Bound external result file reads

internal/adapters/audita/subprocess.go

@@ -14,6 +14,12 @@ import (
 	"gitea.maximumdirect.net/eric/narratio/internal/fileops"
 )
 
+// MaxProcessedOutputBytes bounds Audita's processed-transcript JSON result.
+const MaxProcessedOutputBytes int64 = 64 * 1024 * 1024
+
+// MaxReportOutputBytes bounds Audita's optional report JSON result.
+const MaxReportOutputBytes int64 = 16 * 1024 * 1024
+
 // SubprocessRunnerConfig defines deterministic settings for Audita CLI execution.
 type SubprocessRunnerConfig struct {
 	Binary                   string
@@ -375,9 +381,9 @@ func (r *SubprocessRunner) writeInvocationConfig(req PolishRequest, args []strin
 }
 
 func validateProcessedOutput(path string) error {
-	data, err := os.ReadFile(path)
+	data, err := readAuditaResult(path, MaxProcessedOutputBytes, "processed transcript")
 	if err != nil {
-		return fmt.Errorf("read file: %w", err)
+		return err
 	}
 
 	var payload map[string]any
@@ -406,9 +412,9 @@ func addSubprocessStreamHint(message string, runErr error) string {
 }
 
 func validateJSONFile(path string) error {
-	data, err := os.ReadFile(path)
+	data, err := readAuditaResult(path, MaxReportOutputBytes, "report")
 	if err != nil {
-		return fmt.Errorf("read file: %w", err)
+		return err
 	}
 	var v any
 	if err := json.Unmarshal(data, &v); err != nil {
@@ -416,3 +422,11 @@ func validateJSONFile(path string) error {
 	}
 	return nil
 }
+
+func readAuditaResult(path string, limit int64, category string) ([]byte, error) {
+	data, err := fileops.ReadRegularFile(path, limit)
+	if err != nil {
+		return nil, fmt.Errorf("audita %s result exceeds or cannot be read within %d-byte limit: %w", category, limit, err)
+	}
+	return data, nil
+}

internal/adapters/notarius/subprocess.go

@@ -5,12 +5,12 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
-	"io"
 	"os"
 	"path/filepath"
 	"strings"
 
 	"gitea.maximumdirect.net/eric/narratio/internal/adapters/subprocess"
+	"gitea.maximumdirect.net/eric/narratio/internal/fileops"
 	"gitea.maximumdirect.net/eric/narratio/internal/pathsafe"
 )
 
@@ -388,32 +388,9 @@ func loadWarnings(path string) ([]WarningSummary, error) {
 }
 
 func decodeBoundedJSON(path string, limit int64, destination any) error {
-	inspected, err := os.Lstat(path)
+	data, err := fileops.ReadRegularFile(path, limit)
 	if err != nil {
-		return err
-	}
-	if inspected.Mode()&os.ModeSymlink != 0 || !inspected.Mode().IsRegular() {
-		return fmt.Errorf("path %q must be a regular file without symlinks", path)
-	}
-	file, err := os.Open(path)
-	if err != nil {
-		return err
-	}
-	defer func() { _ = file.Close() }()
-	opened, err := file.Stat()
-	if err != nil {
-		return err
-	}
-	if !opened.Mode().IsRegular() || !os.SameFile(inspected, opened) {
-		return fmt.Errorf("file %q changed before it could be read", path)
-	}
-	reader := io.LimitReader(file, limit+1)
-	data, err := io.ReadAll(reader)
-	if err != nil {
-		return err
-	}
-	if int64(len(data)) > limit {
-		return fmt.Errorf("file %q exceeds %d-byte limit", path, limit)
+		return fmt.Errorf("notarius JSON result exceeds or cannot be read within %d-byte limit: %w", limit, err)
 	}
 	if err := json.Unmarshal(data, destination); err != nil {
 		return err

internal/adapters/scriptorium/subprocess.go

@@ -12,6 +12,9 @@ import (
 	"gitea.maximumdirect.net/eric/narratio/internal/fileops"
 )
 
+// MaxOutputFileBytes bounds one Scriptorium artifact result.
+const MaxOutputFileBytes int64 = 64 * 1024 * 1024
+
 // SubprocessRunner invokes Scriptorium through its public CLI.
 type SubprocessRunner struct{}
 
@@ -326,14 +329,11 @@ func writeInvocationConfig(path string, payload invocationPayload) error {
 }
 
 func validateNonEmptyOutput(path string) error {
-	info, err := os.Stat(path)
+	data, err := fileops.ReadRegularFile(path, MaxOutputFileBytes)
 	if err != nil {
-		return fmt.Errorf("stat file: %w", err)
+		return fmt.Errorf("scriptorium artifact output exceeds or cannot be read within %d-byte limit: %w", MaxOutputFileBytes, err)
 	}
-	if info.IsDir() {
-		return fmt.Errorf("path is a directory")
-	}
-	if info.Size() <= 0 {
+	if len(data) == 0 {
 		return fmt.Errorf("file is empty")
 	}
 	return nil

internal/adapters/seriatim/subprocess.go

@@ -4,7 +4,6 @@ import (
 	"context"
 	"encoding/json"
 	"fmt"
-	"os"
 	"strconv"
 	"strings"
 	"time"
@@ -14,6 +13,9 @@ import (
 	"gitea.maximumdirect.net/eric/narratio/internal/fileops"
 )
 
+// MaxOutputFileBytes bounds each Seriatim JSON or rendered-text result.
+const MaxOutputFileBytes int64 = 64 * 1024 * 1024
+
 // EnvConfig defines optional Seriatim environment tuning values.
 type EnvConfig struct {
 	OverlapWordRunGap           *float64
@@ -636,9 +638,9 @@ func writeRenderInvocationConfig(req RenderRequest, args []string, binary string
 }
 
 func validateJSONFile(path string) error {
-	data, err := os.ReadFile(path)
+	data, err := readSeriatimResult(path, "JSON output")
 	if err != nil {
-		return fmt.Errorf("read file: %w", err)
+		return err
 	}
 	var v any
 	if err := json.Unmarshal(data, &v); err != nil {
@@ -648,9 +650,9 @@ func validateJSONFile(path string) error {
 }
 
 func validateJSONFileWithSegments(path string) error {
-	data, err := os.ReadFile(path)
+	data, err := readSeriatimResult(path, "transcript JSON output")
 	if err != nil {
-		return fmt.Errorf("read file: %w", err)
+		return err
 	}
 
 	var payload map[string]any
@@ -669,9 +671,9 @@ func validateJSONFileWithSegments(path string) error {
 }
 
 func validateNonEmptyTextFile(path string) error {
-	data, err := os.ReadFile(path)
+	data, err := readSeriatimResult(path, "rendered text output")
 	if err != nil {
-		return fmt.Errorf("read file: %w", err)
+		return err
 	}
 	if len(data) == 0 {
 		return fmt.Errorf("file is empty")
@@ -684,3 +686,11 @@ func validateNonEmptyTextFile(path string) error {
 	}
 	return nil
 }
+
+func readSeriatimResult(path, category string) ([]byte, error) {
+	data, err := fileops.ReadRegularFile(path, MaxOutputFileBytes)
+	if err != nil {
+		return nil, fmt.Errorf("seriatim %s exceeds or cannot be read within %d-byte limit: %w", category, MaxOutputFileBytes, err)
+	}
+	return data, nil
+}