Bound external result file reads

This commit is contained in:
2026-08-10 18:30:28 +00:00
parent 99b2e1cd81
commit ab5a7e8e3d
22 changed files with 215 additions and 84 deletions

View File

@@ -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
}