Adding Additional Rules via XML

[erik-buchholz]

I was wondering if there is any way to add custom rules via XML to the bundled LanguageTool server?

For instance, I might have an XML rule discouraging the use of "it is":

<rule id="IT_IS" name="it is">
 <pattern>
  <token>it</token>
  <token>is</token>
 </pattern>
 <message>Try avoiding "it is" if possible.</message>
</rule>

This is possible when running a standalone LanguageTool server on localhost and adding the rules to grammar_custom.xml; however, I prefer the convenience of using the bundled LanguageTool server in LTeX.

[spitzerd]

The LanguageTool HTTP interface does not support custom rules, see https://languagetool.org/http-api/swagger-ui/#!/default/post_check
According to my knowledge, same is true for the Java interface.

You have to open an issue or pull request at the LanguageTool repo first. Afterwards, we can implement the feature into LTeX+

As you correctly wrote, you can use a standalone LanguageTool server instead of the bundled LanguageTool if you want to use custom rules.

[erik-buchholz]

Thanks for the quick response @spitzerd!

If anyone finds this topic in the future, here is a guide on how you can make LTeX(+) work with custom grammar rules.
This guide is for macOS, but will work similarly on Windows. You just have to change the paths.

Use LanguageTool with Custom Rules

The following guide is based on pl-lt-rules but has beed adapted to work with the latest LanguageTool version (6.5).

1. Install LanguageTool

See: Official Guide.

On macOS:

brew install languagetool

2. Get the path to the grammar.xml file

On macOS, the path is usually:

ls /opt/homebrew/Cellar/languagetool/*/libexec/org/languagetool/rules/en/grammar.xml

This path is different on Linux and Windows, make sure to adjust the path accordingly.

3. Set the environment variable PATH_TO_GRAMMAR to the output of the previous command

export PATH_TO_GRAMMAR=$(ls /opt/homebrew/Cellar/languagetool/*/libexec/org/languagetool/rules/en/grammar.xml)

Again, adjust the path if you are not on macOS.

4. Install the patch to the grammar.xml file

sed -e "s|home/username|$HOME|" lt.patch > lt.patch.tmp
patch -p1 $PATH_TO_GRAMMAR lt.patch.tmp
rm lt.patch.tmp

lt.patch: (Note that the patch has changed since the guide by Joey Eremondi)

--- /tmp/b	2022-06-06 16:37:42.000000000 -0700
+++ /tmp/a	2022-06-06 17:00:59.000000000 -0700
@@ -29,6 +29,7 @@
 -->

 <!DOCTYPE rules [
+    <!ENTITY UserRules SYSTEM "file:///home/username/.language-tool-rules.xml">
     <!ENTITY fractions "½|↉|⅓|⅔|¼|¾|⅕|⅖|⅗|⅘|⅙|⅚|⅐|⅛|⅜|⅝|⅞|⅑|⅒">
     <!ENTITY particles_of "de|von|van|of|d'|di|du|da|ibn|des|'s">
     <!ENTITY exceptions_multitoken_sensitive "(?-i)[A-Z]+\d+|[a-z][a-z]?\d+">
@@ -598,6 +599,10 @@
     </phrase>
   </phrases>

+  <category id="CUSTOM_USER_RULES" name="Custom User Rules" default="on">
+    &UserRules;
+  </category>
+
     <!--
     <category id="APRIL" name="April 1st">
         <rule id="APRIL_1ST" name="good advice">

You could change file:///home/username/.language-tool-rules.xml to the path of your custom rules so that you don't have to link the file below. In this case, you can skip steps 5-7.

5. Define path to custom rules

export CUSTOM_RULES="PATH/TO/CUSTOM/RULES"

For example, you can use the rules from pl-lt-rules. I might create a repository with my custom rules in the future, too. You can certainly find some inspiration how to write your own rules in the pl-lt-rules repository.

6. Link the custom rules to the specified location

ln -s $CUSTOM_RULES $HOME/.language-tool-rules.xml

7. Check that file exists

head -n 10 $HOME/.language-tool-rules.xml

8. [Optional] Define Path to N-gram model

export PATH_TO_NGRAM_MODEL=".../n-gram"

Replace the path.
You can download the n-gram model from here.
The directory structure should look like this:

.../n-gram
├── en
│   ├── 1grams
│   ├── 2grams
│   └── 3grams

9. [Optional] Define Path to LanguageTool Configuration

export PATH_TO_LT_CONFIG="/PATH/TO/LANGUAGETOOL/CONFIG"

You can see the available configuration options by running languagetool-server --help.

10. Start LanguageTool Server

# Define the configuration parameters (only use --languageModel if you have a n-gram model)
languagetool-server --languageModel $PATH_TO_NGRAM_MODEL --port 8081 --config $PATH_TO_LT_CONFIG --verbose

11. Use this server in the LTeX config (VSCode settings.json)

{
    "ltex.languageToolHttpServerUri": "http://localhost:8081/",
}