Add config setting to disable cross language search

docs/locale/es/config.md.po

@@ -188,3 +188,13 @@ msgstr ""
 
 msgid "Content"
 msgstr "Contenido"
+
+msgid ""
+"By default is enabled. You can disable it to restrict the search to the "
+"active language."
+msgstr ""
+"Por defecto está habilitado. Puedes deshabilitarlo para restringir la "
+"búsqueda al idioma activo."
+
+msgid "Currently is only compatible with the [mkdocs-material] theme."
+msgstr "Actualmente sólo es compatible con el tema [mkdocs-material]."

docs/src/config.md

@@ -203,6 +203,14 @@ File extensions that are ignored from being added to site directory, defaults to
 You can ignore certain messages from being dumped into PO files adding them to
 this list.
 
+<!-- mdpo-disable-next-line -->
+### **`cross_language_search`** (*bool*)
+
+By default is enabled. You can disable it to restrict the search to the active
+language.
+
+Currently is only compatible with the [mkdocs-material] theme.
+
 [iso-369]: https://en.wikipedia.org/wiki/ISO_639
 [mkdocs-material]: https://squidfunk.github.io/mkdocs-material
 [mkdocs-material-site-language]: https://squidfunk.github.io/mkdocs-material/setup/changing-the-language/#site-language

mkdocs.yml

@@ -81,6 +81,7 @@ plugins:
       root_domain: true
   - mdpo:
       locale_dir: ../locale
+      cross_language_search: false
   - minify:
       minify_html: true
 

mkdocs_mdpo_plugin/config.py

@@ -19,6 +19,7 @@
     ),
     ('ignore_extensions', Type(list, default=['.po', '.pot', '.mo'])),
     ('ignore_msgids', Type(list, default=[])),
+    ('cross_language_search', Type(bool, default=True)),
 )
 
 

mkdocs_mdpo_plugin/plugin.py

@@ -24,7 +24,11 @@
     MkdocsBuild,
     set_on_build_error_event,
 )
-from mkdocs_mdpo_plugin.translations import Translation, Translations
+from mkdocs_mdpo_plugin.translations import (
+    Translation,
+    Translations,
+    TranslationSearchIndexes,
+)
 
 
 class MdpoPlugin(mkdocs.plugins.BasePlugin):
@@ -329,6 +333,7 @@ def on_page_markdown(self, markdown, page, config, files):
                 _translated_entries_msgstrs.append(entry.msgstr)
                 _translated_entries_msgids.append(entry.msgid)
 
+            # create translation object
             translation = Translation(
                 language,
                 po,
@@ -349,6 +354,7 @@ def on_page_markdown(self, markdown, page, config, files):
             self.translations.nav[page.title][language] = [
                 translated_page_title, new_page.file.url,
             ]
+
             mkdocs.commands.build._populate_page(
                 new_page,
                 config,
@@ -398,6 +404,20 @@ def on_post_page(self, output, page, config):
     def on_post_build(self, config):
         self.translations.tempdir.cleanup()
 
+        if (
+            not self.config['cross_language_search'] and
+            'search' in config['plugins']
+        ):
+            # cross language search is disabled, so build indexes
+            # for each language and patch the 'site_dir' directory
+            search_indexes = TranslationSearchIndexes(
+                config['site_dir'],
+                self.config['languages'],
+                self.config['default_language'],
+            )
+            search_indexes.patch_site_dir()
+
+        # save PO files
         for translations in self.translations.all.values():
             for translation in translations:
                 translation.po.save(translation.po_filepath)

mkdocs_mdpo_plugin/translations.py

@@ -1,3 +1,6 @@
+import json
+import os
+import re
 import tempfile
 
 
@@ -96,3 +99,125 @@ def __str__(self):  # pragma: no cover
             f' current={current}'
             ')'
         )
+
+
+class TranslationSearchIndexes:
+    def __init__(self, site_dir, languages, default_language):
+        self.site_dir = site_dir
+        self.search_index_json_path = os.path.join(
+            site_dir,
+            'search',
+            'search_index.json',
+        )
+
+        with open(self.search_index_json_path) as f:
+            self.search_index_json = json.load(f)
+
+        self.search_index = self.search_index_json['docs']
+
+        self.languages = languages
+        self.default_language = default_language
+
+        self.lang_search_indexes = {default_language: []}  # {lang: [records]}
+
+        self.index_loader_js = self._search_index_loader_js_filepath()
+
+    def patch_site_dir(self):
+        for language in self.languages:
+            # build indexes for languages
+            self.lang_search_indexes[language] = []
+
+            if language == self.default_language:
+                lang_matchers = tuple(f'{lang}/' for lang in self.languages)
+                for record in self.search_index:
+                    if not record['location'].startswith(lang_matchers):
+                        self.lang_search_indexes[language].append(record)
+            else:
+                for record in self.search_index:
+                    if record['location'].startswith(f'{language}/'):
+                        self.lang_search_indexes[language].append(record)
+
+            lang_search_index_path, lang_search_index = (
+                self._lang_search_index_json(
+                    language,
+                    self.lang_search_indexes[language],
+                )
+            )
+
+            # create JSON indexes
+            with open(lang_search_index_path, 'w') as f:
+                f.write(lang_search_index)
+
+            # create javascript assets by language to load custom
+            # search_index_{language}.json files depending on the
+            # theme used
+
+            # generate js loader for the language
+            lang_index_loader_js_path, lang_index_loader_js = (
+                self._lang_index_loader_js(language)
+            )
+            with open(lang_index_loader_js_path, 'w') as f:
+                f.write(lang_index_loader_js)
+
+        self._patch_html_files()
+
+    def _lang_search_index_json(self, language, records):
+        search_index = self.search_index_json
+        search_index['docs'] = records
+        if 'config' in search_index and 'lang' in search_index['config']:
+            search_index['config']['lang'] = [language]
+        return (
+            self.search_index_json_path.rstrip('.json') + f'_{language}.json',
+            json.dumps(search_index),
+        )
+
+    def _lang_index_loader_js(self, language):
+        return (
+            self.index_loader_js['path'].rstrip('.js') + f'_{language}.js',
+            self.index_loader_js['content'].replace(
+                '/search/search_index.json',
+                f'/search/search_index_{language}.json',
+            ),
+        )
+
+    def _patch_html_files(self):
+        for root, _, files in os.walk(self.site_dir):
+            for fname in files:
+                if fname.endswith('.html'):
+                    fpath = os.path.join(root, fname)
+                    with open(fpath) as f:
+                        content = f.read()
+                    match = re.search(r'lang="([^"]+)"', content)
+                    if match:
+                        language = match.group(1)
+                        index_loader_js_fname = os.path.basename(
+                            self.index_loader_js['path'],
+                        )
+                        index_loader_js_fname_lang = (
+                            index_loader_js_fname.rstrip('.js')
+                            + f'_{language}.js'
+                        )
+                        new_content = content.replace(
+                            f'assets/javascripts/{index_loader_js_fname}"',
+                            (
+                                'assets/javascripts/'
+                                f'{index_loader_js_fname_lang}"'
+                            ),
+                        )
+                        with open(fpath, 'w') as f:
+                            f.write(new_content)
+
+    def _search_index_loader_js_filepath(self):
+        javascripts_dir = os.path.join(self.site_dir, 'assets', 'javascripts')
+        js_loader_filepath, js_loader_content = None, None
+        if os.path.isdir(javascripts_dir):
+            for fname in os.listdir(javascripts_dir):
+                if fname.endswith('.js'):
+                    fpath = os.path.join(javascripts_dir, fname)
+                    with open(fpath) as f:
+                        content = f.read()
+                    if '/search/search_index.json' in content:
+                        js_loader_filepath = fpath
+                        js_loader_content = content
+                        continue
+        return {'path': js_loader_filepath, 'content': js_loader_content}