Batch replacement with a keyword table
Short answer: Put the words to change into a find-and-replace table, import it, and every keyword is found in one pass. Each scanned page is recognized only once, and all replacements work from the original text, so replaced text is never changed a second time.
Three ways to import
- Excel (.xlsx): column A is the text to find, column B is the replacement, one pair per row. The first row can be a header. Cells are read as they are, so spaces, commas and quotes are kept. “Export…” saves the current list as an Excel file you can use as a template.
- Paste two columns copied from Excel or WPS: the two columns are separated by a tab automatically, and the content is kept as it is.
- Text or CSV: one pair per row, with a separator between the find and replace text. The default separator is ”=>”. CSV follows the standard rules: a cell that contains a comma must be in double quotes.
For format details, see Keyword table format. You can also prepare the table in the online keyword table editor and export it.
All keywords in one pass
Each scanned page is recognized once, and then all keywords are searched together. A 13-page file with 60 keywords takes about ten to twenty seconds. When the search is done, the results list shows every match by page, and you can untick any of them. The summary lists the keywords that were not found.
Replacement rules
- At the same time, no chaining: every replacement works from the original text. With both “A → B” and “B → C”, an original A becomes B and an original B becomes C.
- Longer keywords first: when keywords overlap, the longer one is used.
- Conflicts are errors: if the same find text has two different replacements, an error is shown. Keep only one.
- Deletion: leave the replacement empty to delete the word.
Multiple rounds
After saving, click “Continue on result” to run another set of keywords on the output you just saved. Each round has its own operation record (manifest.json).
Batch processing a folder
The batch window accepts files or whole folders (including subfolders). It supports PDF and JPG, PNG, TIFF, BMP and WebP images:
- Each file is saved separately to the output folder with a suffix added to its name. If a file with the same name exists, a number is added. Nothing is overwritten.
- Risk rules apply to each file: official or certified files are skipped, and signed or issued files are processed only if you tick a separate option.
- Files with no matches, a wrong password or an error are listed, and the other files continue.
- At the end, a “processing report_date.xlsx” file is created in the output folder. It lists each file’s result, number of replacements, read-back verification, risk level and time taken.
The command line supports the same workflow. See Command-line mode.
FAQ
What happens if the table has both "A → B" and "B → C"?
All replacements work from the original text at the same time, so they do not chain. An original A becomes B, an original B becomes C, and a B produced by a replacement does not become C.
Which keyword wins when one contains another?
The longer keyword. For example, with both "3456 → 6789" and "123456 → 00000", an original 123456 becomes 00000. It is not first changed to 126789 by 3456.
What if the keyword table has a mistake?
The editor checks the table as you type. Rows with errors are marked in red with the reason. If error rows remain when you run Find, you are asked to skip them or go back and fix them. Nothing is silently split the wrong way.
Can it process several files?
Yes. The batch window accepts files or whole folders. Each file is saved separately, and a processing report is created at the end.