Find and replace text in a PDF, every page in one pass

Open the PDF and replace in 1 pass

Edit PDFReal text editing, objects, annotations, forms. Free, runs on your device.Open

1. Drop the PDF on the editor above, or open /edit.
2. Press Ctrl+F to open the find bar (Cmd+F on a Mac). Ctrl+H opens it with the replace row already showing; otherwise click Replace… in the bar.
3. Type the text to find. Every match on every page is highlighted, and the counter shows where you are, for example “1 of 500”.
4. Type the new text in Replace with.
5. Choose Replace all, or step through the hits first (next section).
6. Save. The edited file downloads with “-edited” added to its name.

We ran this on our 500-page test file, which says “Page 1 of 500” to “Page 500 of 500”: the search for “of 500” found 500 matches, Replace all changed every one to “of five hundred”, and after saving, a search in the new file found the new wording 500 times and the old wording 0 times. The Replace all took about 2.3 seconds on a laptop.

Check each hit before you change it

Replace all is right when every occurrence is wrong. When some must stay (a quoted original, a citation, a name in a list of former staff), go one at a time:

1. Press Enter to jump to the next match and Shift+Enter to go back. The page scrolls to each one and the current match is marked.
2. For a match that must change, choose Replace. That one changes, and the counter moves on to the next.
3. For a match that must stay, press Enter to step past it.

There is no separate skip button; stepping past a match is the skip. If you replaced one too many, Ctrl+Z puts it back.

Match case and whole words: why “May” needs both

Two switches sit in the find bar:

Switch	What it does	When you need it
Aa (match case)	“may” no longer finds “May”	Names and months that are also ordinary words
ab (whole words only)	“Pag” no longer finds the “Pag” inside “Page”	Short words that live inside longer ones
Neither	Every occurrence of the letters, anywhere	Codes and numbers like an invoice number

On the 500-page file, “Pag” finds 500 matches with whole words off (the start of every “Page”) and 0 with it on. Searching for the month “May” with both switches on finds the month and leaves “maybe” and “Mayfield” alone.

The search is PDFium’s own, the same engine Chrome uses to read PDFs, so it finds what a reader’s search would find, including a word the author’s software split into 2 pieces.

What a replacement does to the layout

A PDF page is not a word processor page. Each line of text is placed at a fixed point, and the replacement takes the old text’s place on that same line:

• Same length or shorter: nothing visible moves except the changed words.
• Longer: the line grows to the right from where it starts. It does not wrap onto the next line, and the next line does not move down. A short name swapped for a longer one usually fits; “Jane Cooper” to “Jana Čermák” is a non-event. A word swapped for a whole sentence can run into the margin or into text beside it.
• Empty replacement: the matched text is deleted, and a line left with nothing in it is taken off the page.

After a Replace all with a longer replacement, step through the hits once and look. For long rewrites, edit the line by hand in the editor; Edit a PDF shows how, headers and footers included.

Fonts

When the file’s font holds every letter the new text needs, the replacement is set in that font. PDFs often carry only the letters the author used, so a name with a letter the original never had (an “č” in an English document, say) is set in a bundled font of the same family instead: Liberation Serif, Sans or Mono for the common Times, Arial and Courier shapes. Liberation is built to the same letter widths as those fonts, so spacing holds, and a note on screen names the font it used; Edit a PDF explains what you see when it happens. Liberation covers Latin, Greek, Cyrillic and Vietnamese. A replacement in a script none of the bundled fonts carries (Arabic, Hebrew, Chinese, Japanese, Korean, Thai, Indic scripts) is refused with a message naming the script, rather than written as empty boxes.

The limits, honestly

Scanned pages have nothing to find. A scan is a photo of text. The counter says “No matches” because there are none. Run OCR PDF first; the scanned PDF guide covers what recognised text looks like when edited.

Some matches are found but not changed. The replace skips a match when its text sits inside a group (some authoring tools wrap a logo, a chart or a whole header into 1 grouped object), when the file writes it letter by letter instead of as a run of text, or when it belongs to an OCR layer. It counts them and says why, for example “21 replaced; 2 could not be changed here, because the text sits inside a group, is written letter by letter, or is an OCR layer. Click into those to edit them by hand.” Step to each with Enter and fix it on the page: double-click the line to retype it, and where a click selects a whole group or the scan picture instead, cover the old words with a white rectangle and add new text, or fix the source file.

Text drawn as outlines is not text. Some designers convert headings to shapes before export. Those letters are pictures of letters, and no search finds them.

No regular expressions. Find matches exactly what you type, with the 2 switches above. There is no wildcard or pattern search.

Replacing reaches the text, not everything else. The editor writes the saved file anew, so a replaced name leaves no earlier copy of the page behind: we changed an invoice number on our test file, saved, and the old number was nowhere in the file’s bytes. The exception is a digitally signed file. To keep its signature valid, the editor appends the changes after the signed bytes and says so on saving; the signed version, with the old wording, is still inside. Beyond the page text, a name can also sit in the document’s properties (author, title), in bookmarks, in a picture or in a form field, and find reaches none of those. If a person must disappear from a file, use Redact PDF for what is on the pages, clear the properties with Remove PDF metadata, and read the privacy and protection guide first.

One more boundary: correct documents you are allowed to change. A signed contract or an official certificate should be reissued by whoever issued it.

When the source file is the better fix

If you have the Word or Google Docs file the PDF came from, its find and replace handles wrapping, page flow and every running header, and exporting a fresh PDF costs a minute. Use the PDF route when the source is lost, belongs to someone else, or the change is small enough that re-exporting would cost more. Everything else about changing a PDF lives in the editing guide.