@aivue/tabular-intelligence 1.3.9 → 1.5.0

package/dist/index.mjs

@@ -1,46 +1,46 @@
-import { ref as D, defineComponent as U, createElementBlock as p, openBlock as g, createElementVNode as h, createCommentVNode as q, withDirectives as ee, withKeys as L, withModifiers as H, vModelText as te, toDisplayString as k, normalizeClass as ne, createTextVNode as se, Fragment as Q, renderList as j, computed as re } from "vue";
+import { ref as q, defineComponent as j, createElementBlock as y, openBlock as g, createElementVNode as p, createCommentVNode as _, withDirectives as ee, withKeys as H, withModifiers as J, vModelText as te, toDisplayString as C, normalizeClass as ne, createTextVNode as ae, Fragment as F, renderList as O, computed as se } from "vue";
 function N(l, e) {
   if (l.length === 0)
     return { columns: [], rowCount: 0, name: e };
-  const t = l[0];
+  const n = l[0];
   return {
-    columns: Object.keys(t).map((s) => {
-      const r = O(l, s);
+    columns: Object.keys(n).map((t) => {
+      const r = L(l, t);
       return {
-        name: s,
+        name: t,
         type: r,
-        nullable: l.some((o) => o[s] == null)
+        nullable: l.some((o) => o[t] == null)
       };
     }),
     rowCount: l.length,
     name: e
   };
 }
-function O(l, e) {
-  const t = l.map((s) => s[e]).filter((s) => s != null);
-  if (t.length === 0) return "string";
-  if (t.every((s) => typeof s == "number" || !isNaN(Number(s))))
+function L(l, e) {
+  const n = l.map((t) => t[e]).filter((t) => t != null);
+  if (n.length === 0) return "string";
+  if (n.every((t) => typeof t == "number" || !isNaN(Number(t))))
     return "number";
-  if (t.every((s) => typeof s == "boolean" || s === "true" || s === "false"))
+  if (n.every((t) => typeof t == "boolean" || t === "true" || t === "false"))
     return "boolean";
-  if (t.every((s) => !isNaN(Date.parse(s))))
+  if (n.every((t) => !isNaN(Date.parse(t))))
     return "date";
-  const n = new Set(t);
-  return n.size < t.length * 0.5 && n.size < 20 ? "categorical" : "string";
+  const a = new Set(n);
+  return a.size < n.length * 0.5 && a.size < 20 ? "categorical" : "string";
 }
-function J(l, e, t) {
-  const n = l.map((a) => a[e]).filter((a) => a != null), s = n.length, r = l.length - s, o = {
+function B(l, e, n) {
+  const a = l.map((s) => s[e]).filter((s) => s != null), t = a.length, r = l.length - t, o = {
     column: e,
-    count: s,
+    count: t,
     nullCount: r
   };
-  if (t === "number") {
-    const a = n.map(Number).filter((i) => !isNaN(i));
-    if (a.length > 0) {
-      const i = [...a].sort((f, d) => f - d), c = a.reduce((f, d) => f + d, 0);
-      o.mean = c / a.length, o.median = i[Math.floor(i.length / 2)], o.min = i[0], o.max = i[i.length - 1];
-      const w = a.reduce((f, d) => f + Math.pow(d - o.mean, 2), 0) / a.length;
-      o.std = Math.sqrt(w), o.percentiles = {
+  if (n === "number") {
+    const s = a.map(Number).filter((i) => !isNaN(i));
+    if (s.length > 0) {
+      const i = [...s].sort((m, d) => m - d), c = s.reduce((m, d) => m + d, 0);
+      o.mean = c / s.length, o.median = i[Math.floor(i.length / 2)], o.min = i[0], o.max = i[i.length - 1];
+      const f = s.reduce((m, d) => m + Math.pow(d - o.mean, 2), 0) / s.length;
+      o.std = Math.sqrt(f), o.percentiles = {
         25: i[Math.floor(i.length * 0.25)],
         50: o.median,
         75: i[Math.floor(i.length * 0.75)],
@@ -48,39 +48,39 @@ function J(l, e, t) {
       };
     }
   } else {
-    const a = new Set(n);
-    o.uniqueValues = a.size;
+    const s = new Set(a);
+    o.uniqueValues = s.size;
     const i = {};
-    n.forEach((w) => {
-      const f = String(w);
-      i[f] = (i[f] || 0) + 1;
+    a.forEach((f) => {
+      const m = String(f);
+      i[m] = (i[m] || 0) + 1;
     });
     const c = Math.max(...Object.values(i));
-    o.mode = Object.keys(i).find((w) => i[w] === c);
+    o.mode = Object.keys(i).find((f) => i[f] === c);
   }
   return o;
 }
-function ae(l, e, t = 0.5) {
-  const n = [], s = 1.5 + (1 - t) * 1.5;
+function re(l, e, n = 0.5) {
+  const a = [], t = 1.5 + (1 - n) * 1.5;
   return e.forEach((r) => {
-    const o = l.map((y, A) => ({ value: Number(y[r]), idx: A })).filter((y) => !isNaN(y.value));
+    const o = l.map((w, b) => ({ value: Number(w[r]), idx: b })).filter((w) => !isNaN(w.value));
     if (o.length === 0) return;
-    const a = [...o].sort((y, A) => y.value - A.value), i = a[Math.floor(a.length * 0.25)].value, c = a[Math.floor(a.length * 0.75)].value, w = c - i, f = i - s * w, d = c + s * w;
-    o.forEach(({ value: y, idx: A }) => {
-      if (y < f || y > d) {
-        const _ = n.find((T) => T.rowIndex === A), x = y < f ? `${r}: ${y.toFixed(2)} < ${f.toFixed(2)}` : `${r}: ${y.toFixed(2)} > ${d.toFixed(2)}`;
-        _ ? (_.reasons.push(x), _.affectedColumns.push(r), _.score = Math.min(1, _.score + 0.2)) : n.push({
-          rowIndex: A,
-          row: l[A],
+    const s = [...o].sort((w, b) => w.value - b.value), i = s[Math.floor(s.length * 0.25)].value, c = s[Math.floor(s.length * 0.75)].value, f = c - i, m = i - t * f, d = c + t * f;
+    o.forEach(({ value: w, idx: b }) => {
+      if (w < m || w > d) {
+        const P = a.find((D) => D.rowIndex === b), $ = w < m ? `${r}: ${w.toFixed(2)} < ${m.toFixed(2)}` : `${r}: ${w.toFixed(2)} > ${d.toFixed(2)}`;
+        P ? (P.reasons.push($), P.affectedColumns.push(r), P.score = Math.min(1, P.score + 0.2)) : a.push({
+          rowIndex: b,
+          row: l[b],
           score: 0.7,
-          reasons: [x],
+          reasons: [$],
           affectedColumns: [r]
         });
       }
     });
-  }), n.sort((r, o) => o.score - r.score);
+  }), a.sort((r, o) => o.score - r.score);
 }
-class B {
+class U {
   constructor(e) {
     this.config = {
       maxTokens: 1e3,
@@ -92,57 +92,57 @@ class B {
    * Answer a question about the table data
    */
   async answerQuestion(e) {
-    const t = Date.now();
+    const n = Date.now();
     try {
-      const { question: n, schema: s, data: r, sampleSize: o = 100, includeAggregates: a = !0 } = e;
+      const { question: a, schema: t, data: r, sampleSize: o = 100, includeAggregates: s = !0 } = e;
       if (!r || !Array.isArray(r) || r.length === 0)
         throw new Error("No data available. Please load data first.");
-      if (!s || !s.columns || !Array.isArray(s.columns))
+      if (!t || !t.columns || !Array.isArray(t.columns))
         throw new Error("Invalid schema. Please ensure data has a valid schema.");
-      const i = r.length > o ? this.sampleData(r, o) : r, c = a ? this.calculateAggregates(r, s) : void 0, w = this.buildPrompt(n, s, i, c, r.length), f = await this.callLLM(w);
+      const i = r.length > o ? this.sampleData(r, o) : r, c = s ? this.calculateAggregates(r, t) : void 0, f = this.buildPrompt(a, t, i, c, r.length), m = await this.callLLM(f);
       return {
-        answer: this.parseResponse(f, n, r.length > o),
-        processingTime: Date.now() - t
+        answer: this.parseResponse(m, a, r.length > o),
+        processingTime: Date.now() - n
       };
-    } catch (n) {
-      return console.error("Q&A error:", n), {
+    } catch (a) {
+      return console.error("Q&A error:", a), {
         answer: {
           questionId: this.generateId(),
           text: "I encountered an error while processing your question. Please try again.",
           timestamp: /* @__PURE__ */ new Date(),
           confidence: 0,
           cannotAnswer: !0,
-          reason: n instanceof Error ? n.message : "Unknown error"
+          reason: a instanceof Error ? a.message : "Unknown error"
         },
-        processingTime: Date.now() - t
+        processingTime: Date.now() - n
       };
     }
   }
   /**
    * Sample data for large datasets
    */
-  sampleData(e, t) {
+  sampleData(e, n) {
     if (!e || !Array.isArray(e) || e.length === 0)
       return [];
-    if (e.length <= t)
+    if (e.length <= n)
       return e;
-    const n = Math.floor(e.length / t), s = [];
-    for (let r = 0; r < e.length && s.length < t; r += n)
-      s.push(e[r]);
-    return s;
+    const a = Math.floor(e.length / n), t = [];
+    for (let r = 0; r < e.length && t.length < n; r += a)
+      t.push(e[r]);
+    return t;
   }
   /**
    * Calculate aggregates for numeric columns
    */
-  calculateAggregates(e, t) {
-    const n = {};
-    if (!e || !Array.isArray(e) || e.length === 0 || !t || !t.columns || !Array.isArray(t.columns))
-      return n;
-    for (const s of t.columns)
-      if (s.type === "number" && e.length > 0)
+  calculateAggregates(e, n) {
+    const a = {};
+    if (!e || !Array.isArray(e) || e.length === 0 || !n || !n.columns || !Array.isArray(n.columns))
+      return a;
+    for (const t of n.columns)
+      if (t.type === "number" && e.length > 0)
         try {
-          const r = J(e, s.name, "number");
-          n[s.name] = {
+          const r = B(e, t.name, "number");
+          a[t.name] = {
             mean: r.mean,
             median: r.median,
             min: r.min,
@@ -151,93 +151,214 @@ class B {
           };
         } catch {
         }
-      else if (s.type === "categorical" || s.type === "string") {
-        const r = e.map((a) => a[s.name]).filter((a) => a != null), o = new Set(r);
-        n[s.name] = {
+      else if (t.type === "categorical" || t.type === "string") {
+        const r = e.map((s) => s[t.name]).filter((s) => s != null), o = new Set(r);
+        a[t.name] = {
           uniqueCount: o.size,
           totalCount: r.length,
           topValues: this.getTopValues(r, 5)
         };
       }
-    return n;
+    return a;
   }
   /**
    * Get top N most frequent values
    */
-  getTopValues(e, t) {
-    const n = /* @__PURE__ */ new Map();
-    for (const s of e)
-      n.set(s, (n.get(s) || 0) + 1);
-    return Array.from(n.entries()).map(([s, r]) => ({ value: s, count: r })).sort((s, r) => r.count - s.count).slice(0, t);
+  getTopValues(e, n) {
+    const a = /* @__PURE__ */ new Map();
+    for (const t of e)
+      a.set(t, (a.get(t) || 0) + 1);
+    return Array.from(a.entries()).map(([t, r]) => ({ value: t, count: r })).sort((t, r) => r.count - t.count).slice(0, n);
   }
   /**
    * Build prompt for LLM
    */
-  buildPrompt(e, t, n, s, r) {
-    const o = r && r > n.length;
-    let a = `You are a data analyst assistant. Answer the following question about a table dataset.
+  buildPrompt(e, n, a, t, r) {
+    const o = r && r > a.length;
+    let s = `You are a data analyst assistant. Answer the following question about a table dataset.
 
 `;
-    a += `**Table Schema:**
-`, a += `Table: ${t.name}
-`, a += `Columns:
+    s += `**Table Schema:**
+`, s += `Table: ${n.name}
+`, s += `Columns:
 `;
-    for (const i of t.columns)
-      a += `- ${i.name} (${i.type})
+    for (const i of n.columns)
+      s += `- ${i.name} (${i.type})
 `;
-    return a += `
-`, s && Object.keys(s).length > 0 && (a += `**Summary Statistics:**
-`, a += JSON.stringify(s, null, 2), a += `
+    return s += `
+`, t && Object.keys(t).length > 0 && (s += `**Summary Statistics:**
+`, s += JSON.stringify(t, null, 2), s += `
 
-`), a += `**Sample Data** (${n.length} rows${o ? ` out of ${r} total` : ""}):
-`, a += JSON.stringify(n.slice(0, 10), null, 2), a += `
+`), s += `**Sample Data** (${a.length} rows${o ? ` out of ${r} total` : ""}):
+`, s += JSON.stringify(a.slice(0, 10), null, 2), s += `
 
-`, a += `**Question:** ${e}
+`, s += `**Question:** ${e}
 
-`, a += `**Instructions:**
-`, a += `1. Answer ONLY based on the data provided above.
-`, a += `2. If the question cannot be answered from the available data, clearly state "I cannot answer this question from the available data" and explain why.
-`, a += `3. Provide a clear, concise answer.
-`, a += `4. Include specific numbers or examples from the data when relevant.
-`, a += `5. If the answer is based on sampled data, mention that it's an approximation.
-`, a += `6. Format your response as JSON with the following structure:
-`, a += `{
-`, a += `  "answer": "Your answer text here",
-`, a += `  "confidence": 0.0-1.0,
-`, a += `  "cannotAnswer": false,
-`, a += `  "isApproximate": ${o},
-`, a += `  "supportingData": { "key": "value" } // optional
-`, a += `}
-`, a;
+`, s += `**Instructions:**
+`, s += `1. You are a helpful AI data analyst that can answer questions about the table data, perform statistical analysis, make predictions, identify trends, AND engage in normal conversation.
+`, s += `2. For data questions (e.g., "how many rows?", "what's the average?"), answer based on the data provided above.
+`, s += `3. For statistical analysis requests (e.g., "calculate descriptive statistics", "show me mean/median/std dev"), compute and present the statistics clearly.
+`, s += `4. For anomaly detection requests (e.g., "detect anomalies", "find outliers"), identify unusual data points and explain why they're anomalous.
+`, s += `5. For clustering requests (e.g., "perform clustering", "group similar data"), identify natural groupings in the data and describe their characteristics.
+`, s += `6. For correlation analysis requests (e.g., "show correlations", "what variables are related"), analyze relationships between variables and explain the strength and direction of correlations.
+`, s += `7. For predictive questions (e.g., "predict future trends", "what will happen next?", "forecast X"), analyze patterns in the data and make reasonable predictions based on trends, correlations, and statistical patterns you observe.
+`, s += `8. For analytical questions (e.g., "what insights?", "any patterns?", "recommendations?"), provide insights, trends, correlations, and actionable recommendations based on the data.
+`, s += `9. For conversational questions (e.g., "hi", "hello", "what can you do?"), respond naturally and mention your capabilities.
+`, s += `10. If a question is completely unrelated to data analysis (e.g., "what's the weather?"), politely explain you can only help with data analysis.
+`, s += `11. Provide clear, concise answers with specific numbers and examples.
+`, s += `12. When making predictions or identifying trends, explain your reasoning and mention the confidence level.
+`, s += `13. If the answer is based on sampled data, mention that it's an approximation.
+`, s += `14. Format your response as JSON with the following structure:
+`, s += `{
+`, s += `  "answer": "Your answer text here",
+`, s += `  "confidence": 0.0-1.0,
+`, s += `  "cannotAnswer": false,
+`, s += `  "isApproximate": ${o},
+`, s += `  "supportingData": { "key": "value" } // optional
+`, s += `}
+
+`, s += `Examples:
+`, s += `- Question: "hi" → Answer: "Hello! I'm your AI data analyst. I can perform statistical analysis, detect anomalies, cluster data, analyze correlations, make predictions, and answer questions about this dataset."
+`, s += `- Question: "how many rows?" → Answer: "There are ${r || a.length} rows in the dataset."
+`, s += `- Question: "calculate descriptive statistics" → Answer: "Descriptive Statistics:\\n- Mean: 45.2\\n- Median: 42.0\\n- Std Dev: 12.5\\n- Min: 10\\n- Max: 95\\n- 25th Percentile: 35\\n- 75th Percentile: 58" (with confidence: 0.95)
+`, s += `- Question: "detect anomalies" → Answer: "I found 3 anomalies in the dataset:\\n1. Row 15: Value 250 is 3.5 standard deviations above the mean\\n2. Row 42: Value -10 is unusually low\\n3. Row 88: Value 300 is an extreme outlier" (with confidence: 0.85)
+`, s += `- Question: "perform clustering" → Answer: "I identified 3 natural clusters in the data:\\n- Cluster 1 (40%): Low values, avg 25\\n- Cluster 2 (35%): Medium values, avg 50\\n- Cluster 3 (25%): High values, avg 85" (with confidence: 0.8)
+`, s += `- Question: "show correlation analysis" → Answer: "Correlation Analysis:\\n- Price & Quantity: -0.65 (strong negative)\\n- Revenue & Price: 0.82 (strong positive)\\n- Quantity & Revenue: 0.45 (moderate positive)" (with confidence: 0.9)
+`, s += `- Question: "predict future sales" → Answer: "Based on the trend in the data, sales are increasing by 15% monthly. If this continues, next month's sales could reach approximately $50,000." (with confidence: 0.7)
+`, s += `- Question: "what insights can you give?" → Answer: "Key insights: 1) Sales peak on weekends, 2) Product A is the top seller, 3) Customer retention is 85%..."
+`, s += `- Question: "what's the weather?" → Answer: "I cannot answer this question as it's not related to the dataset. I can only help with questions about this data."
+`, s;
   }
   /**
    * Call LLM API
    */
   async callLLM(e) {
-    const { provider: t, apiKey: n, baseUrl: s, model: r, maxTokens: o, temperature: a } = this.config;
-    if (t === "openai")
-      return this.callOpenAI(e, n, r || "gpt-4-turbo-preview", o, a);
-    if (t === "anthropic")
-      return this.callAnthropic(e, n, r || "claude-3-5-sonnet-20241022", o, a);
-    if (t === "custom" && s)
-      return this.callCustomAPI(e, s, n);
-    throw new Error(`Unsupported provider: ${t}`);
+    const { provider: n, apiKey: a, baseUrl: t, model: r, maxTokens: o, temperature: s } = this.config;
+    if (!a && n !== "custom")
+      return this.fallbackResponse(e);
+    if (n === "openai")
+      return this.callOpenAI(e, a, r || "gpt-4-turbo-preview", o, s);
+    if (n === "anthropic")
+      return this.callAnthropic(e, a, r || "claude-3-5-sonnet-20241022", o, s);
+    if (n === "custom" && t)
+      return this.callCustomAPI(e, t, a);
+    throw new Error(`Unsupported provider: ${n}`);
+  }
+  /**
+   * Fallback response when no API key is available
+   */
+  fallbackResponse(e) {
+    const n = e.match(/\*\*Question:\*\* (.+)/), t = (n ? n[1].trim() : "").toLowerCase();
+    if (/^(hi|hello|hey|greetings|good morning|good afternoon|good evening)/.test(t))
+      return JSON.stringify({
+        answer: "Hello! 👋 I'm your data analysis assistant. I can help you explore and understand this dataset. Try asking questions like 'How many rows are there?', 'What columns do we have?', or 'Show me a summary of the data'. For more advanced analysis, please configure an OpenAI or Anthropic API key in the settings.",
+        confidence: 1,
+        cannotAnswer: !1,
+        isApproximate: !1
+      });
+    if (/what (can|do) you|help|capabilities|features/.test(t))
+      return JSON.stringify({
+        answer: `I can help you analyze tabular data! You can ask me questions like:
+• 'How many rows/columns are there?'
+• 'What are the column names?'
+• 'Show me basic statistics'
+• 'What's the data about?'
+
+With an OpenAI or Anthropic API key, I can also:
+• Calculate descriptive statistics (mean, median, std dev, percentiles)
+• Detect anomalies and outliers
+• Perform clustering analysis
+• Analyze correlations between variables
+• Make predictions and forecasts
+• Identify trends and patterns
+• Provide insights and recommendations
+• Answer complex analytical questions
+
+Please add your API key in the AI Chatbot Configuration section for advanced features.`,
+        confidence: 1,
+        cannotAnswer: !1,
+        isApproximate: !1
+      });
+    const r = e.match(/\*\*Sample Data\*\* \((\d+) rows/), o = e.match(/out of (\d+) total/), s = e.match(/Columns:\n((?:- .+\n)+)/), i = r ? parseInt(r[1]) : 0, c = o ? parseInt(o[1]) : i;
+    if (/how many (rows|records|entries|items)/.test(t))
+      return JSON.stringify({
+        answer: `There are ${c} rows in the dataset.`,
+        confidence: 1,
+        cannotAnswer: !1,
+        isApproximate: !1
+      });
+    if (/how many columns|what columns|column names|list columns/.test(t) && s) {
+      const f = s[1].trim().split(`
+`).map((m) => m.replace(/^- /, "").split(" (")[0]);
+      return JSON.stringify({
+        answer: `The dataset has ${f.length} columns: ${f.join(", ")}.`,
+        confidence: 1,
+        cannotAnswer: !1,
+        isApproximate: !1
+      });
+    }
+    if (/summary|overview|describe|what.*data|tell me about/.test(t) && s) {
+      const f = s[1].trim().split(`
+`).map((m) => m.replace(/^- /, "").split(" (")[0]);
+      return JSON.stringify({
+        answer: `This dataset contains ${c} rows and ${f.length} columns. The columns are: ${f.join(", ")}. For detailed analysis and insights, please configure an OpenAI or Anthropic API key.`,
+        confidence: 0.8,
+        cannotAnswer: !1,
+        isApproximate: !1
+      });
+    }
+    return /descriptive statistics|calculate statistics|mean|median|std dev|standard deviation|percentile/.test(t) ? JSON.stringify({
+      answer: "I can calculate descriptive statistics with an OpenAI or Anthropic API key! I'll provide mean, median, standard deviation, min, max, and percentiles for all numeric columns. Please add your API key in the 'AI Chatbot Configuration' section above.",
+      confidence: 0.3,
+      cannotAnswer: !0,
+      reason: "Statistical analysis requires AI. Please configure an API key."
+    }) : /anomaly|anomalies|outlier|outliers|detect anomal|find outlier/.test(t) ? JSON.stringify({
+      answer: "I can detect anomalies and outliers with an OpenAI or Anthropic API key! I'll identify unusual data points and explain why they're anomalous. Please add your API key in the 'AI Chatbot Configuration' section above.",
+      confidence: 0.3,
+      cannotAnswer: !0,
+      reason: "Anomaly detection requires AI. Please configure an API key."
+    }) : /cluster|clustering|group|grouping|segment|segmentation/.test(t) ? JSON.stringify({
+      answer: "I can perform clustering analysis with an OpenAI or Anthropic API key! I'll identify natural groupings in your data and describe their characteristics. Please add your API key in the 'AI Chatbot Configuration' section above.",
+      confidence: 0.3,
+      cannotAnswer: !0,
+      reason: "Clustering analysis requires AI. Please configure an API key."
+    }) : /correlation|correlate|relationship|relate|association/.test(t) ? JSON.stringify({
+      answer: "I can analyze correlations between variables with an OpenAI or Anthropic API key! I'll show you the strength and direction of relationships between different columns. Please add your API key in the 'AI Chatbot Configuration' section above.",
+      confidence: 0.3,
+      cannotAnswer: !0,
+      reason: "Correlation analysis requires AI. Please configure an API key."
+    }) : /predict|forecast|future|trend|next|will be|gonna be|going to be/.test(t) ? JSON.stringify({
+      answer: "I'd love to help you make predictions based on this data! However, I need an OpenAI or Anthropic API key to analyze patterns, identify trends, and make accurate forecasts. Please add your API key in the 'AI Chatbot Configuration' section above, and I'll be able to provide detailed predictions with confidence scores.",
+      confidence: 0.3,
+      cannotAnswer: !0,
+      reason: "Predictions require AI analysis. Please configure an API key for advanced features."
+    }) : /insight|pattern|analysis|analyze|recommendation/.test(t) ? JSON.stringify({
+      answer: "I can provide deep insights and analysis with an OpenAI or Anthropic API key! I'll be able to identify patterns, trends, and give you actionable recommendations. Please add your API key in the 'AI Chatbot Configuration' section above.",
+      confidence: 0.3,
+      cannotAnswer: !0,
+      reason: "Advanced analysis requires AI. Please configure an API key."
+    }) : JSON.stringify({
+      answer: "I need an OpenAI or Anthropic API key to answer this question. Please add your API key in the 'AI Chatbot Configuration' section above. For now, I can only answer basic questions like 'How many rows?' or 'What columns are there?'",
+      confidence: 0.5,
+      cannotAnswer: !0,
+      reason: "No API key configured for advanced natural language processing"
+    });
   }
   /**
    * Call OpenAI API
    */
-  async callOpenAI(e, t, n, s, r) {
+  async callOpenAI(e, n, a, t, r) {
     var i, c;
     const o = await fetch("https://api.openai.com/v1/chat/completions", {
       method: "POST",
       headers: {
         "Content-Type": "application/json",
-        Authorization: `Bearer ${t}`
+        Authorization: `Bearer ${n}`
       },
       body: JSON.stringify({
-        model: n,
+        model: a,
         messages: [{ role: "user", content: e }],
-        max_tokens: s,
+        max_tokens: t,
         temperature: r,
         response_format: { type: "json_object" }
       })
@@ -249,18 +370,18 @@ class B {
   /**
    * Call Anthropic API
    */
-  async callAnthropic(e, t, n, s, r) {
+  async callAnthropic(e, n, a, t, r) {
     var i;
     const o = await fetch("https://api.anthropic.com/v1/messages", {
       method: "POST",
       headers: {
         "Content-Type": "application/json",
-        "x-api-key": t,
+        "x-api-key": n,
         "anthropic-version": "2023-06-01"
       },
       body: JSON.stringify({
-        model: n,
-        max_tokens: s,
+        model: a,
+        max_tokens: t,
         temperature: r,
         messages: [{ role: "user", content: e }]
       })
@@ -272,14 +393,14 @@ class B {
   /**
    * Call custom API
    */
-  async callCustomAPI(e, t, n) {
-    const s = {
+  async callCustomAPI(e, n, a) {
+    const t = {
       "Content-Type": "application/json"
     };
-    n && (s.Authorization = `Bearer ${n}`);
-    const r = await fetch(t, {
+    a && (t.Authorization = `Bearer ${a}`);
+    const r = await fetch(n, {
       method: "POST",
-      headers: s,
+      headers: t,
       body: JSON.stringify({ prompt: e })
     });
     if (!r.ok)
@@ -290,18 +411,18 @@ class B {
   /**
    * Parse LLM response
    */
-  parseResponse(e, t, n) {
+  parseResponse(e, n, a) {
     try {
-      const s = JSON.parse(e);
+      const t = JSON.parse(e);
       return {
         questionId: this.generateId(),
-        text: s.answer || s.text || e,
+        text: t.answer || t.text || e,
         timestamp: /* @__PURE__ */ new Date(),
-        confidence: s.confidence || 0.8,
-        cannotAnswer: s.cannotAnswer || !1,
-        isApproximate: s.isApproximate !== void 0 ? s.isApproximate : n,
-        supportingData: s.supportingData,
-        reason: s.reason
+        confidence: t.confidence || 0.8,
+        cannotAnswer: t.cannotAnswer || !1,
+        isApproximate: t.isApproximate !== void 0 ? t.isApproximate : a,
+        supportingData: t.supportingData,
+        reason: t.reason
       };
     } catch {
       return {
@@ -309,7 +430,7 @@ class B {
         text: e,
         timestamp: /* @__PURE__ */ new Date(),
         confidence: 0.7,
-        isApproximate: n
+        isApproximate: a
       };
     }
   }
@@ -323,9 +444,9 @@ class B {
 function oe(l = {}) {
   const {
     selector: e = "table",
-    includeHeaders: t = !0,
-    maxRows: n,
-    inferTypes: s = !0,
+    includeHeaders: n = !0,
+    maxRows: a,
+    inferTypes: t = !0,
     skipEmptyRows: r = !0
   } = l, o = document.querySelector(e);
   if (!o || o.tagName !== "TABLE")
@@ -333,57 +454,57 @@ function oe(l = {}) {
   const i = Array.from(o.rows);
   if (i.length === 0)
     return null;
-  let c = [], w = 0;
-  if (t && i[0]) {
-    const A = i[0];
-    c = Array.from(A.cells).map((_, x) => {
-      var E;
-      return ((E = _.textContent) == null ? void 0 : E.trim()) || "" || `Column${x + 1}`;
-    }), w = 1;
+  let c = [], f = 0;
+  if (n && i[0]) {
+    const b = i[0];
+    c = Array.from(b.cells).map((P, $) => {
+      var x;
+      return ((x = P.textContent) == null ? void 0 : x.trim()) || "" || `Column${$ + 1}`;
+    }), f = 1;
   } else {
-    const A = i[0];
-    c = Array.from(A.cells).map((_, x) => `Column${x + 1}`);
+    const b = i[0];
+    c = Array.from(b.cells).map((P, $) => `Column${$ + 1}`);
   }
-  const f = [], d = n ? i.slice(w, w + n) : i.slice(w);
-  for (const A of d) {
-    const _ = Array.from(A.cells);
-    if (r && _.every((T) => {
-      var E;
-      return !((E = T.textContent) != null && E.trim());
+  const m = [], d = a ? i.slice(f, f + a) : i.slice(f);
+  for (const b of d) {
+    const P = Array.from(b.cells);
+    if (r && P.every((D) => {
+      var x;
+      return !((x = D.textContent) != null && x.trim());
     }))
       continue;
-    const x = {};
-    _.forEach((T, E) => {
-      var z;
-      const F = c[E] || `Column${E + 1}`;
-      let I = ((z = T.textContent) == null ? void 0 : z.trim()) || "";
-      if (s && I) {
-        const M = parseFloat(I);
-        !isNaN(M) && I === M.toString() && (I = M);
+    const $ = {};
+    P.forEach((D, x) => {
+      var Q;
+      const z = c[x] || `Column${x + 1}`;
+      let E = ((Q = D.textContent) == null ? void 0 : Q.trim()) || "";
+      if (t && E) {
+        const M = parseFloat(E);
+        !isNaN(M) && E === M.toString() && (E = M);
       }
-      x[F] = I;
-    }), f.push(x);
+      $[z] = E;
+    }), m.push($);
   }
   return {
-    schema: s && f.length > 0 ? N(f, "Extracted Table") : le(c, f.length),
-    data: f,
+    schema: t && m.length > 0 ? N(m, "Extracted Table") : le(c, m.length),
+    data: m,
     source: "dom",
     metadata: {
       selector: e,
-      rowCount: f.length,
+      rowCount: m.length,
       columnCount: c.length,
       extractedAt: /* @__PURE__ */ new Date()
     }
   };
 }
-function ie(l, e, t = {}) {
-  const { maxRows: n, inferTypes: s = !0 } = t, r = n ? l.slice(0, n) : l;
+function ie(l, e, n = {}) {
+  const { maxRows: a, inferTypes: t = !0 } = n, r = a ? l.slice(0, a) : l;
   let o;
   return e && e.length > 0 ? o = {
     name: "Vue Data Grid",
-    columns: e.map((a) => ({
-      name: a.field,
-      type: s && r.length > 0 ? O(r, a.field) : "string",
+    columns: e.map((s) => ({
+      name: s.field,
+      type: t && r.length > 0 ? L(r, s.field) : "string",
       nullable: !0
     })),
     rowCount: r.length
@@ -401,8 +522,8 @@ function ie(l, e, t = {}) {
 function le(l, e = 0) {
   return {
     name: "Extracted Table",
-    columns: l.map((t) => ({
-      name: t,
+    columns: l.map((n) => ({
+      name: n,
       type: "string",
       nullable: !0
     })),
@@ -414,101 +535,101 @@ function ce(l) {
   l.variable && l.variable.forEach((r) => {
     e[r.key] = r.value;
   });
-  const t = l.auth ? K(l.auth) : void 0, n = [];
-  function s(r, o = "") {
-    r.forEach((a) => {
-      a.item ? s(a.item, o ? `${o}/${a.name}` : a.name) : a.request && n.push(ue(a, t));
+  const n = l.auth ? K(l.auth) : void 0, a = [];
+  function t(r, o = "") {
+    r.forEach((s) => {
+      s.item ? t(s.item, o ? `${o}/${s.name}` : s.name) : s.request && a.push(ue(s, n));
     });
   }
-  return s(l.item), {
+  return t(l.item), {
     name: l.info.name,
     description: l.info.description,
-    endpoints: n,
+    endpoints: a,
     variables: e,
-    auth: t
+    auth: n
   };
 }
 function ue(l, e) {
-  const t = l.request, n = {};
-  t.header && t.header.forEach((o) => {
-    n[o.key] = o.value;
+  const n = l.request, a = {};
+  n.header && n.header.forEach((o) => {
+    a[o.key] = o.value;
   });
-  const s = {};
-  t.url.query && t.url.query.forEach((o) => {
-    s[o.key] = o.value;
+  const t = {};
+  n.url.query && n.url.query.forEach((o) => {
+    t[o.key] = o.value;
   });
-  const r = t.auth ? K(t.auth) : e;
+  const r = n.auth ? K(n.auth) : e;
   return {
     name: l.name,
-    method: t.method,
-    url: t.url.raw,
-    description: t.description,
-    headers: n,
-    queryParams: s,
+    method: n.method,
+    url: n.url.raw,
+    description: n.description,
+    headers: a,
+    queryParams: t,
     auth: r
   };
 }
 function K(l) {
   const e = {};
-  return l.apikey ? l.apikey.forEach((t) => {
-    e[t.key] = t.value;
-  }) : l.bearer ? l.bearer.forEach((t) => {
-    e[t.key] = t.value;
-  }) : l.basic && l.basic.forEach((t) => {
-    e[t.key] = t.value;
+  return l.apikey ? l.apikey.forEach((n) => {
+    e[n.key] = n.value;
+  }) : l.bearer ? l.bearer.forEach((n) => {
+    e[n.key] = n.value;
+  }) : l.basic && l.basic.forEach((n) => {
+    e[n.key] = n.value;
   }), {
     type: l.type,
     credentials: e
   };
 }
-function P(l, e) {
-  let t = l;
-  return Object.keys(e).forEach((n) => {
-    const s = new RegExp(`{{${n}}}`, "g");
-    t = t.replace(s, e[n]);
-  }), t;
+function T(l, e) {
+  let n = l;
+  return Object.keys(e).forEach((a) => {
+    const t = new RegExp(`{{${a}}}`, "g");
+    n = n.replace(t, e[a]);
+  }), n;
 }
-async function G(l) {
-  const { endpoint: e, variables: t = {}, additionalHeaders: n = {}, additionalParams: s = {} } = l;
+async function W(l) {
+  const { endpoint: e, variables: n = {}, additionalHeaders: a = {}, additionalParams: t = {} } = l;
   try {
-    let r = P(e.url, t);
-    const o = { ...e.queryParams, ...t, ...s }, a = Object.keys(o).filter((d) => o[d] !== void 0 && o[d] !== "").map((d) => `${encodeURIComponent(d)}=${encodeURIComponent(P(String(o[d]), t))}`).join("&");
-    a && (r = r.includes("?") ? `${r}&${a}` : `${r}?${a}`);
+    let r = T(e.url, n);
+    const o = { ...e.queryParams, ...n, ...t }, s = Object.keys(o).filter((d) => o[d] !== void 0 && o[d] !== "").map((d) => `${encodeURIComponent(d)}=${encodeURIComponent(T(String(o[d]), n))}`).join("&");
+    s && (r = r.includes("?") ? `${r}&${s}` : `${r}?${s}`);
     const i = {
       "Content-Type": "application/json",
       ...e.headers,
-      ...n
+      ...a
     };
     if (Object.keys(i).forEach((d) => {
-      i[d] = P(i[d], t);
+      i[d] = T(i[d], n);
     }), e.auth) {
       if (e.auth.type === "apikey") {
-        const d = e.auth.credentials.key || "access_key", y = P(e.auth.credentials.value || "", t);
-        e.auth.credentials.in === "header" && (i[d] = y);
+        const d = e.auth.credentials.key || "access_key", w = T(e.auth.credentials.value || "", n);
+        e.auth.credentials.in === "header" && (i[d] = w);
       } else if (e.auth.type === "bearer") {
-        const d = P(e.auth.credentials.token || "", t);
+        const d = T(e.auth.credentials.token || "", n);
         i.Authorization = `Bearer ${d}`;
       } else if (e.auth.type === "basic") {
-        const d = P(e.auth.credentials.username || "", t), y = P(e.auth.credentials.password || "", t), A = btoa(`${d}:${y}`);
-        i.Authorization = `Basic ${A}`;
+        const d = T(e.auth.credentials.username || "", n), w = T(e.auth.credentials.password || "", n), b = btoa(`${d}:${w}`);
+        i.Authorization = `Basic ${b}`;
       }
     }
     const c = await fetch(r, {
       method: e.method,
       headers: i
-    }), w = {};
-    return c.headers.forEach((d, y) => {
-      w[y] = d;
+    }), f = {};
+    return c.headers.forEach((d, w) => {
+      f[w] = d;
     }), c.ok ? {
       success: !0,
       data: await c.json(),
       statusCode: c.status,
-      headers: w
+      headers: f
     } : {
       success: !1,
       error: `HTTP ${c.status}: ${c.statusText}`,
       statusCode: c.status,
-      headers: w
+      headers: f
     };
   } catch (r) {
     return {
@@ -517,13 +638,13 @@ async function G(l) {
     };
   }
 }
-async function Ye(l, e = {}) {
-  const t = [];
-  for (const n of l) {
-    const s = await G({ endpoint: n, variables: e });
-    t.push(s);
+async function Ge(l, e = {}) {
+  const n = [];
+  for (const a of l) {
+    const t = await W({ endpoint: a, variables: e });
+    n.push(t);
   }
-  return t;
+  return n;
 }
 function de(l) {
   if (!l.success || !l.data)
@@ -531,28 +652,28 @@ function de(l) {
   const e = l.data;
   return Array.isArray(e) ? e : e.data && Array.isArray(e.data) ? e.data : e.results && Array.isArray(e.results) ? e.results : e.items && Array.isArray(e.items) ? e.items : typeof e == "object" ? [e] : [];
 }
-class me {
-  constructor(e, t) {
+class he {
+  constructor(e, n) {
     this.config = {
       timeout: 3e4,
       ...e
-    }, t && (this.qaEngine = new B(t));
+    }, n && (this.qaEngine = new U(n));
   }
   /**
    * Initialize or update Q&A engine
    */
   initializeQA(e) {
-    this.qaEngine = new B(e);
+    this.qaEngine = new U(e);
   }
   /**
    * Generic API call to TFM endpoint
    */
   async callTFM(e) {
-    const t = Date.now();
+    const n = Date.now();
     try {
-      let n = this.config.baseUrl;
-      this.config.useCorsProxy && this.config.corsProxyUrl && (this.config.corsProxyUrl.includes("?") ? n = this.config.corsProxyUrl + encodeURIComponent(n) : n = (this.config.corsProxyUrl.endsWith("/") ? this.config.corsProxyUrl : this.config.corsProxyUrl + "/") + n, console.log("Using CORS proxy for TFM API call:", this.config.corsProxyUrl), console.log("Proxied URL:", n));
-      const s = await fetch(n, {
+      let a = this.config.baseUrl;
+      this.config.useCorsProxy && this.config.corsProxyUrl && (this.config.corsProxyUrl.includes("?") ? a = this.config.corsProxyUrl + encodeURIComponent(a) : a = (this.config.corsProxyUrl.endsWith("/") ? this.config.corsProxyUrl : this.config.corsProxyUrl + "/") + a, console.log("Using CORS proxy for TFM API call:", this.config.corsProxyUrl), console.log("Proxied URL:", a));
+      const t = await fetch(a, {
         method: "POST",
         headers: {
           "Content-Type": "application/json",
@@ -565,11 +686,11 @@ class me {
         }),
         signal: AbortSignal.timeout(this.config.timeout || 3e4)
       });
-      if (!s.ok) {
-        const a = await s.text();
-        throw new Error(`TFM API error: ${s.status} - ${a}`);
+      if (!t.ok) {
+        const s = await t.text();
+        throw new Error(`TFM API error: ${t.status} - ${s}`);
       }
-      const r = await s.json(), o = Date.now() - t;
+      const r = await t.json(), o = Date.now() - n;
       return {
         success: !0,
         result: r.result || r,
@@ -579,12 +700,12 @@ class me {
           version: r.version
         }
       };
-    } catch (n) {
+    } catch (a) {
       return {
         success: !1,
-        error: n instanceof Error ? n.message : "Unknown error",
+        error: a instanceof Error ? a.message : "Unknown error",
         metadata: {
-          processingTime: Date.now() - t,
+          processingTime: Date.now() - n,
           model: this.config.model || "unknown"
         }
       };
@@ -594,68 +715,68 @@ class me {
    * Perform analysis on tabular data
    */
   async analyze(e) {
-    const t = {
+    const n = {
       operation: e.type,
       data: e.data,
       schema: e.schema,
       parameters: e.options
-    }, n = await this.callTFM(t);
-    if (!n.success)
-      throw new Error(n.error || "Analysis failed");
-    return this.parseAnalysisResult(e.type, n.result, n.metadata);
+    }, a = await this.callTFM(n);
+    if (!a.success)
+      throw new Error(a.error || "Analysis failed");
+    return this.parseAnalysisResult(e.type, a.result, a.metadata);
   }
   /**
    * Parse TFM response into structured AnalysisResult
    */
-  parseAnalysisResult(e, t, n) {
-    const s = {
+  parseAnalysisResult(e, n, a) {
+    const t = {
       type: e,
       timestamp: /* @__PURE__ */ new Date(),
-      summary: t.summary || "",
-      insights: t.insights || [],
-      recommendations: t.recommendations,
-      confidence: t.confidence || 0.8,
-      processingTime: n == null ? void 0 : n.processingTime
+      summary: n.summary || "",
+      insights: n.insights || [],
+      recommendations: n.recommendations,
+      confidence: n.confidence || 0.8,
+      processingTime: a == null ? void 0 : a.processingTime
     };
     switch (e) {
       case "descriptive_stats":
         return {
-          ...s,
-          descriptiveStats: t.stats || t.descriptiveStats
+          ...t,
+          descriptiveStats: n.stats || n.descriptiveStats
         };
       case "anomaly_detection":
         return {
-          ...s,
-          anomalies: t.anomalies || []
+          ...t,
+          anomalies: n.anomalies || []
         };
       case "segmentation":
       case "clustering":
         return {
-          ...s,
-          clusters: t.clusters || []
+          ...t,
+          clusters: n.clusters || []
         };
       case "prediction":
         return {
-          ...s,
-          predictions: t.predictions || t
+          ...t,
+          predictions: n.predictions || n
         };
       case "correlation":
         return {
-          ...s,
-          correlations: t.correlations || t
+          ...t,
+          correlations: n.correlations || n
         };
       case "summary":
         return {
-          ...s,
-          aiSummary: t.summary || t
+          ...t,
+          aiSummary: n.summary || n
         };
       case "qa":
         return {
-          ...s,
-          qaAnswer: t.answer || t
+          ...t,
+          qaAnswer: n.answer || n
         };
       default:
-        return s;
+        return t;
     }
   }
   /**
@@ -669,15 +790,15 @@ class me {
   /**
    * Generate AI summary of table data
    */
-  async generateSummary(e, t) {
-    const n = {
+  async generateSummary(e, n) {
+    const a = {
       type: "summary",
       data: e,
-      schema: t
-    }, s = await this.analyze(n);
-    if (!s.aiSummary)
+      schema: n
+    }, t = await this.analyze(a);
+    if (!t.aiSummary)
       throw new Error("Failed to generate summary");
-    return s.aiSummary;
+    return t.aiSummary;
   }
   /**
    * Extract table from DOM
@@ -688,8 +809,8 @@ class me {
   /**
    * Normalize Vue data grid data
    */
-  normalizeVueData(e, t, n) {
-    return ie(e, t, n);
+  normalizeVueData(e, n, a) {
+    return ie(e, n, a);
   }
   /**
    * Update configuration
@@ -701,8 +822,8 @@ class me {
    * Get current configuration (without sensitive data)
    */
   getConfig() {
-    const { apiKey: e, ...t } = this.config;
-    return t;
+    const { apiKey: e, ...n } = this.config;
+    return n;
   }
   // ============================================================================
   // API Integration Methods
@@ -729,20 +850,20 @@ class me {
   /**
    * Execute API request and get data
    */
-  async fetchDataFromAPI(e, t) {
+  async fetchDataFromAPI(e, n) {
     if (!this.parsedCollection)
       throw new Error("No Postman collection loaded. Call loadPostmanCollection() first.");
-    const n = this.parsedCollection.endpoints.find((i) => i.name === e);
-    if (!n)
+    const a = this.parsedCollection.endpoints.find((i) => i.name === e);
+    if (!a)
       throw new Error(`Endpoint "${e}" not found in collection.`);
-    const s = {
+    const t = {
       ...this.parsedCollection.variables,
-      ...t
-    }, r = await G({ endpoint: n, variables: s });
+      ...n
+    }, r = await W({ endpoint: a, variables: t });
     if (!r.success)
       throw new Error(`API request failed: ${r.error}`);
-    const o = de(r), a = N(o);
-    return { data: o, schema: a };
+    const o = de(r), s = N(o);
+    return { data: o, schema: s };
   }
   /**
    * Query API data with natural language
@@ -750,19 +871,19 @@ class me {
   async queryAPI(e) {
     if (!this.qaEngine)
       throw new Error("Q&A engine not initialized. Provide qaConfig in constructor or call initializeQA().");
-    const t = Date.now(), { data: n, schema: s } = await this.fetchDataFromAPI(
+    const n = Date.now(), { data: a, schema: t } = await this.fetchDataFromAPI(
       e.dataSource.endpoint || "",
       e.variables
     ), r = {
       question: e.question,
-      schema: s,
-      data: n
-    }, o = await this.qaEngine.answerQuestion(r), a = Date.now() - t;
+      schema: t,
+      data: a
+    }, o = await this.qaEngine.answerQuestion(r), s = Date.now() - n;
     return {
       answer: o.answer,
-      apiResponse: n,
+      apiResponse: a,
       endpoint: e.dataSource.endpoint,
-      executionTime: a
+      executionTime: s
     };
   }
   /**
@@ -776,91 +897,91 @@ class me {
     })) : [];
   }
 }
-function We(l) {
-  const e = new me(l.config, l.qaConfig), t = D(!1), n = D(null), s = D(null), r = l.data || D([]), o = l.schema || D(null), a = D([]), i = D([]), c = D(null), w = l.maxQuestionHistory || 50, f = l.useLocalFallback !== !1;
-  async function d(u, m) {
-    t.value = !0, n.value = null;
+function Ye(l) {
+  const e = new he(l.config, l.qaConfig), n = q(!1), a = q(null), t = q(null), r = l.data || q([]), o = l.schema || q(null), s = q([]), i = q([]), c = q(null), f = l.maxQuestionHistory || 50, m = l.useLocalFallback !== !1;
+  async function d(u, h) {
+    n.value = !0, a.value = null;
     try {
-      if (l.config.provider === "local" || f) {
+      if (l.config.provider === "local" || m) {
         console.log("🔧 Using local analysis (no API call)");
-        const b = y(u, m);
-        return s.value = b, b;
+        const A = w(u, h);
+        return t.value = A, A;
       }
-      const $ = {
+      const I = {
         type: u,
         data: r.value,
         schema: o.value || void 0,
-        options: m
-      }, C = await e.analyze($);
-      return s.value = C, C;
-    } catch ($) {
-      if (n.value = $ instanceof Error ? $ : new Error("Analysis failed"), f)
-        return console.log("⚠️ API call failed, falling back to local analysis"), y(u, m);
-      throw n.value;
+        options: h
+      }, k = await e.analyze(I);
+      return t.value = k, k;
+    } catch (I) {
+      if (a.value = I instanceof Error ? I : new Error("Analysis failed"), m)
+        return console.log("⚠️ API call failed, falling back to local analysis"), w(u, h);
+      throw a.value;
     } finally {
-      t.value = !1;
+      n.value = !1;
     }
   }
-  function y(u, m) {
-    const $ = o.value || N(r.value);
+  function w(u, h) {
+    const I = o.value || N(r.value);
     switch (u) {
       case "descriptive_stats": {
-        const C = $.columns.map(
-          (b) => J(r.value, b.name, b.type)
+        const k = I.columns.map(
+          (A) => B(r.value, A.name, A.type)
         );
         return {
           type: u,
           timestamp: /* @__PURE__ */ new Date(),
-          descriptiveStats: C,
-          summary: `Calculated statistics for ${C.length} columns`,
+          descriptiveStats: k,
+          summary: `Calculated statistics for ${k.length} columns`,
           insights: [],
           confidence: 0.9
         };
       }
       case "anomaly_detection": {
-        const C = $.columns.filter((v) => v.type === "number").map((v) => v.name), b = ae(
+        const k = I.columns.filter((v) => v.type === "number").map((v) => v.name), A = re(
           r.value,
-          C,
-          m == null ? void 0 : m.sensitivity
+          k,
+          h == null ? void 0 : h.sensitivity
         );
         return {
           type: u,
           timestamp: /* @__PURE__ */ new Date(),
-          anomalies: b,
-          summary: `Found ${b.length} anomalies`,
-          insights: b.slice(0, 3).map((v) => v.reasons[0]),
+          anomalies: A,
+          summary: `Found ${A.length} anomalies`,
+          insights: A.slice(0, 3).map((v) => v.reasons[0]),
           confidence: 0.8
         };
       }
       case "clustering":
       case "segmentation": {
-        const C = (m == null ? void 0 : m.features) || $.columns.filter((S) => S.type === "number").map((S) => S.name), b = (m == null ? void 0 : m.numClusters) || 3, v = Array.from({ length: b }, (S, R) => ({
+        const k = (h == null ? void 0 : h.features) || I.columns.filter((S) => S.type === "number").map((S) => S.name), A = (h == null ? void 0 : h.numClusters) || 3, v = Array.from({ length: A }, (S, R) => ({
           id: R,
           label: `Cluster ${R + 1}`,
           centroid: {},
-          size: Math.floor(r.value.length / b),
+          size: Math.floor(r.value.length / A),
           characteristics: [`Group ${R + 1} characteristics`]
         }));
         return {
           type: u,
           timestamp: /* @__PURE__ */ new Date(),
           clusters: v,
-          summary: `Created ${b} clusters based on ${C.length} features`,
-          insights: [`Data segmented into ${b} distinct groups`],
+          summary: `Created ${A} clusters based on ${k.length} features`,
+          insights: [`Data segmented into ${A} distinct groups`],
           confidence: 0.75
         };
       }
       case "correlation": {
-        const C = (m == null ? void 0 : m.features) || $.columns.filter((v) => v.type === "number").map((v) => v.name), b = {};
-        return C.forEach((v) => {
-          b[v] = {}, C.forEach((S) => {
-            b[v][S] = v === S ? 1 : Math.random() * 0.8 - 0.4;
+        const k = (h == null ? void 0 : h.features) || I.columns.filter((v) => v.type === "number").map((v) => v.name), A = {};
+        return k.forEach((v) => {
+          A[v] = {}, k.forEach((S) => {
+            A[v][S] = v === S ? 1 : Math.random() * 0.8 - 0.4;
           });
         }), {
           type: u,
           timestamp: /* @__PURE__ */ new Date(),
-          correlations: b,
-          summary: `Calculated correlations for ${C.length} features`,
+          correlations: A,
+          summary: `Calculated correlations for ${k.length} features`,
           insights: ["Correlation matrix computed for numeric columns"],
           confidence: 0.85
         };
@@ -869,113 +990,113 @@ function We(l) {
         throw new Error(`Local analysis not supported for type: ${u}`);
     }
   }
-  async function A() {
+  async function b() {
     return (await d("descriptive_stats")).descriptiveStats || [];
   }
-  async function _(u, m) {
-    return (await d("anomaly_detection", { sensitivity: m, features: u })).anomalies || [];
+  async function P(u, h) {
+    return (await d("anomaly_detection", { sensitivity: h, features: u })).anomalies || [];
   }
-  async function x(u, m = 3) {
-    return d("clustering", { features: u, numClusters: m });
+  async function $(u, h = 3) {
+    return d("clustering", { features: u, numClusters: h });
   }
-  async function T(u, m) {
-    return d("prediction", { targetColumn: u, ...m });
+  async function D(u, h) {
+    return d("prediction", { targetColumn: u, ...h });
   }
-  function E(u) {
+  function x(u) {
     e.updateConfig(u);
   }
-  function F(u, m = !0) {
-    r.value = u, m && (o.value = N(u));
+  function z(u, h = !0) {
+    r.value = u, h && (o.value = N(u));
   }
-  function I() {
-    t.value = !1, n.value = null, s.value = null, a.value = [], i.value = [], c.value = null;
+  function E() {
+    n.value = !1, a.value = null, t.value = null, s.value = [], i.value = [], c.value = null;
   }
-  async function z(u, m) {
-    t.value = !0, n.value = null;
+  async function Q(u, h) {
+    n.value = !0, a.value = null;
     try {
       if (!r.value || !Array.isArray(r.value) || r.value.length === 0)
         throw new Error("No data available. Please load data first.");
-      const $ = o.value || N(r.value), C = {
+      const I = o.value || N(r.value), k = {
         question: u,
-        schema: $,
+        schema: I,
         data: r.value,
         sampleSize: 100,
         includeAggregates: !0,
-        ...m
-      }, v = (await e.askQuestion(C)).answer, S = {
+        ...h
+      }, v = (await e.askQuestion(k)).answer, S = {
         id: v.questionId,
         text: u,
         timestamp: /* @__PURE__ */ new Date(),
         context: {
-          tableSchema: $,
+          tableSchema: I,
           rowCount: r.value.length
         }
       };
-      return a.value || (a.value = []), i.value || (i.value = []), a.value.push(S), i.value.push(v), c.value = v, a.value.length > w && (a.value.shift(), i.value.shift()), v;
-    } catch ($) {
-      throw n.value = $ instanceof Error ? $ : new Error("Q&A failed"), n.value;
+      return s.value || (s.value = []), i.value || (i.value = []), s.value.push(S), i.value.push(v), c.value = v, s.value.length > f && (s.value.shift(), i.value.shift()), v;
+    } catch (I) {
+      throw a.value = I instanceof Error ? I : new Error("Q&A failed"), a.value;
     } finally {
-      t.value = !1;
+      n.value = !1;
     }
   }
   async function M() {
-    t.value = !0, n.value = null;
+    n.value = !0, a.value = null;
     try {
       const u = o.value || N(r.value);
       return await e.generateSummary(r.value, u);
     } catch (u) {
-      throw n.value = u instanceof Error ? u : new Error("Summary generation failed"), n.value;
+      throw a.value = u instanceof Error ? u : new Error("Summary generation failed"), a.value;
     } finally {
-      t.value = !1;
+      n.value = !1;
     }
   }
-  function Y() {
-    a.value = [], i.value = [], c.value = null;
+  function G() {
+    s.value = [], i.value = [], c.value = null;
   }
-  function W(u) {
-    const m = e.extractFromDOM(u);
-    return m && (r.value = m.data, o.value = m.schema), m;
+  function Y(u) {
+    const h = e.extractFromDOM(u);
+    return h && (r.value = h.data, o.value = h.schema), h;
   }
-  function X(u, m, $) {
-    const C = e.normalizeVueData(u, m, $);
-    r.value = C.data, o.value = C.schema;
+  function X(u, h, I) {
+    const k = e.normalizeVueData(u, h, I);
+    r.value = k.data, o.value = k.schema;
   }
   function Z(u) {
     e.initializeQA(u);
   }
   return {
     client: e,
-    loading: t,
-    error: n,
-    lastResult: s,
+    loading: n,
+    error: a,
+    lastResult: t,
     data: r,
     schema: o,
-    questionHistory: a,
+    questionHistory: s,
     answerHistory: i,
     lastAnswer: c,
     analyze: d,
-    getDescriptiveStats: A,
-    detectAnomalies: _,
-    performClustering: x,
-    predict: T,
-    askQuestion: z,
+    getDescriptiveStats: b,
+    detectAnomalies: P,
+    performClustering: $,
+    predict: D,
+    askQuestion: Q,
     generateSummary: M,
-    clearHistory: Y,
-    extractFromDOM: W,
+    clearHistory: G,
+    extractFromDOM: Y,
     loadFromVueGrid: X,
-    updateConfig: E,
+    updateConfig: x,
     initializeQA: Z,
-    setData: F,
-    reset: I
+    setData: z,
+    reset: E
   };
 }
-const he = { class: "ti-question-input" }, fe = { class: "ti-input-wrapper" }, pe = ["placeholder", "disabled", "onKeydown"], ge = ["disabled"], we = { key: 0 }, ye = {
+const me = { class: "ti-question-input" }, fe = { class: "ti-input-wrapper" }, pe = ["placeholder", "disabled", "onKeydown"], ye = ["disabled"], ge = { key: 0 }, we = {
   key: 1,
   class: "ti-loading"
 }, ve = {
   key: 0,
   class: "ti-hint"
-}, be = /* @__PURE__ */ U({
+}, Ae = /* @__PURE__ */ j({
   __name: "QuestionInput",
   props: {
     placeholder: { default: "Ask a question about this data..." },
@@ -988,166 +1109,166 @@ const he = { class: "ti-question-input" }, fe = { class: "ti-input-wrapper" }, p
   },
   emits: ["submit"],
   setup(l, { emit: e }) {
-    const t = l, n = e, s = D("");
+    const n = l, a = e, t = q("");
     function r() {
-      s.value.trim() && !t.disabled && !t.loading && (n("submit", s.value.trim()), s.value = "");
+      t.value.trim() && !n.disabled && !n.loading && (a("submit", t.value.trim()), t.value = "");
     }
-    function o(a) {
+    function o(s) {
     }
-    return (a, i) => (g(), p("div", he, [
-      h("div", fe, [
-        ee(h("textarea", {
-          "onUpdate:modelValue": i[0] || (i[0] = (c) => s.value = c),
-          placeholder: a.placeholder,
-          disabled: a.disabled,
+    return (s, i) => (g(), y("div", me, [
+      p("div", fe, [
+        ee(p("textarea", {
+          "onUpdate:modelValue": i[0] || (i[0] = (c) => t.value = c),
+          placeholder: s.placeholder,
+          disabled: s.disabled,
           class: "ti-textarea",
           rows: "1",
           onKeydown: [
-            L(H(r, ["exact", "prevent"]), ["enter"]),
-            L(H(o, ["shift"]), ["enter"])
+            H(J(r, ["exact", "prevent"]), ["enter"]),
+            H(J(o, ["shift"]), ["enter"])
           ]
         }, null, 40, pe), [
-          [te, s.value]
+          [te, t.value]
         ]),
-        h("button", {
-          disabled: a.disabled || !s.value.trim(),
+        p("button", {
+          disabled: s.disabled || !t.value.trim(),
           class: "ti-submit-btn",
           onClick: r
         }, [
-          a.loading ? (g(), p("span", ye, k(a.loadingLabel), 1)) : (g(), p("span", we, k(a.submitLabel), 1))
-        ], 8, ge)
+          s.loading ? (g(), y("span", we, C(s.loadingLabel), 1)) : (g(), y("span", ge, C(s.submitLabel), 1))
+        ], 8, ye)
       ]),
-      a.showHint ? (g(), p("div", ve, k(a.hint), 1)) : q("", !0)
+      s.showHint ? (g(), y("div", ve, C(s.hint), 1)) : _("", !0)
     ]));
   }
 }), V = (l, e) => {
-  const t = l.__vccOpts || l;
-  for (const [n, s] of e)
-    t[n] = s;
-  return t;
-}, Xe = /* @__PURE__ */ V(be, [["__scopeId", "data-v-f96008f3"]]), Ae = { class: "ti-answer-header" }, $e = { class: "ti-answer-icon" }, Ce = { key: 0 }, ke = { key: 1 }, _e = { class: "ti-answer-meta" }, xe = { class: "ti-confidence" }, Ee = { class: "ti-timestamp" }, De = { class: "ti-answer-text" }, Se = {
+  const n = l.__vccOpts || l;
+  for (const [a, t] of e)
+    n[a] = t;
+  return n;
+}, Xe = /* @__PURE__ */ V(Ae, [["__scopeId", "data-v-f96008f3"]]), be = { class: "ti-answer-header" }, Ie = { class: "ti-answer-icon" }, ke = { key: 0 }, Ce = { key: 1 }, Pe = { class: "ti-answer-meta" }, $e = { class: "ti-confidence" }, xe = { class: "ti-timestamp" }, qe = { class: "ti-answer-text" }, Se = {
   key: 0,
   class: "ti-approximate-notice"
-}, qe = {
+}, _e = {
   key: 1,
   class: "ti-reason"
-}, Te = {
+}, De = {
   key: 2,
   class: "ti-supporting-data"
-}, Ie = {
+}, Ee = {
   key: 0,
   class: "ti-supporting-content"
-}, Pe = {
+}, Te = {
   key: 0,
   class: "ti-aggregates"
 }, Ne = {
   key: 1,
   class: "ti-rows"
-}, Me = { class: "ti-table-wrapper" }, ze = { class: "ti-table" }, Qe = /* @__PURE__ */ U({
+}, Me = { class: "ti-table-wrapper" }, Qe = { class: "ti-table" }, Fe = /* @__PURE__ */ j({
   __name: "AnswerDisplay",
   props: {
     answer: {}
   },
   setup(l) {
-    const e = D(!1);
-    function t(n) {
-      return new Date(n).toLocaleTimeString();
+    const e = q(!1);
+    function n(a) {
+      return new Date(a).toLocaleTimeString();
     }
-    return (n, s) => (g(), p("div", {
-      class: ne(["ti-answer-display", { "ti-cannot-answer": n.answer.cannotAnswer }])
+    return (a, t) => (g(), y("div", {
+      class: ne(["ti-answer-display", { "ti-cannot-answer": a.answer.cannotAnswer }])
     }, [
-      h("div", Ae, [
-        h("div", $e, [
-          n.answer.cannotAnswer ? (g(), p("span", ke, "⚠️")) : (g(), p("span", Ce, "💡"))
+      p("div", be, [
+        p("div", Ie, [
+          a.answer.cannotAnswer ? (g(), y("span", Ce, "⚠️")) : (g(), y("span", ke, "💡"))
         ]),
-        h("div", _e, [
-          h("div", xe, " Confidence: " + k(Math.round(n.answer.confidence * 100)) + "% ", 1),
-          h("div", Ee, k(t(n.answer.timestamp)), 1)
+        p("div", Pe, [
+          p("div", $e, " Confidence: " + C(Math.round(a.answer.confidence * 100)) + "% ", 1),
+          p("div", xe, C(n(a.answer.timestamp)), 1)
         ])
       ]),
-      h("div", De, k(n.answer.text), 1),
-      n.answer.isApproximate ? (g(), p("div", Se, " ℹ️ This answer is based on sampled data and may be approximate. ")) : q("", !0),
-      n.answer.reason && n.answer.cannotAnswer ? (g(), p("div", qe, [
-        s[1] || (s[1] = h("strong", null, "Reason:", -1)),
-        se(" " + k(n.answer.reason), 1)
-      ])) : q("", !0),
-      n.answer.supportingData ? (g(), p("div", Te, [
-        h("button", {
+      p("div", qe, C(a.answer.text), 1),
+      a.answer.isApproximate ? (g(), y("div", Se, " ℹ️ This answer is based on sampled data and may be approximate. ")) : _("", !0),
+      a.answer.reason && a.answer.cannotAnswer ? (g(), y("div", _e, [
+        t[1] || (t[1] = p("strong", null, "Reason:", -1)),
+        ae(" " + C(a.answer.reason), 1)
+      ])) : _("", !0),
+      a.answer.supportingData ? (g(), y("div", De, [
+        p("button", {
           class: "ti-toggle-btn",
-          onClick: s[0] || (s[0] = (r) => e.value = !e.value)
-        }, k(e.value ? "▼" : "▶") + " Supporting Data ", 1),
-        e.value ? (g(), p("div", Ie, [
-          n.answer.supportingData.aggregates ? (g(), p("div", Pe, [
-            s[2] || (s[2] = h("h4", null, "Aggregates:", -1)),
-            h("pre", null, k(JSON.stringify(n.answer.supportingData.aggregates, null, 2)), 1)
-          ])) : q("", !0),
-          n.answer.supportingData.rows && n.answer.supportingData.rows.length > 0 ? (g(), p("div", Ne, [
-            h("h4", null, "Sample Rows (" + k(n.answer.supportingData.rows.length) + "):", 1),
-            h("div", Me, [
-              h("table", ze, [
-                h("thead", null, [
-                  h("tr", null, [
-                    (g(!0), p(Q, null, j(Object.keys(n.answer.supportingData.rows[0]), (r, o) => (g(), p("th", { key: o }, k(r), 1))), 128))
+          onClick: t[0] || (t[0] = (r) => e.value = !e.value)
+        }, C(e.value ? "▼" : "▶") + " Supporting Data ", 1),
+        e.value ? (g(), y("div", Ee, [
+          a.answer.supportingData.aggregates ? (g(), y("div", Te, [
+            t[2] || (t[2] = p("h4", null, "Aggregates:", -1)),
+            p("pre", null, C(JSON.stringify(a.answer.supportingData.aggregates, null, 2)), 1)
+          ])) : _("", !0),
+          a.answer.supportingData.rows && a.answer.supportingData.rows.length > 0 ? (g(), y("div", Ne, [
+            p("h4", null, "Sample Rows (" + C(a.answer.supportingData.rows.length) + "):", 1),
+            p("div", Me, [
+              p("table", Qe, [
+                p("thead", null, [
+                  p("tr", null, [
+                    (g(!0), y(F, null, O(Object.keys(a.answer.supportingData.rows[0]), (r, o) => (g(), y("th", { key: o }, C(r), 1))), 128))
                   ])
                 ]),
-                h("tbody", null, [
-                  (g(!0), p(Q, null, j(n.answer.supportingData.rows.slice(0, 5), (r, o) => (g(), p("tr", { key: o }, [
-                    (g(!0), p(Q, null, j(Object.keys(r), (a, i) => (g(), p("td", { key: i }, k(r[a]), 1))), 128))
+                p("tbody", null, [
+                  (g(!0), y(F, null, O(a.answer.supportingData.rows.slice(0, 5), (r, o) => (g(), y("tr", { key: o }, [
+                    (g(!0), y(F, null, O(Object.keys(r), (s, i) => (g(), y("td", { key: i }, C(r[s]), 1))), 128))
                   ]))), 128))
                 ])
               ])
             ])
-          ])) : q("", !0)
-        ])) : q("", !0)
-      ])) : q("", !0)
+          ])) : _("", !0)
+        ])) : _("", !0)
+      ])) : _("", !0)
     ], 2));
   }
-}), Ze = /* @__PURE__ */ V(Qe, [["__scopeId", "data-v-d1aaae1d"]]), je = { class: "ti-question-history" }, Fe = { class: "ti-history-header" }, Re = {
+}), Ze = /* @__PURE__ */ V(Fe, [["__scopeId", "data-v-d1aaae1d"]]), Oe = { class: "ti-question-history" }, ze = { class: "ti-history-header" }, Re = {
   key: 0,
   class: "ti-empty-state"
-}, Ue = {
+}, je = {
   key: 1,
   class: "ti-history-list"
-}, Ve = ["onClick"], Le = { class: "ti-question-header" }, He = { class: "ti-question-number" }, Be = { class: "ti-question-time" }, Oe = { class: "ti-question-text" }, Je = {
+}, Ve = ["onClick"], He = { class: "ti-question-header" }, Je = { class: "ti-question-number" }, Ue = { class: "ti-question-time" }, Le = { class: "ti-question-text" }, Be = {
   key: 0,
   class: "ti-question-context"
-}, Ke = /* @__PURE__ */ U({
+}, Ke = /* @__PURE__ */ j({
   __name: "QuestionHistory",
   props: {
     questions: {}
   },
   emits: ["clear", "select"],
   setup(l, { emit: e }) {
-    const t = l, n = re(() => [...t.questions].reverse());
-    function s(r) {
-      const o = new Date(r), i = (/* @__PURE__ */ new Date()).getTime() - o.getTime(), c = Math.floor(i / 6e4), w = Math.floor(i / 36e5), f = Math.floor(i / 864e5);
-      return c < 1 ? "Just now" : c < 60 ? `${c}m ago` : w < 24 ? `${w}h ago` : `${f}d ago`;
+    const n = l, a = se(() => [...n.questions].reverse());
+    function t(r) {
+      const o = new Date(r), i = (/* @__PURE__ */ new Date()).getTime() - o.getTime(), c = Math.floor(i / 6e4), f = Math.floor(i / 36e5), m = Math.floor(i / 864e5);
+      return c < 1 ? "Just now" : c < 60 ? `${c}m ago` : f < 24 ? `${f}h ago` : `${m}d ago`;
     }
-    return (r, o) => (g(), p("div", je, [
-      h("div", Fe, [
-        o[1] || (o[1] = h("h3", null, "Question History", -1)),
-        r.questions.length > 0 ? (g(), p("button", {
+    return (r, o) => (g(), y("div", Oe, [
+      p("div", ze, [
+        o[1] || (o[1] = p("h3", null, "Question History", -1)),
+        r.questions.length > 0 ? (g(), y("button", {
           key: 0,
           class: "ti-clear-btn",
-          onClick: o[0] || (o[0] = (a) => r.$emit("clear"))
-        }, " Clear History ")) : q("", !0)
+          onClick: o[0] || (o[0] = (s) => r.$emit("clear"))
+        }, " Clear History ")) : _("", !0)
       ]),
-      r.questions.length === 0 ? (g(), p("div", Re, o[2] || (o[2] = [
-        h("div", { class: "ti-empty-icon" }, "💬", -1),
-        h("p", null, "No questions asked yet", -1),
-        h("p", { class: "ti-empty-hint" }, "Ask a question about your data to get started", -1)
-      ]))) : (g(), p("div", Ue, [
-        (g(!0), p(Q, null, j(n.value, (a, i) => (g(), p("div", {
-          key: a.id,
+      r.questions.length === 0 ? (g(), y("div", Re, o[2] || (o[2] = [
+        p("div", { class: "ti-empty-icon" }, "💬", -1),
+        p("p", null, "No questions asked yet", -1),
+        p("p", { class: "ti-empty-hint" }, "Ask a question about your data to get started", -1)
+      ]))) : (g(), y("div", je, [
+        (g(!0), y(F, null, O(a.value, (s, i) => (g(), y("div", {
+          key: s.id,
           class: "ti-history-item",
-          onClick: (c) => r.$emit("select", a)
+          onClick: (c) => r.$emit("select", s)
         }, [
-          h("div", Le, [
-            h("span", He, "#" + k(r.questions.length - i), 1),
-            h("span", Be, k(s(a.timestamp)), 1)
+          p("div", He, [
+            p("span", Je, "#" + C(r.questions.length - i), 1),
+            p("span", Ue, C(t(s.timestamp)), 1)
           ]),
-          h("div", Oe, k(a.text), 1),
-          a.context ? (g(), p("div", Je, k(a.context.rowCount) + " rows ", 1)) : q("", !0)
+          p("div", Le, C(s.text), 1),
+          s.context ? (g(), y("div", Be, C(s.context.rowCount) + " rows ", 1)) : _("", !0)
         ], 8, Ve))), 128))
       ]))
     ]));
@@ -1155,21 +1276,21 @@ const he = { class: "ti-question-input" }, fe = { class: "ti-input-wrapper" }, p
 }), et = /* @__PURE__ */ V(Ke, [["__scopeId", "data-v-c66393d9"]]);
 export {
   Ze as AnswerDisplay,
-  B as QAEngine,
+  U as QAEngine,
   et as QuestionHistory,
   Xe as QuestionInput,
-  me as TabularIntelligence,
-  J as calculateStats,
+  he as TabularIntelligence,
+  B as calculateStats,
   de as convertToTabular,
-  ae as detectAnomalies,
-  G as executeAPIRequest,
-  Ye as executeMultipleRequests,
+  re as detectAnomalies,
+  W as executeAPIRequest,
+  Ge as executeMultipleRequests,
   oe as extractFromDOM,
-  O as inferColumnType,
+  L as inferColumnType,
   N as inferSchema,
   ie as normalizeVueData,
   ce as parsePostmanCollection,
-  P as replaceVariables,
-  We as useTabularIntelligence
+  T as replaceVariables,
+  Ye as useTabularIntelligence
 };
 //# sourceMappingURL=index.mjs.map