adding analyzer script

analyzer.sh

@@ -0,0 +1,76 @@
+#!/bin/bash
+
+# Error handling
+set -eo pipefail
+trap 'echo "❌ Error occurred on line $LINENO. Exiting..."; exit 1' ERR
+
+# Function to display error messages
+error_msg() {
+    echo "❌ Error: $1" >&2
+    exit 1
+}
+
+# Check if input file is provided
+if [ $# -ne 1 ]; then
+    error_msg "Usage: $0 <csv_file>"
+fi
+
+INPUT_FILE="$1"
+
+# Validate input file exists and is readable
+[ ! -f "$INPUT_FILE" ] && error_msg "Input file '$INPUT_FILE' does not exist!"
+[ ! -r "$INPUT_FILE" ] && error_msg "Input file '$INPUT_FILE' is not readable!"
+
+# Create temporary file for processing
+TEMP_FILE=$(mktemp)
+trap 'rm -f "$TEMP_FILE"' EXIT
+
+# Check if ollama is installed
+command -v ollama >/dev/null 2>&1 || error_msg "ollama is not installed!"
+
+# Process the CSV file
+while IFS=, read -r chargeName remainder || [ -n "$chargeName" ]; do
+    # Skip empty lines
+    [ -z "$chargeName" ] && continue
+
+    # Remove any trailing/leading whitespace
+    chargeName=$(echo "$chargeName" | sed -e 's/^[[:space:]]*//' -e 's/[[:space:]]*$//')
+
+    # Skip if first column is empty
+    [ -z "$chargeName" ] && continue
+
+    echo "🔄 Processing charge: $chargeName"
+
+    # Call ollama and capture output, then remove commas
+    aiDesc=$(printf '%s' "$chargeName" | ollama run granite3.1-dense:latest "interpret what this credit card charge description might be in simple, plain English. Use no more than three sentences: $chargeName" 2>/dev/null | tr -d ',' || echo "AI processing failed")
+
+    # If AI processing failed, add error message
+    if [ "$aiDesc" = "AI processing failed" ]; then
+        echo "⚠️ Warning: Failed to process '$chargeName'"
+    fi
+
+    # Properly escape the AI description for sed
+    aiDesc_escaped=$(printf '%s\n' "$aiDesc" | sed 's/[\/&]/\\&/g')
+
+    # Write the processed line to temp file
+    printf '%s,%s\n' "$chargeName" "$aiDesc" >> "$TEMP_FILE"
+
+    echo "✅ Processed entry successfully"
+done < "$INPUT_FILE"
+
+# Move temp file to original file
+mv "$TEMP_FILE" "$INPUT_FILE"
+
+# **Apply regex fix to remove stray newline followed by a double quote**
+# _Using perl for reliable multiline regex handling_
+perl -i -0777 -pe 's/\n"/,\n/g' "$INPUT_FILE"
+
+# Display success message with some flair
+cat << "EOF"
+✨✨✨✨✨✨✨✨✨✨✨✨✨✨✨✨✨
+🎉 CSV Processing Complete! 🎉
+✨ All entries have been processed
+🔍 AI descriptions have been added
+💾 File has been updated successfully
+✨✨✨✨✨✨✨✨✨✨✨✨✨✨✨✨✨
+EOF