@earvinpiamonte/pagasa-tcb-parser/dist/utils/text-utils.js

A TypeScript library for parsing PAGASA weather bulletin PDF files

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.extractMunicipalities = exports.splitPreservingParentheses = void 0;
const patterns_1 = require("../constants/patterns");
const splitPreservingParentheses = (text) => {
    const result = [];
    let current = "";
    let parenthesesDepth = 0;
    let i = 0;
    while (i < text.length) {
        const char = text[i];
        const nextChars = text.slice(i, i + 5).toLowerCase();
        if (char === "(") {
            parenthesesDepth++;
            current += char;
        }
        else if (char === ")") {
            parenthesesDepth--;
            current += char;
        }
        else if (parenthesesDepth === 0 && (char === "," || char === ";")) {
            if (current.trim()) {
                result.push(current.trim());
            }
            current = "";
        }
        else if (parenthesesDepth === 0 && nextChars === " and ") {
            if (current.trim()) {
                result.push(current.trim());
            }
            current = "";
            i += 4;
        }
        else {
            current += char;
        }
        i++;
    }
    if (current.trim()) {
        result.push(current.trim());
    }
    return result.filter((segment) => segment.length > 0);
};
exports.splitPreservingParentheses = splitPreservingParentheses;
const extractMunicipalities = (areaText) => {
    const municipalities = [];
    let match;
    // Reset pattern to start from beginning
    patterns_1.PATTERNS.parentheses.lastIndex = 0;
    while ((match = patterns_1.PATTERNS.parentheses.exec(areaText)) !== null) {
        const munis = match[1]
            .split(",")
            .map((m) => m.trim())
            .filter((m) => m.length > 0 && !/^\d+$/.test(m));
        municipalities.push(...munis);
    }
    const cleanName = areaText
        .replace(/\s*\([^)]*\)/g, "")
        .replace(patterns_1.PATTERNS.cleanExtra, "")
        .replace(patterns_1.PATTERNS.restPattern, "")
        .replace(/^the\s+/i, "")
        .trim();
    return { name: cleanName, municipalities };
};
exports.extractMunicipalities = extractMunicipalities;
//# sourceMappingURL=text-utils.js.map