'use strict';

var pdf_mjs = require('pdfjs-dist/legacy/build/pdf.mjs');
var fs = require('fs');

class Cell {
  constructor(options = {}) {
    this.options = options;
    this.text = "";
    this.x1 = 9999;
    this.y1 = 9999;
    this.x2 = 0;
    this.y2 = 0;
    this.fontHeight = 8;
    this.fontWidth = 4;
    this.lineHeightRatio = options.lineHeight || 1.67;
    this.count = 0;
    this.prevX = 0;
    this.prevY = 0;
    this.prevX2 = 0;
    this.prevY2 = 0;
    this.hasSpan = false;
    this.inserted = false;
    this.id = 0;
  }
  get lineHeight() {
    return this.fontHeight * this.lineHeightRatio;
  }
  addItem(item, index) {
    this.count++;
    this.id = index;
    if (item.str) this.text += item.str;
    if (item.hasEOL) this.text += this.options.newlines ? "\n" : " ";
    const x = item.transform[4];
    const y = item.transform[5];
    const w = item.width;
    const h = item.height;
    if (x < this.x1) this.x1 = x;
    if (y < this.y1) this.y1 = y;
    if (x + w > this.x2) this.x2 = x + w;
    if (y + h > this.y2) this.y2 = y + h;
    const fh = item.transform[0];
    const fw = item.str ? item.width / item.str.length : 0;
    if (fh > this.fontHeight) this.fontHeight = fh;
    if (fw > this.fontWidth) this.fontWidth = fw;
    this.prevX = x;
    this.prevY = y;
    this.prevX2 = x + w;
    this.prevY2 = y + h;
  }
  isSameLine(cell) {
    let same = 0;
    if (cell.y1 - 1 > this.y2)
      same = 1;
    else if (cell.y2 + 1 < this.y1)
      same = -1;
    return same;
  }
  isOutputLine(cell) {
    let yOverlaps = cell.y1 >= this.y1 && cell.y1 <= this.y2 || this.y1 >= cell.y1 && this.y1 <= cell.y2;
    if (yOverlaps) {
      if (this.x1 < cell.x1 && this.y2 - this.y1 < cell.y2 - cell.y1)
        yOverlaps = false;
    }
    return yOverlaps;
  }
  isAdjacent(item) {
    const x = item.transform[4];
    const y = item.transform[5];
    const w = item.width;
    item.height;
    if (Math.abs(y - this.prevY) <= this.lineHeight * 0.125 && x - this.prevX2 < this.fontWidth)
      return true;
    if (this.hasSpan && this.prevY - y > this.lineHeight * 0.75 && this.prevY - y <= this.lineHeight * 1.25 && (x >= this.x1 && x <= this.x2 || this.x1 >= x && this.x1 <= x + w))
      return true;
    return false;
  }
  alignment(item) {
    const aligns = {
      top: false,
      bottom: false,
      left: false,
      right: false,
      adjacent: false
    };
    if (this.count === 0) return aligns;
    const x = item.transform[4];
    const y = item.transform[5];
    if (Math.abs(y - this.y1) <= 2) aligns.bottom = true;
    if (Math.abs(y + item.height - this.y2) <= 2) aligns.top = true;
    if (Math.abs(x - this.x1) <= 2) aligns.left = true;
    if (Math.abs(x + item.width - this.x2) <= 2) aligns.right = true;
    if ((aligns.top || aligns.bottom) && Math.abs(x - this.x2) < this.fontWidth)
      aligns.adjacent = true;
    if ((aligns.left || aligns.right) && Math.abs(y + item.height - this.y1) < this.fontWidth)
      aligns.adjacent = true;
    return aligns;
  }
}

const parseMarkedPage = async (page, options) => {
  let cell = null;
  let markedContent = "";
  let paragraph = false;
  let span = false;
  let cells = [];
  let content = await page.getTextContent({
    includeMarkedContent: true,
    disableNormalization: false,
    disableCombineTextItems: false
  });
  content.items.forEach((item, index) => {
    if (item.type === "beginMarkedContent") {
      switch (item.tag) {
        case "Artifact":
          markedContent = "Artifact";
          cells.push(cell);
          cell = null;
          break;
        default:
          console.warn("unknown content tag: " + item.tag);
      }
    } else if (item.type === "beginMarkedContentProps") {
      switch (item.tag) {
        case "P":
          markedContent = "P";
          paragraph = true;
          break;
        case "Span":
          markedContent = "Span";
          span = true;
          break;
      }
    } else if (item.type === "endMarkedContent") {
      switch (markedContent) {
        case "Artifact":
          if (options.artifacts) cells.push(cell);
          cell = null;
          break;
      }
      markedContent = "";
    } else if (item.type) {
      console.warn("Warning: unknown content type: " + item.type);
    } else {
      if (item.dir !== "ltr")
        console.warn("Warning: text direction is: " + item.dir);
      if (paragraph || span) {
        if (item.str === "" && item.width === 0 && paragraph && item.hasEOL)
          return;
        if (item.str === " " && (paragraph || item.width > (cell == null ? void 0 : cell.fontWidth)))
          return;
        if (cell && cell.count > 0) {
          cell.hasSpan = cell.hasSpan || span;
          if (!cell.isAdjacent(item)) {
            cells.push(cell);
            cell = null;
          }
        }
      }
      if (!cell) cell = new Cell(options);
      cell.addItem(item, index);
      paragraph = false;
      span = false;
    }
  });
  if (cell) cells.push(cell);
  if ((options == null ? void 0 : options.sort) === "Y1") {
    cells.sort((a, b) => b.y1 - a.y1);
  } else if ((options == null ? void 0 : options.sort) === "Y2") {
    cells.sort((a, b) => b.y2 - a.y2);
  }
  return cells;
};

const parseLinedPage = async (page, options) => {
  let cell = new Cell(options);
  let wasEOL = false;
  let cells = [];
  let content = await page.getTextContent({
    disableNormalization: false
  });
  content.items.forEach((item, index) => {
    if (item.dir !== "ltr")
      console.warn(item.dir);
    let aligns = cell.alignment(item);
    if (!aligns.adjacent && cell.count > 0) {
      cells.push(cell);
      cell = new Cell(options);
    }
    if (wasEOL && (aligns.top || (aligns.left || aligns.right) && aligns.adjacent)) {
      wasEOL = false;
    }
    if (wasEOL && cell.count > 0) {
      cells.push(cell);
      cell = new Cell(options);
    }
    if (item.height > 0 || item.str === " " && item.width < (cell == null ? void 0 : cell.fontWidth))
      cell.addItem(item, index);
    wasEOL = item.hasEOL;
  });
  if (cell.count > 0) cells.push(cell);
  if ((options == null ? void 0 : options.sort) === "Y1") {
    cells.sort((a, b) => b.y1 - a.y1);
  } else if ((options == null ? void 0 : options.sort) === "Y2") {
    cells.sort((a, b) => b.y2 - a.y2);
  }
  return cells;
};

function parseLineRange(line) {
  const ranges = line.split(",");
  const result = [];
  for (const range of ranges) {
    const [start, end] = range.split("-").map(Number);
    if (isNaN(start) || end !== void 0 && isNaN(end)) {
      throw new Error(`Invalid line range: ${range}`);
    }
    if (end !== void 0) {
      result.push(
        ...Array.from({ length: end - start + 1 }, (_, i) => start + i)
      );
    } else {
      result.push(start);
    }
  }
  return result;
}
function getSkipLines(options, currentPage) {
  var _a, _b, _c;
  let skipLines = [];
  if ((_a = options.skip) == null ? void 0 : _a.global) {
    if (typeof options.skip.global.lines === "string") {
      skipLines = parseLineRange(options.skip.global.lines);
    } else {
      skipLines = options.skip.global.lines;
    }
  }
  if ((_b = options.skip) == null ? void 0 : _b.pageSpecific) {
    const page = (_c = options.skip.pageSpecific) == null ? void 0 : _c.find(
      (page2) => page2.page === currentPage
    );
    if (page) {
      if (typeof page.lines === "string") {
        skipLines = parseLineRange(page.lines);
      } else {
        skipLines = page.lines;
      }
    }
  }
  return skipLines;
}
function checkTextMatch(text, matchText, matchType) {
  switch (matchType) {
    case "exact":
      return text === matchText;
    case "contain":
      return text.includes(matchText);
    case "startWith":
      return text.startsWith(matchText);
    case "regex":
      return new RegExp(matchText).test(text);
    // return RegExp(matchText).test(text);
    default:
      return false;
  }
}
function shouldDeleteItem(item, textFilters, nextItem) {
  if (!textFilters) return false;
  return textFilters.some(({ value, match = "contain", nextLine }) => {
    const matchF = checkTextMatch(item.text, value, match);
    return matchF && (!nextLine || nextItem && checkTextMatch(nextItem.text, nextLine.match, nextLine.match));
  });
}
function filter(items, options, currentPage) {
  var _a, _b, _c, _d;
  const skipLines = getSkipLines(options, currentPage);
  const cells = items.filter((item, index) => {
    var _a2;
    const currentLine = index + 1;
    if (skipLines == null ? void 0 : skipLines.includes(currentLine)) {
      return false;
    }
    const nextItem = index + 1 < items.length ? items[index + 1] : void 0;
    return !shouldDeleteItem(item, (_a2 = options.skip) == null ? void 0 : _a2.text, nextItem);
  });
  const page = (_b = (_a = options.skip) == null ? void 0 : _a.pageSpecific) == null ? void 0 : _b.find(
    (page2) => page2.page === currentPage
  );
  if (page == null ? void 0 : page.lastLines) {
    cells.splice(-page.lastLines);
  } else if ((_d = (_c = options == null ? void 0 : options.skip) == null ? void 0 : _c.global) == null ? void 0 : _d.lastLines) {
    cells.splice(-options.skip.global.lastLines);
  }
  let isSkipping = false;
  return cells.map((item) => {
    var _a2;
    if ((_a2 = options == null ? void 0 : options.skip) == null ? void 0 : _a2.ranges) {
      for (const range of options.skip.ranges) {
        const startMatch = checkTextMatch(
          item.text,
          range.start.value,
          range.start.match || "contain"
        );
        if (startMatch) {
          isSkipping = true;
        }
        const endMatch = checkTextMatch(
          item.text,
          range.end.value,
          range.end.match || "contain"
        );
        if (endMatch) {
          isSkipping = false;
          return null;
        }
      }
    }
    if (!isSkipping) {
      return item.text.trim();
    }
  }).filter(Boolean).join("\n");
}

const parsePagesOption = (pagesOption, totalPages) => {
  const pages = /* @__PURE__ */ new Set();
  if (Array.isArray(pagesOption)) {
    pagesOption.forEach((page) => {
      if (page >= 1 && page <= totalPages) {
        pages.add(page);
      }
    });
  } else if (typeof pagesOption === "string") {
    if (pagesOption.trim() === "all") {
      for (let i = 1; i <= totalPages; i++) {
        pages.add(i);
      }
    } else {
      const ranges = pagesOption.split(",");
      ranges.forEach((range) => {
        const [start, end] = range.split("-").map(Number);
        if (end) {
          for (let i = start; i <= end; i++) {
            if (i >= 1 && i <= totalPages) {
              pages.add(i);
            }
          }
        } else if (start >= 1 && start <= totalPages) {
          pages.add(start);
        }
      });
    }
  }
  return [...pages];
};

const pdf = async (pdfPath, options = {
  threshold: 5,
  lineHeight: 1.67,
  pages: "1"
}) => {
  const finalOptions = Object.assign(
    {
      threshold: 5,
      lineHeight: 1.67,
      pages: "1",
      sort: "Y2"
    },
    options
  );
  const raw = [];
  const text = [];
  const textContentArray = [];
  const data = new Uint8Array(fs.readFileSync(pdfPath));
  try {
    const doc = await pdf_mjs.getDocument({
      data,
      standardFontDataUrl: "../node_modules/pdfjs-dist/standard_fonts/",
      verbosity: 0,
      password: finalOptions.password
    }).promise;
    const markInfo = await doc.getMarkInfo();
    const numPages = doc.numPages;
    const pagesToProcess = parsePagesOption(finalOptions.pages, numPages);
    console.log("Extracting...");
    for (const i of pagesToProcess) {
      console.log("Page", i, "of", numPages);
      const page = await doc.getPage(i);
      const textContent = await page.getTextContent();
      const pageTextContent = textContent.items.map((item) => item.str).join("\n");
      textContentArray.push(pageTextContent);
      let temp = [];
      if (markInfo == null ? void 0 : markInfo.Marked) {
        temp = await parseMarkedPage(page, finalOptions);
      } else {
        temp = await parseLinedPage(page, finalOptions);
      }
      raw.push(
        ...temp.map((cell) => ({
          text: cell.text,
          x1: cell.x1,
          y1: cell.y1,
          x2: cell.x2,
          y2: cell.y2,
          fontHeight: cell.fontHeight,
          fontWidth: cell.fontWidth
        }))
      );
      text.push(filter(temp, finalOptions, i));
    }
    console.log("done");
    return {
      getRaw: () => raw,
      getText: () => text,
      getPages: () => numPages,
      getTextContent: () => textContentArray
    };
  } catch (e) {
    console.error("Error reading pdf", e);
    throw e;
  }
};

exports.pdf = pdf;