Compare two PDF documents side by side, detect modified text, added or removed pages, and visual differences locally in your browser.
Client-Side PDF Comparison Studio examines two versions of a PDF document directly within your browser memory using PDF.js and a Longest Common Subsequence (LCS) line- and word-level diffing engine. It allows contract reviewers, legal auditors, engineers, and students to verify revisions, spot stealthy clause modifications, and track page insertions or deletions without transmitting sensitive documents across third-party cloud networks.
The console offers three complementary inspection modes: Side-by-Side (synchronized dual page rendering), Text Diff (semantic line-by-line and word-level additions in green and deletions in red with line numbering), and Visual Overlay (pixel-level canvas comparison highlighting spatial rendering changes in magenta). Pages are processed lazily on demand to manage device memory, and an integrated technical summary report (.txt) can be exported locally.
Concrete Scenario: A legal counsel receives an updated 12-page commercial service agreement. Instead of manually proofreading all 12 pages against the signed draft, the counsel drops both files into Document A and Document B. In 340 milliseconds, the tool flags that pages 1 through 7 and 9 through 12 are identical, while page 8 contains a modified liability limitation clause changing '$50,000' to '$5,000', with 1 line modified (+2 words, -2 words). The counsel reviews the side-by-side view, exports the diff audit log, and resolves the issue immediately.
Technical Scope & Boundaries: Semantic text comparison requires extractable text streams and valid font encodings (ToUnicode CMaps). Scanned or image-only pages lack text streams; the tool explicitly identifies them as scanned pages and offers visual canvas comparison or redirects to OCR. Visual pixel differences do not represent semantic text meaning and may be affected by scanner skew or raster compression artifacts.
LCS Edit Script: dp[i][j] = (a[i] === b[j]) ? dp[i+1][j+1] + 1 : max(dp[i+1][j], dp[i][j+1])
Dynamic programming matrix calculates the Longest Common Subsequence of text tokens, producing minimal edit scripts of additions, deletions, and unchanged lines.