import { Component, OnInit } from '@angular/core';
import { ChartOptions } from 'chart.js';
import { ApiService } from '../services/api.service';
import { animate, style, transition, trigger } from '@angular/animations';
import cloneDeep from 'lodash/cloneDeep';
import omit from 'lodash/omit';
import { DataService } from '../services/data.service';
import { LoginService } from '../services/login.service';
import { ModelItemModel } from '../models/model-item.model';
import * as FileSaver from 'file-saver';
import * as pdfMake from "pdfmake/build/pdfmake";
import * as pdfFonts from 'pdfmake/build/vfs_fonts';

(<any>pdfMake).vfs = pdfFonts.pdfMake.vfs;


@Component({
  selector: 'app-prediction',
  templateUrl: './prediction.component.html',
  styleUrls: ['./prediction.component.scss'],
  animations: [
    trigger(
      'inOut', [
        transition(':enter', [
          style({height: 0, opacity: 0}),
          animate('0.5s ease-out', style({height: '*', opacity: 1}))
        ]),
        transition(':leave', [
          style({height: '*', opacity: 1}),
          animate('0.5s ease-in', style({height: 0, opacity: 0}))
        ])
      ]
    )
  ]
})
export class PredictionComponent implements OnInit {

  result: { predictions: any[], mean: any[] };  // Prediction result from python container
  loading = false;
  activeGroup: ModelItemModel = new ModelItemModel();
  activeModelIndex = 0;
  // If true, spectra belong to different samples, otherwise multiple spectra from the same sample are given
  multipleSamples = false;
  spectrumNames: string[] = [];
  spectrum: string[][] = [[]];
  flattenedSpectra = [];
  chart = [];
  readonly chartInit = {
    data: [],
    label: 'Spectrum',
    showLine: true,
    fill: false,
    pointRadius: 0,
    borderColor: '#00a8b0',
    borderWidth: 2
  };
  readonly chartOptions: ChartOptions = {
    scales: {
      xAxes: [{ticks: {min: 400, max: 4000, stepSize: 400, reverse: true}}],
      yAxes: [{ticks: {}}]
    },
    responsive: true,
    tooltips: {enabled: false},
    hover: {mode: null},
    maintainAspectRatio: true,
    plugins: {datalabels: {display: false}}
  };

  constructor(
    private api: ApiService,
    public d: DataService,
    public login: LoginService
  ) {
    this.chart[0] = cloneDeep(this.chartInit);
  }

  ngOnInit(): void {
    this.d.load('modelGroups', () => {
      this.activeGroup = this.d.arr.modelGroups[0];
    });
  }

  fileToArray(files) {
    this.loading = true;
    this.flattenedSpectra = [];
    this.chart = [];
    let load = files.length;
    this.spectrumNames = files.map(e => e.name);
    for (const i in files) {
      if (files.hasOwnProperty(i)) {
        const fileReader = new FileReader();
        fileReader.onload = () => {
          // Parse to database spectrum representation
          this.spectrum = fileReader.result.toString().split('\r\n').map(e => e.split(',').map(el => parseFloat(el)))
            .filter(el => el.length === 2) as any;
          // Flatten to format needed for prediction
          this.flattenedSpectra[i] = {labels: this.spectrum.map(e => e[0]), values: this.spectrum.map(e => e[1])};
          // Add to chart
          this.chart[i] = cloneDeep(this.chartInit);
          this.chart[i].data = this.spectrum.map(e => ({x: parseFloat(e[0]), y: parseFloat(e[1])}));
          load --;
          if (load <= 0) {  // All loaded
            this.loadPrediction();
          }
        };
        fileReader.readAsText(files[i]);
      }
    }
  }

  loadPrediction() {
    this.loading = true;
    this.api.post<any>(this.activeGroup.models[this.activeModelIndex].url, this.flattenedSpectra, data => {
      let tmp = Object.entries(omit(data, ['mean', 'std', 'label']))  // Form: [[label, [{value, color}]]]
          .map((entry: any) => entry[1].map(e => ({category: entry[0], label: data.label[entry[0]], value: e.value, color: e.color})));  // Form: [[{category, label, value, color}]]
      this.result = {
        predictions: tmp[0].map((ignore, columnIndex) => tmp.map(row => row[columnIndex])),  // Transpose tmp
        mean: Object.entries(data.mean)
          .map((entry:any) => ({category: entry[0], label: data.label[entry[0]], value: entry[1].value, color: entry[1].color, std: data.std[entry[0]]}))  // Form: [{category, label, value, color}]
      };
      this.loading = false;
    });
  }

  groupChange(index) {  // Group was changed
    this.activeGroup = this.d.arr.modelGroups[index];
    this.activeModelIndex = 0;
    this.result = undefined;
  }

  // Aggregates spectrum names and prediction values into an associative array
  prepareExport() {
    const zip = (a, b) => a.map((k, i) => [k, b[i]]);
    const values = this.result.predictions
      .map(prediction => prediction
           .filter(field => field.category === 'Prediction')
           .map(field => field.value));
    return zip(this.spectrumNames, values);
  }

  // Converts the prediction results to a CSV file
  exportCSV() {
    const predictions = this.prepareExport();
    const csv = predictions.map(line => line.join(";")).join("\n");
    FileSaver.saveAs(new Blob([csv], { type: 'text/csv;charset=utf-8' }), "predictions.csv");
  }

  // Converts the prediction results to a PDF file
  exportPDF() {
    const dd = {
      content: [
        {
          text: 'DeFinMa - Decoding the Fingerprint of Materials by AI',
          style: 'header'
        },
        {
          table: {
            body: [
              [new Date().toLocaleDateString(), 'Security class', 'Person in charge', 'Phone'],
              [this.activeGroup.group, 'Intern',
                {
                  stack: [
                    'CR/APS1-Lotter',
                    'CR/APS1-Lingenfelser'
                  ]
                },
                {
                  stack: [
                    '0711/811-49017',
                    '0711/811-6897'
                  ]
                }
              ]
            ]
          }
        },
        {
          bold: true,
          text: 'Customer'
        },
        this.login.username,
        {
          text: 'Prediction of ' + this.activeGroup.group + ' (' + this.activeGroup.models[this.activeModelIndex].name + ')*',
          style: 'subheader'
        },
        {
          text: this.result.mean.map(e => e.category + ' ' + e.value + ' ' + e.label + ' ' + (e.std !== '' ? (' (standard deviation: ' + e.std + ')') : ''))
        },
        {
          table: {
            widths: ['*', 'auto'],
            body: [
              ['Input Data / Sample Name', 'Prediction*']
            ].concat(this.prepareExport())
          }
        },
        {
          text: 'Reference Data',
          style: 'subheader'
        },
        {
          image: document.getElementsByTagName('canvas')[0].toDataURL('image/png'),
          width: 500
        },
        {
          text: '*Disclaimer: This tool is still under development and Testing',
          style: 'subsubheader'
        },
        {
          text: [
            'The prediction and classification of material parameters are validated only for certain conditions.',
            'These results may therefore under no circumstances be used to evaluate quality-relevant issues.',
            'For more details please contact ',
            {
              text: 'CR/APS1-Lingenfelder',
              link: 'mailto:dominic.lingenfelser@bosch.com'
            },
            '.'
          ]
        },
        {
          table: {
            body: [
              ['Pr\u00fcfung', 'Freigabe', 'Datum'],
              ['CR/APS1-Lotter', 'CR/APS1-Lingenfelser', new Date().toLocaleDateString()]
            ]
          }
        },
        '\u00a9 Alle Rechte bei Robert Bosch GmbH, auch f\u00fcr den Fall von Schutzreichtsanmeldungen. Jede Verf\u00fcgungsbefugnis, wie Kopier- und Weitergaberecht, bei uns.'
      ],
      styles: {
        header: {
          fontSize: 18,
          bold: true
        },
        subheader: {
          fontSize: 15,
          bold: true
        },
        subsubheader: {
          bold: true
        }
      }
    };

    pdfMake.createPdf(dd).download();
  }
}