Add comprehensive error handling for file validation and CSV parsing

Co-authored-by: rajbos <6085745+rajbos@users.noreply.github.com>

Author: Copilot

src/App.tsx

@@ -33,11 +33,32 @@ function App() {
   const processFile = useCallback((file: File) => {
     if (!file) return;
 
+    // Add basic file validation
+    if (!file.type.includes('text') && !file.type.includes('csv') && !file.name.endsWith('.csv')) {
+      toast.error("Please upload a valid CSV or text file.");
+      return;
+    }
+    
     const reader = new FileReader();
+    
+    reader.onerror = () => {
+      toast.error("Failed to read the file. The file may be corrupted or unreadable.");
+    };
+    
     reader.onload = (e) => {
       try {
         const csvContent = e.target?.result as string;
-        if (!csvContent) throw new Error("Failed to read file");
+        if (!csvContent) throw new Error("Failed to read file content");
+        
+        // Check if the content looks like text (not binary)
+        if (csvContent.includes('\0')) {
+          throw new Error("File appears to be binary. Please upload a text-based CSV file.");
+        }
+        
+        // Check for minimum content
+        if (csvContent.trim().length === 0) {
+          throw new Error("File is empty. Please upload a CSV file with data.");
+        }
 
         const parsedData = parseCSV(csvContent);
         setData(parsedData);
@@ -61,7 +82,32 @@ function App() {
         toast.success(`Loaded ${parsedData.length} records successfully`);
       } catch (error) {
         console.error("Error parsing CSV:", error);
-        toast.error("Failed to parse CSV file. Please check the format.");
+        let errorMessage = "Failed to parse CSV file. Please check the format.";
+        
+        if (error instanceof Error) {
+          // Provide more specific error messages based on the error type
+          if (error.message.includes("missing required columns")) {
+            errorMessage = "Invalid CSV format: " + error.message;
+          } else if (error.message.includes("Invalid timestamp")) {
+            errorMessage = "Invalid data format: " + error.message;
+          } else if (error.message.includes("Invalid requests used")) {
+            errorMessage = "Invalid data format: " + error.message;
+          } else if (error.message.includes("Invalid exceeds quota")) {
+            errorMessage = "Invalid data format: " + error.message;
+          } else if (error.message.includes("Invalid CSV row format")) {
+            errorMessage = "Invalid CSV structure: " + error.message;
+          } else if (error.message.includes("binary")) {
+            errorMessage = error.message;
+          } else if (error.message.includes("empty")) {
+            errorMessage = error.message;
+          } else if (error.message.includes("header")) {
+            errorMessage = "Invalid CSV format: " + error.message;
+          } else {
+            errorMessage = error.message;
+          }
+        }
+        
+        toast.error(errorMessage);
         setData(null);
         setAggregatedData([]);
         setModelSummary([]);
@@ -107,10 +153,10 @@ function App() {
     const files = e.dataTransfer?.files;
     if (files && files.length > 0) {
       const file = files[0];
-      if (file.type === "text/csv" || file.name.endsWith('.csv')) {
+      if (file.type === "text/csv" || file.name.endsWith('.csv') || file.type.includes('text')) {
         processFile(file);
       } else {
-        toast.error("Please upload a CSV file.");
+        toast.error("Please upload a CSV file. Supported formats: .csv files or text files with CSV content.");
       }
     }
   }, [processFile]);

src/lib/utils.ts

@@ -27,13 +27,38 @@ export function parseCSV(csv: string): CopilotUsageData[] {
     throw new Error('CSV must contain a header row and at least one data row');
   }
 
+  // Validate header row
+  const headerLine = lines[0];
+  const expectedHeaders = ['Timestamp', 'User', 'Model', 'Requests Used', 'Exceeds Monthly Quota', 'Total Monthly Quota'];
+  
+  // Parse header row to check for expected columns
+  const headerMatches = headerLine.match(/("([^"]*)"|([^,]*))(,|$)/g);
+  if (!headerMatches || headerMatches.length < 6) {
+    throw new Error('CSV header must contain at least 6 columns');
+  }
+  
+  const headers = headerMatches.map(m => 
+    m.endsWith(',') 
+      ? m.slice(0, -1).replace(/^"(.*)"$/, '$1') 
+      : m.replace(/^"(.*)"$/, '$1')
+  );
+  
+  // Check if all expected headers are present (case-insensitive)
+  const missingHeaders = expectedHeaders.filter(expected => 
+    !headers.some(header => header.toLowerCase().includes(expected.toLowerCase()))
+  );
+  
+  if (missingHeaders.length > 0) {
+    throw new Error(`CSV is missing required columns: ${missingHeaders.join(', ')}. Expected columns: ${expectedHeaders.join(', ')}`);
+  }
+  
   // Skip the header row and process data rows
-  return lines.slice(1).map(line => {
+  return lines.slice(1).map((line, index) => {
     // Handle quoted CSV properly
     const matches = line.match(/("([^"]*)"|([^,]*))(,|$)/g);
 
     if (!matches || matches.length < 6) {
-      throw new Error('Invalid CSV row format');
+      throw new Error(`Invalid CSV row format at line ${index + 2}: expected 6 columns, got ${matches ? matches.length : 0}`);
     }
 
     const values = matches.map(m => 
@@ -42,12 +67,30 @@ export function parseCSV(csv: string): CopilotUsageData[] {
         : m.replace(/^"(.*)"$/, '$1')
     );
 
+    // Validate timestamp
+    const timestamp = new Date(values[0]);
+    if (isNaN(timestamp.getTime())) {
+      throw new Error(`Invalid timestamp format at line ${index + 2}: "${values[0]}"`);
+    }
+    
+    // Validate requests used
+    const requestsUsed = parseFloat(values[3]);
+    if (isNaN(requestsUsed)) {
+      throw new Error(`Invalid requests used value at line ${index + 2}: "${values[3]}" must be a number`);
+    }
+    
+    // Validate exceeds quota
+    const exceedsQuotaValue = values[4].toLowerCase();
+    if (exceedsQuotaValue !== 'true' && exceedsQuotaValue !== 'false') {
+      throw new Error(`Invalid exceeds quota value at line ${index + 2}: "${values[4]}" must be "true" or "false"`);
+    }
+    
     return {
-      timestamp: new Date(values[0]),
+      timestamp,
       user: values[1],
       model: values[2],
-      requestsUsed: parseFloat(values[3]),
-      exceedsQuota: values[4].toLowerCase() === "true",
+      requestsUsed,
+      exceedsQuota: exceedsQuotaValue === "true",
       totalMonthlyQuota: values[5],
     };
   });