DockerCLI/cli-plugins/plugin/plugin.go

178 lines
5.2 KiB
Go
Raw Normal View History

package plugin
import (
"encoding/json"
"fmt"
"os"
"sync"
"github.com/docker/cli/cli"
"github.com/docker/cli/cli-plugins/manager"
"github.com/docker/cli/cli/command"
"github.com/docker/cli/cli/connhelper"
"github.com/docker/docker/client"
"github.com/spf13/cobra"
)
// PersistentPreRunE must be called by any plugin command (or
// subcommand) which uses the cobra `PersistentPreRun*` hook. Plugins
// which do not make use of `PersistentPreRun*` do not need to call
// this (although it remains safe to do so). Plugins are recommended
// to use `PersistenPreRunE` to enable the error to be
// returned. Should not be called outside of a command's
// PersistentPreRunE hook and must not be run unless Run has been
// called.
var PersistentPreRunE func(*cobra.Command, []string) error
// RunPlugin executes the specified plugin command
func RunPlugin(dockerCli *command.DockerCli, plugin *cobra.Command, meta manager.Metadata) error {
allow plugins to have argument which match a top-level flag. The issue with plugin options clashing with globals is that when cobra is parsing the command line and it comes across an argument which doesn't start with a `-` it (in the absence of plugins) distinguishes between "argument to current command" and "new subcommand" based on the list of registered sub commands. Plugins breaks that model. When presented with `docker -D plugin -c foo` cobra parses up to the `plugin`, sees it isn't a registered sub-command of the top-level docker (because it isn't, it's a plugin) so it accumulates it as an argument to the top-level `docker` command. Then it sees the `-c`, and thinks it is the global `-c` (for AKA `--context`) option and tries to treat it as that, which fails. In the specific case of the top-level `docker` subcommand we know that it has no arguments which aren't `--flags` (or `-f` short flags) and so anything which doesn't start with a `-` must either be a (known) subcommand or an attempt to execute a plugin. We could simply scan for and register all installed plugins at start of day, so that cobra can do the right thing, but we want to avoid that since it would involve executing each plugin to fetch the metadata, even if the command wasn't going to end up hitting a plugin. Instead we can parse the initial set of global arguments separately before hitting the main cobra `Execute` path, which works here exactly because we know that the top-level has no non-flag arguments. One slight wrinkle is that the top-level `PersistentPreRunE` is no longer called on the plugins path (since it no longer goes via `Execute`), so we arrange for the initialisation done there (which has to be done after global flags are parsed to handle e.g. `--config`) to happen explictly after the global flags are parsed. Rather than make `newDockerCommand` return the complicated set of results needed to make this happen, instead return a closure which achieves this. The new functionality is introduced via a common `TopLevelCommand` abstraction which lets us adjust the plugin entrypoint to use the same strategy for parsing the global arguments. This isn't strictly required (in this case the stuff in cobra's `Execute` works fine) but doing it this way avoids the possibility of subtle differences in behaviour. Fixes #1699, and also, as a side-effect, the first item in #1661. Signed-off-by: Ian Campbell <ijc@docker.com>
2019-03-06 05:29:42 -05:00
tcmd := newPluginCommand(dockerCli, plugin, meta)
var persistentPreRunOnce sync.Once
PersistentPreRunE = func(_ *cobra.Command, _ []string) error {
var err error
persistentPreRunOnce.Do(func() {
var opts []command.InitializeOpt
if os.Getenv("DOCKER_CLI_PLUGIN_USE_DIAL_STDIO") != "" {
opts = append(opts, withPluginClientConn(plugin.Name()))
}
err = tcmd.Initialize(opts...)
})
return err
allow plugins to have argument which match a top-level flag. The issue with plugin options clashing with globals is that when cobra is parsing the command line and it comes across an argument which doesn't start with a `-` it (in the absence of plugins) distinguishes between "argument to current command" and "new subcommand" based on the list of registered sub commands. Plugins breaks that model. When presented with `docker -D plugin -c foo` cobra parses up to the `plugin`, sees it isn't a registered sub-command of the top-level docker (because it isn't, it's a plugin) so it accumulates it as an argument to the top-level `docker` command. Then it sees the `-c`, and thinks it is the global `-c` (for AKA `--context`) option and tries to treat it as that, which fails. In the specific case of the top-level `docker` subcommand we know that it has no arguments which aren't `--flags` (or `-f` short flags) and so anything which doesn't start with a `-` must either be a (known) subcommand or an attempt to execute a plugin. We could simply scan for and register all installed plugins at start of day, so that cobra can do the right thing, but we want to avoid that since it would involve executing each plugin to fetch the metadata, even if the command wasn't going to end up hitting a plugin. Instead we can parse the initial set of global arguments separately before hitting the main cobra `Execute` path, which works here exactly because we know that the top-level has no non-flag arguments. One slight wrinkle is that the top-level `PersistentPreRunE` is no longer called on the plugins path (since it no longer goes via `Execute`), so we arrange for the initialisation done there (which has to be done after global flags are parsed to handle e.g. `--config`) to happen explictly after the global flags are parsed. Rather than make `newDockerCommand` return the complicated set of results needed to make this happen, instead return a closure which achieves this. The new functionality is introduced via a common `TopLevelCommand` abstraction which lets us adjust the plugin entrypoint to use the same strategy for parsing the global arguments. This isn't strictly required (in this case the stuff in cobra's `Execute` works fine) but doing it this way avoids the possibility of subtle differences in behaviour. Fixes #1699, and also, as a side-effect, the first item in #1661. Signed-off-by: Ian Campbell <ijc@docker.com>
2019-03-06 05:29:42 -05:00
}
cmd, args, err := tcmd.HandleGlobalFlags()
allow plugins to have argument which match a top-level flag. The issue with plugin options clashing with globals is that when cobra is parsing the command line and it comes across an argument which doesn't start with a `-` it (in the absence of plugins) distinguishes between "argument to current command" and "new subcommand" based on the list of registered sub commands. Plugins breaks that model. When presented with `docker -D plugin -c foo` cobra parses up to the `plugin`, sees it isn't a registered sub-command of the top-level docker (because it isn't, it's a plugin) so it accumulates it as an argument to the top-level `docker` command. Then it sees the `-c`, and thinks it is the global `-c` (for AKA `--context`) option and tries to treat it as that, which fails. In the specific case of the top-level `docker` subcommand we know that it has no arguments which aren't `--flags` (or `-f` short flags) and so anything which doesn't start with a `-` must either be a (known) subcommand or an attempt to execute a plugin. We could simply scan for and register all installed plugins at start of day, so that cobra can do the right thing, but we want to avoid that since it would involve executing each plugin to fetch the metadata, even if the command wasn't going to end up hitting a plugin. Instead we can parse the initial set of global arguments separately before hitting the main cobra `Execute` path, which works here exactly because we know that the top-level has no non-flag arguments. One slight wrinkle is that the top-level `PersistentPreRunE` is no longer called on the plugins path (since it no longer goes via `Execute`), so we arrange for the initialisation done there (which has to be done after global flags are parsed to handle e.g. `--config`) to happen explictly after the global flags are parsed. Rather than make `newDockerCommand` return the complicated set of results needed to make this happen, instead return a closure which achieves this. The new functionality is introduced via a common `TopLevelCommand` abstraction which lets us adjust the plugin entrypoint to use the same strategy for parsing the global arguments. This isn't strictly required (in this case the stuff in cobra's `Execute` works fine) but doing it this way avoids the possibility of subtle differences in behaviour. Fixes #1699, and also, as a side-effect, the first item in #1661. Signed-off-by: Ian Campbell <ijc@docker.com>
2019-03-06 05:29:42 -05:00
if err != nil {
return err
}
// We've parsed global args already, so reset args to those
// which remain.
cmd.SetArgs(args)
allow plugins to have argument which match a top-level flag. The issue with plugin options clashing with globals is that when cobra is parsing the command line and it comes across an argument which doesn't start with a `-` it (in the absence of plugins) distinguishes between "argument to current command" and "new subcommand" based on the list of registered sub commands. Plugins breaks that model. When presented with `docker -D plugin -c foo` cobra parses up to the `plugin`, sees it isn't a registered sub-command of the top-level docker (because it isn't, it's a plugin) so it accumulates it as an argument to the top-level `docker` command. Then it sees the `-c`, and thinks it is the global `-c` (for AKA `--context`) option and tries to treat it as that, which fails. In the specific case of the top-level `docker` subcommand we know that it has no arguments which aren't `--flags` (or `-f` short flags) and so anything which doesn't start with a `-` must either be a (known) subcommand or an attempt to execute a plugin. We could simply scan for and register all installed plugins at start of day, so that cobra can do the right thing, but we want to avoid that since it would involve executing each plugin to fetch the metadata, even if the command wasn't going to end up hitting a plugin. Instead we can parse the initial set of global arguments separately before hitting the main cobra `Execute` path, which works here exactly because we know that the top-level has no non-flag arguments. One slight wrinkle is that the top-level `PersistentPreRunE` is no longer called on the plugins path (since it no longer goes via `Execute`), so we arrange for the initialisation done there (which has to be done after global flags are parsed to handle e.g. `--config`) to happen explictly after the global flags are parsed. Rather than make `newDockerCommand` return the complicated set of results needed to make this happen, instead return a closure which achieves this. The new functionality is introduced via a common `TopLevelCommand` abstraction which lets us adjust the plugin entrypoint to use the same strategy for parsing the global arguments. This isn't strictly required (in this case the stuff in cobra's `Execute` works fine) but doing it this way avoids the possibility of subtle differences in behaviour. Fixes #1699, and also, as a side-effect, the first item in #1661. Signed-off-by: Ian Campbell <ijc@docker.com>
2019-03-06 05:29:42 -05:00
return cmd.Execute()
}
// Run is the top-level entry point to the CLI plugin framework. It should be called from your plugin's `main()` function.
func Run(makeCmd func(command.Cli) *cobra.Command, meta manager.Metadata) {
dockerCli, err := command.NewDockerCli()
if err != nil {
fmt.Fprintln(os.Stderr, err)
os.Exit(1)
}
plugin := makeCmd(dockerCli)
if err := RunPlugin(dockerCli, plugin, meta); err != nil {
if sterr, ok := err.(cli.StatusError); ok {
if sterr.Status != "" {
fmt.Fprintln(dockerCli.Err(), sterr.Status)
}
// StatusError should only be used for errors, and all errors should
// have a non-zero exit status, so never exit with 0
if sterr.StatusCode == 0 {
os.Exit(1)
}
os.Exit(sterr.StatusCode)
}
fmt.Fprintln(dockerCli.Err(), err)
os.Exit(1)
}
}
func withPluginClientConn(name string) command.InitializeOpt {
return command.WithInitializeClient(func(dockerCli *command.DockerCli) (client.APIClient, error) {
cmd := "docker"
if x := os.Getenv(manager.ReexecEnvvar); x != "" {
cmd = x
}
var flags []string
// Accumulate all the global arguments, that is those
// up to (but not including) the plugin's name. This
// ensures that `docker system dial-stdio` is
// evaluating the same set of `--config`, `--tls*` etc
// global options as the plugin was called with, which
// in turn is the same as what the original docker
// invocation was passed.
for _, a := range os.Args[1:] {
if a == name {
break
}
flags = append(flags, a)
}
flags = append(flags, "system", "dial-stdio")
helper, err := connhelper.GetCommandConnectionHelper(cmd, flags...)
if err != nil {
return nil, err
}
return client.NewClientWithOpts(client.WithDialContext(helper.Dialer))
})
}
allow plugins to have argument which match a top-level flag. The issue with plugin options clashing with globals is that when cobra is parsing the command line and it comes across an argument which doesn't start with a `-` it (in the absence of plugins) distinguishes between "argument to current command" and "new subcommand" based on the list of registered sub commands. Plugins breaks that model. When presented with `docker -D plugin -c foo` cobra parses up to the `plugin`, sees it isn't a registered sub-command of the top-level docker (because it isn't, it's a plugin) so it accumulates it as an argument to the top-level `docker` command. Then it sees the `-c`, and thinks it is the global `-c` (for AKA `--context`) option and tries to treat it as that, which fails. In the specific case of the top-level `docker` subcommand we know that it has no arguments which aren't `--flags` (or `-f` short flags) and so anything which doesn't start with a `-` must either be a (known) subcommand or an attempt to execute a plugin. We could simply scan for and register all installed plugins at start of day, so that cobra can do the right thing, but we want to avoid that since it would involve executing each plugin to fetch the metadata, even if the command wasn't going to end up hitting a plugin. Instead we can parse the initial set of global arguments separately before hitting the main cobra `Execute` path, which works here exactly because we know that the top-level has no non-flag arguments. One slight wrinkle is that the top-level `PersistentPreRunE` is no longer called on the plugins path (since it no longer goes via `Execute`), so we arrange for the initialisation done there (which has to be done after global flags are parsed to handle e.g. `--config`) to happen explictly after the global flags are parsed. Rather than make `newDockerCommand` return the complicated set of results needed to make this happen, instead return a closure which achieves this. The new functionality is introduced via a common `TopLevelCommand` abstraction which lets us adjust the plugin entrypoint to use the same strategy for parsing the global arguments. This isn't strictly required (in this case the stuff in cobra's `Execute` works fine) but doing it this way avoids the possibility of subtle differences in behaviour. Fixes #1699, and also, as a side-effect, the first item in #1661. Signed-off-by: Ian Campbell <ijc@docker.com>
2019-03-06 05:29:42 -05:00
func newPluginCommand(dockerCli *command.DockerCli, plugin *cobra.Command, meta manager.Metadata) *cli.TopLevelCommand {
name := plugin.Name()
fullname := manager.NamePrefix + name
cmd := &cobra.Command{
Use: fmt.Sprintf("docker [OPTIONS] %s [ARG...]", name),
Short: fullname + " is a Docker CLI plugin",
SilenceUsage: true,
SilenceErrors: true,
PersistentPreRunE: func(cmd *cobra.Command, args []string) error {
// We can't use this as the hook directly since it is initialised later (in runPlugin)
return PersistentPreRunE(cmd, args)
},
TraverseChildren: true,
DisableFlagsInUseLine: true,
CompletionOptions: cobra.CompletionOptions{
DisableDefaultCmd: false,
HiddenDefaultCmd: true,
DisableDescriptions: true,
},
}
opts, flags := cli.SetupPluginRootCommand(cmd)
cmd.SetIn(dockerCli.In())
cmd.SetOut(dockerCli.Out())
cmd.SetErr(dockerCli.Err())
cmd.AddCommand(
plugin,
newMetadataSubcommand(plugin, meta),
)
cli.DisableFlagsInUseLine(cmd)
allow plugins to have argument which match a top-level flag. The issue with plugin options clashing with globals is that when cobra is parsing the command line and it comes across an argument which doesn't start with a `-` it (in the absence of plugins) distinguishes between "argument to current command" and "new subcommand" based on the list of registered sub commands. Plugins breaks that model. When presented with `docker -D plugin -c foo` cobra parses up to the `plugin`, sees it isn't a registered sub-command of the top-level docker (because it isn't, it's a plugin) so it accumulates it as an argument to the top-level `docker` command. Then it sees the `-c`, and thinks it is the global `-c` (for AKA `--context`) option and tries to treat it as that, which fails. In the specific case of the top-level `docker` subcommand we know that it has no arguments which aren't `--flags` (or `-f` short flags) and so anything which doesn't start with a `-` must either be a (known) subcommand or an attempt to execute a plugin. We could simply scan for and register all installed plugins at start of day, so that cobra can do the right thing, but we want to avoid that since it would involve executing each plugin to fetch the metadata, even if the command wasn't going to end up hitting a plugin. Instead we can parse the initial set of global arguments separately before hitting the main cobra `Execute` path, which works here exactly because we know that the top-level has no non-flag arguments. One slight wrinkle is that the top-level `PersistentPreRunE` is no longer called on the plugins path (since it no longer goes via `Execute`), so we arrange for the initialisation done there (which has to be done after global flags are parsed to handle e.g. `--config`) to happen explictly after the global flags are parsed. Rather than make `newDockerCommand` return the complicated set of results needed to make this happen, instead return a closure which achieves this. The new functionality is introduced via a common `TopLevelCommand` abstraction which lets us adjust the plugin entrypoint to use the same strategy for parsing the global arguments. This isn't strictly required (in this case the stuff in cobra's `Execute` works fine) but doing it this way avoids the possibility of subtle differences in behaviour. Fixes #1699, and also, as a side-effect, the first item in #1661. Signed-off-by: Ian Campbell <ijc@docker.com>
2019-03-06 05:29:42 -05:00
return cli.NewTopLevelCommand(cmd, dockerCli, opts, flags)
}
func newMetadataSubcommand(plugin *cobra.Command, meta manager.Metadata) *cobra.Command {
if meta.ShortDescription == "" {
meta.ShortDescription = plugin.Short
}
cmd := &cobra.Command{
Use: manager.MetadataSubcommandName,
Hidden: true,
// Suppress the global/parent PersistentPreRunE, which
// needlessly initializes the client and tries to
// connect to the daemon.
PersistentPreRun: func(cmd *cobra.Command, args []string) {},
RunE: func(cmd *cobra.Command, args []string) error {
enc := json.NewEncoder(os.Stdout)
enc.SetEscapeHTML(false)
enc.SetIndent("", " ")
return enc.Encode(meta)
},
}
return cmd
}
// RunningStandalone tells a CLI plugin it is run standalone by direct execution
func RunningStandalone() bool {
if os.Getenv(manager.ReexecEnvvar) != "" {
return false
}
return len(os.Args) < 2 || os.Args[1] != manager.MetadataSubcommandName
}