fix_tooltips.sh

#!/bin/bash

# Script to add model tooltips and fix existing tooltips in all HTML files

# Model tooltip definitions - exact descriptions from cost analysis tab
declare -A model_tooltips
model_tooltips["OpenAI gpt-4o"]="OpenAI's flagship multimodal model optimized for a balance of quality and speed. Features strong performance across diverse tasks with capabilities for complex financial reasoning and instruction following."
model_tooltips["GPT-4o"]="OpenAI's flagship multimodal model optimized for a balance of quality and speed. Features strong performance across diverse tasks with capabilities for complex financial reasoning and instruction following."
model_tooltips["OpenAI o1-mini"]="OpenAI's smaller advanced model balancing efficiency and performance. Demonstrates surprisingly strong results on financial tasks despite its reduced parameter count."
model_tooltips["o1-mini"]="OpenAI's smaller advanced model balancing efficiency and performance. Demonstrates surprisingly strong results on financial tasks despite its reduced parameter count."
model_tooltips["Claude 3.5 Sonnet"]="Anthropic's advanced proprietary language model optimized for complex reasoning and instruction-following. Features enhanced performance on financial tasks with strong text processing capabilities."
model_tooltips["Claude 3 Haiku"]="Anthropic's smaller efficiency-focused model in the Claude family. Designed for speed and lower computational requirements while maintaining reasonable performance on financial tasks."
model_tooltips["Google Gemini 1.5 Pro"]="Google's advanced proprietary multimodal model designed for complex reasoning and instruction-following tasks. Features strong performance across financial domains with advanced reasoning capabilities."
model_tooltips["Gemini 1.5 Pro"]="Google's advanced proprietary multimodal model designed for complex reasoning and instruction-following tasks. Features strong performance across financial domains with advanced reasoning capabilities."
model_tooltips["Cohere Command R 7B"]="Cohere's 7-billion parameter model focused on instruction-following. An efficient model with reasonable financial domain capabilities for its size."
model_tooltips["Cohere Command R +"]="Cohere's enhanced command model with improved instruction-following capabilities. Features advanced reasoning for financial domains with stronger performance than its smaller counterpart."
model_tooltips["DeepSeek R1"]="DeepSeek's premium 671 billion parameter Mixture of Experts (MoE) model representing their most advanced offering. Designed for state-of-the-art performance across complex reasoning and financial tasks."
model_tooltips["DeepSeek-V3"]="DeepSeek's 685 billion parameter Mixture of Experts (MoE) model optimized for advanced reasoning. Strong performance on financial tasks with robust instruction-following capabilities."
model_tooltips["DeepSeek LLM (67B)"]="DeepSeek's 67 billion parameter model optimized for chat applications. Balances performance and efficiency across financial tasks with solid reasoning capabilities."
model_tooltips["Llama 3 70B Instruct"]="Meta's advanced 70 billion parameter dense language model optimized for instruction-following tasks. Available through Together AI and notable for complex reasoning capabilities."
model_tooltips["Llama 3 8B Instruct"]="Meta's efficient 8 billion parameter language model optimized for instruction-following. Balances performance and efficiency for financial tasks with reasonable reasoning capabilities."
model_tooltips["DBRX Instruct"]="Databricks' 132 billion parameter Mixture of Experts (MoE) model focused on advanced reasoning. Demonstrates competitive performance on financial tasks with strong text processing capabilities."
model_tooltips["Mixtral-8x22B Instruct"]="Mistral AI's 141 billion parameter MoE model with eight 22B expert networks. Features robust reasoning capabilities for financial tasks with strong instruction-following performance."
model_tooltips["Mixtral-8x7B Instruct"]="Mistral AI's 47 billion parameter MoE model with eight 7B expert networks. Balances efficiency and performance with reasonable financial reasoning capabilities."
model_tooltips["Mistral (7B) Instruct v0.3"]="Mistral AI's 7 billion parameter instruction-tuned model. Demonstrates impressive efficiency with reasonable performance on financial tasks despite its smaller size."
model_tooltips["Qwen 2 Instruct (72B)"]="Alibaba's 72 billion parameter instruction-following model optimized for reasoning tasks. Features strong performance on financial domains with advanced text processing capabilities."
model_tooltips["WizardLM-2 8x22B"]="A 176 billion parameter MoE model focused on complex reasoning. Designed for advanced instruction-following with strong capabilities across financial tasks."
model_tooltips["Gemma 2 27B"]="Google's open-weight 27 billion parameter model optimized for reasoning tasks. Balances performance and efficiency across financial domains with strong instruction-following."
model_tooltips["Gemma 2 9B"]="Google's efficient open-weight 9 billion parameter model. Demonstrates good performance on financial tasks relative to its smaller size."
model_tooltips["QwQ-32B-Preview"]="Qwen's experimental 32 billion parameter MoE model focused on efficient computation. Features interesting performance characteristics on certain financial tasks."
model_tooltips["Jamba 1.5 Mini"]="A compact variant in the Jamba model series focused on efficiency. Balances performance and computational requirements for financial tasks."
model_tooltips["Jamba 1.5 Large"]="An expanded variant in the Jamba model series with enhanced capabilities. Features stronger reasoning for financial tasks than its smaller counterpart."

# Files to process
files=(
  "text_classification_table.html"
  "sentiment_analysis_table.html"
  "information_retrieval_table.html"
  "causal_analysis_table.html"
  "text_summarization_table.html"
  "qa_table.html"
)

# Fix existing dataset tooltips
# Fix tooltips in information_retrieval_table.html
sed -i 's/tooltip-trigger" data-tooltip="A dataset for information retrieval in the financial domain/tooltip-trigger tooltip-right" data-tooltip="A dataset for information retrieval in the financial domain/g' information_retrieval_table.html

# Fix tooltips in text_classification_table.html
sed -i 's/tooltip-trigger" data-tooltip="An expert-annotated dataset for detecting fine-grained investor claims/tooltip-trigger tooltip-right" data-tooltip="An expert-annotated dataset for detecting fine-grained investor claims/g' text_classification_table.html

# Fix tooltips in causal_analysis_table.html (in case the tooltip-right class isn't working)
sed -i 's/tooltip-trigger tooltip-right" data-tooltip="Determines if a given financial text section contains a causal relation/tooltip-trigger tooltip-right" data-tooltip="Determines if a given financial text section contains a causal relation/g' causal_analysis_table.html

# Fix tooltips in sentiment_analysis_table.html (in case the tooltip-right class isn't working)
sed -i 's/tooltip-trigger tooltip-right" data-tooltip="Manually-annotated dataset focusing on subjectivity/tooltip-trigger tooltip-right" data-tooltip="Manually-annotated dataset focusing on subjectivity/g' sentiment_analysis_table.html

# Fix tooltips in text_summarization_table.html (in case the tooltip-right class isn't working)
sed -i 's/tooltip-trigger tooltip-right" data-tooltip="Financial news summarization dataset with 2,000 financial news articles/tooltip-trigger tooltip-right" data-tooltip="Financial news summarization dataset with 2,000 financial news articles/g' text_summarization_table.html

# Add or update model tooltips to each file
for file in "${files[@]}"; do
  echo "Processing $file..."
  
  # For each model in our list
  for model in "${!model_tooltips[@]}"; do
    # Convert model name to a sed-safe string by escaping special characters
    model_sed_safe=$(echo "$model" | sed 's/[\/&]/\\&/g')
    tooltip_sed_safe=$(echo "${model_tooltips[$model]}" | sed 's/[\/&]/\\&/g')
    
    # First, update existing tooltips if they exist
    sed -i "s/data-title=\"$model_sed_safe\" data-tooltip=\"[^\"]*\"/data-title=\"$model_sed_safe\" data-tooltip=\"$tooltip_sed_safe\"/g" "$file"
    
    # Then, add tooltips to plain model names without tooltips
    sed -i "s/<td>$model_sed_safe<\/td>/<td class=\"tooltip-trigger tooltip-right\" data-title=\"$model_sed_safe\" data-tooltip=\"$tooltip_sed_safe\">$model_sed_safe<\/td>/g" "$file"
  done
  
  # Ensure tooltip script is included at the bottom of the file
  if ! grep -q "tooltips.js" "$file"; then
    echo "<script src=\"static/js/tooltips.js\"></script>" >> "$file"
  fi
  
  if ! grep -q "fixed-tooltips.js" "$file"; then
    echo "<script src=\"static/js/fixed-tooltips.js\"></script>" >> "$file"
  fi
  
  # Add tooltips.css if not already included
  if ! grep -q "tooltips.css" "$file"; then
    sed -i '1i<link rel="stylesheet" href="static/css/tooltips.css">' "$file"
  fi
done

# Also update results.html to ensure proper tooltip handling
echo "Adding tooltip fix to results.html..."

# Copy the model tooltip fixing code for all tabs to a new JS file
cat > static/js/model-tooltips.js << EOF
document.addEventListener('DOMContentLoaded', function() {
  // Fix model tooltips in all tabs
  function fixAllModelTooltips() {
    console.log("Fixing model tooltips in all tabs");
    
    // Find all model name cells (first column in all tables)
    const modelCells = document.querySelectorAll('td:first-child');
    
    // Process each model cell
    modelCells.forEach(cell => {
      // Skip cells that already have tooltips
      if (cell.classList.contains('tooltip-trigger')) {
        return;
      }
      
      // Get the model name
      const modelName = cell.textContent.trim();
      
      // Add tooltip-trigger class and position style
      cell.classList.add('tooltip-trigger');
      cell.style.position = 'relative';
      
      // Add data-title attribute with the model name
      cell.setAttribute('data-title', modelName);
      
      // Add descriptive tooltip based on model
      let tooltipText = "";
      
      // Set descriptive tooltip based on model name
      if (modelName.includes("GPT-4o")) {
        tooltipText = "OpenAI's advanced proprietary closed-source model. One of the top performers across most tasks.";
      } else if (modelName.includes("o1-mini")) {
        tooltipText = "Compact proprietary model from OpenAI. Shows strong performance on causal analysis tasks.";
      } else if (modelName.includes("Claude 3.5 Sonnet")) {
        tooltipText = "Anthropic's model optimized for advanced reasoning. Strong performer on text classification and summarization.";
      } else if (modelName.includes("Claude 3 Haiku")) {
        tooltipText = "Anthropic's smaller, efficiency-focused model in the Claude series.";
      } else if (modelName.includes("Gemini 1.5")) {
        tooltipText = "Google's highly capable proprietary model.";
      } else if (modelName.includes("Command R 7B")) {
        tooltipText = "A 7-billion parameter model from Cohere focused on instruction-following.";
      } else if (modelName.includes("Command R +")) {
        tooltipText = "An improved version of Cohere's Command R model.";
      } else if (modelName.includes("DeepSeek R1")) {
        tooltipText = "Open-weight model from DeepSeek AI with 671B parameters (MoE architecture). One of the top performers in the benchmark.";
      } else if (modelName.includes("DeepSeek-V3") || modelName.includes("DeepSeek V3")) {
        tooltipText = "Open-weight model from DeepSeek AI with 685B parameters (MoE architecture).";
      } else if (modelName.includes("DeepSeek LLM")) {
        tooltipText = "A 67-billion parameter chat-optimized model from DeepSeek AI.";
      } else if (modelName.includes("Llama 3 70B")) {
        tooltipText = "Meta's 70-billion parameter dense model, optimized for instruction-following tasks.";
      } else if (modelName.includes("Llama 3 8B")) {
        tooltipText = "Meta's 8-billion parameter efficient model variant.";
      } else if (modelName.includes("DBRX")) {
        tooltipText = "Databricks' 132B parameter MoE model.";
      } else if (modelName.includes("Mixtral-8x22B")) {
        tooltipText = "141B parameter MoE model from Mistral AI with eight 22-billion parameter sub-models.";
      } else if (modelName.includes("Mixtral-8x7B")) {
        tooltipText = "46.7B parameter MoE model from Mistral AI with eight 7-billion parameter sub-models.";
      } else if (modelName.includes("Mistral")) {
        tooltipText = "A 7-billion parameter instruction-tuned model from Mistral AI.";
      } else if (modelName.includes("Qwen 2")) {
        tooltipText = "Alibaba's 72-billion parameter instruction-following model.";
      } else if (modelName.includes("WizardLM")) {
        tooltipText = "A 176B parameter MoE model focused on complex reasoning.";
      } else if (modelName.includes("Gemma 2 27B")) {
        tooltipText = "Google's open-weight 27B parameter model.";
      } else if (modelName.includes("Gemma 2 9B")) {
        tooltipText = "Google's open-weight 9B parameter efficient model.";
      } else if (modelName.includes("QwQ-32B")) {
        tooltipText = "Qwen's experimental MoE model with 32B parameters.";
      } else if (modelName.includes("Jamba 1.5 Mini")) {
        tooltipText = "A compact variant of the Jamba model series.";
      } else if (modelName.includes("Jamba 1.5 Large")) {
        tooltipText = "An expanded variant of the Jamba model series.";
      } else {
        tooltipText = "A large language model from the FLaME evaluation benchmark.";
      }
      
      // Set the tooltip
      cell.setAttribute('data-tooltip', tooltipText);
    });
    
    // After adding attributes, run the tooltip fix
    if (window.fixProblemTooltips) {
      window.fixProblemTooltips();
    }
  }
  
  // Run on page load
  setTimeout(fixAllModelTooltips, 500);
  
  // Run when tabs are clicked
  const tabs = document.querySelectorAll('.tabs li');
  tabs.forEach(tab => {
    tab.addEventListener('click', () => {
      // Give time for content to be displayed
      setTimeout(fixAllModelTooltips, 200);
    });
  });
});
EOF

# Add script inclusion to results.html if not already there
if ! grep -q "model-tooltips.js" "results.html"; then
  # Add the script link before the closing body tag
  sed -i 's/<\/body>/<script src="static\/js\/model-tooltips.js"><\/script>\n<\/body>/g' "results.html"
fi

# Add tooltip fix to ensure all tabs initialize properly
if ! grep -q "window.fixProblemTooltips" "results.html"; then
  # Add call to fix all tooltips when tabs are clicked
  sed -i '/document\.addEventListener.*DOMContentLoaded/a \
  // Fix all tooltips in all tabs\
  setTimeout(function() {\
    if (window.fixProblemTooltips) {\
      window.fixProblemTooltips();\
    }\
  }, 500);' "results.html"
fi

echo "Fixed tooltips in all HTML files"