From 51480e6b358a0ddf95bf374061e38334a12d80a5 Mon Sep 17 00:00:00 2001
From: Thomas von Dein <tom@vondein.org>
Date: Wed, 8 Jul 2026 14:58:07 +0200
Subject: [PATCH] add 'index du <index>'

---
 cmd/index.go    | 22 ++++++++++++++
 pkg/es/index.go | 79 +++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 101 insertions(+)

diff --git a/cmd/index.go b/cmd/index.go
index 67f19e9..09fc656 100644
--- a/cmd/index.go
+++ b/cmd/index.go
@@ -41,6 +41,7 @@ func Index(conf *cfg.Config) *cli.Command {
 			IndexClose(conf),
 			IndexFields(conf),
 			IndexIlm(conf),
+			IndexDu(conf),
 
 			// sub commands
 			IndexAlias(conf),
@@ -277,3 +278,24 @@ func IndexIlm(conf *cfg.Config) *cli.Command {
 		},
 	}
 }
+
+func IndexDu(conf *cfg.Config) *cli.Command {
+	return &cli.Command{
+		Name:      "du",
+		Usage:     "show index disk usage",
+		UsageText: "index du <index>",
+
+		ShellComplete: func(ctx context.Context, cmd *cli.Command) {
+			complete(conf, cmd, Cindex)
+		},
+
+		Action: func(ctx context.Context, cmd *cli.Command) error {
+			index := cmd.Args().Get(0)
+			if index == "" {
+				return errors.New("no index specified")
+			}
+
+			return es.IndexDiskusage(conf, index)
+		},
+	}
+}
diff --git a/pkg/es/index.go b/pkg/es/index.go
index 7929a7e..e1ab0f7 100644
--- a/pkg/es/index.go
+++ b/pkg/es/index.go
@@ -18,6 +18,8 @@ package es
 
 import (
 	"context"
+	"encoding/json"
+	"errors"
 	"fmt"
 	"log/slog"
 	"regexp"
@@ -28,6 +30,7 @@ import (
 
 	"codeberg.org/scip/esctl/pkg/cfg"
 	"codeberg.org/scip/esctl/pkg/printer"
+	"github.com/charmbracelet/lipgloss"
 	"github.com/elastic/go-elasticsearch/v9/typedapi/cat/indices"
 	"github.com/elastic/go-elasticsearch/v9/typedapi/esdsl"
 	"github.com/elastic/go-elasticsearch/v9/typedapi/types/enums/healthstatus"
@@ -306,3 +309,79 @@ func IndexFields(conf *cfg.Config, index string) error {
 
 	return nil
 }
+
+type Diskusage struct {
+	Total       int64 `json:"total_in_bytes"`
+	Points      int64 `json:"points_in_bytes"`
+	Norms       int64 `json:"norms_in_bytes"`
+	TermVectors int64 `json:"term_vectors_in_bytes"`
+	KnnVectors  int64 `json:"knn_vectors_in_bytes"`
+	BloomFilter int64 `json:"bloom_filter_in_bytes"`
+}
+
+type IndexDiskUsage struct {
+	AllFields Diskusage            `json:"all_fields"`
+	Fields    map[string]Diskusage `json:"fields"`
+}
+
+type ResIndexDiskUsage map[string]IndexDiskUsage
+
+func IndexDiskusage(conf *cfg.Config, index string) error {
+	var bold = lipgloss.NewStyle().Bold(true)
+
+	res, err := conf.DefaultCluster.ES().Indices.DiskUsage(index).
+		RunExpensiveTasks(true).
+		Do(context.Background())
+	if err != nil {
+		return fmt.Errorf("failed to retrieve index disk usage: %w", esErrorString(err))
+	}
+
+	dures := ResIndexDiskUsage{}
+	if err := json.Unmarshal(res, &dures); err != nil {
+		return fmt.Errorf("failed to unmarshal disk usage response: %w", err)
+	}
+
+	du, exists := dures[index]
+	if !exists {
+		return errors.New("no disk usage reported for index")
+	}
+
+	table := printer.NewTableEmpty(conf).
+		WithHeaders("field", "bloom filter", "norms", "points", "term vectors", "knn vectors", "total")
+
+	for name, field := range du.Fields {
+		if strings.HasPrefix(name, "_") || strings.HasSuffix(name, ".keyword") {
+			continue
+		}
+
+		table.AddRow(
+			name,
+			printer.Bytes(field.BloomFilter),
+			printer.Bytes(field.Norms),
+			printer.Bytes(field.Points),
+			printer.Bytes(field.TermVectors),
+			printer.Bytes(field.KnnVectors),
+			printer.Bytes(field.Total),
+		)
+	}
+
+	all := du.AllFields
+
+	table.Sort()
+
+	table.AddRowLate(
+		bold.Render("Summary"),
+		printer.Bytes(all.BloomFilter),
+		printer.Bytes(all.Norms),
+		printer.Bytes(all.Points),
+		printer.Bytes(all.TermVectors),
+		printer.Bytes(all.KnnVectors),
+		printer.Bytes(all.Total),
+	)
+
+	if err := table.Print(); err != nil {
+		return err
+	}
+
+	return nil
+}