How to split text into chunks without separating emoji and accents

Choose the unit before setting the size

A limit of one hundred characters can mean different things in different applications. Snipshed’s counter measures code points; its splitter works with visual characters or lines. One visual character may contain several code points and occupy several UTF-16 units. Bytes and language-model tokens are different measurements again.

Check the destination’s counting rule before using chunks to meet a limit. A hundred visual characters do not guarantee a hundred bytes, UTF-16 units or tokens. If the receiving application has its own counter, check the final chunk there. Do not choose a size only from how long a word appears on screen, and keep the original until the receiving application has accepted the intended text.

Split without separating an emoji’s components

In character mode, the browser segments complete visual clusters. The sequence woman, zero-width joiner and laptop forms 👩‍💻 and stays together. An e followed by U+0301 stays with its combining accent even though the accent is stored as a separate code point. Words themselves may still be split between two chunks.

The sample A👩‍💻éBC contains five visual characters, eight code points and ten UTF-16 units. With size two it produces three chunks: A👩‍💻, éB and C. The last chunk may be shorter. The Unicode inspector exposes the components; it does not normalize input or automatically replace a combining-accent representation with another spelling.

1. Open the Unicode sample and copy its contents without adding lines.
2. Paste it into Split text into chunks.
3. Choose visual characters and size 2.
4. Check all three chunks and the complete emoji.

Interpret the downloaded JSON array

Output is a JSON array of strings, rather than several independent files. Quotes, commas and brackets belong to the array representation. Line breaks are written as escaped sequences within strings. To paste one chunk into a form, use its contents rather than the quotes or escape characters from the JSON document.

To reconstruct character-mode input, read the array as JSON and concatenate its strings with no separator. Do not replace every backslash as plain text: some backslashes may belong to the original input. Use Format JSON to inspect the structure if needed. Keeping the JSON file preserves chunk boundaries for review or further processing, while a plain concatenation deliberately removes those boundaries.

[
  "A👩‍💻",
  "éB",
  "C"
]

Group lines while retaining empty ones

In line mode, size two groups two consecutive lines, including empty ones. With A, a blank line, B and C, the first chunk contains A and a trailing line break; the second contains B and C separated by a line break. A blank line is not discarded merely because it contains no letters.

This mode normalizes CRLF and CR endings to LF. To reconstruct normalized text, read the JSON strings and join them with one LF between chunks. Exact original file bytes are not preserved. If a PDF introduced breaks halfway through sentences, decide whether to clean them first; grouping lines does not automatically rebuild paragraphs or recover the layout from which the text was copied.

Check limits and recover the right text

Input supports up to 100000 UTF-16 units. Chunk size must be between 1 and 10000 elements, and up to 1000 chunks are allowed. If that count is exceeded, increase the size or use a smaller input. Switching to lines may change the result because it changes the unit being counted.

Calculate, inspect the first chunk, a middle boundary and the last chunk, then download when satisfied. Editing text or settings invalidates the previous output. For a complete character-mode check, compare reconstructed text with the original. In line mode, compare with the LF-normalized version. Retain the source if you need its original encoding or line endings; the result array does not record those file properties.