From 75a085b92d1ce5711f04903e6a150d8caa8a2a55 Mon Sep 17 00:00:00 2001
From: Mustafa Cagri Ardic <mardic@Mustafas-MBP.lan>
Date: Thu, 18 Apr 2024 22:29:35 +0100
Subject: [PATCH] add pandas 2.2.2 documentation

---
 docs/file-scrapers.md       |  5 ++++-
 lib/docs/scrapers/pandas.rb | 23 +++++++++++++++++++++++
 2 files changed, 27 insertions(+), 1 deletion(-)

diff --git a/docs/file-scrapers.md b/docs/file-scrapers.md
index 2cc99165..403c8333 100644
--- a/docs/file-scrapers.md
+++ b/docs/file-scrapers.md
@@ -183,10 +183,13 @@ mv ./usr/share/doc/openjdk-16-jre-headless/api/ docs/openjdk~$VERSION
 
 ## Pandas
 
+From the home directory; `devdocs`, execute below:
+
 ```sh
-curl https://pandas.pydata.org/docs/pandas.zip | bsdtar --extract --file - --directory=docs/pandas~1
+curl https://pandas.pydata.org/docs/pandas.zip -o tmp.zip && unzip tmp.zip -d docs/pandas~2 && rm tmp.zip
 ```
 
+
 ## PHP
 Click the link under the "Many HTML files" column on https://www.php.net/download-docs.php, extract the tarball, change its name to `php` and put it in `docs/`.
 
diff --git a/lib/docs/scrapers/pandas.rb b/lib/docs/scrapers/pandas.rb
index d748c297..c95f7561 100644
--- a/lib/docs/scrapers/pandas.rb
+++ b/lib/docs/scrapers/pandas.rb
@@ -16,6 +16,29 @@ module Docs
       Licensed under the 3-clause BSD License.
     HTML
 
+    version '2' do
+      self.release = '2.2.2'
+      self.base_url = "https://pandas.pydata.org/pandas-docs/version/#{self.release}/"
+
+      html_filters.push 'pandas/clean_html', 'pandas/entries'
+
+      options[:container] = 'main section'
+
+      options[:skip_patterns] = [
+        /development/,
+        /getting_started/,
+        /whatsnew/
+      ]
+
+      options[:skip] = [
+        'panel.html',
+        'pandas.pdf',
+        'pandas.zip',
+        'ecosystem.html'
+      ]
+
+    end
+
     version '1' do
       self.release = '1.5.0'
       self.base_url = "https://pandas.pydata.org/pandas-docs/version/#{self.release}/"