mPDF Manual – Fonts & Languages

Automatic font selection

mPDF has two functions which can be used together or separately:

autoScriptToLang - marks up HTML text using the lang attribute, based on the Unicode script block in question, and configurable values in config_script2lang.php.

autoLangToFont - selects the font to use, based on the HTML lang attribute, using configurable values in config_lang2font.php.

For automatic font selection, ideally we would choose the font based on the language in use. However it is actually impossible to determine the language used from a string of HTML text. The Unicode script block can be ascertained, and sometimes this tells us the language e.g. Telugu. However, Cyrillic script is used for example in many different languages. So the best we can do is base it on the script used. However, mPDF does this in two stages via the lang attribute, because this allows the options of using either of the stages alone or together:

English ру́сский язы́к پښتو

              ↓ autoScriptToLang (config_script2lang.php) ↓

English <span lang="und-Cyrl">ру́сский язы́к</span> <span lang="ps">پښتو</span>

             ↓ autoLangToFont (config_lang2fonts.php) ↓

Uses "lang" to select font, and to determine OTL features applied



$mpdf->autoScriptToLang = true;

$mpdf->baseScript = 1;

$mpdf->autoVietnamese = true;

$mpdf->autoArabic = true;

$mpdf-&gt;baseScript = 1; tells mPDF which Script to ignore. It is set by default to “1” which is for Latin script. In this mode, all scripts except Latin script are marked up with lang attribute. To select other scripts as the base, see the file /classes/ucdn.php

Using autoScriptToLang, mPDF detects text runs based on Unicode script block; using the values in config_script2lang.php it then encloses the text run within a span tag with the appropriate language attribute. For many scripts, the language cannot be determined: see the example above which recognises Cyrillic script and marks it up using und-Cyrl, which is a valid IETF tag, coding for language=”undetermined”, script=”Cyrillic”.

Two optional refinements are added: Vietnamese text can often be recognised by the presence of certain characters which do not appear in other Latin script langauges, and similarly analysis of the text can attempt to distinguish Arabic, Farsi, Pashto, Urdu and Sindhi. If active, the text will then be marked with a specific language tag e.g. “vi”, “pa”, “ur”, “fa” etc.

These features can be disabled or enabled (default) using the variables $mpdf-&gt;autoVietnamese $mpdf-&gt;autoArabic, either in config.php or at runtime.



$mpdf->autoLangToFont = true;

You can edit the values in config_lang2font.php to specify which fonts are used for which lang.

Using text with multiple languages

Recommended ways to use multiple languages in mPDF:

  1. If you have full control over the HTML, mark-up the text with the `lang `atribute and use CSS (`:lang` selector preferably); this method means that the language information can also be used by OTL for language dependent substitutions.
  2. If you have no control over (user) HTML input and want to output faithfully, use both `autoScriptToLang` and `autoLangToFont`

It is preferable not to use autoScriptToLang and autoLangToFont unless they are necessary: they will result in increased processing time, and OTL tables will not be able to use language dependent substitutions when undefined languages are set e.g “und-Cyrl”.