poml 0.0.1 → 0.0.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 3e104ca36939275deeacddbc12fd7ad75129a288e15120ee5103928691188a33
-  data.tar.gz: 997315c25a24ec17f3e0281d634fca83cb65535673ab6334a380402783d5a854
+  metadata.gz: 57f8a3904a43267d79cbc6fb76465dd51c05da5923e58635304242eb90b04b96
+  data.tar.gz: 591512861013d2ae3d302b342b4678696a4d3691c98a26a2d9cdcd042ec70422
 SHA512:
-  metadata.gz: 429e8c22d2f9527a537d312a72975700f84d1dcda381aff1a98fd2842f5520130928780b7cf1c40fc60366180425ae61ac14719560996548ec363e777e619dca
-  data.tar.gz: 776c99377cf172f0edc9cbfa277f83e3d7d1fd47442b64c57c416087a56d4817b50abdfe777e92835a6238a65d292d5e6928cd89bb298e52b522dd3fc5d9a3d4
+  metadata.gz: ba9aeb9b1e06de010b5daa9489d847dc41f5f5d73a9a02e70d81432e997b1c3a3a32116cf9d85ee2f4c70755024201272e62951381cf6363be59dc807c6c55b0
+  data.tar.gz: 2f763c41a44b095f64fbe75cc64f58af266b50c8d966e0d44e6c382945a21fdde9abdd83a49b17ab57608c90a8481fb3066eca6dbc3633533e16269af1693f61

data/lib/poml/components/base.rb

@@ -16,7 +16,8 @@ module Poml
 
     def apply_stylesheet
       # Apply stylesheet rules to the element
-      style_rules = @context.stylesheet[@element.tag_name.to_s] || {}
+      stylesheet = @context.respond_to?(:stylesheet) ? @context.stylesheet : {}
+      style_rules = stylesheet[@element.tag_name.to_s] || {}
       style_rules.each do |attr, value|
         @element.attributes[attr] ||= value
       end
@@ -24,7 +25,7 @@ module Poml
       # Apply class-based styles
       class_name = @element.attributes['classname'] || @element.attributes['className']
       if class_name
-        class_rules = @context.stylesheet[".#{class_name}"] || {}
+        class_rules = stylesheet[".#{class_name}"] || {}
         class_rules.each do |attr, value|
           @element.attributes[attr] ||= value
         end
@@ -32,7 +33,11 @@ module Poml
     end
 
     def xml_mode?
-      @context.determine_syntax(@element) == 'xml'
+      if @context.respond_to?(:determine_syntax)
+        @context.determine_syntax(@element) == 'xml'
+      else
+        false
+      end
     end
 
     def render_as_xml(tag_name, content = nil, attributes = {})
@@ -81,13 +86,17 @@ module Poml
       rendered_children.each_with_index do |child_content, index|
         result << child_content
         
-        # Add spacing if current element is text and next element is a component  
+        # Add spacing if current element is text and next element is a block-level component
         if index < rendered_children.length - 1
           current_element = @element.children[index]
           next_element = @element.children[index + 1]
           
+          # Only add spacing for block-level components, not inline components
           if current_element.text? && next_element.component?
-            result << "\n\n"
+            next_component_class = Components::COMPONENT_MAPPING[next_element.tag_name]
+            if next_component_class && !is_inline_component?(next_component_class)
+              result << "\n\n"
+            end
           end
         end
       end
@@ -102,8 +111,9 @@ module Poml
       component_name = self.class.name.split('::').last.gsub('Component', '').downcase
       
       # Check for text transformation in stylesheet - first try component-specific, then "cp" (for captioned paragraph inheritance)
-      transform = @context.stylesheet.dig(component_name, 'captionTextTransform') ||
-                  @context.stylesheet.dig('cp', 'captionTextTransform')
+      stylesheet = @context.respond_to?(:stylesheet) ? @context.stylesheet : {}
+      transform = stylesheet.dig(component_name, 'captionTextTransform') ||
+                  stylesheet.dig('cp', 'captionTextTransform')
       
       case transform
       when 'upper'
@@ -116,6 +126,34 @@ module Poml
         text
       end
     end
+    
+    def inline_component?
+      # Override this in inline components
+      false
+    end
+    
+    def self.inline_component_classes
+      # List of component classes that should be treated as inline
+      @inline_component_classes ||= []
+    end
+    
+    def self.register_inline_component(component_class)
+      inline_component_classes << component_class
+    end
+    
+    private
+    
+    def is_inline_component?(component_class)
+      # Check if the component class is registered as inline
+      # For now, we'll use a simple list of known inline formatting components
+      inline_component_names = %w[
+        BoldComponent ItalicComponent UnderlineComponent StrikethroughComponent 
+        CodeComponent InlineComponent
+      ]
+      
+      component_class_name = component_class.name.split('::').last
+      inline_component_names.include?(component_class_name)
+    end
   end
 
   # Component registry and factory

data/lib/poml/components/data.rb

@@ -12,16 +12,19 @@ module Poml
       columns_attr = get_attribute('columns') 
       parser = get_attribute('parser', 'auto')
       syntax = get_attribute('syntax')
-      selected_columns = get_attribute('selectedColumns')
-      selected_records = get_attribute('selectedRecords')
-      max_records = get_attribute('maxRecords')
-      max_columns = get_attribute('maxColumns')
+      selected_columns = parse_array_attribute('selectedColumns')
+      selected_records = parse_array_attribute('selectedRecords')
+      max_records = parse_integer_attribute('maxRecords')
+      max_columns = parse_integer_attribute('maxColumns')
       
       # Load data from source or use provided records
       data = if src
         load_table_data(src, parser)
       elsif records_attr
         parse_records_attribute(records_attr)
+      elsif @element.children.any? { |child| child.tag_name == :tr }
+        # Handle HTML-style table markup
+        parse_html_table_children
       else
         { records: [], columns: [] }
       end
@@ -133,7 +136,12 @@ module Poml
     def parse_records_attribute(records_attr)
       # Handle string records (JSON) or already parsed arrays
       records = if records_attr.is_a?(String)
-        JSON.parse(records_attr)
+        begin
+          JSON.parse(records_attr)
+        rescue JSON::ParserError => e
+          # Return empty records on parse error
+          return { records: [], columns: [] }
+        end
       else
         records_attr
       end
@@ -147,6 +155,71 @@ module Poml
       { records: records.is_a?(Array) ? records : [records], columns: columns }
     end
     
+    def parse_array_attribute(attr_name)
+      attr_value = get_attribute(attr_name)
+      return nil unless attr_value
+      
+      if attr_value.is_a?(String)
+        begin
+          parsed = JSON.parse(attr_value)
+          parsed.is_a?(Array) ? parsed : nil
+        rescue JSON::ParserError
+          # Try to handle as slice notation
+          attr_value.include?(':') ? attr_value : nil
+        end
+      elsif attr_value.is_a?(Array)
+        attr_value
+      else
+        nil
+      end
+    end
+    
+    def parse_integer_attribute(attr_name)
+      attr_value = get_attribute(attr_name)
+      return nil unless attr_value
+      
+      if attr_value.is_a?(String)
+        attr_value.to_i
+      elsif attr_value.is_a?(Integer)
+        attr_value
+      else
+        nil
+      end
+    end
+    
+    def parse_html_table_children
+      records = []
+      columns = []
+      
+      # Extract rows from tr children
+      @element.children.each do |child|
+        next unless child.tag_name == :tr
+        
+        row_data = {}
+        child.children.each_with_index do |cell, index|
+          next unless cell.tag_name == :td || cell.tag_name == :th
+          
+          # Get cell content (render children to get text)
+          cell_content = cell.children.map do |cell_child|
+            Components.render_element(cell_child, @context)
+          end.join('').strip
+          
+          # Use cell content as content, index as key for now
+          column_key = "col_#{index}"
+          row_data[column_key] = cell_content
+          
+          # Track columns
+          unless columns.any? { |col| col[:field] == column_key }
+            columns << { field: column_key, header: "Column #{index + 1}" }
+          end
+        end
+        
+        records << row_data unless row_data.empty?
+      end
+      
+      { records: records, columns: columns }
+    end
+    
     def apply_selection(data, selected_columns, selected_records, max_records, max_columns)
       records = data[:records]
       columns = data[:columns]
@@ -183,13 +256,15 @@ module Poml
         end
       end
       
-      # Apply max records
-      if max_records && records.length > max_records
-        # Show top half and bottom half with ellipsis
-        top_rows = (max_records / 2.0).ceil
-        bottom_rows = max_records - top_rows
-        ellipsis_record = columns.each_with_object({}) { |col, record| record[col[:field]] = '...' }
-        records = records[0...top_rows] + [ellipsis_record] + records[-bottom_rows..-1]
+      # Apply max records - simple truncation with ellipsis row
+      if max_records && max_records > 0 && records.length > max_records
+        truncated_records = records[0...max_records]
+        # Add ellipsis row if we truncated
+        if records.length > max_records && columns
+          ellipsis_record = columns.each_with_object({}) { |col, record| record[col[:field]] = '...' }
+          truncated_records << ellipsis_record
+        end
+        records = truncated_records
       end
       
       # Apply max columns
@@ -327,6 +402,229 @@ module Poml
     end
   end
 
+  # Object component for displaying structured data
+  class ObjectComponent < Component
+    require 'json'
+    require 'yaml'
+    
+    def render
+      apply_stylesheet
+      
+      data = get_attribute('data')
+      syntax = get_attribute('syntax', 'json')
+      
+      return '' unless data
+      
+      if xml_mode?
+        render_as_xml('obj', serialize_data(data, syntax))
+      else
+        serialize_data(data, syntax)
+      end
+    end
+    
+    private
+    
+    def serialize_data(data, syntax)
+      case syntax.downcase
+      when 'json'
+        JSON.pretty_generate(data)
+      when 'yaml', 'yml'
+        YAML.dump(data)
+      when 'xml'
+        # Simple XML serialization for basic data structures
+        serialize_to_xml(data)
+      else
+        data.to_s
+      end
+    rescue => e
+      "[Error serializing data: #{e.message}]"
+    end
+    
+    def serialize_to_xml(data, root_name = 'data', indent = 0)
+      spaces = '  ' * indent
+      
+      case data
+      when Hash
+        result = ["#{spaces}<#{root_name}>"]
+        data.each do |key, value|
+          result << serialize_to_xml(value, key, indent + 1)
+        end
+        result << "#{spaces}</#{root_name}>"
+        result.join("\n")
+      when Array
+        result = ["#{spaces}<#{root_name}>"]
+        data.each_with_index do |item, index|
+          result << serialize_to_xml(item, "item#{index}", indent + 1)
+        end
+        result << "#{spaces}</#{root_name}>"
+        result.join("\n")
+      else
+        "#{spaces}<#{root_name}>#{escape_xml(data)}</#{root_name}>"
+      end
+    end
+    
+    def escape_xml(text)
+      text.to_s.gsub('&', '&amp;').gsub('<', '&lt;').gsub('>', '&gt;')
+    end
+  end
+
+  # Webpage component for displaying web content
+  class WebpageComponent < Component
+    def render
+      apply_stylesheet
+      
+      url = get_attribute('url')
+      src = get_attribute('src')
+      buffer = get_attribute('buffer')
+      base64 = get_attribute('base64')
+      extract_text = get_attribute('extractText', false)
+      selector = get_attribute('selector', 'body')
+      syntax = get_attribute('syntax', 'text')
+      
+      content = if url
+        fetch_webpage_content(url, selector, extract_text)
+      elsif src
+        read_html_file(src, selector, extract_text)
+      elsif buffer
+        process_html_content(buffer, selector, extract_text)
+      elsif base64
+        require 'base64'
+        decoded = Base64.decode64(base64)
+        process_html_content(decoded, selector, extract_text)
+      else
+        '[Webpage: no source specified]'
+      end
+      
+      if xml_mode?
+        render_as_xml('webpage', content)
+      else
+        content
+      end
+    end
+    
+    private
+    
+    def fetch_webpage_content(url, selector, extract_text)
+      begin
+        require 'net/http'
+        require 'uri'
+        
+        uri = URI.parse(url)
+        http = Net::HTTP.new(uri.host, uri.port)
+        http.use_ssl = true if uri.scheme == 'https'
+        http.read_timeout = 10
+        
+        request = Net::HTTP::Get.new(uri.request_uri)
+        request['User-Agent'] = 'POML/1.0'
+        
+        response = http.request(request)
+        
+        if response.code == '200'
+          process_html_content(response.body, selector, extract_text)
+        else
+          "[Webpage: HTTP #{response.code} error fetching #{url}]"
+        end
+      rescue => e
+        "[Webpage: Error fetching #{url}: #{e.message}]"
+      end
+    end
+    
+    def read_html_file(file_path, selector, extract_text)
+      begin
+        # Resolve relative paths
+        full_path = if file_path.start_with?('/')
+          file_path
+        else
+          base_path = @context.source_path ? File.dirname(@context.source_path) : Dir.pwd
+          File.join(base_path, file_path)
+        end
+        
+        unless File.exist?(full_path)
+          return "[Webpage: File not found: #{file_path}]"
+        end
+        
+        html_content = File.read(full_path)
+        process_html_content(html_content, selector, extract_text)
+      rescue => e
+        "[Webpage: Error reading file #{file_path}: #{e.message}]"
+      end
+    end
+    
+    def process_html_content(html_content, selector, extract_text)
+      begin
+        require 'nokogiri'
+        
+        doc = Nokogiri::HTML(html_content)
+        
+        # Apply selector if specified
+        if selector && selector != 'body'
+          selected = doc.css(selector).first
+          return "[Webpage: Selector '#{selector}' not found]" unless selected
+          doc = selected
+        end
+        
+        if extract_text
+          # Extract plain text
+          doc.text.strip.gsub(/\s+/, ' ')
+        else
+          # Convert HTML to structured POML-like format
+          convert_html_to_poml(doc)
+        end
+      rescue LoadError
+        # Nokogiri not available, do simple text extraction
+        if extract_text
+          html_content.gsub(/<[^>]*>/, ' ').gsub(/\s+/, ' ').strip
+        else
+          html_content
+        end
+      rescue => e
+        "[Webpage: Error processing HTML: #{e.message}]"
+      end
+    end
+    
+    def convert_html_to_poml(element)
+      case element.name.downcase
+      when 'h1', 'h2', 'h3', 'h4', 'h5', 'h6'
+        level = element.name[1].to_i
+        "#{('#' * level)} #{element.text.strip}\n\n"
+      when 'p'
+        "#{element.text.strip}\n\n"
+      when 'ul', 'ol'
+        items = element.css('li').map { |li| "- #{li.text.strip}" }
+        "#{items.join("\n")}\n\n"
+      when 'strong', 'b'
+        "**#{element.text.strip}**"
+      when 'em', 'i'
+        "*#{element.text.strip}*"
+      when 'code'
+        "`#{element.text.strip}`"
+      when 'pre'
+        "```\n#{element.text.strip}\n```\n\n"
+      when 'blockquote'
+        lines = element.text.strip.split("\n")
+        quoted = lines.map { |line| "> #{line}" }
+        "#{quoted.join("\n")}\n\n"
+      when 'a'
+        href = element['href']
+        text = element.text.strip
+        href ? "[#{text}](#{href})" : text
+      when 'img'
+        alt = element['alt'] || 'Image'
+        src = element['src']
+        src ? "![#{alt}](#{src})" : "[#{alt}]"
+      else
+        # For other elements, process children recursively
+        if element.children.any?
+          element.children.map { |child| 
+            child.text? ? child.text : convert_html_to_poml(child) 
+          }.join('')
+        else
+          element.text.strip
+        end
+      end
+    end
+  end
+
   # Image component
   class ImageComponent < Component
     def render

data/lib/poml/components/examples.rb

@@ -5,7 +5,38 @@ module Poml
       apply_stylesheet
       
       content = @element.content.empty? ? render_children : @element.content
-      "#{content}\n\n"
+      caption = get_attribute('caption', 'Example')
+      caption_style = get_attribute('captionStyle', 'hidden')
+      chat = get_attribute('chat', nil)
+      
+      if xml_mode?
+        render_as_xml('example', content)
+      else
+        # Determine if chat format should be used
+        if chat.nil?
+          # Auto-detect: use chat format for markup syntaxes by default
+          use_chat = @context.determine_syntax(@element) != 'xml'
+        else
+          use_chat = chat
+        end
+        
+        if use_chat && caption_style == 'hidden'
+          content
+        else
+          case caption_style
+          when 'header'
+            "## #{caption}\n\n#{content}\n\n"
+          when 'bold'
+            "**#{caption}:** #{content}\n\n"
+          when 'plain'
+            "#{caption}: #{content}\n\n"
+          when 'hidden'
+            "#{content}\n\n"
+          else
+            "#{content}\n\n"
+          end
+        end
+      end
     end
   end
 
@@ -15,7 +46,26 @@ module Poml
       apply_stylesheet
       
       content = @element.content.empty? ? render_children : @element.content
-      "#{content}\n\n"
+      caption = get_attribute('caption', 'Input')
+      caption_style = get_attribute('captionStyle', 'hidden')
+      speaker = get_attribute('speaker', 'human')
+      
+      if xml_mode?
+        render_as_xml('input', content, { speaker: speaker })
+      else
+        case caption_style
+        when 'header'
+          "## #{caption}\n\n#{content}\n\n"
+        when 'bold'
+          "**#{caption}:** #{content}\n\n"
+        when 'plain'
+          "#{caption}: #{content}\n\n"
+        when 'hidden'
+          "#{content}\n\n"
+        else
+          "#{content}\n\n"
+        end
+      end
     end
   end
 
@@ -25,7 +75,71 @@ module Poml
       apply_stylesheet
       
       content = @element.content.empty? ? render_children : @element.content
-      "#{content}\n\n"
+      caption = get_attribute('caption', 'Output')
+      caption_style = get_attribute('captionStyle', 'hidden')
+      speaker = get_attribute('speaker', 'ai')
+      
+      if xml_mode?
+        render_as_xml('output', content, { speaker: speaker })
+      else
+        case caption_style
+        when 'header'
+          "## #{caption}\n\n#{content}\n\n"
+        when 'bold'
+          "**#{caption}:** #{content}\n\n"
+        when 'plain'
+          "#{caption}: #{content}\n\n"
+        when 'hidden'
+          "#{content}\n\n"
+        else
+          "#{content}\n\n"
+        end
+      end
+    end
+  end
+
+  # Example set component for managing multiple examples
+  class ExampleSetComponent < Component
+    def render
+      apply_stylesheet
+      
+      caption = get_attribute('caption', 'Examples')
+      caption_style = get_attribute('captionStyle', 'header')
+      chat = get_attribute('chat', true)
+      introducer = get_attribute('introducer', '')
+      
+      content = @context.with_chat_context(chat) { render_children }
+      
+      if xml_mode?
+        render_as_xml('examples', content)
+      else
+        result = []
+        
+        case caption_style
+        when 'header'
+          result << "# #{caption}"
+        when 'bold'
+          result << "**#{caption}:**"
+        when 'plain'
+          result << "#{caption}:"
+        when 'hidden'
+          # No caption
+        else
+          result << "# #{caption}"
+        end
+        
+        result << "" unless result.empty? # Add blank line after caption
+        
+        unless introducer.empty?
+          result << introducer
+          result << ""
+        end
+        
+        result << content
+        result << ""
+        
+        result.join("\n")
+      end
     end
   end
 
@@ -38,17 +152,49 @@ module Poml
       caption_style = get_attribute('captionStyle', 'header')
       content = @element.content.empty? ? render_children : @element.content
 
-      case caption_style
-      when 'header'
-        "# #{caption}\n\n#{content}\n\n"
-      when 'bold'
-        "**#{caption}:** #{content}\n\n"
-      when 'plain'
-        "#{caption}: #{content}\n\n"
-      when 'hidden'
-        "#{content}\n\n"
+      if xml_mode?
+        render_as_xml('outputFormat', content)
+      else
+        case caption_style
+        when 'header'
+          "# #{caption}\n\n#{content}\n\n"
+        when 'bold'
+          "**#{caption}:** #{content}\n\n"
+        when 'plain'
+          "#{caption}: #{content}\n\n"
+        when 'hidden'
+          "#{content}\n\n"
+        else
+          "# #{caption}\n\n#{content}\n\n"
+        end
+      end
+    end
+  end
+
+  # Introducer component
+  class IntroducerComponent < Component
+    def render
+      apply_stylesheet
+      
+      content = @element.content.empty? ? render_children : @element.content
+      caption = get_attribute('caption', 'Introducer')
+      caption_style = get_attribute('captionStyle', 'hidden')
+      
+      if xml_mode?
+        render_as_xml('introducer', content)
       else
-        "# #{caption}\n\n#{content}\n\n"
+        case caption_style
+        when 'header'
+          "# #{caption}\n\n#{content}\n\n"
+        when 'bold'
+          "**#{caption}:** #{content}\n\n"
+        when 'plain'
+          "#{caption}: #{content}\n\n"
+        when 'hidden'
+          "#{content}\n\n"
+        else
+          "#{content}\n\n"
+        end
       end
     end
   end