@tricoteuses/arbre-de-la-loi/lib/cleaners.mjs

Generate ASTs from the French bills & laws; manipulate & export them to Markdown, etc.

import deepEqual from "deep-equal";
import { HastType } from "./hast";
export function mergeSpanElementsWithSameAttributes(node) {
    let children = node.children;
    if (children !== undefined) {
        let previous = children[0];
        let index = 1;
        for (const child of [...children.slice(1)]) {
            if (previous.type === HastType.Element &&
                previous.tagName === "span" &&
                previous.children.length === 1 &&
                previous.children[0].type === HastType.Text &&
                child.type === HastType.Element &&
                child.tagName === "span" &&
                child.children.length === 1 &&
                child.children[0].type === HastType.Text &&
                deepEqual(previous.properties, child.properties)) {
                previous.children[0].value += child.children[0].value;
                children.splice(index, 1);
            }
            else {
                previous = child;
                index += 1;
            }
        }
        for (const child of children) {
            mergeSpanElementsWithSameAttributes(child);
        }
    }
}
export function mergeTexts(node) {
    let children = node.children;
    if (children !== undefined) {
        let previous = children[0];
        let index = 1;
        for (const child of [...children.slice(1)]) {
            if (previous.type === HastType.Text &&
                child.type === HastType.Text) {
                previous.value += child.value;
                children.splice(index, 1);
            }
            else {
                previous = child;
                index += 1;
            }
        }
        for (const child of children) {
            mergeTexts(child);
        }
    }
}
/// Useful for documents generated from PDF.
/// Example: https://www.assemblee-nationale.fr/dyn/opendata/PRJLANR5L15B0235.html
/// where each character is a span with a letter-spacing style.
export function removeLetterSpacingStyleAttributes(html) {
    return html.replace(/;?\s*letter-spacing:\s*(-?\d+(\.\d*)?\s*pt|normal)/gm, "");
}
export function removeEmptyStyleAttributes(html) {
    return html.replace(/\s*style=""/g, "");
}
export function removeStyleElements(node) {
    let children = node.children;
    if (children !== undefined) {
        ;
        node.children = children = children.filter((child) => child.type !== HastType.Element ||
            child.tagName !== "style");
        for (const child of children) {
            removeStyleElements(child);
        }
    }
}
export function replaceSpanElementsContainingOnlySpacesWithTexts(node) {
    let children = node.children;
    if (children !== undefined) {
        ;
        node.children = children = children.map((child) => child.type === HastType.Element &&
            child.tagName === "span" &&
            child.children.length === 1 &&
            child.children[0].type === HastType.Text &&
            child.children[0].value.match(/^[ \t\n\f\r\xa0]*$/)
            ? child.children[0] : child);
        for (const child of children) {
            replaceSpanElementsContainingOnlySpacesWithTexts(child);
        }
    }
}
export function replaceSpanElementsWithoutPropertiesWithTexts(node) {
    let children = node.children;
    if (children !== undefined) {
        ;
        node.children = children = children.map((child) => child.type === HastType.Element &&
            child.tagName === "span" &&
            (child.properties === undefined || Object.keys(child.properties).length === 0) &&
            child.children.length === 1 &&
            child.children[0].type === HastType.Text
            ? child.children[0] : child);
        for (const child of children) {
            replaceSpanElementsWithoutPropertiesWithTexts(child);
        }
    }
}