#Text Splitter

The Text Splitter chunks long text for use cases like preparing LLM context (ChatGPT, DeepSeek, Claude), document editing, and social media content prep. It offers four splitting modes — symbol, length, Chinese paragraph, English paragraph — plus batch export and ZIP packaging.

#Four splitting modes

Modes are picked via the Segmented at the top of the Configuration card, labelled "Symbol / Length / Chinese / English" — hover any label for its full description. Each mode only shows its own config, so the panels stay focused.

#1. Symbol (default)

Pure delimiter splitting: the text is cut at every delimiter occurrence, the delimiter stays at the end of its segment, and no character limit applies.

• Delimiters: required — they decide where the cuts happen
• Separators per segment: defaults to 1 (cut at every delimiter). Set 2 to merge two adjacent units into one segment, 3 for three, and so on. Useful when you want sentence-based splitting but 2–3 sentences per segment

Repeated or line-leading delimiters never produce empty segments — stray delimiters are folded into the neighbouring segment, so exporting the merged text reproduces your input.

#2. Length

Chunks text into blocks of a given character count. Default is 2000 (minimum 1) — a reasonable block size for LLM context windows. The delimiter field is optional; filling it in is what activates it, no extra toggle needed:

• Delimiter filled: chunk by length, with boundaries snapped to the nearest delimiter so sentences aren't cut mid-way
• Delimiter empty: hard chunk by length, ignoring sentence boundaries

The delimiter input accepts:

• Manual entry, e.g. 。 ？ ！ (separate multiple delimiters with a space)
• Click "CJK" to fill 。 ？ ！
• Click "Latin" to fill . ? !
• Escape sequences: \n, \r, \t, \s, \\

Delimiter splits keep the delimiter at the end of each segment.

#3. Chinese

Detects Chinese paragraph boundaries using punctuation and line-break rules; paragraph interiors are preserved. Good for articles, blog posts, and other content where paragraph integrity matters. No extra config.

#4. English

Uses the compromise NLP library's English sentence algorithm to detect boundaries, one sentence per segment. Good for documents, emails, and papers. No extra config.

#Shared settings

All modes share these:

• Hide results: when there are many segments (auto-triggers above 500) enable this to hide the preview and keep the page responsive. Export still works.
• Export filename: base name for exported files. Defaults to the uploaded file's name, or split_text if nothing was uploaded. .txt suffix is added unless you type another.

#Export

• Export merged text: combine all segments into a single file separated by double newlines.
• Batch export: ≤ 3 segments download individually; > 3 are packaged as a ZIP automatically.
• Per-segment copy: each card has a copy button; the icon flips to a checkmark on success.

#Common patterns

• LLM context prep: Length mode, 2000–4000 chars with delimiters 。 ？ ！ or . ? ! — boundaries auto-snap to sentence ends
• Sentence-level split: Symbol mode with delimiters 。 ？ ！ (or . ? !), interval 1
• Three sentences per segment: Symbol mode with those delimiters and separators per segment set to 3
• Paragraph-level split: Chinese or English mode
• Code / Markdown split: Length mode with delimiter \n\n (literal \n supported)

Everything runs in the browser — no uploads, no server round trips.