PasteCleanly / Text workshop

Compare two texts by line, word or character

Paste the original and revised versions, then see what changed: whole lines, individual words, punctuation or spaces.

Text and opened files stay in this tab. How privacy works

Try a sample

Original
Friday delivery
Room 204
Bring a copy of the agenda.
Changed lines
− Room 204
+ Room 208
+ Sign in at the front desk.

Paste the earlier version in Original and the updated version in Revised text. Choose Lines, Words or Characters under Compare by. In line mode, + marks an addition and − marks a removal. The other modes show additions underlined and removals crossed out, next to their context.

Choose the right level of detail

  • Lines show which paragraphs, records or code lines changed. Editing one letter can make a whole line appear as removed and added.
  • Words isolate changed words, punctuation and spaces. If “Review the draft” becomes “Review the plan”, the unchanged lead-in stays visible while the two nouns are marked. The count includes punctuation and whitespace segments, not just words.
  • Characters help find a single digit, a repeated space or a punctuation change. In supported browsers an accented letter or a joined emoji sequence is compared as a single visible unit; older browsers compare code points instead.

Read the space and punctuation changes

A changed space appears as ␠, a tab as ⇥, a line break as ↵, and a nonbreaking space as ⍽. For example, switching from Room 208 to Room 208 can be invisible in a document. Character mode marks the replacement even though the wording looks the same. Line endings from Windows and Unix are treated alike; other Unicode differences, such as two ways of encoding an accented letter, are not normalized away.

Copy the report when you need a record

Use Show changes only when you want to skip unchanged text. Copy or download a plain-text report. In word and character modes, changed segments are quoted in the report so tabs, line breaks and repeated spaces stay identifiable. A long preview may be shortened on screen; the export keeps all the segments in the selected view.

How very large changes are handled

When there is too much changed text to align in detail, the editor groups the changed area and says so. If the combined input exceeds the detailed mode limit (60,000 UTF-16 units), it shows a line report instead. Compare smaller corresponding sections if you need a more precise result. A diff reports what changed, not whether the edit was intentional.

Review edits without losing the original wording

Use line mode to find the area, then zoom in

Start with Lines when you need to locate a change in a long draft. An added instruction, a deleted blank line and a rewritten heading will each appear in the report with nearby context. The whole line can be marked even if just one digit changed. That is useful for finding where to look but not necessarily for deciding which characters were edited.

Take a short pair like Please confirm Room 204 by Friday and Please confirm Room 208 by Thursday. In line mode it is easy to find the changed instruction. In word mode you can identify both the room number and the day while keeping the unchanged wording around them. In character mode you can see the difference between 204 and 208 at the last digit. The most detailed mode is not always easiest to read: replacing Friday with Thursday can produce a fine-grained alignment that shares letters even though the actual edit was a replacement of a word.

If the work is important, compare the whole document by lines and take the relevant lines into a second comparison for detail. That makes the output easier to explain to a reviewer. Write down both the changed wording and its purpose: for example, “Meeting moved to room 208 on Thursday.” The diff describes a change, not the decision that caused it.

Try the same edit in all three modes

Prepare equivalent sections before comparing

Place the version you consider the baseline in Original and the candidate in Revised text. Reverse the two and an insertion becomes a removal, so label the versions before copying them. If one version includes an entire article and the other only a section, the difference will include the missing context. Select equivalent sections to avoid burying the real edit in a much larger report.

A paste from a document may contain hard line breaks that a different paste from the same document does not. In that case the line report can show broad replacements even when the words are the same. Decide whether the line structure is part of the thing you are reviewing. For prose copied from a PDF, you can create two normalized working copies with Remove line breaks, then compare those copies. Keep the unmodified originals if the spacing or line layout is evidence you might need later.

A formatting-only change such as bolding part of a sentence may not show up at all. The editor compares text, not font, color, track-changes metadata or link targets that were not pasted. If the actual change concerns document styles or the target of a hyperlink, inspect the source application. A clean text diff is not proof that two formatted files are identical.

Tell invisible text changes from rendering differences

If two strings look the same and the report shows a change, switch to Characters and inspect the marked position. A tab, a normal space and a nonbreaking space can have different effects on wrapping or matching. The detailed view marks changed whitespace with symbols. It does not change the input or turn all spaces into a different character. The exported text quotes segments so you can tell where significant spaces were included.

An apparent change to an accented letter can also be a change of representation. The same visible accented character may be written as one code point or as a base letter plus a combining mark. The editor preserves that distinction unless you normalize the inputs first. Use the character inspector for a small pair and the Unicode normalizer if your task specifically calls for treating equivalent forms alike. Compare raw versions first when exact strings are important.

The tool treats common Windows and Unix line endings alike. If you are trying to verify exact bytes, encoding or line-ending conventions in an exported file, use a file comparison tool for that part of the review. The text report answers what changed in the content you supplied, subject to that line-ending behavior.

Read removals, additions and the context around them

A removal represents text from the original that was not matched in the revised version. An addition represents text from the revised version without a matched counterpart. Moving a paragraph can look like a removal in one place and an addition in another. Repeated lines or words may allow more than one plausible alignment. Read the location and surrounding wording before deciding whether the report represents a move, a replacement or two unrelated edits.

Use Show changes only when you need a compact report of a change you have already located. Leave it off for the initial review if the context tells you which of several identical lines was edited. If you share a report with the context hidden, say which section or version it belongs to. Otherwise a short removal can be ambiguous.

The detailed counts include comparison units, not editorial actions. A single human decision to rewrite a sentence may show as several added and removed words, including spaces or punctuation. A zero result means the selected text matched under the tool's comparison behavior; it does not tell you that the latest version was approved. For a task like checking whether two unordered groups contain the same IDs, use Compare two lists instead and decide independently whether duplicates matter.

Use a diff as a second check after a bulk replacement

Before making an edit, save a source copy. Then run the intended change in Find and replace on a working copy. Paste the before and after versions into the comparison tool. Start with an ordinary occurrence, one at the beginning of a line and one next to punctuation. Check for an unintended change inside a longer word, in a link, or in an identifier that should have remained unchanged.

A correct replacement count is helpful but not sufficient. Three changes can be the right number while still changing one wrong occurrence and missing one intended occurrence. Read the edits in context and examine the first and last change before approving the result. If you cannot explain what a highlighted piece represents, go back to the original and verify the rule before accepting a revised file.

Save a short summary that names the intentional change and any exceptions. For example, say that a product's visible name changed but its internal SKU did not. The guide to reviewing text edits gives a broader workflow for checking a replacement, comparing versions and handing off an editable result. Copy or download a report if you need it, and keep the two originals as the reference instead of using the report as a replacement for either version.

Review long inputs in corresponding pieces

When a comparison is too large for precise alignment, the editor indicates that it grouped a changed region or fell back to a line report. Do not read a grouped block as evidence that every word inside it changed. Take a section bounded by a heading, paragraph marker or other stable reference and compare that same section in both versions. This gives the algorithm less unrelated text to align.

For a review of several sections, use the same naming scheme on both sides and start with any sections that appear only in one version. For sections present in both, record which parts were checked and with what granularity. A count of differences across multiple partial comparisons is less useful than an explicit list of the changes a person should review.

The screen can abbreviate a long report while the export contains the full selected report. Before forwarding it, check that it was generated using the view you intended. A copy with unchanged context is easier to verify; a changes-only version is easier to skim. If two reports describe the same edit, label the shorter one as a summary rather than letting another person assume it contains the full text.