Borg/cmd/collect_website.go
google-labs-jules[bot] 3398fabb14 feat: Add trix encryption and format
This commit introduces the `Enchantrix` library to add support for the `.trix` encrypted file format.

The main changes are:

- The `matrix` format has been renamed to `tim` (Terminal Isolation Matrix).
- The `.tim` format is now a specialized `.trix` file.
- A new `decode` command has been added to decode `.trix` and `.tim` files.
- The `collect` commands now support the `trix` and `tim` formats.
- A `--password` flag has been added to the `collect` commands for encryption.
- A `--i-am-in-isolation` flag has been added to the `decode` command for safely decoding `.tim` files.
- The decryption functionality is currently disabled due to a bug in the `Enchantrix` library. A follow-up PR will be created to re-enable it.
2025-11-14 13:47:27 +00:00

108 lines
3.2 KiB
Go

package cmd
import (
"fmt"
"os"
"github.com/schollz/progressbar/v3"
"github.com/Snider/Borg/pkg/compress"
"github.com/Snider/Borg/pkg/tim"
"github.com/Snider/Borg/pkg/trix"
"github.com/Snider/Borg/pkg/ui"
"github.com/Snider/Borg/pkg/website"
"github.com/spf13/cobra"
)
// collectWebsiteCmd represents the collect website command
var collectWebsiteCmd = NewCollectWebsiteCmd()
func init() {
GetCollectCmd().AddCommand(GetCollectWebsiteCmd())
}
func GetCollectWebsiteCmd() *cobra.Command {
return collectWebsiteCmd
}
func NewCollectWebsiteCmd() *cobra.Command {
collectWebsiteCmd := &cobra.Command{
Use: "website [url]",
Short: "Collect a single website",
Long: `Collect a single website and store it in a DataNode.`,
Args: cobra.ExactArgs(1),
RunE: func(cmd *cobra.Command, args []string) error {
websiteURL := args[0]
outputFile, _ := cmd.Flags().GetString("output")
depth, _ := cmd.Flags().GetInt("depth")
format, _ := cmd.Flags().GetString("format")
compression, _ := cmd.Flags().GetString("compression")
password, _ := cmd.Flags().GetString("password")
if format != "datanode" && format != "tim" && format != "trix" {
return fmt.Errorf("invalid format: %s (must be 'datanode', 'tim', or 'trix')", format)
}
prompter := ui.NewNonInteractivePrompter(ui.GetWebsiteQuote)
prompter.Start()
defer prompter.Stop()
var bar *progressbar.ProgressBar
if prompter.IsInteractive() {
bar = ui.NewProgressBar(-1, "Crawling website")
}
dn, err := website.DownloadAndPackageWebsite(websiteURL, depth, bar)
if err != nil {
return fmt.Errorf("error downloading and packaging website: %w", err)
}
var data []byte
if format == "tim" {
tim, err := tim.FromDataNode(dn)
if err != nil {
return fmt.Errorf("error creating tim: %w", err)
}
data, err = tim.ToTar()
if err != nil {
return fmt.Errorf("error serializing tim: %w", err)
}
} else if format == "trix" {
data, err = trix.ToTrix(dn, password)
if err != nil {
return fmt.Errorf("error serializing trix: %w", err)
}
} else {
data, err = dn.ToTar()
if err != nil {
return fmt.Errorf("error serializing DataNode: %w", err)
}
}
compressedData, err := compress.Compress(data, compression)
if err != nil {
return fmt.Errorf("error compressing data: %w", err)
}
if outputFile == "" {
outputFile = "website." + format
if compression != "none" {
outputFile += "." + compression
}
}
err = os.WriteFile(outputFile, compressedData, 0644)
if err != nil {
return fmt.Errorf("error writing website to file: %w", err)
}
fmt.Fprintln(cmd.OutOrStdout(), "Website saved to", outputFile)
return nil
},
}
collectWebsiteCmd.PersistentFlags().String("output", "", "Output file for the DataNode")
collectWebsiteCmd.PersistentFlags().Int("depth", 2, "Recursion depth for downloading")
collectWebsiteCmd.PersistentFlags().String("format", "datanode", "Output format (datanode, tim, or trix)")
collectWebsiteCmd.PersistentFlags().String("compression", "none", "Compression format (none, gz, or xz)")
collectWebsiteCmd.PersistentFlags().String("password", "", "Password for encryption")
return collectWebsiteCmd
}