omniai-tools 0.4.0 → 0.5.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f361e26584723aa8d30ba41edaefafb44a7e08c5c6345656dbf25e37c965448d
-  data.tar.gz: c718feeb39dfdd1c2c5ff9a2976ece26eec48aae12f52405724fc21209101d32
+  metadata.gz: d27871f61c1d70be893cce085bf33c0b3d9280686ff1a0ba64f35835b7bf10d4
+  data.tar.gz: 6dfddb639a3cbb99069cbb890bb5d8529bf152e5c246bcc90aaf344505758ea9
 SHA512:
-  metadata.gz: fa14b3333edc0b235e046d9f408f75f201ea37be76435d897873fb7585999cd0aeac3d36c86c16e72a772ab6a28f48dd40560074767720736583de79a80d8847
-  data.tar.gz: 6ec511fc27bae83abdf0d827572ad2784419039b0d37d128d5670b19d58645a8e6efd82bfb7132397473181b56897f966314396a76172aaec33883826ff14f12
+  metadata.gz: ff02138674bc5176e869badb4268530d025e5543ca993eaea8aa74ee2ad114592bd5ac3e8ecd8ad8a882ec0af6dd57b9d74448b04669d3fe8295db3c9176e7b8
+  data.tar.gz: c8a79aa2542b81475148b878ab7785d8bea1b32e4649718bd2a5d4684ba897e010da3d392ce9f3da38374f51e10ea2fc074a3a07a2392172a6c68e93f9298aa5

data/Gemfile

@@ -6,6 +6,11 @@ gemspec
 
 gem "factory_bot"
 gem "irb"
+gem "nokogiri"
+gem "omniai-anthropic"
+gem "omniai-google"
+gem "omniai-mistral"
+gem "omniai-openai"
 gem "rake"
 gem "redcarpet"
 gem "rspec"
@@ -17,4 +22,5 @@ gem "rubocop-rake"
 gem "rubocop-rspec"
 gem "simplecov"
 gem "sqlite3"
+gem "watir"
 gem "yard"

data/README.md

@@ -8,6 +8,60 @@
 
 `OmniAI::Tools` is a library of pre-built tools to simplify integrating common tasks with [OmniAI](https://github.com/ksylvest/omniai).
 
+## Browser
+
+Database tools are focused on running SQL statements:
+
+```ruby
+require "omniai/openai"
+require "omniai/tools"
+
+require "watir"
+
+browser = Watir::Browser.new(:chrome)
+
+client = OmniAI::OpenAI::Client.new
+logger = Logger.new($stdout)
+
+tools = [
+  OmniAI::Tools::Browser::VisitTool,
+  OmniAI::Tools::Browser::InspectTool,
+  OmniAI::Tools::Browser::ButtonClickTool,
+  OmniAI::Tools::Browser::LinkClickTool,
+  OmniAI::Tools::Browser::TextFieldAreaSetTool,
+].map { |klass| klass.new(browser:, logger:) }
+
+puts "Type 'exit' or 'quit' to leave."
+
+prompt = OmniAI::Chat::Prompt.build do |builder|
+  builder.system <<~TEXT
+    You are tasked with assisting a user in browsing the web.
+  TEXT
+end
+
+loop do
+  print "# "
+  text = gets.strip
+  break if %w[exit quit].include?(text)
+
+  prompt.user(text)
+  response = client.chat(prompt, stream: $stdout, tools:)
+  prompt.assistant(response.text)
+end
+```
+
+```
+Type 'exit' or 'quit' to leave.
+# Visit news.ycombinator.com and list the top 5 posts.
+
+[browser] OmniAI::Tools::Browser::VisitTool#execute url="https://news.ycombinator.com"
+[browser] OmniAI::Tools::Browser::InspectTool#execute
+
+Here are the top 5 posts on Hacker News right now:
+
+...
+```
+
 ## Database
 
 Database tools are focused on running SQL statements:

data/lib/omniai/tools/browser/base_tool.rb

@@ -0,0 +1,23 @@
+# frozen_string_literal: true
+
+require "watir"
+
+module OmniAI
+  module Tools
+    module Browser
+      # @example
+      #   class SeleniumTool < BaseTool
+      #     # ...
+      #   end
+      class BaseTool < OmniAI::Tool
+        # @param logger [IO] An optional logger for debugging executed commands.
+        # @param browser [Watir::Browser]
+        def initialize(browser:, logger: Logger.new(IO::NULL))
+          super()
+          @logger = logger
+          @browser = browser
+        end
+      end
+    end
+  end
+end

data/lib/omniai/tools/browser/button_click_tool.rb

@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+require "sqlite3"
+
+module OmniAI
+  module Tools
+    module Browser
+      # @example
+      #   browser = Watir::Browser.new(:chrome)
+      #   tool = OmniAI::Tools::Browser::VisitTool.new(browser:)
+      #   tool.click_link(selector: "link_id")
+      class ButtonClickTool < BaseTool
+        description "A browser automation tool for clicking a specific button."
+
+        parameter :selector, :string, description: "The ID or text of the button to interact with."
+
+        required %i[selector]
+
+        # @param to [String] The URL to navigate to.
+        def execute(selector:)
+          @logger.info("#{self.class.name}##{__method__} selector=#{selector.inspect}")
+
+          element = find(text: selector) || find(id: selector)
+
+          return { error: "unknown selector=#{selector}" } if element.nil?
+
+          element.click
+        end
+
+      protected
+
+        # @return [Watir::Anchor, nil]
+        def find(selector)
+          element = @browser.button(selector)
+          element if element.exists?
+        end
+      end
+    end
+  end
+end

data/lib/omniai/tools/browser/inspect_tool.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+require "nokogiri"
+
+module OmniAI
+  module Tools
+    module Browser
+      # @example
+      #   browser = Watir::Browser.new(:chrome)
+      #   tool = OmniAI::Tools::Browser::InspectTool.new(browser:)
+      #   tool.execute
+      class InspectTool < BaseTool
+        description "A browser automation tool for viewing the HTML for the browser."
+
+        # @return [String]
+        def execute
+          @logger.info("#{self.class.name}##{__method__}")
+
+          html = @browser.html
+          doc = Nokogiri::HTML(html)
+
+          doc.css("link").each(&:remove)
+          doc.css("style").each(&:remove)
+          doc.css("script").each(&:remove)
+
+          doc.to_html
+        end
+      end
+    end
+  end
+end

data/lib/omniai/tools/browser/link_click_tool.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+require "sqlite3"
+
+module OmniAI
+  module Tools
+    module Browser
+      # @example
+      #   browser = Watir::Browser.new(:chrome)
+      #   tool = OmniAI::Tools::Browser::VisitTool.new(browser:)
+      #   tool.click_link(selector: "link_id")
+      class LinkClickTool < BaseTool
+        description "A browser automation tool for clicking a specific link."
+
+        parameter :selector, :string, description: "The ID or text of the link to interact with."
+
+        # @param to [String] The ID or text of the link to interact with.
+        def execute(selector:)
+          @logger.info("#{self.class.name}##{__method__} selector=#{selector.inspect}")
+
+          element = find(text: selector) || find(value: selector) || find(id: selector)
+
+          return { error: "unknown selector=#{selector}" } if element.nil?
+
+          element.click
+        end
+
+      protected
+
+        # @return [Watir::Anchor, nil]
+        def find(selector)
+          element = @browser.a(selector)
+          element if element.exists?
+        end
+      end
+    end
+  end
+end

data/lib/omniai/tools/browser/text_field_area_set_tool.rb

@@ -0,0 +1,59 @@
+# frozen_string_literal: true
+
+require "sqlite3"
+
+module OmniAI
+  module Tools
+    module Browser
+      # @example
+      #   browser = Watir::Browser.new(:chrome)
+      #   tool = OmniAI::Tools::Browser::TextFieldSetTool.new(browser:)
+      #   tool.execute(selector: "...", text: "...")
+      class TextFieldAreaSetTool < BaseTool
+        description "A browser automation tool for clicking a specific link."
+
+        parameter :selector, :string, description: "The ID / name of the text field / area to interact with."
+        parameter :text, :string, description: "The text to set."
+
+        required %i[selector]
+
+        # @param selector [String] The ID / name of the text field / text area to interact with.
+        # @param text [String] The text to set.
+        def execute(selector:, text:)
+          @logger.info("#{self.class.name}##{__method__} selector=#{selector.inspect} text=#{text.inspect}")
+
+          element = find(id: selector) || find(name: selector)
+
+          return { error: "unknown selector=#{selector}" } if element.nil?
+
+          element.set(text)
+        end
+
+      protected
+
+        # @param selector [Hash]
+        #
+        # @return [Watir::TextArea, Watir::TextField, nil]
+        def find(selector)
+          find_text_area(selector) || find_text_field(selector)
+        end
+
+        # @param selector [Hash]
+        #
+        # @return [Watir::TextArea, nil]
+        def find_text_area(selector)
+          element = @browser.textarea(selector)
+          element if element.exists?
+        end
+
+        # @param selector [Hash]
+        #
+        # @return [Watir::TextField, nil]
+        def find_text_field(selector)
+          element = @browser.text_field(selector)
+          element if element.exists?
+        end
+      end
+    end
+  end
+end

data/lib/omniai/tools/browser/visit_tool.rb

@@ -0,0 +1,28 @@
+# frozen_string_literal: true
+
+require "sqlite3"
+
+module OmniAI
+  module Tools
+    module Browser
+      # @example
+      #   driver = Selenium::WebDriver.for :chrome
+      #   tool = OmniAI::Tools::Browser::VisitTool.new(driver:)
+      #   tool.execute(to: "https://news.ycombinator.com")
+      class VisitTool < BaseTool
+        description "A browser automation tool for navigating to a specific URL."
+
+        parameter :url, :string, description: "A URL (e.g. https://news.ycombinator.com)."
+
+        required %i[url]
+
+        # @param url [String] A URL (e.g. https://news.ycombinator.com).
+        def execute(url:)
+          @logger.info("#{self.class.name}##{__method__} url=#{url.inspect}")
+
+          @browser.goto(url)
+        end
+      end
+    end
+  end
+end

data/lib/omniai/tools/database/sqlite_tool.rb

@@ -79,6 +79,8 @@ module OmniAI
         #
         # @return [Array<Hash>]
         def execute(statements:)
+          @logger.info("#{self.class.name}#{__method__} statements=#{statements.inspect}")
+
           [].tap do |executions|
             statements.map do |statement|
               execution = perform(statement:)
@@ -94,8 +96,6 @@ module OmniAI
         #
         # @return [Hash]
         def perform(statement:)
-          @logger.info(%(#{self.class.name}#perform statement=#{statement.inspect}))
-
           result = @db.execute2(statement)
 
           { status: :ok, statement:, result: }

data/lib/omniai/tools/version.rb

@@ -2,6 +2,6 @@
 
 module OmniAI
   module Tools
-    VERSION = "0.4.0"
+    VERSION = "0.5.0"
   end
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification
 name: omniai-tools
 version: !ruby/object:Gem::Version
-  version: 0.4.0
+  version: 0.5.0
 platform: ruby
 authors:
 - Kevin Sylvestre
 bindir: exe
 cert_chain: []
-date: 2025-05-14 00:00:00.000000000 Z
+date: 2025-05-15 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: omniai
@@ -49,6 +49,12 @@ files:
 - bin/console
 - bin/setup
 - lib/omniai/tools.rb
+- lib/omniai/tools/browser/base_tool.rb
+- lib/omniai/tools/browser/button_click_tool.rb
+- lib/omniai/tools/browser/inspect_tool.rb
+- lib/omniai/tools/browser/link_click_tool.rb
+- lib/omniai/tools/browser/text_field_area_set_tool.rb
+- lib/omniai/tools/browser/visit_tool.rb
 - lib/omniai/tools/database/base_tool.rb
 - lib/omniai/tools/database/sqlite_tool.rb
 - lib/omniai/tools/disk/base_tool.rb