found matching sample
      sampleDevices[regexRes[2]] = regexRes[1] === 'plastics' ? 'rng01' : regexRes[1].toLowerCase();
    }
  }
}

function stripSpaces(s) {
  return s ? s.replace(/ /g,'') : '';
}

function trim(s) {
  return s.replace(/(^\s+|\s+$)/gm, '');
}