Checks and fixups#

Custom automatic fixups#

You can also implement your own automatic fixup in addition to the standard ones and include them in AUTOFIX_LIST.

The automatic fixes are powerful, but can also cause damage; be careful when writing one.

For example, the following automatic fixup would replace every occurrence of the string foo in a translation with bar:

# Copyright © Michal Čihař <michal@weblate.org>
#
# SPDX-License-Identifier: GPL-3.0-or-later

from django.utils.translation import gettext_lazy

from weblate.trans.autofixes.base import AutoFix


class ReplaceFooWithBar(AutoFix):
    """Replace foo with bar."""

    name = gettext_lazy("Foobar")

    def fix_single_target(self, target, source, unit):
        if "foo" in target:
            return target.replace("foo", "bar"), True
        return target, False

To install custom checks, provide a fully-qualified path to the Python class in the AUTOFIX_LIST, see Custom quality checks, add-ons and auto-fixes.

Customizing behavior using flags#

You can fine-tune Weblate’s behavior by using flags. The flags provide visual feedback to the translators and help them to improve their translation. This can be done on the source string level (see Additional info on source strings), or in the Component configuration (Translation flags). Some file formats also allow to specify flags directly in the format (see Supported file formats).

The flags are comma-separated; if they have parameters, they are separated with colon. You can use quotes to include whitespaces or special characters in the string. For example:

placeholders:"special:value":"other value", regex:.*

Both single and double quotes are accepted, special characters are being escaped using backslash:

placeholders:"quoted \"string\"":'single \'quoted\''

placeholders:r"^#*"

To verify that translators do not change the heading of a Markdown document: A failing check will be triggered if the string ‘### Index’ is translated as ‘# Indice’

placeholders:r"\]\([^h].*?\)"

To make sure that internal links are not being translated (i.e. [test](../checks) does not become [test](../chequeos).

Here is a list of flags currently accepted:

rst-text: Treat a text as an reStructuredText document, affects Unchanged translation.
dos-eol: Uses DOS end-of-line markers instead of Unix ones (\r\n instead of \n).
read-only: The string is read-only and should not be edited in Weblate, see Read-only strings.
terminology: Used in Glossary. Copies the string into all glossary languages so it can be used consistently in all translations. Also useful in combination with read-only, for example in product names.
priority:N: Priority of the string. Higher priority strings are presented first for translation. The default priority is 100, the higher priority a string has, the earlier it is offered for translation.
max-length:N: Limit the maximal length for a string to N characters, see Maximum length of translation.
xml-text: Treat text as XML document, affects XML syntax and XML markup.
font-family:NAME: Define font-family for rendering checks, see Managing fonts.
font-weight:WEIGHT: Define font-weight for rendering checks, see Managing fonts.
font-size:SIZE: Define font-size for rendering checks, see Managing fonts.
font-spacing:SPACING: Define letter spacing for rendering checks, see Managing fonts.
icu-flags:FLAGS: Define flags for customizing the behavior of the ICU MessageFormat quality check.
icu-tag-prefix:PREFIX: Set a required prefix for XML tags for the ICU MessageFormat quality check.
placeholders:NAME:NAME2:...: Placeholder strings expected in translation, see Placeholders.
replacements:FROM:TO:FROM2:TO2...: Replacements to perform when checking resulting text parameters (for example in Maximum size of translation or Maximum length of translation). The typical use case for this is to expand placeables to ensure that the text fits even with long values, for example: replacements:%s:"John Doe".
variants:SOURCE: Mark this string as a variant of string with matching source. See String variants.
regex:REGEX: Regular expression to match translation, see Regular expression.
forbidden: Indicates forbidden translation in a glossary, see Forbidden translations.
strict-same: Make “Unchanged translation” avoid using built-in words blacklist, see Unchanged translation.
strict-format: Make format checks enforce using format even for plural forms with a single value, see Formatted strings.
check-glossary: Enable the Does not follow glossary quality check.
angularjs-format: Enable the AngularJS interpolation string quality check.
c-format: Enable the C format quality check.
c-sharp-format: Enable the C# format quality check.
es-format: Enable the ECMAScript template literals quality check.
i18next-interpolation: Enable the i18next interpolation quality check.
icu-message-format: Enable the ICU MessageFormat quality check.
java-printf-format: Enable the Java format quality check.
java-format: Enable the Java MessageFormat quality check.
javascript-format: Enable the JavaScript format quality check.
lua-format: Enable the Lua format quality check.
object-pascal-format: Enable the Object Pascal format quality check.
percent-placeholders: Enable the Percent placeholders quality check.
perl-format: Enable the Perl format quality check.
php-format: Enable the PHP format quality check.
python-brace-format: Enable the Python brace format quality check.
python-format: Enable the Python format quality check.
qt-format: Enable the Qt format quality check.
qt-plural-format: Enable the Qt plural format quality check.
ruby-format: Enable the Ruby format quality check.
scheme-format: Enable the Scheme format quality check.
vue-format: Enable the Vue I18n formatting quality check.
md-text: Treat text as a Markdown document, and provide Markdown syntax highlighting on the translation text area. Enables Markdown links, Markdown references, and Markdown syntax quality checks.
case-insensitive: Adjust checks behavior to be case-insensitive. Currently affects only Placeholders quality check.
safe-html: Enable the Unsafe HTML quality check.
url: The string should consist of only a URL. Enable the URL quality check.
ignore-all-checks: Ignore all quality checks.
ignore-bbcode: Skip the BBCode markup quality check.
ignore-duplicate: Skip the Consecutive duplicated words quality check.
ignore-check-glossary: Skip the Does not follow glossary quality check.
ignore-double-space: Skip the Double space quality check.
ignore-angularjs-format: Skip the AngularJS interpolation string quality check.
ignore-c-format: Skip the C format quality check.
ignore-c-sharp-format: Skip the C# format quality check.
ignore-es-format: Skip the ECMAScript template literals quality check.
ignore-i18next-interpolation: Skip the i18next interpolation quality check.
ignore-icu-message-format: Skip the ICU MessageFormat quality check.
ignore-java-printf-format: Skip the Java format quality check.
ignore-java-format: Skip the Java MessageFormat quality check.
ignore-javascript-format: Skip the JavaScript format quality check.
ignore-lua-format: Skip the Lua format quality check.
ignore-object-pascal-format: Skip the Object Pascal format quality check.
ignore-percent-placeholders: Skip the Percent placeholders quality check.
ignore-perl-format: Skip the Perl format quality check.
ignore-php-format: Skip the PHP format quality check.
ignore-python-brace-format: Skip the Python brace format quality check.
ignore-python-format: Skip the Python format quality check.
ignore-qt-format: Skip the Qt format quality check.
ignore-qt-plural-format: Skip the Qt plural format quality check.
ignore-ruby-format: Skip the Ruby format quality check.
ignore-scheme-format: Skip the Scheme format quality check.
ignore-vue-format: Skip the Vue I18n formatting quality check.
ignore-translated: Skip the Has been translated quality check.
ignore-inconsistent: Skip the Inconsistent quality check.
ignore-kashida: Skip the Kashida letter used quality check.
ignore-md-link: Skip the Markdown links quality check.
ignore-md-reflink: Skip the Markdown references quality check.
ignore-md-syntax: Skip the Markdown syntax quality check.
ignore-max-length: Skip the Maximum length of translation quality check.
ignore-max-size: Skip the Maximum size of translation quality check.
ignore-escaped-newline: Skip the Mismatched \n quality check.
ignore-end-colon: Skip the Mismatched colon quality check.
ignore-end-ellipsis: Skip the Mismatched ellipsis quality check.
ignore-end-exclamation: Skip the Mismatched exclamation mark quality check.
ignore-end-stop: Skip the Mismatched full stop quality check.
ignore-end-question: Skip the Mismatched question mark quality check.
ignore-end-semicolon: Skip the Mismatched semicolon quality check.
ignore-newline-count: Skip the Mismatching line breaks quality check.
ignore-plurals: Skip the Missing plurals quality check.
ignore-placeholders: Skip the Placeholders quality check.
ignore-punctuation-spacing: Skip the Punctuation spacing quality check.
ignore-regex: Skip the Regular expression quality check.
ignore-reused: Skip the Reused translation quality check.
ignore-same-plurals: Skip the Same plurals quality check.
ignore-begin-newline: Skip the Starting newline quality check.
ignore-begin-space: Skip the Starting spaces quality check.
ignore-end-newline: Skip the Trailing newline quality check.
ignore-end-space: Skip the Trailing space quality check.
ignore-same: Skip the Unchanged translation quality check.
ignore-safe-html: Skip the Unsafe HTML quality check.
ignore-url: Skip the URL quality check.
ignore-xml-tags: Skip the XML markup quality check.
ignore-xml-invalid: Skip the XML syntax quality check.
ignore-zero-width-space: Skip the Zero-width space quality check.
ignore-ellipsis: Skip the Ellipsis quality check.
ignore-icu-message-format-syntax: Skip the ICU MessageFormat syntax quality check.
ignore-long-untranslated: Skip the Long untranslated quality check.
ignore-multiple-failures: Skip the Multiple failing checks quality check.
ignore-unnamed-format: Skip the Multiple unnamed variables quality check.
ignore-optional-plural: Skip the Unpluralised quality check.

Note

Generally the rule is named ignore-* for any check, using its identifier, so you can use this even for your custom checks.

These flags are understood both in Component configuration settings, per source string settings and in the translation file itself (for example in GNU gettext).

Enforcing checks#

You can configure a list of checks which can not be ignored by setting Enforced checks in Component configuration. Each listed check can not be dismissed in the user interface and any string failing this check is marked as Needs editing (see Translation states).

Note

Turning on check enforcing doesn’t enable it automatically. The check can be turned on by adding the corresponding flag to string or component flags.

Managing fonts#

Hint

Fonts uploaded into Weblate are used purely for purposes of the Maximum size of translation check, they do not have an effect in Weblate user interface.

The Maximum size of translation check used to calculate dimensions of the rendered text needs font to be loaded into Weblate and selected using a translation flag (see Customizing behavior using flags).

Weblate font management tool in Fonts under the Manage menu of your translation project provides interface to upload and manage fonts. TrueType or OpenType fonts can be uploaded, set up font-groups and use those in the check.

The font-groups allow you to define different fonts for different languages, which is typically needed for non-latin languages:

The font-groups are identified by name, which can not contain whitespace or special characters, so that it can be easily used in the check definition:

Font-family and style is automatically recognized after uploading them:

You can have a number of fonts loaded into Weblate:

To use the fonts for checking the string length, pass it the appropriate flags (see Customizing behavior using flags). You will probably need the following ones:

max-size:500: Defines maximal width in pixels.
font-family:ubuntu: Defines font group to use by specifying its identifier.
font-size:22: Defines font size in pixels.

Writing own checks#

A wide range of quality checks are built-in, (see Quality checks), though they might not cover everything you want to check. The list of performed checks can be adjusted using CHECK_LIST, and you can also add custom checks.

Subclass the weblate.checks.Check
Set a few attributes.
Implement either the check (if you want to deal with plurals in your code) or the check_single method (which does it for you).

Some examples:

To install custom checks, provide a fully-qualified path to the Python class in the CHECK_LIST, see Custom quality checks, add-ons and auto-fixes.

Checking translation text does not contain “foo”#

This is a pretty simple check which just checks whether the translation is missing the string “foo”.

# Copyright © Michal Čihař <michal@weblate.org>
#
# SPDX-License-Identifier: GPL-3.0-or-later

"""Simple quality check example."""

from django.utils.translation import gettext_lazy

from weblate.checks.base import TargetCheck


class FooCheck(TargetCheck):
    # Used as identifier for check, should be unique
    # Has to be shorter than 50 characters
    check_id = "foo"

    # Short name used to display failing check
    name = gettext_lazy("Foo check")

    # Description for failing check
    description = gettext_lazy("Your translation is foo")

    # Real check code
    def check_single(self, source, target, unit):
        return "foo" in target

Checking that Czech translation text plurals differ#

Check using language info to verify the two plural forms in Czech language are not same.

# Copyright © Michal Čihař <michal@weblate.org>
#
# SPDX-License-Identifier: GPL-3.0-or-later

"""Quality check example for Czech plurals."""

from django.utils.translation import gettext_lazy

from weblate.checks.base import TargetCheck


class PluralCzechCheck(TargetCheck):
    # Used as identifier for check, should be unique
    # Has to be shorter than 50 characters
    check_id = "foo"

    # Short name used to display failing check
    name = gettext_lazy("Foo check")

    # Description for failing check
    description = gettext_lazy("Your translation is foo")

    # Real check code
    def check_target_unit(self, sources, targets, unit):
        if self.is_language(unit, ("cs",)):
            return targets[1] == targets[2]
        return False

    def check_single(self, source, target, unit):
        """We don't check target strings here."""
        return False