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 _ = require('lodash');

const stages = {
  materials: true,
  samples: true,
  dpt: true
}

const docs = [
  "C:\\Users\\vle2fe\\Documents\\Data\\All_200717\\Metadata__AnP2.csv",
  "C:\\Users\\vle2fe\\Documents\\Data\\All_200717\\Metadata__AnP2_A.csv",
  "C:\\Users\\vle2fe\\Documents\\Data\\All_200717\\Metadata__AnP2_B.csv",
  "C:\\Users\\vle2fe\\Documents\\Data\\All_200717\\Metadata_Ap.csv",
  "C:\\Users\\vle2fe\\Documents\\Data\\All_200717\\Metadata_Bj.csv",
  "C:\\Users\\vle2fe\\Documents\\Data\\All_200717\\Metadata_Eh.csv",
  "C:\\Users\\vle2fe\\Documents\\Data\\All_200717\\Metadata_Eh_B.csv",
  "C:\\Users\\vle2fe\\Documents\\Data\\All_200717\\Metadata_Eh_Duroplasten.csv",
  "C:\\Users\\vle2fe\\Documents\\Data\\All_200717\\Metadata_Rng_aktuell.csv",
  "C:\\Users\\vle2fe\\Documents\\Data\\All_200717\\Metadata_Rng_aktuell_A.csv",
  "C:\\Users\\vle2fe\\Documents\\Data\\All_200717\\Metadata_Rng_aktuell_B.csv",
  "C:\\Users\\vle2fe\\Documents\\Data\\All_200717\\Metadata_WaP.csv",
];
const errors = [];
const nmDocs = 'C:\\Users\\vle2fe\\Documents\\Data\\All_200717\\nmDocs';  // NormMaster Documents
const dptFiles = 'C:\\Users\\vle2fe\\Documents\\Data\\All_200717\\DPT';  // Spectrum files
const host = 'http://localhost:3000';
// const host = 'https://definma-api.apps.de1.bosch-iot-cloud.com';
const requiredProperties = ['samplenumber','materialnumber','materialname','supplier','reinforcementmaterial','material','granulate/part','color','charge/batch','comments'];
dict = {  // dictionary
  'Granulat': 'granulate',
  'Zugstab': 'tension rod',
  'Stecker': 'plug'
};
let data = [];  // metadata contents
let materials = {};
let numberToColor = {};
let samples = [];
let normMaster = {};
let sampleDevices = {};
const sampleReferences = [];  // references to other samples in format {sample, referencedSample, relation}
let commentsLog = [];
let customFieldsLog = [];
const vnValues = {};  // vn values from comments
const dptLog = [];
const dptSampleAddLog = [];  // log samples created during dpt insertion
const typeLog = [];

// TODO: conditions

main();

async function main() {
  if (stages.materials) {  // materials
    await getNormMaster();
    for (let i in docs) {
      await importCsv(docs[i]);
      await allMaterials();
      await saveMaterials();
    }
    fs.writeFileSync('./data_import/numberToColor.json', JSON.stringify(numberToColor));
  }
  if (stages.samples) {  // samples
    sampleDeviceMap();
    numberToColor = JSON.parse(fs.readFileSync('./data_import/numberToColor.json'), 'utf-8');
    for (let i in docs) {
      await importCsv(docs[i]);
      await allSamples();
      await saveSamples();
      await allMcVn();
    }
    // write logs
    fs.writeFileSync('./data_import/comments.txt', commentsLog.join('\r\n'));
    fs.writeFileSync('./data_import/typeLog.txt', typeLog.join('\r\n'));
    fs.writeFileSync('./data_import/customFields.txt', customFieldsLog.join('\r\n'));
    fs.writeFileSync('./data_import/sampleReferences.txt', sampleReferences.map(e => JSON.stringify(e)).join('\r\n'));
    fs.writeFileSync('./data_import/sampleReferences.json', JSON.stringify(sampleReferences));

    await sampleReferencesSave();
  }
  if (stages.dpt) {  // DPT
    await allDpts();
    fs.writeFileSync('./data_import/sdptLog.txt', dptLog.join('\r\n'));
    fs.writeFileSync('./data_import/dptSampleAddLog.txt', dptSampleAddLog.join('\r\n'));
  }
  if (0) {  // pdf test
    console.log(await readPdf('N28_BN05-OX023_2019-07-16.pdf'));
  }
  if (errors.length) {
    // console.log(errors);
    fs.writeFileSync('./data_import/errors/errors_' + new Date().getTime() + '.txt', errors.join('\r\n'));
  }
}

async function importCsv(doc) {
  // Uniform name                 samplenumber  materialnumber  materialname  supplier material plastic reinforcingmaterial  granulate/part color charge/batch                comments  vz(ml/g)  kfingew%       degradation(%) reinforcingmaterialcontent          stabwn
  // Metadata__AnP2.csv           Sample number,Material number,Material name,Supplier,Material,Plastic,Reinforcing material,granulate/Part,Color,Charge/ Batch,              Comments
  // Metadata__AnP2_A.csv         Sample number,Material number,Material name,Supplier,         Plastic,Reinforcing material,Granulate/Part,                                  Comments,           Humidity [ppm]
  // Metadata__AnP2_B.csv         Sample number,Material number,Material name,Supplier,         Plastic,Reinforcing material,Granulate/Part,                                            VZ [ml/g],                              glass fibre content
  // Metadata_Ap.csv              Sample number,Material number,Material name,Supplier,         Plastic,Reinforcing material,Granulate/Part,Color,Charge/Batch,               Comments
  // Metadata_Bj.csv              Sample number,Material number,Material name,Supplier,Material,Plastic,Reinforcing material,Granulate/Part,Color,Charge/batch granulate/part,Comments
  // Metadata_Eh.csv              Sample number,Material number,Material name,Supplier,Material,        Reinforcing material,Granulate/Part,Color,Charge/Batch granulate/part,Comments, VZ [cm³/g],                                                                        Spalte1
  // Metadata_Eh_B.csv            Sample number,                Material name,Supplier,         Plastic,Reinforcing material,Granulate/Part,Color,                            Comments, VZ [cm³/g]
  // Metadata_Eh_Duroplasten.csv  Sample number,Material number,Material name,Supplier,Material,        Reinforcing material,Granulate/Part,Color,Charge/Batch granulate/part,Comments
  // Metadata_Rng_aktuell.csv     Sample number,Material number,Material name,Supplier,Material,Plastic,Reinforcing material,Granulate/Part,Color,Charge/batch granulate/part,Comments, VZ (ml/g),               Degradation(%),Glas fibre content (%)
  // Metadata_Rng_aktuell_A.csv   Sample number,Material number,Material name,Supplier,Material,Plastic,Reinforcing material,Granulate/Part,Farbe,Charge/batch granulate/part,Comments,           KF in Gew%,                   Reinforcing material (content in %),Stabwn
  // Metadata_Rng_aktuell_B.csv   Sample number,                Material name,Supplier,         Plastic,                     Granulate/Part,                                  Comments, VZ (ml/g),               Degradation (%),                                          Alterungszeit in h
  // Metadata_WaP.csv             Probennummer,                 Name,         Firma,   Material,                             Teil/Rohstoff,       Charge,                     Anmerkung,VZ (ml/g),               Abbau (%),     Verstärkungsstoffgehalt (%),               Versuchsnummer
  const nameCorrection = {  // map to right column names
    'probennummer': 'samplenumber',
    'name': 'materialname',
    'firma': 'supplier',
    'teil/rohstoff': 'granulate/part',
    'charge/batchgranulate/part': 'charge/batch',
    'charge': 'charge/batch',
    'anmerkung': 'comments',
    'vz[ml/g]': 'vz(ml/g)',
    'vz[cm³/g]': 'vz(ml/g)',
    'abbau(%)': 'degradation(%)',
    'glassfibrecontent': 'reinforcingmaterialcontent',
    'glasfibrecontent(%)': 'reinforcingmaterialcontent',
    'reinforcingmaterial(contentin%)': 'reinforcingmaterialcontent',
    'verstärkungsstoffgehalt(%)': 'reinforcingmaterialcontent'
  };
  const missingFieldsFill = [  // column names to fill if they do not exist
    'color',
    'charge/batch',
    'comments',
    'materialnumber',
    'reinforcementmaterial'
  ]
  console.info('importing ' + doc);
  data = [];
  await new Promise(resolve => {
    fs.createReadStream(doc)
      .pipe(iconv.decodeStream('win1252'))
      .pipe(csv())
      .on('data', (row) => {
        data.push(row);
      })
      .on('end', () => {
        data = data.map(e => {
          const newE = {};
          Object.keys(e).forEach(key => {
            newE[key.toLowerCase().replace(/ /g, '')] = e[key];
          });
          // replace wrong column names
          Object.keys(newE).forEach(key => {
            if (nameCorrection.hasOwnProperty(key)) {
              newE[nameCorrection[key]] = newE[key];
              delete newE[key];
            }
          });

          // add missing fields with empty values
          missingFieldsFill.forEach(field => {
            if (!newE.hasOwnProperty(field)) {
              newE[field] = '';
            }
          });
          if(newE['materialname'] === '') {
            newE['materialname'] = newE['material'];
          }
          if (newE['supplier'] === '') {  // empty supplier fields
            newE['supplier'] = 'unknown';
          }
          if (!newE.hasOwnProperty('material')) {
            newE['material'] = newE['plastic'].indexOf(' GF') >= 0 ? newE['plastic'].split(' ')[0] : newE['plastic'];
          }
          return newE;
        }).filter(e => {
          const missingProperties = requiredProperties.filter(el => !e.hasOwnProperty(el));
          if (e['materialname'] === '') {
            missingProperties.push('materialname');
          }
          if (e['samplenumber'] === '') {  // empty row
            return false;
          }
          else if (missingProperties.length > 0) {  // incomplete sample
            errors.push(`${doc}: ${JSON.stringify(e)}is missing the required properties ${missingProperties}`);
            return false;
          }
          return true;
        });
        console.info('CSV file successfully processed');
        resolve();
      });
  });
}

async function allDpts() {
  let res = await axios({
    method: 'get',
    url: host + '/template/measurements',
    auth: {
      username: 'admin',
      password: 'Abc123!#'
    }
  });
  const measurement_templates = res.data.filter(e => e.name === 'spectrum');
  const measurement_template = measurement_templates[measurement_templates.length - 1]._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+|[a-zA-Z0-9]+[_.]\d+)(_JDX)?[.]{1,2}(DPT|csv|CSV|JDX)/;
  const dpts = fs.readdirSync(dptFiles);
  for (let i in dpts) {
    let regexInput;
    const bjRes = /^(Bj[FT]?)\s?([a-z0-9_]*)_JDX.DPT/.exec(dpts[i]);
    if (bjRes) {  // correct Bj numbers with space
      regexInput = `Bj01_${bjRes[1]}${bjRes[2]}_0.DPT`;
    }
    else {  // remove _JDX from name
      regexInput = dpts[i].replace(/_JDX.*\./, '.');
    }
    const regexRes = dptRegex.exec(regexInput);
    if (regexRes && !sampleIds[regexRes[2]]) {  // when sample number includes an additional _x instead of having _x_x for spectrum description
      regexRes[2] = `${regexRes[2]}_${regexRes[3].split('_')[0]}`;
    }
    let baseSample = null;
    if (regexRes) {
      baseSample = regexRes[2].split('_')[0];
      if (baseSample === 'Wa11') {  // as Wa11 samples use all the same material
        baseSample = 'Wa11_B0_1';
      }
    }
    if (regexRes && !sampleIds[regexRes[2]] && sampleIds[baseSample]) {  // when number_abx does not exist but number
      dptSampleAddLog.push(`Trying to find ${baseSample}`);
      dptSampleAddLog.push(host + '/sample/' + sampleIds[baseSample]);
      res = await axios({  // get base sample
        method: 'get',
        url: host + '/sample/' + stripSpaces(sampleIds[baseSample]),
        auth: {
          username: 'admin',
          password: 'Abc123!#'
        }
      }).catch(err => {
        if (err.response) {
          console.error(err.response.data);
          errors.push(`DPT Could not fetch sample ${baseSample}: ${JSON.stringify(err.response.data)}`);
        }
      });
      if (res) {
        const data = _.merge(_.pick(res.data, ['color', 'type', 'batch']),
          {number: regexRes[2], condition: {}, notes: {}, material_id: res.data.material._id});
        res = await axios({
          method: 'post',
          url: host + '/sample/new',
          auth: {
            username: res.data.user,
            password: res.data.user === 'admin' ? 'Abc123!#' : '2020DeFinMachen!'
          },
          data
        }).catch(err => {
          if (err.response) {
            console.error(err.response.data);
            errors.push(`DPT Could not save sample ${data}: ${err.response.data}`);
          }
        });
        if (res.data) {
          dptSampleAddLog.push(`${regexRes[2]} from ${baseSample}`)
          sampleIds[regexRes[2]] = res.data._id;
        }
      }
    }
    if (regexRes && sampleIds[regexRes[2]]) {  // found matching sample
      console.log(`${i}/${dpts.length}  ${dpts[i]} -> ${regexRes[2]}`);
      dptLog.push(`${dpts[i]}, ${regexRes[2]}`);
      const f = fs.readFileSync(dptFiles + '\\' + dpts[i], 'utf-8');
      const data = {
        sample_id: sampleIds[regexRes[2]],
        values: {},
        measurement_template
      };
      data.values.device = regexRes[1];
      data.values.filename = dpts[i];
      data.values.dpt = f.split('\r\n').map(e => e.split(',').map(e => parseFloat(e)));
      let rescale = false;
      for (let i in data.values.dpt) {
        if (data.values.dpt[i][1] > 10) {
          rescale = true;
          break;
        }
      }
      if (rescale) {
        data.values.dpt = data.values.dpt.map(e => [e[0], e[1] / 100]);
      }
      await axios({
        method: 'post',
        url: host + '/measurement/new',
        auth: {
          username: 'admin',
          password: 'Abc123!#'
        },
        data
      }).catch(err => {
        console.log(dpts[i]);
        if (err.response) {
          console.error(err.response.data);
          errors.push(`Could not upload ${dpts[i]} for sample ${regexRes[2]}: ${err.response.data}`);
        }
        else {
          console.error(err);
          errors.push(`Could not upload ${dpts[i]} for sample ${regexRes[2]}: ${JSON.stringify(err)}`);
        }
      });
    }
    else {
      console.log(`Could not find sample for ${dpts[i]}`);
      if (regexRes) {
        errors.push(`Could not find sample for ${dpts[i]}; [DEBUG] ${regexRes[2]}, ${!sampleIds[regexRes[2]]}, ${sampleIds[baseSample]}`);
      }
      else {
        errors.push(`Could not find sample for ${dpts[i]} (did not match RegEx)`);
      }
    }
  }
}

async function allMcVn() {
  let res = await axios({
    method: 'get',
    url: host + '/template/measurements',
    auth: {
      username: 'admin',
      password: 'Abc123!#'
    }
  });
  const mc_template = res.data.filter(e => e.name === 'moisture content').sort((a, b) => b.version - a.version)[0]._id;
  const vn_template = res.data.filter(e => e.name === 'vn').sort((a, b) => b.version - a.version)[0]._id;
  const rmc_template = res.data.filter(e => e.name === 'reinforcement material content').sort((a, b) => b.version - a.version)[0]._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(`MC/VN ${index}/${data.length}`);
    let sample = data[index];
    sample['samplenumber'] = sample['samplenumber'].replace(/[A-Z][a-z]0\d_/, '');
    let credentials = ['admin', 'Abc123!#'];
    if (sampleDevices[sample['samplenumber']]) {
      credentials = [sampleDevices[sample['samplenumber']], '2020DeFinMachen!']
    }
    if (!sample['vz(ml/g)'] && vnValues[sample['samplenumber']]) {  // fill in VN values from comments
      sample['vz(ml/g)'] = vnValues[sample['samplenumber']];
    }
    if (sample['kfingew%']) {
      await axios({
        method: 'post',
        url: host + '/measurement/new',
        auth: {
          username: credentials[0],
          password: credentials[1]
        },
        data: {
          sample_id: sampleIds[sample['samplenumber']],
          measurement_template: mc_template,
          values: {
            'weight %': sample['kfingew%'],
            'standard deviation': sample['stabwn']
          }
        }
      }).catch(err => {
        console.log(sample['samplenumber']);
        console.error(err.response.data);
        errors.push(`MC/VN upload for ${JSON.stringify(sample)} failed: ${JSON.stringify(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['samplenumber']],
          measurement_template: vn_template,
          values: {
            vn: sample['vz(ml/g)']
          }
        }
      }).catch(err => {
        console.log(sample['samplenumber']);
        console.error(err.response.data);
        errors.push(`MC/VN upload for ${JSON.stringify(sample)} failed: ${JSON.stringify(err.response.data)}`);
      });
    }
    if (sample['reinforcingmaterialcontent']) {
      await axios({
        method: 'post',
        url: host + '/measurement/new',
        auth: {
          username: credentials[0],
          password: credentials[1]
        },
        data: {
          sample_id: sampleIds[sample['samplenumber']],
          measurement_template: rmc_template,
          values: {
            percentage: Number(sample['reinforcingmaterialcontent'].replace('%', '').replace(',', '.'))
          }
        }
      }).catch(err => {
        console.log(sample['samplenumber']);
        console.error(err.response.data);
        errors.push(`MC/VN upload for ${JSON.stringify(sample)} failed: ${JSON.stringify(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 => {
    m.numbers = m.numbers.map(e => ({number: e, color: numberToColor[e]}));
    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(`SAMPLE LOAD ${index}/${data.length}`);
    let sample = data[index];
    if (sample['granulate/Part'] === '') {  // empty supplier fields
      sample['granulate/Part'] = 'unknown';
    }
    const material = dbMaterials[trim(sample['materialname'])];
    if (!material) { // could not find material, skipping sample
      errors.push(`Could not find a material for ${JSON.stringify(sample)}`);
      continue;
    }
    samples.push({
      number: sample['samplenumber'].replace(/[A-Z][a-z]0\d_/, ''),  // remove leading An_01 and Eh_01
      type: sampleType(sample['granulate/part']),
      batch: sample['charge/batch'],
      material_id: material._id,
      notes: {
        custom_fields: customFields(sample['comments'], sample['samplenumber'])
      }
    });
    // if (sample['comments']) {
    //   comments.push(sample['samplenumber'] + '   ' + sample['comments']);
    // }
    const si = samples.length - 1;  // sample index
    if (samples[si].notes.custom_fields.hasOwnProperty('xRest')) {  // reroute xRest property to comment
      samples[si].notes.comment = samples[si].notes.custom_fields.xRest;
      commentsLog.push(sample['samplenumber'] + '   ' + samples[si].notes.comment);
      delete samples[si].notes.custom_fields.xRest;
    }
    if (Object.keys(samples[si].notes.custom_fields).length === 0) {  // delete empty custom fields
      delete samples[si].notes.custom_fields;
    }
    else {
      customFieldsLog.push(sample['samplenumber'] + '   ' + JSON.stringify(samples[si].notes.custom_fields));
    }
    if (sample['materialnumber'] !== '' && material.numbers.find(e => e.number === sample['materialnumber'])) {
      samples[si].color = material.numbers.find(e => e.number === sample['materialnumber']).color;
    }
    else if (sample['color'] !== '') {  // find color with all edge cases
      let number = material.numbers.find(e => e.color && e.color.indexOf(trim(sample['color'])) >= 0);
      if (!number && /black/.test(sample['color'])) {  // special case bk for black
        console.log(material);
        number = material.numbers.find(e => e.color && e.color.toLowerCase().indexOf('bk') >= 0);
        if (!number) {  // try German word
          number = material.numbers.find(e => e.color && e.color.toLowerCase().indexOf('schwarz') >= 0);
        }
      }
      if (number) {
        samples[si].color = number.color;
      }
    }
    else if (sampleColors[sample['samplenumber'].split('_')[0]]) {  // derive color from main sample for mc/vn
      samples[si].color = sampleColors[sample['samplenumber'].split('_')[0]];
    }
    if (!samples[si].color) {
      samples[si].color = sample['color'];
    }
  }
}

async function saveSamples() {
  for (let i in samples) {
    console.info(`SAMPLE SAVE ${i}/${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);
        errors.push(`Upload for ${JSON.stringify(samples[i])} failed: ${JSON.stringify(err.response.data)}`);
      }
    });
  }
  console.info('saved all samples');
}

async function sampleReferencesSave() {
  for (let i in sampleReferences) {
    console.info(`SAMPLE REFERENCES ${i}/${sampleReferences.length}`);
    let refRes = await axios({
      method: 'get',
      url: host + '/sample/number/' + sampleReferences[i].referencedSample,
      auth: {
        username: 'admin',
        password: 'Abc123!#'
      }
    }).catch(err => {
      console.log(sampleReferences[i].referencedSample);
      console.error(err.response.data);
      errors.push(`Getting reference id for ${JSON.stringify(sampleReferences[i])} failed: ${JSON.stringify(err.response.data)}`);
    });
    if (!refRes) continue;
    let sampleRes = await axios({
      method: 'get',
      url: host + '/sample/number/' + sampleReferences[i].sample,
      auth: {
        username: 'admin',
        password: 'Abc123!#'
      }
    }).catch(err => {
      console.log(sampleReferences[i].sample);
      console.error(err.response.data);
      errors.push(`Getting sample id for ${JSON.stringify(sampleReferences[i])} failed: ${JSON.stringify(err.response.data)}`);
    });
    if (!sampleRes) continue;
    sampleRes.data.notes.sample_references.push({sample_id: refRes.data._id, relation: sampleReferences[i].relation})
    await axios({
      method: 'put',
      url: host + '/sample/' + sampleRes.data._id,
      auth: {
        username: 'admin',
        password: 'Abc123!#'
      },
      data: {notes: {sample_references: sampleRes.data.notes.sample_references}}
    }).catch(err => {
      console.log(sampleRes.data.notes.sample_references);
      if (err.response.data) {
        console.error(err.response.data);
        errors.push(`Saving references for ${JSON.stringify(sampleRes.data)} failed: ${JSON.stringify(err.response.data)}`);
      }
      else {
        console.error(err.response);
        errors.push(`Saving references for ${JSON.stringify(sampleRes.data)} failed: ${JSON.stringify(err.response)}`);
      }
    });
  }
}

async function allMaterials() {
  // materials = {};
  let res = await axios({
    method: 'get',
    url: host + '/template/materials',
    auth: {
      username: 'admin',
      password: 'Abc123!#'
    }
  });
  const materialTemplate = res.data.filter(e => e.name === 'plastic').sort((a, b) => b.version - a.version)[0]._id;

  // process all samples
  for (let index in data) {
    let sample = data[index];
    if (sample['supplier'] === '') {  // empty supplier fields
      sample['supplier'] = 'unknown';
    }
    if (sample['materialname'] === '') {  // empty name fields
      sample['materialname'] = sample['material'];
    }
    sample['materialname'] = trim(sample['materialname']);
    if (materials.hasOwnProperty(sample['materialname'])) {  // material already found at least once
      if (sample['materialnumber'] !== '') {  // material number given
        if (materials[sample['materialname']].numbers.length === 0 || !materials[sample['materialname']].numbers.find(e => e.number === stripSpaces(sample['materialnumber']))) {  // new material number
          if (materials[sample['materialname']].numbers.find(e => e.color === sample['color'] && e.number === '')) {  // color already in list, only number missing
            materials[sample['materialname']].numbers.find(e => e.color === sample['color'] && e.number === '').number = stripSpaces(sample['materialnumber']);
          }
          else {  // completely new number entry
            materials[sample['materialname']].numbers.push({color: trim(sample['color']), number: stripSpaces(sample['materialnumber'])});
          }
        }
      }
      else if (sample['color'] !== '') {  // color given
        if (!materials[sample['materialname']].numbers.find(e => e.color === stripSpaces(sample['color']))) {  // new material color
          materials[sample['materialname']].numbers.push({color: trim(sample['color']), number: ''});
        }
      }
    }
    else {  // new material
      console.info(`MATERIAL LOAD ${index}/${data.length}  ${sample['materialname']}`);
      materials[sample['materialname']] = {
        name: trim(sample['materialname']),
        supplier: trim(sample['supplier']),
        group: trim(sample['material'])
      };
      materials[sample['materialname']].numbers = await numbersFetch(sample);

      // material properties
      materials[sample['materialname']].properties = {material_template: materialTemplate};
      let tmp = /M(\d+)/.exec(sample['reinforcingmaterial']);
      materials[sample['materialname']].properties.mineral = tmp ? tmp[1] : 0;
      tmp = /GF(\d+)/.exec(sample['reinforcingmaterial']);
      materials[sample['materialname']].properties.glass_fiber = tmp ? tmp[1] : 0;
      tmp = /CF(\d+)/.exec(sample['reinforcingmaterial']);
      materials[sample['materialname']].properties.carbon_fiber = tmp ? tmp[1] : 0;
    }
  }

  // Fill numberToColor array
  Object.keys(materials).forEach(mKey => {
    materials[mKey].numbers.forEach(number => {
      if (number.number && number.color) {
        numberToColor[number.number] = number.color;
      }
    })
  });
}

async function saveMaterials() {
  const mKeys = Object.keys(materials)
  for (let i in mKeys) {
    console.info(`MATERIAL SAVE ${i}/${mKeys.length}`);
    const material = _.cloneDeep(materials[mKeys[i]]);
    material.numbers = material.numbers.map(e => e.number).filter(e => e !== '').map(e => e.replace(/ /g, ''));
    await axios({
      method: 'post',
      url: host + '/material/new',
      auth: {
        username: 'admin',
        password: 'Abc123!#'
      },
      data: material
    }).catch(err => {
      if (err.response.data.status && err.response.data.status !== 'Material name already taken') {
        console.info(material);
        console.error(err.response.data);
        errors.push(`Upload for ${JSON.stringify(material)} failed: ${JSON.stringify(err.response.data)}`)
      }
    });
  }
  console.info('saved all materials');
}

async function numbersFetch(sample) {
  let nm = [];
  let res = [];
  if (sample['materialnumber']) {  // sample has a material number
    nm = normMaster[stripSpaces(sample['materialnumber'])]? [normMaster[stripSpaces(sample['materialnumber'])]] : [];
  }
  else {  // try finding via material name
    nm = Object.keys(normMaster).filter(e => normMaster[e].nameSpaceless === stripSpaces(sample['materialnumber'])).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) {
        errors.push(`Download of ${nm[i].url.replace(/ /g, '%20')} for material number ${sample['materialnumber']} failed`);
        errors.push(nm[i].doc.replace(/ /g, '_'));
      }
    }
  }
  if (res.length === 0) {  // no results
    if (sample['color'] !== '' || sample['materialnumber'] !== '') {  // information in data available
      return [{color: trim(sample['color']), number: sample['materialnumber']}];
    }
    else {
      return [];
    }
  }
  else {
    if (!res.find(e => e.number === sample['materialnumber'])) {  // sometimes norm master does not include sample number even if listed
      res.push({color: trim(sample['color']), number: sample['materialnumber']});
    }
    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('colordesignationsuppl') >= 0) || (stripSpaces(lastLastText + lastText + item.text).toLowerCase().indexOf('colordesignatiomsuppl') >= 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 customFields (comment, sampleNumber) {
  const customFields = [
    {docKey: 'Versuchsreihe', dbKey: 'test series', regex: /Versuchsreihe (\d+),/, category: 'customField'},
    {docKey: 'Stillstand', dbKey: 'idle', regex: /Stillstand (\d+ min):/, category: 'customField'},
    {docKey: 'Serienzyklus', dbKey: 'cycle', regex: /(\d+.) Serienzyklus (\(.*?\))/, category: 'customField'},
    {docKey: 'Berstdruck', dbKey: 'bursting pressure', regex: /Berstdruck: (.*?bar);/, category: 'customField'},
    {docKey: 'gemessen am', dbKey: 'measured at', regex: /gemessen am (.*20\d\d)/, category: 'customField'},
    {docKey: 'used for', dbKey: 'used for', regex: /used for (.*)/, category: 'customField'},
    {docKey: 'Stabilized', dbKey: 'stabilized', regex: /Stabilized, (.*)/, category: 'customField'},
    {docKey: 'parts from field', dbKey: 'parts from field', regex: null, category: 'customField'},
    {docKey: 'side', dbKey: 'side', regex: /(\S*?) side/, category: 'customField'},
    {docKey: 'Creep test', dbKey: 'creep test', regex: null, category: 'customField'},
    {docKey: 'Variante', dbKey: 'variant', regex: /(.*)/, category: 'customField'},
    {docKey: 'Parameter', dbKey: 'parameter', regex: /Parameter (\d)/, category: 'customField'},
    {docKey: 'days without cooling', dbKey: 'days without cooling', regex: /(\d+) days without cooling/, category: 'customField'},
    {docKey: 'Zyklus', dbKey: 'cycle', regex: /Zyklus (\d+ s)/, category: 'customField'},
    {docKey: 'fast cure', dbKey: 'fast cure', regex: null, category: 'customField'},
    {docKey: 'Stoff gesperrt', dbKey: 'material blocked', regex: null, category: 'customField'},
    {docKey: 'anwendungsbeschränkt', dbKey: 'limited application', regex: null, category: 'customField'},
    {docKey: 'für Neuanwendungen gesperrt', dbKey: 'blocked for new applications', regex: null, category: 'customField'},
    {docKey: 'V', dbKey: 'test', regex: /V(\d+-\d+);/, category: 'customField'},
    {docKey: 'Twz', dbKey: 'twz', regex: /Twz \(°C\): (\d+);/, category: 'customField'},
    {docKey: 'Pnach', dbKey: 'pressure after', regex: /Pnach \(bar\): (\d+);/, category: 'customField'},
    {docKey: 'Vein', dbKey: 'volume in', regex: /Vein \(ccm\/s\): (\d+)/, category: 'customField'},
    {docKey: 'low emission', dbKey: 'low emission', regex: /low emission (\S+)[;]?/, category: 'customField'},
    {docKey: 'aus', dbKey: 'from', regex: /aus (.*)/, category: 'customField'},
    {docKey: 'Erprobung', dbKey: 'trial', regex: /Erprobung (.*?);/, category: 'customField'},
    {docKey: 'Auftragsnummer', dbKey: 'job number', regex: /Auftragsnummer: (\S+)[;]?/, category: 'customField'},
    {docKey: 'Wärmealterung', dbKey: 'heat aging', regex: /Wärmealterung: (.*)/, category: 'customField'},
    {docKey: 'A: Wandung außen / I: Wandung innen / S: Wandung Steg', dbKey: 'outer wall', regex: /Steg.*?A: (\d+)/, category: 'customField'},
    {docKey: 'A: Wandung außen / I: Wandung innen / S: Wandung Steg', dbKey: 'inner wall', regex: /Steg.*?I: (\d+)/, category: 'customField'},
    {docKey: 'A: Wandung außen / I: Wandung innen / S: Wandung Steg', dbKey: 'support wall', regex: /Steg.*?S: (\d+)/, category: 'customField'},
    {docKey: 'A: Wandung außen / I: Wandung innen / S: Wandung Steg', dbKey: 'outer wall degraded', regex: /Degradation:.*?A: (\d+)/, category: 'customField'},
    {docKey: 'A: Wandung außen / I: Wandung innen / S: Wandung Steg', dbKey: 'inner wall degraded', regex: /Degradation:.*?I: (\d+)/, category: 'customField'},
    {docKey: 'A: Wandung außen / I: Wandung innen / S: Wandung Steg', dbKey: 'support wall degraded', regex: /Degradation:.*?S: (\d+)/, category: 'customField'},
    {docKey: 'Reines Polymer', dbKey: 'pure polymer', regex: null, category: 'customField'},
    {docKey: 'Rücksendung erforderlich', dbKey: 'return needed', regex: /(.*?,) Rücksendung erforderlich, (.*)/, category: 'customField'},
    {docKey: 'Prio', dbKey: 'priority', regex: /Prio (\d+)/, category: 'customField'},
    {docKey: 'beanstandet', dbKey: 'faulty', regex: null, category: 'customField'},
    {docKey: 'aged', dbKey: 'aged', regex: /aged: (.*)/, category: 'customField'},
    {docKey: 'DOPPELT!!', dbKey: 'double', regex: null, category: 'customField'},
    {docKey: 'Bauteil', dbKey: 'construction part', regex: /Bauteil (\S+)/, category: 'customField'},
    {docKey: 'T =', dbKey: 'temperature', regex: /T = (\S+)/, category: 'customField'},
    {docKey: 'nicht vorgealtert', dbKey: 'not preaged', regex: /nicht vorgealtert (.*)/, category: 'customField'},
    {docKey: 'TS119', dbKey: 'TS119', regex: /TS119 (W\S+);/, category: 'customField'},
    {docKey: 'GF vom Datenblatt', dbKey: 'glass fibre from data sheet', regex: null, category: 'customField'},
    {docKey: 'nach Datensatz', dbKey: 'according to dataset', regex: null, category: 'customField'},
    {docKey: 'Dosiergeschw', dbKey: 'metering speed', regex: /Dosiergeschw.*? (.*?min)/, category: 'customField'},
    {docKey: 'Einspritzgeschw', dbKey: 'injection speed', regex: /Einspritzgeschw.*? (.*\/s)/, category: 'customField'},
    {docKey: 'Heizbänder', dbKey: 'heating lines', regex: /Heizbänder (.*)/, category: 'customField'},
    {docKey: 'Verweilzeit', dbKey: 'dwell time', regex: /Verweilzeit (.*?min)/, category: 'customField'},
    {docKey: 'Probe', dbKey: 'belongs to', regex: /Probe (\S*\d+)/, category: 'reference'},
    {docKey: 'zu', dbKey: 'belongs to', regex: /zu (\S*\d+)/, category: 'reference'},
    {docKey: 'granulate zu', dbKey: 'granulate to', regex: /granulate zu.* (\S*\d+)/, category: 'reference'},
    {docKey: 'construction part', dbKey: 'construction part', regex: /(?<!granulate)construction part.* (\S*\d+)/, category: 'reference'},
    {docKey: 'VZ =', dbKey: 'vn', regex: /VZ = (\d+) cm³\/g/, category: 'vn'},
    {docKey: 'VWZ', dbKey: 'vwz', regex: /(\d+ min) VWZ \//, category: 'customField'},
    {docKey: 'VZ:', dbKey: 'vn', regex: /VZ: ([0-9.,]+) mL\/g[;]?/, category: 'vn'}
  ];
  const res = {};  // returned result
  const usedParts = [];  // all substrings used for custom fields, subtract at the end, as some parts are used multiple times
  customFields.forEach(cField => {
    if (comment.indexOf(cField.docKey) >= 0) {  // comment contains docKey
      if (cField.regex !== null) {
        const regexRes = cField.regex.exec(comment);
        if (regexRes) {
          usedParts.push(regexRes[0]);
          if (cField.category === 'reference') {
            sampleReferences.push({sample: sampleNumber, referencedSample: regexRes[1], relation: cField.dbKey});
          }
          else if (cField.category === 'vn') {
            vnValues[sampleNumber] = regexRes[1];
          }
          else {
            res[cField.dbKey] = regexRes.filter((e, i) => i > 0).join(' ');
          }
        }
      }
      else {
        usedParts.push(cField.docKey);
        res[cField.dbKey] = true;
      }
    }
  });
  usedParts.forEach(part => {
    const index = comment.indexOf(part);
    if (index >= 0) {
      comment = comment.slice(0, index) + comment.slice(index + part.length);
    }
  });
  if (/\w+/.test(comment)) {
    res.xRest = comment;
  }
  return res;
}

function sampleType (type) {
  type = stripSpaces(type).toLowerCase();
  const allowedTypes = {'tension rod': 'tension rod', 'Zugstab': 'tension rod', 'part': 'part', 'granulate': 'granulate'};
  if (!allowedTypes[type]) {
    typeLog.push(type);
  }
  return allowedTypes[type] ? allowedTypes[type] : 'part';
}

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

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