feat(inputs.exec): Convert 'commands' to already parsed argument lists

plugins/inputs/exec/README.md

@@ -23,7 +23,11 @@ plugin ordering. See [CONFIGURATION.md][CONFIGURATION.md] for more details.
 ```toml @sample.conf
 # Read metrics from one or more commands that can output to stdout
 [[inputs.exec]]
-  ## Commands array
+  ## Commands array with each command being an array of the command and its
+  ## flags, e.g.
+  ##   commands = [['C:\Windows\System32\w32tm.exe', "/stripchart", "/computer:SERVER1", "/dataonly", "/samples:3"]]
+  ## or
+  ##   commands = [["echo", "value 1"], ["echo", "value 2"]]
   commands = []
 
   ## Environment variables

plugins/inputs/exec/exec.go

@@ -5,12 +5,16 @@ import (
 	"bufio"
 	"bytes"
 	_ "embed"
+	"errors"
 	"fmt"
 	"path/filepath"
+	"strconv"
 	"strings"
 	"sync"
 	"time"
 
+	"github.com/kballard/go-shellquote"
+
 	"github.com/influxdata/telegraf"
 	"github.com/influxdata/telegraf/config"
 	"github.com/influxdata/telegraf/internal"
@@ -27,7 +31,7 @@ var once sync.Once
 const maxStderrBytes int = 512
 
 type Exec struct {
-	Commands    []string        `toml:"commands"`
+	Commands    []interface{}   `toml:"commands"`
 	Command     string          `toml:"command"`
 	Environment []string        `toml:"environment"`
 	IgnoreError bool            `toml:"ignore_error"`
@@ -38,6 +42,7 @@ type Exec struct {
 	parser telegraf.Parser
 
 	runner runner
+	cmds   [][]string
 
 	// Allow post-processing of command exit codes
 	exitCodeHandler   exitCodeHandlerFunc
@@ -47,7 +52,7 @@ type Exec struct {
 type exitCodeHandlerFunc func([]telegraf.Metric, error, []byte) []telegraf.Metric
 
 type runner interface {
-	run(string) ([]byte, []byte, error)
+	run([]string) ([]byte, []byte, error)
 }
 
 type commandRunner struct {
@@ -66,6 +71,64 @@ func (e *Exec) Init() error {
 		e.Commands = append(e.Commands, e.Command)
 	}
 
+	if len(e.Commands) == 0 {
+		return errors.New("no command specified")
+	}
+
+	e.cmds = make([][]string, 0, len(e.Commands))
+	for _, raw := range e.Commands {
+		switch c := raw.(type) {
+		case string:
+			// Legacy single string command setting
+			if c == "" {
+				return errors.New("command string cannot be empty")
+			}
+
+			// Convert the legacy command string to a string list and output
+			// deprecation notice
+			cmd, err := shellquote.Split(c)
+			if err != nil {
+				return fmt.Errorf("unable to parse command %q: %w", c, err)
+			}
+			if len(cmd) == 0 {
+				return errors.New("command cannot be empty")
+			}
+			// Create the corresponding command in the new syntax to ease migration
+			suggestion := make([]string, 0, len(cmd))
+			for _, a := range cmd {
+				suggestion = append(suggestion, strconv.Quote(a))
+			}
+			config.PrintOptionValueDeprecationNotice("inputs.exec", "command", c, telegraf.DeprecationInfo{
+				Since:     "1.39.0",
+				RemovalIn: "1.45.0",
+				Notice:    fmt.Sprintf("Use array syntax instead: [%s]", strings.Join(suggestion, ",")),
+			})
+			e.cmds = append(e.cmds, cmd)
+		case []string:
+			if len(c) == 0 {
+				return errors.New("command cannot be empty")
+			}
+			e.cmds = append(e.cmds, c)
+		case []interface{}:
+			if len(c) == 0 {
+				return errors.New("command cannot be empty")
+			}
+
+			// Convert the entries to a string list
+			converted := make([]string, 0, len(c))
+			for _, r := range c {
+				v, ok := r.(string)
+				if !ok {
+					return fmt.Errorf("command %v has invalid entry %v of type %T, expected string", raw, r, r)
+				}
+				converted = append(converted, v)
+			}
+			e.cmds = append(e.cmds, converted)
+		default:
+			return fmt.Errorf("command %v has invalid type %T, expected string list", raw, raw)
+		}
+	}
+
 	e.runner = &commandRunner{
 		environment: e.Environment,
 		timeout:     time.Duration(e.Timeout),
@@ -92,56 +155,50 @@ func (e *Exec) Gather(acc telegraf.Accumulator) error {
 	commands := e.updateRunners()
 
 	var wg sync.WaitGroup
-	for _, cmd := range commands {
+	for _, item := range commands {
 		wg.Add(1)
-
-		go func(c string) {
+		go func(c []string) {
 			defer wg.Done()
 			acc.AddError(e.processCommand(acc, c))
-		}(cmd)
+		}(item)
 	}
 	wg.Wait()
 	return nil
 }
 
-func (e *Exec) updateRunners() []string {
-	commands := make([]string, 0, len(e.Commands))
-	for _, pattern := range e.Commands {
-		if pattern == "" {
-			continue
-		}
-
+func (e *Exec) updateRunners() [][]string {
+	commands := make([][]string, 0, len(e.cmds))
+	for _, cmd := range e.cmds {
 		// Try to expand globbing expressions
-		cmd, args, found := strings.Cut(pattern, " ")
-		matches, err := filepath.Glob(cmd)
+		matches, err := filepath.Glob(cmd[0])
 		if err != nil {
-			e.Log.Errorf("Matching command %q failed: %v", cmd, err)
+			e.Log.Errorf("Matching command %q failed: %v", cmd[0], err)
 			continue
 		}
 
 		if len(matches) == 0 {
 			// There were no matches with the glob pattern, so let's assume
 			// the command is in PATH and just run it as it is
-			commands = append(commands, pattern)
+			commands = append(commands, cmd)
 		} else {
 			// There were matches, so we'll append each match together with
 			// the arguments to the commands slice
 			for _, match := range matches {
-				if found {
-					match += " " + args
-				}
-				commands = append(commands, match)
+				expanded := make([]string, 0, len(cmd))
+				expanded = append(expanded, match)
+				expanded = append(expanded, cmd[1:]...)
+				commands = append(commands, expanded)
 			}
 		}
 	}
 
 	return commands
 }
 
-func (e *Exec) processCommand(acc telegraf.Accumulator, cmd string) error {
+func (e *Exec) processCommand(acc telegraf.Accumulator, cmd []string) error {
 	out, errBuf, runErr := e.runner.run(cmd)
 	if !e.IgnoreError && !e.parseDespiteError && runErr != nil {
-		return fmt.Errorf("exec: %w for command %q: %s", runErr, cmd, string(errBuf))
+		return fmt.Errorf("exec: %w for command %q: %s", runErr, strings.Join(cmd, " "), string(errBuf))
 	}
 
 	// Log output in stderr