The symbol generator

Use it when you need a string of a given length built from a specific set of characters — a token, a coupon code, a random suffix, or "noise" in a particular script. symbol picks one character at a time out of a set you define. (For a whole word from a list, use text instead.)

You define the character set in one of two ways — never both at once:

• value — spell out your own set, right here;
• alphabet — name a built-in Unicode alphabet.

Then length says how many characters to produce, and include / exclude tweak the set without rewriting it.

Example outputs on this page are illustrative — the exact characters depend on the seed and can differ between core versions. The rules they demonstrate are stable.

At a glance

Attribute	Required	What it does
value	one of the two	Your own character set (literals + [x-y] ranges)
alphabet	one of the two	Name of a built-in Unicode alphabet
length	no	String length; default 1, maximum 1024
include	no	Add characters to the set (same grammar as value)
exclude	no	Remove characters from the set

Your own set with value

The simplest route: list the characters you want in value. No regular expressions required — you write the set, symbol draws from it.

<gen type="symbol" value="ABCDEF" length="4"/>        <!-- literals -->
<gen type="symbol" value="[a-z]" length="6"/>         <!-- a range -->
<gen type="symbol" value="[A-Z0-9]" length="8"/>      <!-- two ranges in one group -->
<gen type="symbol" value="[a-f]xY[0-9]" length="5"/>  <!-- ranges + literals mixed -->

./run demo.tdc

value="ABCDEF" length="4"      → CDFA
value="[a-z]" length="6"       → tqmboa
value="[A-Z0-9]" length="8"    → K7QW2ZP4
value="[a-f]xY[0-9]" length="5" → x3bYa

The rules for a set

• Literals — just write the characters: value="ABCDEF" → the set {A, B, C, D, E, F}.
• Ranges — in square brackets: [a-z], [A-Z], [0-9]. You can put several ranges in one group: [a-z0-9_].
• Commas and spaces outside brackets are ignored — they're there only for readability: value="[a-f], [0-5]" is the same set as value="[a-f][0-5]". To use a comma or a space as a character, put it in brackets: [,], [ ].
• Duplicates are removed; order is preserved. value="AABB" is just {A, B}.

Characters from any language work — you're not limited to ASCII. See Named alphabets below for a shortcut to whole scripts, and Unicode scripts by hand for typing them directly.

One character, not one word

symbol picks a single character at a time. To choose a whole word from a list, use text: <gen type="text" value="red,green,blue"/>.

Length

length sets how many characters to produce. It defaults to 1 and caps at 1024. Use it whenever you need a fixed-width token — a 6-digit code, a 32-char key, and so on.

<gen type="symbol" value="[A-Z0-9]"/>            <!-- length omitted → 1 -->
<gen type="symbol" value="[A-Z0-9]" length="6"/>
<gen type="symbol" value="[A-Z0-9]" length="16"/>

./run demo.tdc

length omitted → K
length="6"     → Q7ZW2P
length="16"    → K7QW2ZP4M9RXB3TA

Only an exact length is supported here. If you need a variable-width string, number accepts length ranges and groups; symbol does not.

Extend the set with include

Rather than rewriting a whole set to add a few characters, hand them to include. It uses the same grammar as value (literals and [x-y] ranges). This is most useful with a named alphabet, which is otherwise fixed — for example, letters plus digits for a login-style string.

<!-- your own letters, plus two specific digits -->
<gen type="symbol" value="[a-z]" include="2,4" length="8"/>

<!-- a named alphabet, extended with the digits it lacks -->
<gen type="symbol" alphabet="latin.lower" include="[0-9]" length="8"/>

./run demo.tdc

value="[a-z]" include="2,4" length="8"   → t2qm4boa
alphabet="latin.lower" include="[0-9]"   → k7qw2zp4

Trim the set with exclude

The mirror image: exclude removes characters from the set. Use it to drop look-alike characters from a code (no O/0, no l/1), or trimming a couple of letters from a named alphabet.

<!-- letters and digits, minus the confusing pairs -->
<gen type="symbol" value="[A-Z0-9]" exclude="O0Il1" length="8"/>

<!-- letters, minus one vowel -->
<gen type="symbol" value="[a-z]" exclude="y" length="8"/>

./run demo.tdc

value="[A-Z0-9]" exclude="O0Il1" length="8" → K7QW2ZP4
value="[a-z]" exclude="y" length="8"        → tqmboade

How include and exclude combine

The final set is (base ∪ include) − exclude, so exclude has the last word: a character that is both added by include and removed by exclude will not appear. If the set ends up empty after the modifiers, that's error TDC099.

<!-- 4 is included, then excluded → it loses; 2 survives -->
<gen type="symbol" value="[a-z]" include="2,4" exclude="4" length="8"/>

./run demo.tdc

value="[a-z]" include="2,4" exclude="4" → t2qmboad   (no 4 anywhere)

Named alphabets with alphabet

Use it when you need a random string from a specific writing system — Cyrillic for Russian-style logins, kana for Japanese test data, Arabic or Hebrew letters to exercise right-to-left layout. Typing the Unicode range by hand ([а-я]) is error-prone: you forget ё, or you get the boundary wrong. A named alphabet is checked by name and contains exactly the right characters.

alphabet replaces value — you give a name from the registry, plus a length.

<gen type="symbol" alphabet="cyrillic.ru.letters" length="10"/>
<gen type="symbol" alphabet="kana.hiragana" length="8"/>
<gen type="symbol" alphabet="arabic.letters" length="6"/>

./run demo.tdc

cyrillic.ru.letters length="10" → рнБСпВЖЧжХ
kana.hiragana length="8"        → ゃまぃすめいおち
arabic.letters length="6"       → فؿآحـآ

This is a deliberate Unicode and localization demo: the output is non-Latin on purpose, to show that the same generator produces whatever script you name.

Live examples of each script

The same generator, one line per alphabet (length="10"), so you can compare the scripts side by side:

alphabet	Example output
latin.lower	usahtbcjpi
latin.upper	USAHTBCJPI
digits.fullwidth	７７０２７０１３６３
cyrillic.ru.letters	рнБСпВЖЧжХ
greek.letters	ξμΒΟνΓΖΤζΡ
hebrew.letters	פףאחפבגךני
arabic.letters	فؿآحـآإذغد
kana.hiragana	ゃまぃすめいおちふそ
kana.katakana	ユメィズヤイオヂヘタ
cjk.unified.basic	货袪倱料護冣囱沌耣楿
roman.upper	DDIXDIIXCX

All supported names

alphabet	Contains
latin.lower	ASCII a-z
latin.upper	ASCII A-Z
latin.letters	ASCII A-Z and a-z
digits.ascii	ASCII digits 0-9
digits.fullwidth	Full-width digits ０-９
cyrillic.ru.lower	Russian а-я plus ё
cyrillic.ru.upper	Russian А-Я plus Ё
cyrillic.ru.letters	Russian Cyrillic, both cases, including ё
greek.letters	Basic Greek letters
hebrew.letters	Hebrew א-ת
arabic.letters	Arabic letters ء-ي
kana.hiragana	Japanese hiragana ぁ-ゖ
kana.katakana	Japanese katakana ァ-ヺ
cjk.unified.basic	CJK Unified Ideographs U+4E00..U+9FFF
roman.upper	Roman-numeral letters I V X L C D M
roman.lower	Roman-numeral letters i v x l c d m

All 16 names are validated: each one resolves and produces characters from its own script. Because the set is fixed, use include / exclude to adjust it — for example alphabet="cyrillic.ru.letters" exclude="ъь".

Unicode scripts by hand

You don't have to use a named alphabet — because value accepts any characters, you can also type a script (or mix several) directly. Named alphabets are still preferred where one exists: they're documented, validated, and include the awkward characters a plain range would miss, such as Russian ё.

<gen type="symbol" value="[А-Я]" length="4"/>            <!-- Cyrillic range -->
<gen type="symbol" value="कखगघचछ" length="3"/>          <!-- Devanagari literals -->
<gen type="symbol" value="あア[0-9][A-F]" length="6"/>   <!-- mixed scripts + ranges -->

./run demo.tdc

value="[А-Я]" length="4"          → ШФПР
value="कखगघचछ" length="3"         → चचग
value="あア[0-9][A-F]" length="6" → アB4あ7ア

The same alphabets in regex

The alphabet registry is also available inside regex and advanced_regex through the escape \a{name}, so you can drop a named script into a larger pattern:

<gen type="regex" value="\a{kana.hiragana}{5}"/>
<gen type="advanced_regex" value="(?%{70:\a{latin.upper}{2};30:\a{cyrillic.ru.upper}{2}})-[0-9]{4}"/>

./run demo.tdc

kana.hiragana{5}          → まぃすめい
weighted upper + -[0-9]4  → KM-8042

Plain BMP ranges such as [а-я]{8} also work inside regex, but the named alphabets are preferred for the same reasons as above — they're documented, validated by name, and cover characters that a bare range would drop.

See also

• alphabet and length in the attribute reference.
• Regex — when the string has structure, not just a character set.
• Masks and case — reshape a generated string after the fact.