Changes dependency of HTML Importer from Mechanize to just Nokogiri

Gemfile

@@ -7,8 +7,8 @@ group :optional do
   gem 'dbd-sqlite3'
   gem 'dbi'
   gem 'jsonpath'
-  gem 'mechanize'
   gem 'mongo'
+  gem 'nokogiri'
   gem 'redis'
   gem 'roo', '~> 2.7.0'
   gem 'rsruby'

README.md

@@ -192,7 +192,7 @@ Imports a **Daru::DataFrame** from a **.xlsx** file.
 Imports an **Array** of **Daru::DataFrame**s from a **.html** file or website.
 
 - **Docs**: [rubydoc.info](http://www.rubydoc.info/github/athityakumar/daru-io/master/Daru/IO/Importers/HTML)
-- **Gem Dependencies**: `mechanize` gem
+- **Gem Dependencies**: `nokogiri` gem
 - **Usage**:
     ```ruby
     #! Partially require just HTML Importer

lib/daru/io/importers/html.rb

@@ -14,7 +14,8 @@ class HTML < Base
 
         # Checks for required gem dependencies of HTML Importer
         def initialize
-          optional_gem 'mechanize'
+          require 'open-uri'
+          optional_gem 'nokogiri'
         end
 
         # Reads from a html file / website
@@ -29,7 +30,7 @@ def initialize
         # @example Reading from a website url file
         #   instance = Daru::IO::Importers::HTML.read('http://www.moneycontrol.com/')
         def read(path)
-          @file_data = Mechanize.new.get(path)
+          @file_data = Nokogiri.parse(open(path).read)
           self
         end
 
@@ -72,25 +73,23 @@ def read(path)
         #   #   3        ITC     315.85       6.75     621.12
         #   #   4       HDFC    1598.85      50.95     553.91
         def call(match: nil, order: nil, index: nil, name: nil)
-          @match = match
-          @options = {name: name, order: order, index: index}
+          @match   = match
+          @options = {name: name, index: index, order: order}
 
           @file_data
-            .search('table').map { |table| parse_table table }
-            .keep_if { |table| search table }
+            .search('table')
+            .map { |table| parse_table(table) }
             .compact
-            .map { |table| decide_values table, @options }
-            .map { |table| table_to_dataframe table }
+            .keep_if { |table| satisfy_dimension(table) && search(table) }
+            .map { |table| decide_values(table, @options) }
+            .map { |table| table_to_dataframe(table) }
         end
 
         private
 
         # Allows user to override the scraped order / index / data
-        def decide_values(scraped_val={}, user_val={})
-          %I[data index name order].each do |key|
-            user_val[key] ||= scraped_val[key]
-          end
-          user_val
+        def decide_values(scraped_val, user_val)
+          scraped_val.merge(user_val) { |_key, scraped, user| user || scraped }
         end
 
         # Splits headers (all th tags) into order and index. Wherein,
@@ -121,15 +120,23 @@ def scrape_tag(table, tag)
           [arr, size]
         end
 
+        def satisfy_dimension(table)
+          return false if @options[:order] && table[:data].first.size != @options[:order].size
+          return false if @options[:index] && table[:data].size != @options[:index].size
+          true
+        end
+
         def search(table)
-          @match.nil? ? true : (table.to_s.include? @match)
+          @match.nil? ? true : table.to_s.include?(@match)
         end
 
         def table_to_dataframe(table)
-          Daru::DataFrame.rows table[:data],
+          Daru::DataFrame.rows(
+            table[:data],
             index: table[:index],
             order: table[:order],
             name: table[:name]
+          )
         end
       end
     end

spec/daru/io/importers/html_spec.rb

@@ -5,10 +5,10 @@
   let(:df_index) { 0  }
 
   context 'in wiki info table' do
-    let(:path)  { "file://#{Dir.pwd}/spec/fixtures/html/wiki_table_info.html" }
-    let(:order) { %w[FName LName Age]                                         }
-    let(:index) { %w[One Two Three Four Five Six Seven]                       }
-    let(:name)  { 'Wikipedia Information Table'                               }
+    let(:path)  { 'spec/fixtures/html/wiki_table_info.html' }
+    let(:order) { %w[FName LName Age]                       }
+    let(:index) { %w[One Two Three Four Five Six Seven]     }
+    let(:name)  { 'Wikipedia Information Table'             }
 
     context 'returns default dataframe' do
       it_behaves_like 'exact daru dataframe',
@@ -40,7 +40,7 @@
   end
 
   context 'in wiki climate data' do
-    let(:path) { "file://#{Dir.pwd}/spec/fixtures/html/wiki_climate.html" }
+    let(:path) { 'spec/fixtures/html/wiki_climate.html' }
 
     context 'returns default dataframe' do
       it_behaves_like 'exact daru dataframe',
@@ -58,9 +58,9 @@
   end
 
   context 'with valid html table markups' do
-    let(:path)  { "file://#{Dir.pwd}/spec/fixtures/html/valid_markup.html" }
-    let(:index) { %w[W X Y Z]                                              }
-    let(:name)  { 'Small HTML table with index'                            }
+    let(:path)  { 'spec/fixtures/html/valid_markup.html' }
+    let(:index) { %w[W X Y Z]                            }
+    let(:name)  { 'Small HTML table with index'          }
 
     context 'returns user-modified dataframe' do
       let(:opts) { {index: index, name: name} }
@@ -76,9 +76,9 @@
   end
 
   context 'in year-wise passengers figure' do
-    let(:path)  { "file://#{Dir.pwd}/spec/fixtures/html/macau.html" }
-    let(:match) { '2001'                                            }
-    let(:name)  { 'Year-wise Passengers Figure'                     }
+    let(:path)  { 'spec/fixtures/html/macau.html' }
+    let(:match) { '2001'                          }
+    let(:name)  { 'Year-wise Passengers Figure'   }
 
     context 'returns matching dataframes with index' do
       let(:opts) { {match: match, name: name} }
@@ -108,9 +108,9 @@
   end
 
   context 'in share market data' do
-    let(:path)  { "file://#{Dir.pwd}/spec/fixtures/html/moneycontrol.html" }
-    let(:match) { 'Sun Pharma'                                             }
-    let(:index) { %w[Alpha Beta Gamma Delta Misc]                          }
+    let(:path)  { 'spec/fixtures/html/moneycontrol.html' }
+    let(:match) { 'Sun Pharma'                           }
+    let(:index) { %w[Alpha Beta Gamma Delta Misc]        }
     let(:name)  { 'Share Market Analysis' }
 
     context 'returns matching dataframes' do
@@ -149,7 +149,7 @@
   end
 
   context 'in election results data' do
-    let(:path) { "file://#{Dir.pwd}/spec/fixtures/html/eciresults.html" }
+    let(:path) { 'spec/fixtures/html/eciresults.html' }
 
     context 'returns default dataframes' do
       it_behaves_like 'exact daru dataframe',

spec/spec_helper.rb

@@ -19,7 +19,7 @@
 require 'redis'
 require 'dbi'
 require 'jsonpath'
-require 'mechanize'
+require 'nokogiri'
 require 'mongo'
 require 'spreadsheet'
 require 'sqlite3'