From 3c12e3a7c9934fbb55165979fcb3e313c3f1316a Mon Sep 17 00:00:00 2001
From: Dave Machado <dave@davemachado.com>
Date: Sat, 23 Dec 2017 19:24:15 -0500
Subject: [PATCH] Replace Ruby link validation with Python

---
 build/validate_links.py | 53 +++++++++++++++++++++++++++
 build/validate_links.rb | 81 -----------------------------------------
 2 files changed, 53 insertions(+), 81 deletions(-)
 create mode 100755 build/validate_links.py
 delete mode 100755 build/validate_links.rb

diff --git a/build/validate_links.py b/build/validate_links.py
new file mode 100755
index 00000000..49bb4f88
--- /dev/null
+++ b/build/validate_links.py
@@ -0,0 +1,53 @@
+#!/usr/bin/env python3
+
+import httplib2
+import json
+import socket
+import sys
+
+
+def parse_links(filename):
+    """Returns a list of links from JSON object"""
+    data = json.load(open(filename))
+    links = []
+    for entry in data['entries']:
+        link = entry['Link']
+        https = True if link.startswith('https') else False
+        x = {
+            'link': link,
+            'https': https,
+        }
+        links.append(x)
+    return links
+
+
+def validate_links(links):
+    """Checks each entry in JSON file for live link"""
+    print('Validating {} links...'.format(len(links)))
+    errors = []
+    for each in links:
+        link = each['link']
+        h = httplib2.Http(disable_ssl_certificate_validation=True, timeout=5)
+        try:
+            resp = h.request(link, 'HEAD')
+            code = int(resp[0]['status'])
+            # check if status code is a client or server error
+            if code >= 404:
+                errors.append('{}: {}'.format(code, link))
+        except TimeoutError:
+            errors.append("TMO: " + link)
+        except socket.error as socketerror:
+            errors.append("SOC: {} : {}".format(socketerror, link))
+    return errors
+
+if __name__ == "__main__":
+    num_args = len(sys.argv)
+    if num_args < 2:
+        print("No .json file passed")
+        sys.exit(1)
+    errors = validate_links(parse_links(sys.argv[1]))
+    if len(errors) > 0:
+        for err in errors:
+            print(err)
+        sys.exit(1)
+
diff --git a/build/validate_links.rb b/build/validate_links.rb
deleted file mode 100755
index d236e7c5..00000000
--- a/build/validate_links.rb
+++ /dev/null
@@ -1,81 +0,0 @@
-#!/usr/bin/env ruby
-require 'httparty'
-require 'ruby-progressbar'
-require 'uri'
-allowed_codes = [200, 302, 403, 429]
-allowed_links = ["https://www.yelp.com/developers/documentation/v3"]
-args = ARGV
-filename = args[0]
-contents = File.open(filename, 'rb') { |f| f.read }
-raw_links = URI.extract(contents, ['http', 'https'])
-# Remove trailing ')' from entry URLs
-links = []
-raw_links.each do |link|
-    if link.end_with?(')')
-        links.push(link[0...-1])
-    else
-        links.push(link)
-    end
-end
-if links.length <= 0
-    puts "no links to check"
-    exit(0)
-end
-fails = []
-# Fail on any duplicate elements
-dup = links.select{|element| links.count(element) > 1}
-if dup.uniq.length > 0
-    dup.uniq.each do |e|
-        fails.push("(DUP): #{e}")
-    end
-end
-# Remove any duplicates from array
-links = links.uniq
-count = 0
-total = links.length
-progressbar = ProgressBar.create(:total => total,
-    :format => "%a %P% | Processed: %c from %C")
-# GET each link and check for valid response code from allowed_codes
-links.each do |link|
-    begin
-        count += 1
-        if allowed_links.include?(link)
-            next
-        end
-        res = HTTParty.get(link, timeout: 10)
-        if res.code.nil?
-            fails.push("(NIL): #{link}")
-            next
-        end
-        if !allowed_codes.include?(res.code)
-            fails.push("(#{res.code}): #{link}")
-        end
-    rescue HTTParty::RedirectionTooDeep
-        fails.push("(RTD): #{link}")
-    rescue Net::ReadTimeout
-        fails.push("(TMO): #{link}")
-    rescue Net::OpenTimeout
-        fails.push("(TMO): #{link}")
-    rescue OpenSSL::SSL::SSLError
-        fails.push("(SSL): #{link}")
-    rescue SocketError
-        fails.push("(SOK): #{link}")
-    rescue Errno::ECONNREFUSED
-        fails.push("(CON): #{link}")
-    rescue Errno::ECONNRESET
-        next
-    end
-    progressbar.increment
-end
-puts "#{count}/#{total} links checked"
-if fails.length <= 0
-    puts "all links valid"
-    exit(0)
-else
-    puts "-- RESULTS --"
-    fails.sort!
-    fails.each do |e|
-        puts e
-    end
-    exit(1)
-end