Simplify code and move download and extract to pandas_web

goanpeca · goanpeca · commit 1c5cb626bb04 · 2025-05-19T18:15:54.000-05:00
diff --git a/web/pandas/_templates/layout.html b/web/pandas/_templates/layout.html
@@ -15,8 +15,82 @@
                   href="{{ base_url }}{{ stylesheet }}">
         {% endfor %}
         <link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/bootstrap-icons@1.11.3/font/bootstrap-icons.min.css">
-        <meta id="languages" data-lang="{{ languages }}">
-        <script src="{{ base_url }}static/js/language_switcher.js"></script>
+        <script type="text/javascript">
+            window.addEventListener("DOMContentLoaded", function() {
+            var absBaseUrl = document.baseURI;
+            var baseUrl = location.protocol + "//" + location.hostname
+            if (location.port) {
+                baseUrl = baseUrl + ":" + location.port
+            }
+            var currentLanguage = document.documentElement.lang;
+            var languages = [
+                {% for lang, name in translations["languages"].items() -%}
+                    "{{ lang }}",
+                {% endfor -%}
+            ]
+            var languageNames = {
+                {% for lang, name in translations["languages"].items() -%}
+                    {{ lang }}: '{{ name }}',
+                {% endfor -%}
+            }
+
+            // Handle preview URLs on github
+            // If preview URL changes, this regex will need to be updated
+            var re = /preview\/pandas-dev\/pandas\/(?<pr>[0-9]*)\//g;
+            var previewUrl = '';
+            for (const match of absBaseUrl.matchAll(re)) {
+                previewUrl = `/preview/pandas-dev/pandas/${match.groups.pr}`;
+            }
+            var pathName = location.pathname.replace(previewUrl, '')
+
+            // Create dropdown menu
+            function makeDropdown(options) {
+                var dropdown = document.createElement("li");
+                dropdown.classList.add("nav-item");
+                dropdown.classList.add("dropdown");
+
+                var link = document.createElement("a");
+                link.classList.add("nav-link");
+                link.classList.add("dropdown-toggle");
+                link.setAttribute("data-bs-toggle", "dropdown");
+                link.setAttribute("href", "#");
+                link.setAttribute("role", "button");
+                link.setAttribute("aria-haspopup", "true");
+                link.setAttribute("aria-expanded", "false");
+                link.textContent = languageNames[currentLanguage];
+
+                var dropdownMenu = document.createElement("div");
+                dropdownMenu.classList.add("dropdown-menu");
+
+                options.forEach(function(i) {
+                var dropdownItem = document.createElement("a");
+                dropdownItem.classList.add("dropdown-item");
+                dropdownItem.textContent = languageNames[i] || i.toUpperCase();
+                dropdownItem.setAttribute("href", "#");
+                dropdownItem.addEventListener("click", function() {
+                    var urlLanguage = '';
+                    if (i !== 'en') {
+                    urlLanguage = '/' + i;
+                    }
+                    pathName = pathName.replace('/' + currentLanguage + '/', '/')
+                    var newUrl = baseUrl + previewUrl + urlLanguage + pathName
+                    window.location.href = newUrl;
+                });
+                dropdownMenu.appendChild(dropdownItem);
+                });
+
+                dropdown.appendChild(link);
+                dropdown.appendChild(dropdownMenu);
+                return dropdown;
+            }
+
+            var container = document.getElementById("language-switcher-container");
+            if (container) {
+                var dropdown = makeDropdown(languages);
+                container.appendChild(dropdown);
+            }
+            });
+        </script>
     </head>
     <body>
         <header>
diff --git a/web/pandas/config.yml b/web/pandas/config.yml
@@ -177,5 +177,8 @@ translations:
   source_path: pandas-translations-main/web/pandas/
   default_language: 'en'
   default_prefix: ''
-  ignore:
-    - docs/
+  languages:
+    en: English
+    es: Español
+    pt: Português
+    # fr: Français
diff --git a/web/pandas/navbar.yml b/web/pandas/navbar.yml
@@ -17,6 +17,7 @@ navbar:
     target: getting_started.html
   - name: "Documentation"
     target: docs/
+    translated: false
   - name: "Community"
     target:
     - name: "Blog"
diff --git a/web/pandas/static/js/language_switcher.js b/web/pandas/static/js/language_switcher.js
diff --git a/web/pandas_web.py b/web/pandas_web.py
@@ -28,6 +28,7 @@
 import collections
 import datetime
 import importlib
+import io
 import itertools
 import json
 import operator
@@ -36,6 +37,7 @@
 import re
 import shutil
 import sys
+import tarfile
 import time
 import typing
 
@@ -74,17 +76,16 @@ def current_year(context: dict) -> dict:
         return context
 
     @staticmethod
-    def navbar_add_info(context: dict, skip: bool = True) -> dict:
+    def navbar_add_info(context: dict) -> dict:
         """
         Items in the main navigation bar can be direct links, or dropdowns with
         subitems. This context preprocessor adds a boolean field
         ``has_subitems`` that tells which one of them every element is. It
         also adds a ``slug`` field to be used as a CSS id.
         """
-        ignore = context["translations"]["ignore"]
-        for language in context["languages"]:
+        for language in context["translations"]["languages"]:
             for i, item in enumerate(context["navbar"][language]):
-                if item["target"] in ignore:
+                if not item.get("translated", True):
                     item["target"] = f"../{item['target']}"
 
                 context["navbar"][language][i] = dict(
@@ -391,9 +392,7 @@ def get_callable(obj_as_str: str) -> object:
     return obj
 
 
-def get_context(
-    config_fname: str, navbar_fname: str, languages: list[str], **kwargs: dict
-) -> dict:
+def get_context(config_fname: str, navbar_fname: str, **kwargs: dict) -> dict:
     """
     Load the config yaml as the base context, and enrich it with the
     information added by the context preprocessors defined in the file.
@@ -403,19 +402,19 @@ def get_context(
 
     context["source_path"] = os.path.dirname(config_fname)
 
-    navbar = {}
-    context["languages"] = languages
     default_language = context["translations"]["default_language"]
     default_prefix = context["translations"]["default_prefix"]
-    for language in languages:
+    translated_languages = context["translations"]["languages"].copy()
+    translated_languages.pop(default_language)
+    context["translated_languages"] = translated_languages
+    download_and_extract_translations(context)
+    navbar = {}
+    for language in context["translations"]["languages"]:
         prefix = default_prefix if language == default_language else language
         navbar_path = os.path.join(context["source_path"], prefix, navbar_fname)
-
         with open(navbar_path, encoding="utf-8") as f:
             navbar_lang = yaml.safe_load(f)
-
         navbar[language] = navbar_lang["navbar"]
-
     context["navbar"] = navbar
 
     context.update(kwargs)
@@ -453,6 +452,28 @@ def extend_base_template(content: str, base_template: str) -> str:
     return result
 
 
+def download_and_extract_translations(context: dict) -> None:
+    """
+    Download the translations from the GitHub repository and extract them.
+    """
+    base_folder = os.path.dirname(__file__)
+    extract_path = os.path.join(base_folder, context["translations"]["folder"])
+    shutil.rmtree(extract_path, ignore_errors=True)
+    response = requests.get(context["translations"]["url"])
+    if response.status_code == 200:
+        doc = io.BytesIO(response.content)
+        with tarfile.open(None, "r:gz", doc) as tar:
+            tar.extractall(os.path.join(base_folder, context["translations"]["folder"]))
+    else:
+        raise Exception(f"Failed to download translations: {response.status_code}")
+    for lang in context["translated_languages"]:
+        shutil.rmtree(os.path.join(context["source_path"], lang), ignore_errors=True)
+        shutil.move(
+            os.path.join(extract_path, context["translations"]["source_path"], lang),
+            os.path.join(base_folder, "pandas", lang),
+        )
+
+
 def main(
     source_path: str,
     target_path: str,
@@ -463,24 +484,14 @@ def main(
     For ``.md`` and ``.html`` files, render them with the context
     before copying them. ``.md`` files are transformed to HTML.
     """
-    base_folder = os.path.dirname(__file__)
-
     shutil.rmtree(target_path, ignore_errors=True)
     os.makedirs(target_path, exist_ok=True)
 
-    # Handle translations
-    sys.path.append(base_folder)
-    trans = importlib.import_module("pandas_translations")
-    translated_languages, languages = trans.process_translations(
-        "config.yml", source_path
-    )
-
     sys.stderr.write("Generating context...\n")
     context = get_context(
         os.path.join(source_path, "config.yml"),
         navbar_fname="navbar.yml",
         target_path=target_path,
-        languages=languages,
     )
     sys.stderr.write("Context generated\n")
 
@@ -490,7 +501,9 @@ def main(
     default_language = context["translations"]["default_language"]
     for fname in get_source_files(source_path):
         selected_language = context["translations"]["default_language"]
-        for language in translated_languages:
+        translated_languages = list(context["translations"]["languages"].keys())
+        translated_languages.remove(default_language)
+        for language in context["translated_languages"]:
             if fname.startswith(language + "/"):
                 selected_language = language
                 break
