Because I'm sick of having to watch Mikha do it by hand! Uses ball-drop technique.

pdf-ball-drop

// ==UserScript==
// @name         PDF column extractor
// @description  extracts PDF pages for Mikha
// @namespace    http://tampermonkey.net/
// @version      2024-06-13
// @author       Patrick Bowen
// @match        https://mozilla.github.io/pdf.js/web/viewer.html
// @grant        none
// ==/UserScript==

let hue = 1;
const button = document.createElement("button");
let ySortedPage = [];
let ySiftedPage = [];
let columns = [];
let column = [];
let ys = [];
let stepTimer = null;

function step(x) {
    if (!ySortedPage.length) {
        columns.push(column);
        column = [];
        ySortedPage = ySiftedPage;
        ySiftedPage = [];
        console.log(columns);
        stepTimer = null;
        button.removeAttribute("disabled");
        return;
    }

    const span = ySortedPage.shift();
    const rect = span.getBoundingClientRect();
    let y = rect.y;
    ys.push(y);

    const isObstacle = rect.x < x && (rect.x + rect.width) > x;
    if (isObstacle) {
        const toLeft = Math.abs(rect.x - x);
        const toRight = Math.abs((rect.x + rect.width) - x);
        const goThrough = toLeft > 32 && toRight > 32;
        if (!goThrough) {
            if (toLeft < toRight) {
                x -= toLeft;
            } else {
                x += toRight;
            }
        }
        y += rect.height;
    }
    if ((rect.x + rect.width / 2) < x) {
        column.push(span);
        span.style.color = `hsl(${(hue % 6) * 60}, 100%, 50%)`;
    } else {
        ySiftedPage.push(span);
    }

    stepTimer = setTimeout(() => step(x), 1);
}


let clickTimer;
const click = ({ clientX }) => {
    if (stepTimer !== null) return;
    clickTimer = setTimeout(() => {
        button.setAttribute("disabled", "disabled");
        ++hue;
        stepTimer = step(clientX);
    }, 250);
};

function activate() {
    const pageWithDupes = [...document.querySelectorAll("span[role=presentation]")];
    pageWithDupes.forEach(span => {
        span.style.color = "";
    });

    //Spacial deduplication
    const dedupedPage = (spans => {
        const page = [];
        spans.forEach(span => {
            if (!span.innerText.trim()) return;
            const similar = page.some(s => {
                if (s.innerText !== span.innerText) return false;
                const a = s.getBoundingClientRect();
                const b = span.getBoundingClientRect();
                const t = 10;
                return Math.abs(a.x - b.x) < t && Math.abs(a.y - b.y) < t;
            });
            if (!similar) page.push(span);
        });
        return page;
    })(pageWithDupes);

    ySortedPage = dedupedPage.toSorted((a, b) => {
        return a.getBoundingClientRect().y - b.getBoundingClientRect().y;
    });

    document.body.addEventListener('click', click, true);
    button.innerHTML = "Copy to clipboard";
    button.removeEventListener('click', activate);
    button.addEventListener('click', done);
}

function done() {
    clearTimeout(clickTimer);
    document.body.removeEventListener('click', click, {capture: true});
    button.removeEventListener('click', done);
    button.innerHTML = "Activate";
    button.addEventListener('click', activate);

    //Group ys into rows of 10px each
    const px = 10;
    const rowYs = [0];
    for (const y of ys) {
        if (rowYs.at(-1) + px < y) {
            rowYs.push(y - 1);
        }
    }

    const rows = rowYs.map(y => {
        const cells = [];
        for (const column of columns) {
            //Find the ideal span in this column for this row Y
            const cellIdx = column.findIndex(s => {
                const Y = s.getBoundingClientRect().y;
                return Y >= y && Y < (y + px);
            });
            if (cellIdx === -1) {
                cells.push("");
            } else {
                cells.push(column[cellIdx].innerText);
                column.splice(cellIdx, 1);
            }
        }
        return cells.join("\t");
    });
    const doc = rows.join("\n");
    navigator.clipboard.writeText(doc);
}

function reset() {
    ySortedPage = [];
    ySiftedPage = [];
    columns = [];
    column = [];
    ys = [];
    button.innerHTML = "Activate";
    button.style.position = "absolute";
    button.style.left = 0;
    button.style.bottom = 0;
    button.style.fontSize = "1.5rem";
    button.addEventListener('click', activate);
    document.body.appendChild(button);
}

(function() {
    reset();
})();