Generated text
Edge-case strings are test data. Some of them look like markup, JSON, paths or SQL on purpose, and they stay plain text in every view and format.
Click a line to copy it. Use the pin to keep it in Saved.
lore-gen [preview|code|items] themes amount formatEdge-case strings are test data. Some of them look like markup, JSON, paths or SQL on purpose, and they stay plain text in every view and format.
Click a line to copy it. Use the pin to keep it in Saved.
Everything is generated in this tab. The same seed and the same settings give the same text on any machine.
Press 1 to 9 or click a preset such as Blog post, Log file, Card grid, API fixtures or Locale file. It sets theme, amount, structure and format, and the result appears at once.
Choose one or more of the 28 themes, from git commits and terminal logs to legal filler, meeting notes and product reviews. The Edge cases group holds test strings for layouts and parsers.
Ask for lines, paragraphs, words or characters. Give every item an exact or a ranged target length, or write your own template with variables such as {name}, {email} or {date}.
Pick one of 23 formats, from plain text and Markdown to JSON, SQL, code arrays and locale files. The Preview tab shows the text as an article, cards, a chat or a log, at phone, tablet or desktop width.
G draws a new seed, C copies, D downloads. Copy as converts to any other format, and Link copies an address that brings back this exact text.
# in the addressLore Gen makes placeholder text that looks like the content you are designing for. A list of git commits, a log viewer, an API reply, a chat thread, a support inbox or a screen full of button labels each behave differently once real words arrive, and a block of Latin does not show you that. You choose one or more of 28 themes (1,648 lines in all), say how much you need, pick an output format and watch the result appear in a rendered preview. Everything is generated in your browser, and everything is invented: names, companies, products and version numbers do not exist.
Developer, product, people and story themes, 4 of them new: meeting notes, support tickets, release notes and product reviews. Mix as many as you like.
Articles, card grids, chat threads, log files with level colours, tables, changelogs, ticket lists, profile cards and UI strings inside mock components. Switch between phone, tablet and desktop width.
Text and markup, JSON and NDJSON, CSV, YAML, XML, SQL, Markdown tables, code arrays for five languages and locale files for four toolchains. Each one is escaped for its syntax.
Ask for 8 lines, 3 paragraphs, 250 words or 1,000 characters, or give every item its own length between a minimum and a maximum.
Write one line with 14 variables and get as many variations as you need. Names, e-mail addresses, dates and ids come from the seed.
9 sets of test strings: long words, accents, right-to-left text, emoji, CJK, invisible characters, inert markup and padded spaces.
The same seed with the same settings gives the same text. A link holds the whole setup, and links made by earlier versions still open the text they showed before.
Pick the theme that stresses the part of the layout you are testing. The lengths in a theme are uneven on purpose, because evenly sized text hides the problems that real text causes.
| You are building | Try | What it shows you |
|---|---|---|
| A commit list or history view | Git Commits, Release Notes | Short subject lines, a few long ones, and one-word messages |
| A log viewer or console | Terminal Logs, DevOps / CI-CD, Error Messages | Level prefixes, timestamps and long unbroken tokens |
| An API client or a mock server | API Responses, Fake Profiles | Payload-shaped strings, invented names and reserved example addresses |
| An inbox, a ticket board or a chat | Support Tickets, Issue Tickets, Team Chat | Short replies next to long complaints |
| Screens, buttons and forms | UI Microcopy, Product Copy | Labels that grow in translation, toasts and empty states |
| A shop or review page | Product Reviews, Product Copy | Star ratings next to one-line and rambling opinions |
| A legal or policy page | Legal Filler | Long clauses and numbered sections. It is placeholder text, not advice, and not a contract |
| A page that must stay neutral | Classic Lorem | Latin, for when the topic of the words should not distract |
Lorem Ipsum itself is not random Latin. According to a write-up of how it was traced, it comes from sections 1.10.32 and 1.10.33 of Cicero's De finibus bonorum et malorum, a treatise on ethics written in 45 BC. The Classic Lorem theme and the Start with the classic switch keep it available for the cases where it is the right choice.
The seed text is turned into a number with the FNV-1a hash, and that number starts a small generator named mulberry32 that makes every choice: which lines, in which order, which names in a template. The theme list, the amount and the other settings are part of that starting point, so two people with the same seed and the same settings get the same bytes. Nothing is left to chance once the seed is set. A new seed itself is drawn from your browser's random source. This is what makes placeholder text safe to use in a snapshot test or a screenshot review: nothing changes unless somebody changes a setting.
Generate and G draw a new seed, unless you have locked it. Roll and S always draw a new one. Shuffle reorders the lines you already have and records that in the link. The seed is shown in the field at all times, so you can copy it. The generator is built for repeatability and is not secure: do not use its output as a password, a token or a key.
The address holds the whole setup after the #: themes, amount, unit, structure, format and seed, plus the options you changed. Browsers do not send that part of an address to a server (MDN, URI fragment). Links made before this version carry no pool marker, so they open with the original text pools and the original escaping and show the text they always showed. New links add v=4, which switches on the extended pools and complete escaping for YAML, XML and JSX. The Text pools switch lets you move between the two.
A quote, a backslash or a line break inside a string can break a file or, worse, change what it means. Every format therefore escapes text by the rules of its own syntax, and the rules differ more than you might expect. These are the ones the page follows, each with its source. The pages were read on 5 November 2026. Standards and tools change, so treat them as the final word.
\u escapes are not allowed (GNU gettext manual).\uXXXX (Properties).@ and ? (string resources).\u{...} for a code point (PHP manual).| Format | Saved as | How the text is escaped |
|---|---|---|
| Plain text, Markdown | .txt, .md | Kept as written. A line break stays a line break |
| HTML, JSX | .html, .jsx | &, < and > become entities. In JSX, braces and line-break characters are wrapped in an expression |
| JSON array, JSON objects, i18n JSON | .json, en.json | JSON string escapes. Control characters become \u00XX |
| NDJSON | .ndjson | One JSON value per line, so a line break inside text is always escaped (see JSON Lines) |
| CSV | .csv | Every text field is quoted and a quote is doubled |
| YAML | .yaml | Text in double quotes with JSON-style escapes |
| XML, Android strings | .xml, strings.xml | Entities for &, <, > and CR. Android adds backslashes and quotes values with edge spaces |
| SQL insert, SQL table + rows | .sql | Single quotes doubled, no backslash escapes |
| Markdown table | .md | Pipes, angle brackets and character references escaped, line breaks turned into spaces |
| JavaScript, TypeScript, Python | .js, .ts, .py | Double-quoted literals; invisible and separator characters written as \uXXXX |
| PHP | .php | Double quotes with \$, \\ and \u{...} |
| Go | .go | Interpreted string literal; \x only below 0x80, \u above |
| gettext | messages.po | C-style string with octal escapes for control characters, one message context per key |
| Java properties | messages.properties | \uXXXX for every non-ASCII character, a backslash before a leading space |
| Flutter ARB | app_en.arb | JSON, with apostrophes and braces quoted for ICU |
The test suite behind this page reads each format back with a real parser and compares the result with what went in, using hostile text with quotes, backslashes, markup, emoji, right-to-left text and control characters. It uses JSON, YAML, XML, CSV and SQLite readers, the Python, PHP, Go, TypeScript and Java toolchains and a gettext reader. It has not compiled a .po file with msgfmt or an Android file with Android's own tools, so for those two, check the file in your project before relying on it. Copy as and the Save menu convert the current text into any other format without changing the seed.
With Target length every item gets a length in characters or in words, either exactly N or a number between a minimum and a maximum that the seed picks for each item. Characters are counted as Unicode code points, so an emoji counts as one. A short line is extended with more text from the selected themes and a long one is cut, so every item has exactly the length it was given. When the amount itself is a number of words or characters, the total length is set instead and the per-item target is off.
A Template is one line of your own with variables in braces, repeated once per item: {n} and {n:3} for a padded counter, {name}, {first}, {last}, {email}, {company}, {date}, {time}, {uuid}, {id}, {num:5-20}, {ip}, {word} and {line}. Within one item, {name} and {email} belong together. E-mail addresses use example.com, example.org or example.net, the names that RFC 2606 reserves for examples. {ip} picks from 192.0.2.0/24, 198.51.100.0/24 and 203.0.113.0/24, the blocks that RFC 5737 reserves for documentation. {uuid} has the layout of a version 4 UUID, with its version and variant bits as RFC 9562 describes them, but it is made from the seeded generator and is not unique or secret.
The Edge cases group holds strings that real text will eventually contain and that tidy sample text never does: very long words with no break opportunity, one-character lines next to paragraph-length ones, accents typed as one character or as a letter plus a combining mark, Arabic and Hebrew alone and inside English sentences, emoji and joined emoji sequences, Chinese, Japanese and Korean text, padded and empty lines, and invisible characters such as the zero-width space (U+200B), the zero-width non-joiner (U+200C), the zero-width joiner (U+200D), the word joiner (U+2060) and the zero width no-break space (U+FEFF), as listed in the Unicode overview. Some strings look like HTML, JSON, URLs, file paths or SQL fragments on purpose. They are inert test data: the page shows them as text in every view and escapes them in every format.
Use Show invisibles in the Code view to see where the hidden characters sit, and the width toggle in the Preview to see how a long word behaves at 390 pixels. The strings are original text written for this page, not copied from any list.
The page has no server of its own. Its code makes no network request: it has no call that sends data anywhere, and what you generate, type or paste is never uploaded. The page does load the JetBrains Mono font from Google Fonts, the site's theme files and the site's analytics script, which counts page views. In your browser's local storage it keeps four small items, all optional: jb-theme (dark or light, from the site's theme switch), jb-lg-prefs (view, preview style and width, wrap, open panels and the seed lock), jb-lg-hist (the last 20 generations as share-link strings) and jb-lg-pins (the lines you pinned). If storage is blocked, everything still works until you close the tab. Copying uses the clipboard of your browser only when you press Copy.
Need a scratch space for your placeholder text? Paste it into the free online notepad to edit it, sort lines or remove duplicates, and find it again next time. Every tab autosaves in your browser.
Placeholder text is for drafts. Replace it before a page goes live, because a published page that still says "lorem" or shows invented reviews misleads visitors. The Product Reviews and Fake Profiles themes contain invented people, and the Legal Filler theme is filler for layout, not legal text. A preview shows how text flows in a generic mock-up, not in your own design, so always check the real component too. The amount is capped at 2,000 items and a target length at 5,000. For a long result the Items view lists the first 400 and the Preview draws the first 300, and both say so. The output itself is never cut.