CopyPruneSmart PDF Text CleanerOpen cleaner
Private-by-design text cleanupNo upload · no signup

Clean text copied
from a PDF.

Restore paragraphs, join broken wraps, repair safe hyphenation, and review every change—right in your browser.

Clean copied text See the method
Prune formatting noise—not your words.
Beforecopied from PDF

Research notes often keep the visual
line wraps used on a printed page.

Afterproofed by CopyPrune

Research notes often keep the visual line wraps used on a printed page.

Cleanup rules 6/6 on
RulesChanges rerun from source
SourceUntouched input
0 words · 0 chars
Your text is processed in this browser.
CleanedReady for result
0 words · 0 chars
Nothing is changed until you choose Clean text.

Paste copied PDF text to begin.

Proof report

See what changed—and what did not.

The report comes from the same deterministic operations that created the result. It does not infer changes after the fact.

0wraps joined
0split words repaired
0structure markers preserved
0needs review
No proof report yet

Load the example or paste your own copied text, then choose Clean text.

Browser-only document handling

We do not intentionally send your pasted text to our servers, analytics, URLs, or local storage. Clear removes it from this page’s runtime state.

01Untouched source

Every rule reruns from what you pasted.

02Visible restraint

Uncertain boundaries stay for review.

03Exact handoff

Copy and download match the result.

How careful cleanup works

A proofing desk,
not a guessing box.

CopyPrune follows a deterministic set of readable rules. The cleaner protects obvious structure, makes only the repairs allowed by your mode and settings, and builds the proof report from the actual operations.

1

Paste the source

Copy selectable text from a PDF reader and paste it into the Source pane. The original remains untouched while you compare results.

2

Choose the restraint

Careful is the default. Standard joins more probable wraps. Flatten is available when you genuinely want one continuous block.

3

Review the proof

See joined wraps, repaired words, normalized glyphs, preserved structure, and every boundary the careful mode declined to guess about.

Transformation pipeline
Normalize→Classify lines→Protect structure→Apply repairs→Build ledger
Read the full methodology →
Native before-and-after examples

Four common PDF copy problems.

The examples below describe exactly what the launch version handles. CopyPrune does not upload PDFs, perform OCR, or reconstruct a multi-column page.

01

Visual line wraps

Before
A printed page wraps text at a fixed
width even when the paragraph continues.
After
A printed page wraps text at a fixed width even when the paragraph continues.

High-confidence wraps become spaces; real blank-line paragraphs remain.

02

Split words

Before
The report demon-
strated a repeatable result.
After
The report demonstrated a repeatable result.

Only confident end-of-line splits lose the hyphen. Uncertain compounds stay visible for review.

03

Lists and headings

Before
NEXT STEPS
1. Review the text
2. Copy the result
After
NEXT STEPS
1. Review the text
2. Copy the result

Obvious structure is protected instead of being flattened into prose.

04

Ligatures and spacing

Before
The final file contains hidden spacing.
After
The final file contains hidden spacing.

Common PDF ligatures and invisible formatting characters become ordinary text.

Three modes

Use the least force that solves the problem.

Careful mode protects uncertainty. The stronger modes exist for messier source text, but they should receive more human review.

Recommended

Careful

Joins high-confidence visual wraps, repairs safer split words, and preserves likely headings, lists, paragraphs, quotes, and table-like rows.

Best for quotations, research notes, and ordinary prose.
More active

Standard

Joins more single line boundaries while still protecting the clearest document structure. Its changes are labeled medium confidence.

Best when the copied paragraph is visibly fragmented.
Destructive structure

Flatten

Replaces every line and paragraph break with a space. The words remain, but all layout structure is intentionally removed.

Best only when one continuous text block is required.
What CopyPrune fixes—and what it cannot

Built for copied text,
not PDF archaeology.

PDFs describe visual placement, not always the reading structure a writer expects. CopyPrune works after a PDF reader has already provided selectable text.

See every rule and limitation →
Handles
  • Visual prose line wraps
  • High-confidence split words
  • Common typographic ligatures
  • Soft hyphens and zero-width characters
  • Repeated spacing outside table-like rows
Does not handle
  • Scanned pages or OCR
  • PDF uploads or extraction
  • Multi-column reading order
  • Tables reconstructed from coordinates
  • Headers and footers without page boundaries
Document handling

Your pasted text stays in this browser.

CopyPrune’s cleaning engine runs on your device. We do not intentionally send document text to our servers, put it in URLs, retain it in local storage, or include it in analytics events.

Google Analytics loads only after a visitor accepts analytics. It may receive ordinary device, network, and usage information, but never the source or cleaned text.

Read the privacy policy →
FAQ

Before you paste.

Can CopyPrune clean an uploaded PDF?+

No. The launch version intentionally accepts only text you have already copied. This avoids misleading promises about OCR, reading order, tables, and scanned pages.

Will it preserve every paragraph perfectly?+

No tool can recover every PDF’s intended structure from copied text alone. Careful mode protects blank-line paragraphs and obvious structure, then flags uncertain boundaries instead of silently joining them.

Does it use AI?+

No. CopyPrune uses deterministic browser-based rules. The same source, mode, and rule settings produce the same output and proof ledger.

Can I use confidential or regulated documents?+

CopyPrune does not claim compliance or suitability for confidential legal, medical, financial, or regulated information. Follow your organization’s data-handling requirements.

What happens when I choose Clear?+

After confirmation, the source, cleaned result, and proof report are removed from this page’s runtime state. Document text is not saved in browser storage.

Method record

Version 1.0 · July 22, 2026

The cleaner is an editorial heuristic, not a language model or laboratory-validated reconstruction system. Its rule definitions and limitations are published so results can be evaluated honestly.

Technical references: Unicode normalization and browser clipboard behavior.