Documentation ¶
Index ¶
- Variables
- func Arg(v string) string
- func DefaultTraceAttributes(e ExecutedCommand) []attribute.KeyValue
- func ExitCode(err error) int
- func LogCommands(ctx context.Context, log LogFunc) context.Context
- func TraceCommands(ctx context.Context, attrs TraceAttributesFunc) context.Context
- type BashOpt
- type Command
- func (c *Command) Dir(dir string) *Command
- func (c *Command) Env(env map[string]string) *Command
- func (c *Command) Environ(environ []string) *Command
- func (c *Command) Input(input io.Reader) *Command
- func (c *Command) ResetInput() *Command
- func (c *Command) Run() Output
- func (c *Command) StdErr() *Command
- func (c *Command) StdOut() *Command
- type ExecutedCommand
- type ExitCoder
- type LineMap
- type LogFunc
- type Output
- type TraceAttributesFunc
Examples ¶
Constants ¶
This section is empty.
Variables ¶
var StrictBashOpts = []BashOpt{BashOptPipeFail, BashOptErrExit}
StrictBashOpts contains options that effectively enforce safe execution of bash commands.
Functions ¶
func Arg ¶ added in v0.8.0
Arg quotes a value such that it gets treated as an argument by a command.
It is currently an alias for shell.Quote
func DefaultTraceAttributes ¶ added in v0.10.0
func DefaultTraceAttributes(e ExecutedCommand) []attribute.KeyValue
DefaultTraceAttributes adds Args and Dir as attributes. Note that Args may contain sensitive data.
func ExitCode ¶ added in v0.2.0
ExitCode returns the exit code associated with err if there is one, otherwise 1. If err is nil, returns 0.
In practice, this replicates the behaviour observed when running commands in the shell, running a command with an incorrect syntax for example will set $? to 1, which is what is expected in script. Not implementing this creates a confusing case where an error such as not finding the binary would either force the code to account for the absence of exit code, which defeats the purpose of this library which is to provide a convenient replacement for shell scripting.
Example ¶
package main import ( "context" "fmt" "github.com/sourcegraph/run" ) func main() { ctx := context.Background() err := run.Bash(ctx, "exit 123").Run().Wait() fmt.Println(run.ExitCode(err)) err = run.Cmd(ctx, "echo", run.Arg("hello world!")).Run().Wait() fmt.Println(run.ExitCode(err)) err = run.Cmd(ctx, "non-existing-binary").Run().Wait() fmt.Println(run.ExitCode(err)) }
Output: 123 0 1
func LogCommands ¶ added in v0.10.0
LogCommands enables logging on all commands executed by sourcegraph/run within this context. Set to nil to disable (default), or use run.DefaultTraceAttributes for some recommended defaults.
Note that arguments and environments may contain sensitive information.
If you use loggers carrying contexts, e.g. via sourcegraph/log, it is recommended that you only enable this within relevant scopes.
func TraceCommands ¶ added in v0.10.0
func TraceCommands(ctx context.Context, attrs TraceAttributesFunc) context.Context
TraceCommands toggles OpenTelemetry tracing on all usages of sourcegraph/run within this context. Set to nil to disable (default).
Note that arguments and environments may contain sensitive information.
Types ¶
type BashOpt ¶ added in v0.11.0
type BashOpt string
BashOpt denotes options for running bash commands. For more options see 'man bash'
type Command ¶
type Command struct {
// contains filtered or unexported fields
}
Command builds a command for execution. Functions modify the underlying command.
func Bash ¶ added in v0.5.0
Bash joins all the parts and builds a command from it to be run by 'bash -c'.
Arguments are not implicitly quoted - to quote arguemnts, you can use Arg.
func BashWith ¶ added in v0.11.0
BashWith appends all the given bash options to the bash command with '-o'. The given parts is then joined together to be executed with 'bash -c'
The final command will have the following format: bash -o option-1 -c command. For recommended strict bash options see StrictBashOpts, which has 'pipefail' and 'errexit' options
func Cmd ¶
Cmd joins all the parts and builds a command from it.
Arguments are not implicitly quoted - to quote arguments, you can use Arg.
func (*Command) Environ ¶
Environ adds the given strings representing the environment (key=value) to the command, for example os.Environ().
func (*Command) Input ¶
Input pipes the given io.Reader to the command. If an input is already set, the given input is appended.
func (*Command) ResetInput ¶ added in v0.2.0
ResetInput sets the command's input to nil.
func (*Command) Run ¶
Run starts command execution and returns Output, which defaults to combined output.
type ExecutedCommand ¶ added in v0.10.0
ExecutedCommand represents a command that has been started.
type ExitCoder ¶ added in v0.2.0
ExitCoder is an error that also denotes an exit code to exit with. Users of Output can check if an error implements this interface to get the underlying exit code of a command execution.
For convenience, the ExitCode function can be used to get the code.
type LineMap ¶ added in v0.6.0
LineMap allows modifications of individual lines from Output and enables callbacks that operate on lines from Output. Bytes written to dst are collected and passed to subsequent LineMaps before being written to output aggregation, e.g. Output.Stream().
The return value mirrors the signature of (Writer).Write(), and should be used to indicate what was written to dst.
Errors interrupt line processing and are returned if and only if the command itself did not exit with an error.
type LogFunc ¶ added in v0.10.0
type LogFunc func(ExecutedCommand)
LogFunc can be used to generate a log entry for the executed command.
type Output ¶
type Output interface { // Map adds a LineMap function to be applied to this Output. // // Deprecated: Use Pipeline instead. Map(f LineMap) Output // Pipeline is similar to Map, but uses a new interface that provides more flexible // ways of manipulating output on the stream. It is only applied at aggregation time // using e.g. Stream, Lines, and so on. Multiple Pipelines are applied sequentially, // with the result of previous Pipelines propagated to subsequent Pipelines. // // For more details, refer to the pipeline.Pipeline documentation. Pipeline(p pipeline.Pipeline) Output // Stream writes mapped output from the command to the destination writer until // command completion. Stream(dst io.Writer) error // StreamLines writes mapped output from the command and sends it line by line to the // destination callback until command completion. StreamLines(dst func(line string)) error // Lines waits for command completion and aggregates mapped output from the command as // a slice of lines. Lines() ([]string, error) // Lines waits for command completion and aggregates mapped output from the command as // a combined string. String() (string, error) // JQ waits for command completion executes a JQ query against the entire output. // // Refer to https://github.com/itchyny/gojq for the specifics of supported syntax. JQ(query string) ([]byte, error) // Reader is implemented so that Output can be provided directly to another Command // using Input(). io.Reader // WriterTo is implemented for convenience when chaining commands in LineMap. io.WriterTo // Wait waits for command completion and returns. Wait() error }
Output configures output and aggregation from a command.
It is behind an interface to more easily enable mock outputs and build different types of outputs, such as multi-outputs and error-only outputs, without complicating the core commandOutput implementation.
func NewErrorOutput ¶ added in v0.4.0
NewErrorOutput creates an Output that just returns error. Useful for allowing function that help run Commands and want to just return an Output even if errors can happen before command execution.
type TraceAttributesFunc ¶ added in v0.10.0
type TraceAttributesFunc func(ExecutedCommand) []attribute.KeyValue
TraceAttributesFunc can be used to generate attributes to attach to a span for the executed command.