Table.getRow() crashes with TypeError when findBottomLineIndex returns -1

[zoe606]

When parsing PDFs with incomplete table borders (e.g., partial lines or missing cell boundaries), Table.getRow() crashes with:

TypeError: Cannot read properties of undefined (reading 'from')
  at Table.getRow (src/pdf-parse/geometry/Table.ts:278)
  at Table.toData (src/pdf-parse/geometry/Table.ts:312)
  at LineStore.getTableData (src/pdf-parse/geometry/LineStore.ts:34)
  at PDFParse.getTable (src/pdf-parse/PDFParse.ts:746)

Root cause

findBottomLineIndex() returns -1 when no matching horizontal line is found at a given x-coordinate. getRow() then accesses this.hLines[-1] which is undefined, causing the crash on bottomHLine.from.y.

// Table.ts — getRow()
const bottomHLineIndex = this.findBottomLineIndex(h2Index, xMiddle);
const bottomHLine = this.hLines[bottomHLineIndex]; // hLines[-1] = undefined
// ...
maxXY: new Point(rightVLine.from.x, bottomHLine.from.y), // 💥 crash

Suggested fix

Add a guard when findBottomLineIndex returns -1:

const bottomHLineIndex = this.findBottomLineIndex(h2Index, xMiddle);
if (bottomHLineIndex === -1) continue; // skip cell with missing bottom border
const bottomHLine = this.hLines[bottomHLineIndex];

This skips individual cells where the bottom border line is missing, while other cells, rows, and tables parse normally.

Environment

• pdf-parse v2.4.5
• Node.js (server-side, NestJS)
• Multiple real-world PDFs with varying page counts (10–87 pages)

Related

• getTable returns empty result #48 (getTable returns empty result — different root cause)
• getTables returns empty tables #82 (getTables returns empty tables — different root cause)