From 2b152d09881559a0330b3ff923e03e715777c6c3 Mon Sep 17 00:00:00 2001
From: Kelly Rauchenberger <fefferburbia@gmail.com>
Date: Wed, 27 Feb 2019 20:45:17 -0500
Subject: Initial commit (by Pink!)

---
 get.rb | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)
 create mode 100644 get.rb

(limited to 'get.rb')

diff --git a/get.rb b/get.rb
new file mode 100644
index 0000000..a2a213e
--- /dev/null
+++ b/get.rb
@@ -0,0 +1,28 @@
+require 'open-uri'
+require 'nokogiri'
+require 'csv'
+
+result = []
+transcripts = open('https://steven-universe.fandom.com/wiki/Category:Transcripts').read
+docTrans = Nokogiri::HTML transcripts
+docTrans.css(".category-page__member-link").each do |node|
+  puts node['href']
+  subpage = open("https://steven-universe.fandom.com" + node['href']).read
+  subpagedoc = Nokogiri::HTML subpage
+  rows = subpagedoc.css(".bgrevo tr")
+  rows.shift
+  rows.pop
+  rows.each do |row|
+    if row.children.length == 2
+      result << ["", row.children[1].content.strip.gsub(/\n/," ")]
+    elsif row.children.length == 3
+      result << [row.children[1].content.strip, row.children[2].content.strip.gsub(/\n/," ")]
+    end
+  end
+end
+
+CSV.open("dialogue.csv", "w") do |csv|
+  result.each do |line|
+    csv << line
+  end
+end
-- 
cgit 1.4.1