Handle the <math> HTML tag

[BoboTiG]

Wiktionary page: https://fr.wiktionary.org/wiki/octonion

Wikicode:

<math>x=x_0+x_1{\rm i}+x_2{\rm j}+x_3{\rm k}+x_4{\rm l}+x_5{\rm il}+x_6{\rm jl}+x_7{\rm kl}</math>

Output:

x=x_0+x_1{\rm i}+x_2{\rm j}+x_3{\rm k}+x_4{\rm l}+x_5{\rm il}+x_6{\rm jl}+x_7{\rm kl}

Expected:

Not sure if we can and how to display it.

[BoboTiG]

Kobo is able to display pictures, we should try to generate ones using Tex or LaTex. It must be included using data URI base64 encoded. See https://fr.m.wikipedia.org/wiki/Aide:Formules_TeX and https://pgaskin.net/dictutil/dicthtml/format.html#gif-jpg-etc.

[BoboTiG]

diff --git a/.github/workflows/auto-updates.yml b/.github/workflows/auto-updates.yml
index 45850b5..f8ab3b7 100644
--- a/.github/workflows/auto-updates.yml
+++ b/.github/workflows/auto-updates.yml
@@ -24,6 +24,9 @@ jobs:
       with:
         python-version: 3.8
 
+    - name: Install LaTeX requirements
+      run: sudo apt install dvipng texlive-latex-base
+
     - name: Install requirements
       run: python -m pip install -r requirements.txt
 
diff --git a/scripts/utils.py b/scripts/utils.py
index 63a28c6..b0da64c 100644
--- a/scripts/utils.py
+++ b/scripts/utils.py
@@ -4,7 +4,7 @@ from contextlib import suppress
 from datetime import datetime
 from functools import lru_cache
 from pathlib import Path
-from typing import Tuple
+from typing import Match, Tuple
 from warnings import warn
 
 from .constants import DOWNLOAD_URL
@@ -277,9 +277,29 @@ def clean(word: str, text: str, locale: str) -> str:
     text = sub(r"\s{2,}", " ", text)
     text = sub(r"\s{1,}\.", ".", text)
 
+    # Handle the <math> case
+    text = sub(r"<math>(.+)</math>", convert_math, text)
+
     return text.strip()
 
 
+def convert_math(match: Match[str]) -> str:
+    """"""
+    expr: str = match.group(1) if isinstance(match, re.Match) else match
+
+    import subprocess
+    cmd = ["./pnglatex", "-f", expr, "-o", "file.png", "-d", "96"]
+    subprocess.check_call(cmd)
+
+    import base64
+    img = '<img src="data:image/png;base64,'
+    with open("file.png", "rb") as f:
+        img += base64.encodebytes(f.read()).decode()
+    img += '"/>'
+
+    return img
+
+
 def transform(word: str, template: str, locale: str) -> str:
     """Convert the data from the *template" template.
     This function also checks for template style.

Needs https://github.com/mneri/pnglatex at the root of the repository.

It works but I am not sure we want that. It will break Kobo using a firmware < 4.20.*.

Screenshot from the browser:

[lasconic]

FWIW

$ cat data/fr/data.json | grep "<math>" | wc -l
59

[BoboTiG]

Yes, but considering all locales, it may be interesting to handle it. For english:

$ cat data/en/data.json | grep "<math>" | wc -l
464

Not a big priority though.

[lasconic]

The tag could also be supported by the same code.

Interesting reading : https://en.wikipedia.org/wiki/Help:Displaying_a_formula

Use of chem in FR : https://fr.wiktionary.org/wiki/dim%C3%A9thylpolysiloxane