Random generation tool for Japanese dummy novels (I am a cat, etc.) | ZeroTools

This dummy text generator is a tool for creating placeholder manuscripts to check the placement of main text in websites, apps, and printed materials.

Even at the stage where the final manuscript is not yet determined, you can pour in long passages of English literary fragments or Lorem Ipsum to check the density of characters, line spacing, margins between paragraphs, and balance with headings.

On the screen, you select the material, generation unit, quantity, and output format, and check the results in the preview on the right.

Because changing the settings rearranges the results from the same fixed material, you can compare the same text while changing the design conditions.

The purpose is not as a generative AI to evaluate the originality of the content or the naturalness as text, but to obtain reproducible test data.

Since the output also includes a book title and author-like display, it can also be used to check long text cards that include an article title or author column.

However, the recorded text is not an original text database intended for citation or editing. Please treat it as material for layout, rather than reprinting it as a published manuscript or the main text of a work.

Specifically, it can be used for tasks such as displaying the same article card on a PC and a smartphone to compare the number of wrapped lines, or checking the spacing with the table of contents or advertisements when the main text increases to 15 paragraphs.

Note that comparing how much a column extends between English and Japanese requires preparing actual target language data separately, as this tool only provides an English corpus.

Because it is not intended for evaluating the quality of the text, it does not perform typographical correction, summarization, style conversion, or SEO evaluation.

The selectable materials are four types: I Am a Cat style, Night on the Galactic Railroad style, Run Melos style, and English Lorem Ipsum.

In the English version, the three Japanese works are a fixed 5-paragraph array of English-translation-style text. For each, five paragraphs are stored as an array within the component.

The generation process takes them out in order starting from the first paragraph, and when it reaches the fifth, it returns to the first again, without randomness or AI generation.

It is different from the Japanese kanji/kana typesetting test.

For example, if you specify 7 paragraphs, it will output the 1st to 5th once each, and then add the 1st and 2nd again. It does not use random numbers to rearrange paragraphs or replace words.

Therefore, if the material, quantity, and format are the same, the result will be the same. The regeneration operation on the screen rearranges the fixed array with the current settings, and is not a function to create a new story.

Markov chains, morphological analysis, natural language processing, external AI, and the Aozora Bunko API are not used. I cannot guarantee translation accuracy, translation rights, or a complete work.

For materials that require avoiding misquotations, please refer to the official original text.

Lorem Ipsum also has 5 fixed paragraphs, and circulates in the same order when exceeding the specified amount. Reselecting the material also switches the book title and author-like notation, but there is no field for the user to input their own original title or author name.

If necessary, replace them with an external editor after copying or saving.

The generation unit is selected from the number of paragraphs and the number of characters. The paragraph count mode specifies 1 to 15 paragraphs in increments of 1 paragraph, and adds the selected number of fixed paragraphs.

It is suitable when you want to check the margins between paragraphs or the height of a long text card. From 6 paragraphs onwards, the 5-paragraph material circulates, so the same paragraph will reappear.

The character count mode specifies 100 to 2,000 characters in increments of 50 characters. The process sequentially adds the length of the main text paragraphs, and if adding the entire next paragraph exceeds the specified value, it slices that paragraph at the necessary position and ends.

Note that the character mode measures UTF-16 length of the main text only, not including title, author, linebreaks, or tags, and may cut mid-sentence.

Because there is no function to find the end of a sentence or word boundary to adjust, it may end before a period or in the middle of a word or sentence.

The specified character count targets only the string of the main text. The book title and author in Text format, HTML tags, Markdown symbols, and line breaks between paragraphs are not included in the calculation.

Therefore, if you measure the entire saved file or copied string with another counter, it will be longer than the number specified on the screen.

It is used as a setting to determine the maximum amount of the main text part, not the exact total number of characters.

The calculation of the number of characters is based on the length and slice of JavaScript strings. This unit is not always the same as the number of graphemes visible on the screen, and in general Unicode strings containing emojis or combining characters, a single character may become multiple units.

Although it is unlikely to be a major issue with the current built-in materials, it is not a function guaranteed as a strict submission character count.

In Text format, the book title and author are placed at the beginning, and the main text paragraphs are lined up with empty lines in between.

It is suitable for pasting into a plain text field or checking only the amount of text. When saving generated stories, the file extension and MIME type match the chosen format instead of defaulting to .txt for all options: plain text saves as .txt (text/plain), HTML saves as .html (text/html), and Markdown saves as .md (text/markdown) using UTF-8 encoding.

Note that exported files contain only the raw story content or body fragment without full document boilerplates such as head tags or external stylesheet links, and UI preview themes or vertical writing modes are excluded from downloads.

In HTML format, an h1, a standard paragraph element output with <p class="author">, and p elements for the main text paragraphs are arranged inside an article container.

The output generated in this view is raw markup code rather than a browser-rendered page, and saving now properly assigns the .html extension with the text/html;charset=utf-8 MIME type.

The author paragraph uses the valid standard HTML class attribute instead of JSX syntax. However, the export remains an excerpted fragment without document-level elements such as <!DOCTYPE html>, <html>, <head>, or embedded CSS rules.

Furthermore, the copy button continues to write plain markup text to your clipboard using the plain text clipboard API rather than rich formatted data, and visual theme styling or vertical reading modes from the preview window are omitted from the downloaded file.

Markdown format makes the book title a level 1 heading, the author in bold notation, and separates the main text with empty lines.

The preview is not a display that interprets HTML or Markdown to make a complete page, but a display to check the source of the selected format.

Since the final appearance differs depending on the destination editor or CMS, please also check it at the paste destination.

The HTML output now directly provides the standard class="author" attribute on the author paragraph rather than JSX syntax, so you no longer need to manually rename attributes before linking stylesheets.

However, because the tool generates only markup fragments without embedded CSS styling or complete page wrappers, you still need to supply your own stylesheet rules to style the layout and .author class in your project.

Similarly, heading hashes and emphasis syntax in Markdown format render visually only when the receiving platform supports Markdown parsing.

If you paste the copied string into a plain text editor, formatting symbols and HTML tags appear as literal text characters, and customized preview themes or vertical alignments shown in the UI are not embedded into the saved files.

There are three preview color schemes: Japanese paper, night reading, and modern. Japanese paper switches to a display with a pale background and brownish text, night reading has a dark background and bright text, and modern follows the site's background and text color.

This is a setting that only changes the CSS of the preview, and does not write the theme CSS to the copied content or saved file.

The theme and vertical writing are only preview CSS and are not included in the text file.

The vertical writing switch can be used for the three Japanese-style works. When turned on, it sets writing-mode to vertical-rl and text-orientation to mixed, making the fixed-height area flow from right to left.

The switch is not displayed for Lorem Ipsum. Vertical writing is also only displayed on the screen, and is not a function to add vertical writing specifications to the Text, HTML, or Markdown output strings.

In an actual web page, fonts, line-breaking rules, ruby, punctuation, container width, and browser differences will affect the result.

This preview is for simple confirmation and does not reproduce the typesetting of e-books or printed materials. It is necessary to check again with the CSS of the final design and the actual device.

Even during vertical writing, the generation amount and output format do not change; only the direction of the display container changes.

Because long sentences spread horizontally within the fixed-height area, you can also check the scroll amount and column arrangement, but there is no function to generate or copy vertical writing CSS for the entire page.

When copying generated story text, the application awaits the system clipboard API asynchronously and displays an active pending state while writing.

If clipboard access is restricted or unavailable, an accessible manual copy fallback appears with keyboard instructions so you can copy the exact text directly.

Any adjustment to text style, quantity, mode, or format immediately invalidates previous copy feedback and regenerates the content.

Regenerating with unchanged parameters is deterministic based on the selected story and does not produce random variations. Presentation settings such as book themes and vertical layout adapt the preview styling while preserving the underlying text formatting for copy and download.

It may not be possible to copy depending on browser permissions or the execution environment. The save button creates a Blob from the string and downloads it, but also provides no guarantee of file saving completion.

There is no function to save the input history or generation history to IndexedDB or localStorage.

The maximum amount that can be handled at one time is 15 paragraphs or 2,000 characters of main text. It does not support paragraph randomization, temperature settings, automatic heading creation, image generation, adding material from files, or batch saving of multiple manuscripts.

If you need text that changes under the same conditions, choose another generation method instead of this tool with fixed materials.

The saved file name includes the identifier of the selected material, and there is no process to overwrite the original file. The download destination and handling of files with the same name follow the browser settings.

Because it does not output a configuration file to return the generated result to the same state later, record the material, unit, quantity, and format if you want to reproduce it.

Processing is performed in the displaying browser, but this does not mean that the entire page operates completely offline, or that there is absolutely no communication including site loading.

Also, please use it understanding that the Japanese-style materials should not be mistaken for original texts and published, that mid-sentence truncation occurs with length specification, and that themes and vertical writing are not reflected in the saved content.

After the actual manuscript arrives, please check for replacement, including whether it is published as dummy text, the title, alt text, and search meta information.