sakti-parser-fa/dist/parser.js

Simple TypeScript library for parsing Indonesian government budget Excel files (Sakti FA)

import ExcelJS from 'exceljs';
/**
 * Universal SAKTI Excel parser that handles files directly
 */
export class SaktiParser {
    /**
     * Parse SAKTI Excel file from various input types
     */
    async parse(input) {
        const worksheet = await this.loadWorksheet(input);
        const metadata = this.extractMetadata(worksheet);
        const data = this.extractData(worksheet);
        return { metadata, data };
    }
    /**
     * Load worksheet from various input types
     */
    async loadWorksheet(input) {
        const workbook = new ExcelJS.Workbook();
        if (typeof input === 'string') {
            // File path (Node.js)
            await workbook.xlsx.readFile(input);
        }
        else if (input instanceof File) {
            // File object (Browser)
            const arrayBuffer = await input.arrayBuffer();
            await workbook.xlsx.load(arrayBuffer);
        }
        else if (input instanceof ArrayBuffer) {
            // ArrayBuffer (Browser/Node.js)
            await workbook.xlsx.load(input);
        }
        else {
            throw new Error('Unsupported input type. Use file path (string), File object, or ArrayBuffer.');
        }
        const worksheet = workbook.getWorksheet(1);
        if (!worksheet) {
            throw new Error('No worksheet found in the Excel file.');
        }
        return worksheet;
    }
    extractMetadata(worksheet) {
        const getValue = (address) => {
            const cell = worksheet.getCell(address);
            return this.getCellValue(cell)?.toString() || '';
        };
        // Extract metadata from standard SAKTI header positions
        const judul = getValue('B1');
        const periodeText = getValue('B2');
        const kementerianInfo = getValue('B3');
        const unitInfo = getValue('B4');
        const satuanKerjaInfo = getValue('B5');
        // Parse periode
        const { bulan, tahun } = this.parsePeriode(periodeText);
        // Parse kementerian
        const { kode: kementerianKode, nama: kementerianNama } = this.parseKodeNama(kementerianInfo);
        // Parse unit
        const { kode: unitKode, nama: unitNama } = this.parseKodeNama(unitInfo);
        // Parse satuan kerja
        const { kode: satuanKerjaKode, nama: satuanKerjaNama } = this.parseKodeNama(satuanKerjaInfo);
        return {
            judul,
            periode_text: periodeText,
            periode_bulan: bulan,
            periode_tahun: tahun,
            kementerian_kode: kementerianKode,
            kementerian_nama: kementerianNama,
            unit_kode: unitKode,
            unit_nama: unitNama,
            satuan_kerja_kode: satuanKerjaKode,
            satuan_kerja_nama: satuanKerjaNama
        };
    }
    extractData(worksheet) {
        const data = [];
        const maxRow = worksheet.rowCount;
        // Find data start row (after header)
        let dataStartRow = 8; // Default, adjust based on actual structure
        for (let rowNumber = dataStartRow; rowNumber <= maxRow; rowNumber++) {
            const row = worksheet.getRow(rowNumber);
            // Skip empty rows
            if (this.isEmptyRow(row))
                continue;
            const parsedRow = this.parseDataRow(row, rowNumber);
            if (parsedRow) {
                data.push(parsedRow);
            }
        }
        return data;
    }
    parseDataRow(row, rowNumber) {
        const getValue = (col) => {
            const cell = row.getCell(col);
            return this.getCellValue(cell)?.toString() || '';
        };
        const getNumber = (col) => {
            const cell = row.getCell(col);
            const value = this.getCellValue(cell);
            return typeof value === 'number' ? value : 0;
        };
        // Determine level based on indentation or cell position
        const level = this.determineLevel(row);
        // Extract kode and uraian from appropriate columns
        const kodeCol = this.findKodeColumn(row, level);
        const uraianCol = this.findUraianColumn(row, level);
        const kode = getValue(kodeCol);
        const uraian = getValue(uraianCol);
        if (!kode && !uraian)
            return null;
        // Initialize data structure
        const parsedData = {
            _level: level,
            _parent: this.determineParent(level),
            mata_anggaran: '',
            level_1_kode: '', level_1_uraian: '',
            level_2_kode: '', level_2_uraian: '',
            level_3_kode: '', level_3_uraian: '',
            level_4_kode: '', level_4_uraian: '',
            level_5_kode: '', level_5_uraian: '',
            level_6_kode: '', level_6_uraian: '',
            level_7_kode: '', level_7_uraian: '',
            level_8_kode: '', level_8_uraian: '',
            level_9_kode: '', level_9_uraian: '',
            level_10_kode: '', level_10_uraian: '',
            level_11_kode: '', level_11_uraian: '',
            nilai_rupiah: getNumber(this.findNilaiColumn(row)),
            nilai_total: getNumber(this.findNilaiColumn(row))
        };
        // Set the appropriate level data
        parsedData[`level_${level}_kode`] = kode;
        parsedData[`level_${level}_uraian`] = uraian;
        return parsedData;
    }
    getCellValue(cell) {
        if (!cell || cell.value === null || cell.value === undefined) {
            return null;
        }
        // Handle rich text
        if (cell.value && typeof cell.value === 'object' && 'richText' in cell.value) {
            return cell.value.richText.map((rt) => rt.text).join('');
        }
        // Handle formulas
        if (cell.value && typeof cell.value === 'object' && 'result' in cell.value) {
            return cell.value.result;
        }
        return cell.value;
    }
    parsePeriode(periodeText) {
        // Parse month and year from periode text
        const match = periodeText.match(/(\\d{1,2}).*?(\\d{4})/);
        if (match) {
            return {
                bulan: parseInt(match[1], 10),
                tahun: parseInt(match[2], 10)
            };
        }
        return { bulan: 0, tahun: 0 };
    }
    parseKodeNama(info) {
        // Parse "KODE - NAMA" format
        const parts = info.split(' - ');
        return {
            kode: parts[0]?.trim() || '',
            nama: parts[1]?.trim() || ''
        };
    }
    isEmptyRow(row) {
        return !row.hasValues;
    }
    determineLevel(row) {
        // Simple level determination based on cell positions
        // This is a simplified approach - adjust based on actual SAKTI format
        for (let col = 1; col <= 15; col++) {
            const cell = row.getCell(col);
            if (this.getCellValue(cell)) {
                return Math.min(col, 11); // Max 11 levels
            }
        }
        return 1;
    }
    findKodeColumn(row, level) {
        // Find the column containing the kode for this level
        return level; // Simplified
    }
    findUraianColumn(row, level) {
        // Find the column containing the uraian for this level  
        return level + 1; // Simplified
    }
    findNilaiColumn(row) {
        // Find the rightmost column with numeric values
        for (let col = row.cellCount; col >= 1; col--) {
            const cell = row.getCell(col);
            const value = this.getCellValue(cell);
            if (typeof value === 'number') {
                return col;
            }
        }
        return row.cellCount; // Default to last column
    }
    determineParent(level) {
        // Simplified parent determination
        return level > 1 ? `level_${level - 1}` : '';
    }
}
//# sourceMappingURL=parser.js.map