crm_formatter 2.4 → 2.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 83777a72332f62d45de21bbd2caa03eb61b05017b84fb57169c879074785c7d5
-  data.tar.gz: 0a7ff7909b01d40fdd053116105a88093f80aed75c25973e3799d8d205b4e6cc
+  metadata.gz: 3272fc47f3c2ccd991ce89ab4e7252b7783f306ebc86a3d03a0af49084509d92
+  data.tar.gz: f0854aa02ed045fde909d4e2240e9cb382240b8e8b2367f618faebb0dc8962de
 SHA512:
-  metadata.gz: b7c6eabaf39cd839ab291548141e74faec397b5d9405ef911a923010205e5ecab2881308deb20001f9cd8d1c4d555d5b0e923e6c872a4028ee2816fec2f1806c
-  data.tar.gz: fe6be76a0cf51ac51189423769871cf7c267a367bbae245626aa3c5e2837f1d1b9f30d8b4d69a08a82d673fe96e5405123cd4e76fa06804bcdb808e38092b189
+  metadata.gz: 55aee04e100de2b9862d9c09129599e9e5cb1460c1c078c1a1a15a649bea7e42a495f436f3034263e01560633640aae1e312daaf39ddd27ca8750aa30d7c0546
+  data.tar.gz: b5e4dd5923b92a24f7b7f235b1059e00f66b8bf20bfde316cbbe83982cfe25e2d2073dcfdbd68c80824c02784944695a177ef40a7712c788dd3ec8c2005cf838

data/README.md

@@ -28,8 +28,125 @@ Or install it yourself as:
 ## Usage
 
 ### I. Basic Usage
+Basic methods available are:
+```
+format_addresses(array_of_addresses)
+format_phones(array_of_phones)
+format_propers(array_of_propers)
+format_urls(array_of_urls)
+```
+
+1. Format Array of Proper Strings:
+Use `format_propers` to format strings with proper nouns, such as (but not limited to):
+
+Business Account Names (123 bmw-world => 123 BMW-World),
+
+Proper Names (adam john booth => Adam John Booth),
+
+Job Titles (marketing director => Marketing Director),
+
+Article Titles (the 15 most useful ruby methods => The 15 Most Useful Ruby Methods)
+
+```
+array_of_propers = [
+  '123 bmw-world',
+  'Car-world Kia',
+  'BUDGET - AUTOMOTORES ZONA & FRANCA, INC',
+  'DOWNTOWN CAR REPAIR, INC',
+  'Young Gmc Trucks',
+  'TEXAS TRAVEL, CO',
+  'youmans Chevrolet',
+  'Hot-Deal auto Insurance',
+  'quick auto approval, inc',
+  'yazell chevy',
+  'quick cAr LUBE',
+  'yAtEs AuTo maLL',
+  'YADKIN VALLEY COLLISION CO',
+  'XIT FORD INC'
+]
+
+formatted_proper_hashes = CrmFormatter.format_propers(array_of_propers)
+```
+
+Formatted Proper Strings:
+
+```
+formatted_proper_hashes = [
+  {
+    :proper_status=>"formatted",
+    :proper=>"123 bmw-world",
+    :proper_f=>"123 BMW-World"
+  },
+   {
+     :proper_status=>"formatted",
+     :proper=>"Car-world Kia",
+     :proper_f=>"Car-World Kia"
+   },
+   {
+     :proper_status=>"formatted",
+     :proper=>"BUDGET - AUTOMOTORES ZONA & FRANCA, INC",
+     :proper_f=>"Budget - Automotores Zona & Franca, Inc"
+   },
+
+   {:proper_status=>"formatted",
+     :proper=>"DOWNTOWN CAR REPAIR, INC",
+     :proper_f=>"Downtown Car Repair, Inc"
+   },
+   {
+     :proper_status=>"formatted",
+     :proper=>"Young Gmc Trucks",
+     :proper_f=>"Young GMC Trucks"
+   },
+   {
+     :proper_status=>"formatted",
+     :proper=>"TEXAS TRAVEL, CO",
+     :proper_f=>"Texas Travel, Co"
+   },
+   {
+     :proper_status=>"formatted",
+     :proper=>"youmans Chevrolet",
+     :proper_f=>"Youmans Chevrolet"
+   },
+   {
+     :proper_status=>"formatted",
+     :proper=>"Hot-Deal auto Insurance",
+     :proper_f=>"Hot-Deal Auto Insurance"
+   },
+   {
+     :proper_status=>"formatted",
+     :proper=>"quick auto approval, inc",
+     :proper_f=>"Quick Auto Approval, Inc"
+   },
+   {
+     :proper_status=>"formatted",
+     :proper=>"yazell chevy",
+     :proper_f=>"Yazell Chevy"
+   },
+   {
+     :proper_status=>"formatted",
+     :proper=>"quick cAr LUBE",
+     :proper_f=>"Quick Car Lube"
+   },
+   {
+     :proper_status=>"formatted",
+     :proper=>"yAtEs AuTo maLL",
+     :proper_f=>"Yates Auto Mall"
+   },
+   {
+     :proper_status=>"formatted",
+     :proper=>"YADKIN VALLEY COLLISION CO",
+     :proper_f=>"Yadkin Valley Collision Co"
+   },
+   {
+     :proper_status=>"formatted",
+     :proper=>"XIT FORD INC",
+     :proper_f=>"Xit Ford Inc"
+   }
+]
+```
+
+2. Format Array of Phone Numbers:
 
-1. Format Array of Phone Numbers:
 ```
 array_of_phones = %w[
   555-457-4391
@@ -44,6 +161,7 @@ formatted_phone_hashes = CrmFormatter.format_phones(array_of_phones)
 ```
 
 Formatted Phone Numbers:
+
 ```
 formatted_phone_hashes = [
   {
@@ -79,7 +197,8 @@ formatted_phone_hashes = [
 ]
 ```
 
-2. Format Array of URLs:
+3. Format Array of URLs:
+
 ```
 array_of_urls = %w[
   sample01.com/staff
@@ -98,6 +217,7 @@ formatted_url_hashes = CrmFormatter.format_urls(array_of_urls)
 ```
 
 Formatted URLs:
+
 ```
 formatted_url_hashes = [
   {
@@ -173,7 +293,8 @@ formatted_url_hashes = [
 ]
 ```
 
-3. Format Array of Addresses (each as a hash):
+4. Format Array of Addresses (each as a hash):
+
 ```
 array_of_addresses = [
   { street: '1234 EAST FAIR BOULEVARD', city: 'AUSTIN', state: 'TEXAS', zip: '78734' },
@@ -184,6 +305,7 @@ formatted_address_hashes = CrmFormatter.format_addresses(array_of_addresses)
 ```
 
 Formatted Addresses:
+
 ```
 formatted_address_hashes = [
   {
@@ -222,11 +344,13 @@ Advanced usage has ability to parse a CSV file or pass large data sets.  It also
 Access advanced usage via `format_with_report(args)` method and pass a csv file_path or data hashes.
 
 1. Parse and Format CSV via File Path (Must be absolute path to root and follow the syntax as below)
+
 ```
 formatted_csv_results = CrmFormatter.format_with_report(file_path: './path/to/your/csv.csv')
 ```
 
 Parsed & Formatted CSV Results:
+
 ```
 formatted_csv_results = {
   stats:
@@ -282,6 +406,7 @@ formatted_csv_results = {
 ```
 
 2. Format Data Hashes
+
 ```
 data_hashes_array = [{ row_id: '1', url: 'abcacura.com/twitter', act_name: "Stanley Chevrolet Kaufman\x99_\xCC", street: '825 East Fair Street', city: 'Kaufman', state: 'Texas', zip: '75142', phone: "555-457-4391\r\n" }]
 
@@ -289,6 +414,7 @@ formatted_data_hash_results = CrmFormatter.format_with_report(data: data_hashes_
 ```
 
 Formatted Data Hashes Results:
+
 ```
 formatted_data_hash_results = { stats:
   {

data/Rakefile

@@ -17,20 +17,38 @@ task :console do
 
   # formatted_data = format_with_report
   # formatted_phones = format_phones
-  formatted_urls = format_urls
+  # formatted_urls = format_urls
+  formatted_propers = format_propers
   # formatted_addresses = format_addresses
+  binding.pry
   IRB.start
 end
 
-def format_with_report
-  data = [{ row_id: '1', url: 'abcacura.com/twitter', act_name: "Stanley Chevrolet Kaufman\x99_\xCC", street: '825 East Fair Street', city: 'Kaufman', state: 'Texas', zip: '75142', phone: "555-457-4391\r\n" }]
-
-  file_path = './lib/crm_formatter/csv/seed.csv'
+#############################################
+def format_propers
+  array_of_propers = [
+  '123 bmw-world',
+  'Car-world Kia',
+  'BUDGET - AUTOMOTORES ZONA & FRANCA, INC',
+  'DOWNTOWN CAR REPAIR, INC',
+  'Young Gmc Trucks',
+  'TEXAS TRAVEL, CO',
+  'youmans Chevrolet',
+  'Hot-Deal auto Insurance',
+  'quick auto approval, inc',
+  'yazell chevy',
+  'quick cAr LUBE',
+  'yAtEs AuTo maLL',
+  'YADKIN VALLEY COLLISION CO',
+  'XIT FORD INC'
+]
 
-  # args = {data: data}
-  args = { file_path: file_path }
-  formatted_data = CrmFormatter.format_with_report(args)
+  formatted_propers = CrmFormatter.format_propers(array_of_propers)
+  formatted_propers
 end
+#############################################
+
+
 
 def format_addresses
   array_of_addresses = [
@@ -41,6 +59,7 @@ def format_addresses
   formatted_addresses = CrmFormatter.format_addresses(array_of_addresses)
 end
 
+
 def format_phones
   array_of_phones = %w[
     555-457-4391 555-888-4391
@@ -51,6 +70,8 @@ def format_phones
   formatted_phones = CrmFormatter.format_phones(array_of_phones)
 end
 
+
+
 def format_urls
   array_of_urls = %w[
     sample01.com/staff
@@ -68,3 +89,14 @@ def format_urls
   array_of_urls = %w[sample01.com/staff www.sample02.net.com http://www.sample3.net www.sample04.net/contact_us http://sample05.net www.sample06.sofake www.sample07.com.sofake example08.not.real www.sample09.net/staff/management www.www.sample10.com]
   formatted_urls = CrmFormatter.format_urls(array_of_urls)
 end
+
+
+def format_with_report
+  data = [{ row_id: '1', url: 'abcacura.com/twitter', act_name: "Stanley Chevrolet Kaufman\x99_\xCC", street: '825 East Fair Street', city: 'Kaufman', state: 'Texas', zip: '75142', phone: "555-457-4391\r\n" }]
+
+  file_path = './lib/crm_formatter/csv/seed.csv'
+
+  # args = {data: data}
+  args = { file_path: file_path }
+  formatted_data = CrmFormatter.format_with_report(args)
+end

data/crm_formatter.gemspec

@@ -33,6 +33,7 @@ Gem::Specification.new do |spec|
   spec.require_paths = ['lib']
 
   spec.required_ruby_version = '~> 2.5.1'
+  spec.add_development_dependency "pry", "~> 0.11.3"
   spec.add_dependency "utf8_sanitizer", "~> 2.0"
   spec.add_dependency 'activesupport', '~> 5.2', '>= 5.2.0'
   # spec.add_dependency "activesupport-inflector", ['~> 0.1.0']
@@ -44,7 +45,6 @@ Gem::Specification.new do |spec|
   spec.add_development_dependency 'rspec', '~> 3.7'
   spec.add_development_dependency 'rubocop', '~> 0.56.0'
   spec.add_development_dependency 'ruby-beautify', '~> 0.97.4'
-  spec.add_development_dependency "pry", "~> 0.11.3"
 
   # spec.add_runtime_dependency 'library', '~> 2.2'
   # spec.add_dependency 'activerecord', '>= 3.0'

data/lib/crm_formatter.rb

@@ -3,21 +3,29 @@
 require 'crm_formatter/address'
 require 'crm_formatter/extensions'
 require 'crm_formatter/phone'
-require 'crm_formatter/version'
+require 'crm_formatter/proper'
+require 'crm_formatter/tools'
 require 'crm_formatter/web'
 require 'crm_formatter/wrap'
 
 # require 'crm_formatter/tools'
 # require 'crm_formatter/seed_criteria'
-require 'pry'
 require 'utf8_sanitizer'
+require 'pry'
 
 module CrmFormatter
-  def self.format_with_report(args={})
-    formatted_data = self::Wrap.new.run(args)
-    formatted_data
+
+  ## Takes array of proper strings, returns array of proper hashes.
+  def self.format_propers(array_of_propers)
+    proper_obj = CrmFormatter::Proper.new
+
+    formatted_proper_hashes = array_of_propers.map do |string|
+      crmf_proper_hsh = proper_obj.format_proper(string)
+    end
+    formatted_proper_hashes
   end
 
+
   ## Takes array of address hashes, returns array of address hashes.
   def self.format_addresses(array_of_addresses)
     address_obj = CrmFormatter::Address.new
@@ -51,4 +59,10 @@ module CrmFormatter
     end
     formatted_url_hashes
   end
+
+  def self.format_with_report(args={})
+    formatted_data = self::Wrap.new.run(args)
+    formatted_data
+  end
+
 end

data/lib/crm_formatter/address.rb

@@ -51,10 +51,7 @@ module CrmFormatter
     def format_street(street)
       street = street&.gsub(/\s/, ' ')&.strip
       return unless street.present?
-      # street = Wrap.new.letter_case_check(street)
-      return unless street.present?
-      # street = CrmFormatter::Tools.new.letter_case_check(street)
-      street = letter_case_check(street)
+      street = CrmFormatter::Tools.new.letter_case_check(street)
 
       street = " #{street} " # Adds white space, to match below, then strip.
       street&.gsub!('.', ' ')
@@ -178,18 +175,5 @@ module CrmFormatter
       zip
     end
 
-    def letter_case_check(str)
-      return unless str.present?
-      flashes = str&.gsub(/[^ A-Za-z]/, '')&.strip&.split(' ')
-      flash = flashes&.reject { |e| e.length < 3 }&.join(' ')
-
-      return str unless flash.present?
-      has_caps = flash.scan(/[A-Z]/).any?
-      has_lows = flash.scan(/[a-z]/).any?
-
-      return str unless !has_caps || !has_lows
-      str = str.split(' ')&.each { |el| el.capitalize! if el.gsub(/[^ A-Za-z]/, '')&.strip&.length > 2 }&.join(' ')
-      str
-    end
   end
 end

data/lib/crm_formatter/proper.rb

@@ -0,0 +1,26 @@
+# frozen_string_literal: false
+
+module CrmFormatter
+  class Proper
+
+    def format_proper(string)
+      str_hsh = { proper_status: nil, proper: string, proper_f: nil }
+      return str_hsh unless string.present?
+      str_hsh[:proper_f] = CrmFormatter::Tools.new.letter_case_check(string)
+
+      str_hsh = check_proper_status(str_hsh)
+      str_hsh
+    end
+
+    ####### COMPARE ORIGINAL AND FORMATTED PROPER ######
+    def check_proper_status(hsh)
+      proper = hsh[:proper]
+      proper_f = hsh[:proper_f]
+      status = 'invalid'
+      status = proper != proper_f ? 'formatted' : 'unchanged' if proper && proper_f
+      hsh[:proper_status] = status if status.present?
+      hsh
+    end
+
+  end
+end

data/lib/crm_formatter/tools.rb

@@ -0,0 +1,37 @@
+# frozen_string_literal: false
+
+module CrmFormatter
+  class Tools
+
+    def letter_case_check(str)
+      return unless str.present?
+      # str = str.upcase
+      str = str.upcase.split(' ')&.each { |el| el.capitalize! if el.gsub(/[^ A-Za-z]/, '')&.strip }&.join(' ')
+      str = capitalize_dashes(str)
+      str = check_for_brands(str)
+      str
+    end
+
+    def capitalize_dashes(str)
+      if str&.include?('-')
+        els = str.split(' ')
+        dash_els = els.select { |el| el != '-' && el.include?('-') }
+
+        dash_els.each do |el|
+          el_cap = el.split('-').map(&:capitalize).join('-')
+          str = str.gsub(el, el_cap)
+        end
+      end
+      str
+    end
+
+    def check_for_brands(str)
+      return unless str.present?
+      ['BMW', 'CDJR', 'CJDR', 'GMC', 'CJD'].map do |brand|
+        str = str.gsub(brand.capitalize, brand)
+      end
+      str
+    end
+
+  end
+end

data/lib/crm_formatter/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: false
 
 module CrmFormatter
-  VERSION = "2.4"
+  VERSION = "2.5"
 end

metadata

@@ -1,15 +1,29 @@
 --- !ruby/object:Gem::Specification
 name: crm_formatter
 version: !ruby/object:Gem::Version
-  version: '2.4'
+  version: '2.5'
 platform: ruby
 authors:
 - Adam Booth
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2018-06-27 00:00:00.000000000 Z
+date: 2018-06-28 00:00:00.000000000 Z
 dependencies:
+- !ruby/object:Gem::Dependency
+  name: pry
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: 0.11.3
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: 0.11.3
 - !ruby/object:Gem::Dependency
   name: utf8_sanitizer
   requirement: !ruby/object:Gem::Requirement
@@ -180,20 +194,6 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: 0.97.4
-- !ruby/object:Gem::Dependency
-  name: pry
-  requirement: !ruby/object:Gem::Requirement
-    requirements:
-    - - "~>"
-      - !ruby/object:Gem::Version
-        version: 0.11.3
-  type: :development
-  prerelease: false
-  version_requirements: !ruby/object:Gem::Requirement
-    requirements:
-    - - "~>"
-      - !ruby/object:Gem::Version
-        version: 0.11.3
 description: |-
   CrmFormatter is perfect for curating high-volume enterprise-scale web scraping, and integrates well with Nokogiri, Mechanize, and asynchronous jobs via Delayed_job or SideKick, to name a few.  Web Scraping and Harvesting often gathers a lot of junk to sift through; presenting unexpected edge cases around each corner.  CrmFormatter has been developed and refined during the past few years to focus on improving that task.
    It's also perfect for processing API data, Web Forms, and routine DB normalizing and scrubbing processes.  Not only does it reformat Address, Phone, and Web data, it can also accept lists to scrub against, then providing detailed reports about how each piece of data compares with your criteria lists.
@@ -220,6 +220,8 @@ files:
 - lib/crm_formatter/csv/seed.csv
 - lib/crm_formatter/extensions.rb
 - lib/crm_formatter/phone.rb
+- lib/crm_formatter/proper.rb
+- lib/crm_formatter/tools.rb
 - lib/crm_formatter/version.rb
 - lib/crm_formatter/web.rb
 - lib/crm_formatter/wrap.rb