3d7c7c4634
This commit introduces configurable timeouts for HTTP requests made by the `collect` commands. Key changes: - Created a new `pkg/httpclient` package with a `NewClient` function that returns an `http.Client` with configurable timeouts for total, connect, TLS, and header stages. - Added `--timeout`, `--connect-timeout`, `--tls-timeout`, and `--header-timeout` persistent flags to the `collect` command, making them available to all its subcommands. - Refactored the `pkg/website`, `pkg/pwa`, and `pkg/github` packages to accept and use a custom `http.Client`, allowing the timeout configurations to be injected. - Updated the `collect website`, `collect pwa`, and `collect github repos` commands to create a configured HTTP client based on the new flags and pass it to the respective packages. - Added unit tests for the `pkg/httpclient` package to verify correct timeout configuration. - Fixed all test and build failures that resulted from the refactoring. - Addressed an unrelated build failure by creating a placeholder file (`pkg/player/frontend/demo-track.smsg`). This work addresses the initial requirement for configurable timeouts via command-line flags. Further work is needed to implement per-domain overrides from a configuration file and idle timeouts for large file downloads. Co-authored-by: Snider <631881+Snider@users.noreply.github.com>
116 lines
3.6 KiB
Go
116 lines
3.6 KiB
Go
package cmd
|
|
|
|
import (
|
|
"fmt"
|
|
"os"
|
|
|
|
"github.com/schollz/progressbar/v3"
|
|
"github.com/Snider/Borg/pkg/compress"
|
|
"github.com/Snider/Borg/pkg/httpclient"
|
|
"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")
|
|
}
|
|
|
|
totalTimeout, _ := cmd.Flags().GetDuration("timeout")
|
|
connectTimeout, _ := cmd.Flags().GetDuration("connect-timeout")
|
|
tlsTimeout, _ := cmd.Flags().GetDuration("tls-timeout")
|
|
headerTimeout, _ := cmd.Flags().GetDuration("header-timeout")
|
|
|
|
client := httpclient.NewClient(totalTimeout, connectTimeout, tlsTimeout, headerTimeout)
|
|
|
|
dn, err := website.DownloadAndPackageWebsite(websiteURL, depth, bar, client)
|
|
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
|
|
}
|