This is an automated email from the ASF dual-hosted git repository.
davsclaus pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/camel.git
The following commit(s) were added to refs/heads/main by this push:
new 50a3e865f077 CAMEL-23861: Add AI usage stats view to TUI AI panel
(#24362)
50a3e865f077 is described below
commit 50a3e865f077dbfbe27c8f80ced979a393467d5c
Author: Claus Ibsen <[email protected]>
AuthorDate: Wed Jul 1 14:40:36 2026 +0200
CAMEL-23861: Add AI usage stats view to TUI AI panel (#24362)
* CAMEL-23861: Add AI usage stats view to TUI AI panel (Ctrl+U)
Co-Authored-By: Claude <[email protected]>
Signed-off-by: Claus Ibsen <[email protected]>
* CAMEL-23861: Use locale-independent dot in token formatter
Co-Authored-By: Claude <[email protected]>
Signed-off-by: Claus Ibsen <[email protected]>
---------
Signed-off-by: Claus Ibsen <[email protected]>
Co-authored-by: Claude <[email protected]>
---
.../camel/dsl/jbang/core/commands/LlmClient.java | 10 +-
.../camel/dsl/jbang/core/commands/tui/AiPanel.java | 240 ++++++++++++++++++++-
2 files changed, 242 insertions(+), 8 deletions(-)
diff --git
a/dsl/camel-jbang/camel-jbang-core/src/main/java/org/apache/camel/dsl/jbang/core/commands/LlmClient.java
b/dsl/camel-jbang/camel-jbang-core/src/main/java/org/apache/camel/dsl/jbang/core/commands/LlmClient.java
index 63488519de91..472c37b4cb0f 100644
---
a/dsl/camel-jbang/camel-jbang-core/src/main/java/org/apache/camel/dsl/jbang/core/commands/LlmClient.java
+++
b/dsl/camel-jbang/camel-jbang-core/src/main/java/org/apache/camel/dsl/jbang/core/commands/LlmClient.java
@@ -111,7 +111,7 @@ public class LlmClient {
if (k == (int) k) {
return (int) k + "k";
}
- return String.format("%.1fk", k);
+ return String.format(java.util.Locale.ROOT, "%.1fk", k);
}
return String.valueOf(tokens);
}
@@ -153,6 +153,14 @@ public class LlmClient {
private String vertexRegion;
private String vertexProjectId;
+ public String model() {
+ return model;
+ }
+
+ public ApiType apiType() {
+ return apiType;
+ }
+
// -- Builder --
public static LlmClient create() {
diff --git
a/dsl/camel-jbang/camel-jbang-plugin-tui/src/main/java/org/apache/camel/dsl/jbang/core/commands/tui/AiPanel.java
b/dsl/camel-jbang/camel-jbang-plugin-tui/src/main/java/org/apache/camel/dsl/jbang/core/commands/tui/AiPanel.java
index ee4bb5c59e59..df2c87e85438 100644
---
a/dsl/camel-jbang/camel-jbang-plugin-tui/src/main/java/org/apache/camel/dsl/jbang/core/commands/tui/AiPanel.java
+++
b/dsl/camel-jbang/camel-jbang-plugin-tui/src/main/java/org/apache/camel/dsl/jbang/core/commands/tui/AiPanel.java
@@ -20,7 +20,9 @@ import java.time.Instant;
import java.time.ZoneId;
import java.time.format.DateTimeFormatter;
import java.util.ArrayList;
+import java.util.LinkedHashMap;
import java.util.List;
+import java.util.Map;
import java.util.concurrent.atomic.AtomicBoolean;
import dev.tamboui.layout.Constraint;
@@ -34,11 +36,18 @@ import dev.tamboui.text.Line;
import dev.tamboui.text.Span;
import dev.tamboui.tui.event.KeyCode;
import dev.tamboui.tui.event.KeyEvent;
+import dev.tamboui.widgets.barchart.Bar;
+import dev.tamboui.widgets.barchart.BarChart;
+import dev.tamboui.widgets.barchart.BarGroup;
import dev.tamboui.widgets.block.Block;
import dev.tamboui.widgets.block.BorderType;
import dev.tamboui.widgets.block.Borders;
import dev.tamboui.widgets.block.Title;
import dev.tamboui.widgets.paragraph.Paragraph;
+import dev.tamboui.widgets.table.Cell;
+import dev.tamboui.widgets.table.Row;
+import dev.tamboui.widgets.table.Table;
+import dev.tamboui.widgets.table.TableState;
import org.apache.camel.dsl.jbang.core.commands.AskTools;
import org.apache.camel.dsl.jbang.core.commands.LlmClient;
@@ -91,12 +100,22 @@ class AiPanel {
// Activity log for AI Log popup
private final List<LogEntry> activityLog = new ArrayList<>();
+ // AI usage stats
+ private final List<AiUsageEntry> usageHistory = new ArrayList<>();
+ private final TableState statsTableState = new TableState();
+ private boolean statsView;
+ private int statsScrollOffset;
+
record ConversationEntry(String role, String text, long elapsedSeconds,
int totalTokens) {
ConversationEntry(String role, String text) {
this(role, text, -1, 0);
}
}
+ record AiUsageEntry(String model, String provider, int inputTokens, int
outputTokens,
+ int totalTokens, long latencyMs, String stopReason, Instant
timestamp) {
+ }
+
void setContext(MonitorContext ctx) {
this.ctx = ctx;
}
@@ -187,12 +206,25 @@ class AiPanel {
close();
return true;
}
+ if (ke.hasCtrl() && ke.isCharIgnoreCase('u')) {
+ statsView = !statsView;
+ statsScrollOffset = 0;
+ return true;
+ }
if (ke.isKey(KeyCode.PAGE_UP)) {
- scrollOffset += 5;
+ if (statsView) {
+ statsScrollOffset += 5;
+ } else {
+ scrollOffset += 5;
+ }
return true;
}
if (ke.isKey(KeyCode.PAGE_DOWN)) {
- scrollOffset = Math.max(0, scrollOffset - 5);
+ if (statsView) {
+ statsScrollOffset = Math.max(0, statsScrollOffset - 5);
+ } else {
+ scrollOffset = Math.max(0, scrollOffset - 5);
+ }
return true;
}
if (thinking.get()) {
@@ -300,7 +332,9 @@ class AiPanel {
throw new InterruptedException();
}
+ long callStart = System.currentTimeMillis();
LlmClient.ChatResponse response =
client.chatWithTools(systemPrompt, messages, tools);
+ long callLatency = System.currentTimeMillis() - callStart;
if (response == null) {
String err = "No response from LLM";
conversation.add(new ConversationEntry("error", err));
@@ -308,6 +342,7 @@ class AiPanel {
return;
}
totalUsage = totalUsage.add(response.usage());
+ recordUsage(response, callLatency);
// check for error response (null text, no tool calls, error stop
reason)
if ("error".equals(response.stopReason())
@@ -358,12 +393,27 @@ class AiPanel {
"Reached maximum iterations (" + MAX_ITERATIONS + ") without a
final answer."));
}
+ private void recordUsage(LlmClient.ChatResponse response, long latencyMs) {
+ if (client == null || response.usage().totalTokens() == 0) {
+ return;
+ }
+ String model = client.model() != null ? client.model() : "unknown";
+ String provider = client.apiType() != null ? client.apiType().name() :
"unknown";
+ usageHistory.add(new AiUsageEntry(
+ model, provider,
+ response.usage().inputTokens(),
response.usage().outputTokens(),
+ response.usage().totalTokens(), latencyMs,
+ response.stopReason(), Instant.now()));
+ }
+
void render(Frame frame, Rect area) {
// At 25% show elapsed and tokens in the title bar to save space
long titleElapsed = lastResponseElapsed();
int titleTokens = lastResponseTokens();
Line titleLine;
- if (splitIndex == 0 && titleElapsed >= 0) {
+ if (statsView) {
+ titleLine = Line.from(Span.styled(" AI Usage ",
Style.EMPTY.bold()));
+ } else if (splitIndex == 0 && titleElapsed >= 0) {
String tokenSuffix = titleTokens > 0 ? ", " +
LlmClient.formatTokens(titleTokens) + " tokens" : "";
titleLine = Line.from(
Span.styled(" AI ", Style.EMPTY.bold()),
@@ -387,6 +437,11 @@ class AiPanel {
return;
}
+ if (statsView) {
+ renderStats(frame, inner);
+ return;
+ }
+
// Split inner area: conversation (fill) + separator (1 row) + input
(1 row)
List<Rect> parts = Layout.vertical()
.constraints(Constraint.fill(), Constraint.length(1),
Constraint.length(1))
@@ -550,12 +605,183 @@ class AiPanel {
void renderFooter(List<Span> spans) {
MonitorContext.hint(spans, "F8", "close");
+ if (statsView) {
+ MonitorContext.hint(spans, "Ctrl+U", "chat");
+ } else {
+ MonitorContext.hint(spans, "Ctrl+U", "usage");
+ }
MonitorContext.hint(spans, "Shift+F8", "resize (" +
SPLIT_PERCENTS[splitIndex] + "%)");
MonitorContext.hint(spans, "PgUp/Dn", "scroll");
- if (!thinking.get()) {
- MonitorContext.hint(spans, "Enter", "send");
- } else {
- MonitorContext.hint(spans, "Ctrl+C", "cancel");
+ if (!statsView) {
+ if (!thinking.get()) {
+ MonitorContext.hint(spans, "Enter", "send");
+ } else {
+ MonitorContext.hint(spans, "Ctrl+C", "cancel");
+ }
+ }
+ }
+
+ private void renderStats(Frame frame, Rect area) {
+ if (area.height() < 3) {
+ return;
+ }
+
+ if (usageHistory.isEmpty()) {
+ frame.renderWidget(
+ Paragraph.from(Line.from(Span.styled("No AI usage data
yet. Ask a question first.", Style.EMPTY.dim()))),
+ area);
+ return;
+ }
+
+ // Compute aggregates
+ int totalInput = 0;
+ int totalOutput = 0;
+ int totalTokens = 0;
+ long totalLatency = 0;
+ for (AiUsageEntry e : usageHistory) {
+ totalInput += e.inputTokens();
+ totalOutput += e.outputTokens();
+ totalTokens += e.totalTokens();
+ totalLatency += e.latencyMs();
+ }
+ int requestCount = usageHistory.size();
+
+ // Per-model aggregation
+ Map<String, int[]> perModel = new LinkedHashMap<>();
+ for (AiUsageEntry e : usageHistory) {
+ String key = e.model() + " (" + e.provider() + ")";
+ int[] stats = perModel.computeIfAbsent(key, k -> new int[4]);
+ stats[0]++; // requests
+ stats[1] += e.inputTokens();
+ stats[2] += e.outputTokens();
+ stats[3] += e.totalTokens();
+ }
+
+ // Per-conversation token totals (group by consecutive runs between
user questions)
+ List<Integer> turnTokens = new ArrayList<>();
+ int currentTurn = 0;
+ int turnIndex = 0;
+ for (AiUsageEntry e : usageHistory) {
+ if (turnIndex > 0) {
+ AiUsageEntry prev = usageHistory.get(turnIndex - 1);
+ long gap = e.timestamp().toEpochMilli() -
prev.timestamp().toEpochMilli();
+ if (gap > 30_000) {
+ turnTokens.add(currentTurn);
+ currentTurn = 0;
+ }
+ }
+ currentTurn += e.totalTokens();
+ turnIndex++;
+ }
+ if (currentTurn > 0) {
+ turnTokens.add(currentTurn);
+ }
+
+ // Layout: summary (2 rows) + model table (header + models + 1 blank)
+ chart (fill)
+ int tableRows = perModel.size() + 1;
+ int summaryRows = 2;
+ int chartMinRows = 4;
+ boolean hasChart = area.height() > summaryRows + tableRows +
chartMinRows + 1;
+
+ List<Constraint> constraints = new ArrayList<>();
+ constraints.add(Constraint.length(summaryRows));
+ constraints.add(Constraint.length(tableRows + 1));
+ if (hasChart) {
+ constraints.add(Constraint.fill());
+ }
+ List<Rect> sections = Layout.vertical()
+ .constraints(constraints)
+ .split(area);
+
+ // --- Summary ---
+ Rect summaryArea = sections.get(0);
+ Style dimStyle = Style.EMPTY.dim();
+ Style cyanStyle = Style.EMPTY.fg(Color.CYAN);
+ List<Line> summaryLines = new ArrayList<>();
+ summaryLines.add(Line.from(
+ Span.styled("Requests: ", dimStyle),
+ Span.styled(String.valueOf(requestCount), cyanStyle),
+ Span.styled(" Total tokens: ", dimStyle),
+ Span.styled(LlmClient.formatTokens(totalTokens), cyanStyle),
+ Span.styled(" (in: ", dimStyle),
+ Span.styled(LlmClient.formatTokens(totalInput),
Style.EMPTY.fg(Color.GREEN)),
+ Span.styled(" / out: ", dimStyle),
+ Span.styled(LlmClient.formatTokens(totalOutput),
Style.EMPTY.fg(Color.YELLOW)),
+ Span.styled(")", dimStyle)));
+ summaryLines.add(Line.from(
+ Span.styled("Avg latency: ", dimStyle),
+ Span.styled((totalLatency / requestCount / 1000) + "s",
cyanStyle),
+ Span.styled(" Total time: ", dimStyle),
+ Span.styled((totalLatency / 1000) + "s", cyanStyle)));
+ frame.renderWidget(
+ Paragraph.from(new dev.tamboui.text.Text(summaryLines,
dev.tamboui.layout.Alignment.LEFT)),
+ summaryArea);
+
+ // --- Per-model table ---
+ Rect tableArea = sections.get(1);
+ List<Row> rows = new ArrayList<>();
+ for (Map.Entry<String, int[]> entry : perModel.entrySet()) {
+ int[] s = entry.getValue();
+ rows.add(Row.from(
+ Cell.from(Span.styled(entry.getKey(), cyanStyle)),
+ Cell.from(String.valueOf(s[0])),
+ Cell.from(LlmClient.formatTokens(s[1])),
+ Cell.from(LlmClient.formatTokens(s[2])),
+ Cell.from(LlmClient.formatTokens(s[3]))));
+ }
+ Table table = Table.builder()
+ .rows(rows)
+ .header(Row.from(
+ Cell.from(Span.styled("MODEL", Style.EMPTY.bold())),
+ Cell.from(Span.styled("REQS", Style.EMPTY.bold())),
+ Cell.from(Span.styled("INPUT", Style.EMPTY.bold())),
+ Cell.from(Span.styled("OUTPUT", Style.EMPTY.bold())),
+ Cell.from(Span.styled("TOTAL", Style.EMPTY.bold()))))
+ .widths(
+ Constraint.fill(),
+ Constraint.length(6),
+ Constraint.length(8),
+ Constraint.length(8),
+ Constraint.length(8))
+ .build();
+ frame.renderStatefulWidget(table, tableArea, statsTableState);
+
+ // --- Token bar chart per turn ---
+ if (hasChart && turnTokens.size() > 1) {
+ Rect chartArea = sections.get(2);
+
+ // Title row + chart
+ List<Rect> chartParts = Layout.vertical()
+ .constraints(Constraint.length(1), Constraint.fill())
+ .split(chartArea);
+ frame.renderWidget(
+ Paragraph.from(Line.from(Span.styled("Tokens per
question:", Style.EMPTY.bold()))),
+ chartParts.get(0));
+
+ Rect barArea = chartParts.get(1);
+ int maxTokensInTurn =
turnTokens.stream().mapToInt(Integer::intValue).max().orElse(1);
+
+ // Limit bars to available width
+ int maxBars = Math.max(1, barArea.width() / 2);
+ int startIdx = Math.max(0, turnTokens.size() - maxBars);
+ List<BarGroup> groups = new ArrayList<>();
+ for (int i = startIdx; i < turnTokens.size(); i++) {
+ groups.add(BarGroup.of(
+ Bar.builder()
+ .value(turnTokens.get(i))
+ .textValue("")
+ .style(Style.EMPTY.fg(Color.CYAN))
+ .build()));
+ }
+
+ BarChart barChart = BarChart.builder()
+ .data(groups)
+ .max(maxTokensInTurn + 2)
+ .barWidth(1)
+ .barGap(1)
+ .groupGap(0)
+ .build();
+ frame.renderWidget(barChart, barArea);
}
}
