[CCK-4] Add plain text files to build process of artifacts

cc_licenses/templates/includes/legalcode_40_license.html

@@ -1,7 +1,7 @@
 {% load bidi i18n license_tags %}
 <div id="legal-code-body" class="padding-larger margin-top-bigger has-text-black">
   {% include 'includes/snippet/icon_header_snippet.html' %}
-  <div>
+  <div id="plain-text-marker"> {# see publish management command #}
     <h3 class="padding-bottom-normal b-header">
       {% if legalcode|is_one_of:"by" %}{# Long full title #}
         {% blocktrans trimmed %}Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International Public License{% endblocktrans %}

cc_licenses/templates/includes/view_legal_code_link_plain_text.html

@@ -1,5 +1,5 @@
 {% load i18n %}
 
 <div id="legal-code-plain-text" style="font-weight: bold;" class="padding-vertical-normal" >
-  <p class="body-big"><a href="#" class="link">{% trans "View Legal Code as plain text" %}</a></p>
+  <p class="body-big"><a href="{{ request.get_full_path }}.txt" class="link" download="{{ legalcode.fat_code }}.txt">{% trans "View Legal Code as plain text" %}</a></p>
 </div>

licenses/management/commands/publish.py

@@ -1,8 +1,12 @@
+import fnmatch
 import os
+import subprocess
+import tempfile
 from argparse import ArgumentParser
 from shutil import rmtree
 
 import git
+from bs4 import BeautifulSoup
 from django.conf import settings
 from django.core.management import BaseCommand, CommandError
 from django_distill.distill import urls_to_distill
@@ -57,6 +61,35 @@ def add_arguments(self, parser: ArgumentParser):
     def _quiet(self, *args, **kwargs):
         pass
 
+    def output_text_files(self, *args, **kwargs):
+        output_dir = getattr(settings, "DISTILL_DIR", None)
+        licenses_dir = f"{output_dir}licenses/"
+        pattern = "legalcode*"
+        for root, dirs, files in os.walk(licenses_dir):
+            for filename in fnmatch.filter(files, pattern):
+                file_path = os.path.join(root, filename)
+                txt_filename = f"{filename}.txt"
+                txt_file_path = os.path.join(root, txt_filename)
+                with open(file_path, "r") as f:
+                    soup = BeautifulSoup(f, "lxml")
+                    plain_text_soup = soup.find(id="plain-text-marker")
+                    plain_text_html = plain_text_soup.prettify(formatter="html")
+                with tempfile.NamedTemporaryFile(mode="w+t") as temp:
+                    temp.write(plain_text_html)
+                    temp.seek(0)
+                    subprocess.run(
+                        [
+                            "pandoc",
+                            "-f",
+                            "html",
+                            temp.name,
+                            "-t",
+                            "plain",
+                            "-o",
+                            txt_file_path,
+                        ]
+                    )
+
     def run_django_distill(self):
         """Outputs static files into the specified directory determined by settings.base.DISTILL_DIR
         """
@@ -79,6 +112,7 @@ def publish_branch(self, branch: str):
         with git.Repo(settings.TRANSLATION_REPOSITORY_DIRECTORY) as repo:
             setup_local_branch(repo, branch, settings.OFFICIAL_GIT_BRANCH)
             self.run_django_distill()
+            self.output_text_files()
             if repo.is_dirty():
                 repo.index.add("build")
                 commit_and_push_changes(repo, "Updated built HTML files")