forgejo/modules/git/command.go

422 lines
13 KiB
Go
Raw Normal View History

2016-11-04 07:16:01 +09:00
// Copyright 2015 The Gogs Authors. All rights reserved.
// Copyright 2016 The Gitea Authors. All rights reserved.
// SPDX-License-Identifier: MIT
2016-11-04 07:16:01 +09:00
package git
import (
"bytes"
"context"
"errors"
2016-11-04 07:16:01 +09:00
"fmt"
"io"
"os"
2016-11-04 07:16:01 +09:00
"os/exec"
"strings"
"time"
"unsafe"
Refactor git command package to improve security and maintainability (#22678) This PR follows #21535 (and replace #22592) ## Review without space diff https://github.com/go-gitea/gitea/pull/22678/files?diff=split&w=1 ## Purpose of this PR 1. Make git module command completely safe (risky user inputs won't be passed as argument option anymore) 2. Avoid low-level mistakes like https://github.com/go-gitea/gitea/pull/22098#discussion_r1045234918 3. Remove deprecated and dirty `CmdArgCheck` function, hide the `CmdArg` type 4. Simplify code when using git command ## The main idea of this PR * Move the `git.CmdArg` to the `internal` package, then no other package except `git` could use it. Then developers could never do `AddArguments(git.CmdArg(userInput))` any more. * Introduce `git.ToTrustedCmdArgs`, it's for user-provided and already trusted arguments. It's only used in a few cases, for example: use git arguments from config file, help unit test with some arguments. * Introduce `AddOptionValues` and `AddOptionFormat`, they make code more clear and simple: * Before: `AddArguments("-m").AddDynamicArguments(message)` * After: `AddOptionValues("-m", message)` * - * Before: `AddArguments(git.CmdArg(fmt.Sprintf("--author='%s <%s>'", sig.Name, sig.Email)))` * After: `AddOptionFormat("--author='%s <%s>'", sig.Name, sig.Email)` ## FAQ ### Why these changes were not done in #21535 ? #21535 is mainly a search&replace, it did its best to not change too much logic. Making the framework better needs a lot of changes, so this separate PR is needed as the second step. ### The naming of `AddOptionXxx` According to git's manual, the `--xxx` part is called `option`. ### How can it guarantee that `internal.CmdArg` won't be not misused? Go's specification guarantees that. Trying to access other package's internal package causes compilation error. And, `golangci-lint` also denies the git/internal package. Only the `git/command.go` can use it carefully. ### There is still a `ToTrustedCmdArgs`, will it still allow developers to make mistakes and pass untrusted arguments? Generally speaking, no. Because when using `ToTrustedCmdArgs`, the code will be very complex (see the changes for examples). Then developers and reviewers can know that something might be unreasonable. ### Why there was a `CmdArgCheck` and why it's removed? At the moment of #21535, to reduce unnecessary changes, `CmdArgCheck` was introduced as a hacky patch. Now, almost all code could be written as `cmd := NewCommand(); cmd.AddXxx(...)`, then there is no need for `CmdArgCheck` anymore. ### Why many codes for `signArg == ""` is deleted? Because in the old code, `signArg` could never be empty string, it's either `-S[key-id]` or `--no-gpg-sign`. So the `signArg == ""` is just dead code. --------- Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
2023-02-04 11:30:43 +09:00
"code.gitea.io/gitea/modules/git/internal" //nolint:depguard // only this file can use the internal type CmdArg, other files and packages should use AddXxx functions
"code.gitea.io/gitea/modules/log"
"code.gitea.io/gitea/modules/process"
"code.gitea.io/gitea/modules/util"
2016-11-04 07:16:01 +09:00
)
Refactor git command package to improve security and maintainability (#22678) This PR follows #21535 (and replace #22592) ## Review without space diff https://github.com/go-gitea/gitea/pull/22678/files?diff=split&w=1 ## Purpose of this PR 1. Make git module command completely safe (risky user inputs won't be passed as argument option anymore) 2. Avoid low-level mistakes like https://github.com/go-gitea/gitea/pull/22098#discussion_r1045234918 3. Remove deprecated and dirty `CmdArgCheck` function, hide the `CmdArg` type 4. Simplify code when using git command ## The main idea of this PR * Move the `git.CmdArg` to the `internal` package, then no other package except `git` could use it. Then developers could never do `AddArguments(git.CmdArg(userInput))` any more. * Introduce `git.ToTrustedCmdArgs`, it's for user-provided and already trusted arguments. It's only used in a few cases, for example: use git arguments from config file, help unit test with some arguments. * Introduce `AddOptionValues` and `AddOptionFormat`, they make code more clear and simple: * Before: `AddArguments("-m").AddDynamicArguments(message)` * After: `AddOptionValues("-m", message)` * - * Before: `AddArguments(git.CmdArg(fmt.Sprintf("--author='%s <%s>'", sig.Name, sig.Email)))` * After: `AddOptionFormat("--author='%s <%s>'", sig.Name, sig.Email)` ## FAQ ### Why these changes were not done in #21535 ? #21535 is mainly a search&replace, it did its best to not change too much logic. Making the framework better needs a lot of changes, so this separate PR is needed as the second step. ### The naming of `AddOptionXxx` According to git's manual, the `--xxx` part is called `option`. ### How can it guarantee that `internal.CmdArg` won't be not misused? Go's specification guarantees that. Trying to access other package's internal package causes compilation error. And, `golangci-lint` also denies the git/internal package. Only the `git/command.go` can use it carefully. ### There is still a `ToTrustedCmdArgs`, will it still allow developers to make mistakes and pass untrusted arguments? Generally speaking, no. Because when using `ToTrustedCmdArgs`, the code will be very complex (see the changes for examples). Then developers and reviewers can know that something might be unreasonable. ### Why there was a `CmdArgCheck` and why it's removed? At the moment of #21535, to reduce unnecessary changes, `CmdArgCheck` was introduced as a hacky patch. Now, almost all code could be written as `cmd := NewCommand(); cmd.AddXxx(...)`, then there is no need for `CmdArgCheck` anymore. ### Why many codes for `signArg == ""` is deleted? Because in the old code, `signArg` could never be empty string, it's either `-S[key-id]` or `--no-gpg-sign`. So the `signArg == ""` is just dead code. --------- Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
2023-02-04 11:30:43 +09:00
// TrustedCmdArgs returns the trusted arguments for git command.
// It's mainly for passing user-provided and trusted arguments to git command
// In most cases, it shouldn't be used. Use AddXxx function instead
type TrustedCmdArgs []internal.CmdArg
2016-12-22 18:30:52 +09:00
var (
// globalCommandArgs global command args for external package setting
Refactor git command package to improve security and maintainability (#22678) This PR follows #21535 (and replace #22592) ## Review without space diff https://github.com/go-gitea/gitea/pull/22678/files?diff=split&w=1 ## Purpose of this PR 1. Make git module command completely safe (risky user inputs won't be passed as argument option anymore) 2. Avoid low-level mistakes like https://github.com/go-gitea/gitea/pull/22098#discussion_r1045234918 3. Remove deprecated and dirty `CmdArgCheck` function, hide the `CmdArg` type 4. Simplify code when using git command ## The main idea of this PR * Move the `git.CmdArg` to the `internal` package, then no other package except `git` could use it. Then developers could never do `AddArguments(git.CmdArg(userInput))` any more. * Introduce `git.ToTrustedCmdArgs`, it's for user-provided and already trusted arguments. It's only used in a few cases, for example: use git arguments from config file, help unit test with some arguments. * Introduce `AddOptionValues` and `AddOptionFormat`, they make code more clear and simple: * Before: `AddArguments("-m").AddDynamicArguments(message)` * After: `AddOptionValues("-m", message)` * - * Before: `AddArguments(git.CmdArg(fmt.Sprintf("--author='%s <%s>'", sig.Name, sig.Email)))` * After: `AddOptionFormat("--author='%s <%s>'", sig.Name, sig.Email)` ## FAQ ### Why these changes were not done in #21535 ? #21535 is mainly a search&replace, it did its best to not change too much logic. Making the framework better needs a lot of changes, so this separate PR is needed as the second step. ### The naming of `AddOptionXxx` According to git's manual, the `--xxx` part is called `option`. ### How can it guarantee that `internal.CmdArg` won't be not misused? Go's specification guarantees that. Trying to access other package's internal package causes compilation error. And, `golangci-lint` also denies the git/internal package. Only the `git/command.go` can use it carefully. ### There is still a `ToTrustedCmdArgs`, will it still allow developers to make mistakes and pass untrusted arguments? Generally speaking, no. Because when using `ToTrustedCmdArgs`, the code will be very complex (see the changes for examples). Then developers and reviewers can know that something might be unreasonable. ### Why there was a `CmdArgCheck` and why it's removed? At the moment of #21535, to reduce unnecessary changes, `CmdArgCheck` was introduced as a hacky patch. Now, almost all code could be written as `cmd := NewCommand(); cmd.AddXxx(...)`, then there is no need for `CmdArgCheck` anymore. ### Why many codes for `signArg == ""` is deleted? Because in the old code, `signArg` could never be empty string, it's either `-S[key-id]` or `--no-gpg-sign`. So the `signArg == ""` is just dead code. --------- Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
2023-02-04 11:30:43 +09:00
globalCommandArgs TrustedCmdArgs
// defaultCommandExecutionTimeout default command execution timeout duration
defaultCommandExecutionTimeout = 360 * time.Second
2016-12-22 18:30:52 +09:00
)
// DefaultLocale is the default LC_ALL to run git commands in.
const DefaultLocale = "C"
2016-11-04 07:16:01 +09:00
// Command represents a command with its subcommands or arguments.
type Command struct {
name string
args []string
parentContext context.Context
desc string
globalArgsLength int
brokenArgs []string
2016-11-04 07:16:01 +09:00
}
func (c *Command) String() string {
if len(c.args) == 0 {
return c.name
}
return fmt.Sprintf("%s %s", c.name, strings.Join(c.args, " "))
}
// NewCommand creates and returns a new Git Command based on given command and arguments.
// Each argument should be safe to be trusted. User-provided arguments should be passed to AddDynamicArguments instead.
Refactor git command package to improve security and maintainability (#22678) This PR follows #21535 (and replace #22592) ## Review without space diff https://github.com/go-gitea/gitea/pull/22678/files?diff=split&w=1 ## Purpose of this PR 1. Make git module command completely safe (risky user inputs won't be passed as argument option anymore) 2. Avoid low-level mistakes like https://github.com/go-gitea/gitea/pull/22098#discussion_r1045234918 3. Remove deprecated and dirty `CmdArgCheck` function, hide the `CmdArg` type 4. Simplify code when using git command ## The main idea of this PR * Move the `git.CmdArg` to the `internal` package, then no other package except `git` could use it. Then developers could never do `AddArguments(git.CmdArg(userInput))` any more. * Introduce `git.ToTrustedCmdArgs`, it's for user-provided and already trusted arguments. It's only used in a few cases, for example: use git arguments from config file, help unit test with some arguments. * Introduce `AddOptionValues` and `AddOptionFormat`, they make code more clear and simple: * Before: `AddArguments("-m").AddDynamicArguments(message)` * After: `AddOptionValues("-m", message)` * - * Before: `AddArguments(git.CmdArg(fmt.Sprintf("--author='%s <%s>'", sig.Name, sig.Email)))` * After: `AddOptionFormat("--author='%s <%s>'", sig.Name, sig.Email)` ## FAQ ### Why these changes were not done in #21535 ? #21535 is mainly a search&replace, it did its best to not change too much logic. Making the framework better needs a lot of changes, so this separate PR is needed as the second step. ### The naming of `AddOptionXxx` According to git's manual, the `--xxx` part is called `option`. ### How can it guarantee that `internal.CmdArg` won't be not misused? Go's specification guarantees that. Trying to access other package's internal package causes compilation error. And, `golangci-lint` also denies the git/internal package. Only the `git/command.go` can use it carefully. ### There is still a `ToTrustedCmdArgs`, will it still allow developers to make mistakes and pass untrusted arguments? Generally speaking, no. Because when using `ToTrustedCmdArgs`, the code will be very complex (see the changes for examples). Then developers and reviewers can know that something might be unreasonable. ### Why there was a `CmdArgCheck` and why it's removed? At the moment of #21535, to reduce unnecessary changes, `CmdArgCheck` was introduced as a hacky patch. Now, almost all code could be written as `cmd := NewCommand(); cmd.AddXxx(...)`, then there is no need for `CmdArgCheck` anymore. ### Why many codes for `signArg == ""` is deleted? Because in the old code, `signArg` could never be empty string, it's either `-S[key-id]` or `--no-gpg-sign`. So the `signArg == ""` is just dead code. --------- Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
2023-02-04 11:30:43 +09:00
func NewCommand(ctx context.Context, args ...internal.CmdArg) *Command {
// Make an explicit copy of globalCommandArgs, otherwise append might overwrite it
cargs := make([]string, 0, len(globalCommandArgs)+len(args))
for _, arg := range globalCommandArgs {
cargs = append(cargs, string(arg))
}
for _, arg := range args {
cargs = append(cargs, string(arg))
}
2016-11-04 07:16:01 +09:00
return &Command{
name: GitExecutable,
args: cargs,
parentContext: ctx,
globalArgsLength: len(globalCommandArgs),
2016-11-04 07:16:01 +09:00
}
}
// NewCommandContextNoGlobals creates and returns a new Git Command based on given command and arguments only with the specify args and don't care global command args
// Each argument should be safe to be trusted. User-provided arguments should be passed to AddDynamicArguments instead.
Refactor git command package to improve security and maintainability (#22678) This PR follows #21535 (and replace #22592) ## Review without space diff https://github.com/go-gitea/gitea/pull/22678/files?diff=split&w=1 ## Purpose of this PR 1. Make git module command completely safe (risky user inputs won't be passed as argument option anymore) 2. Avoid low-level mistakes like https://github.com/go-gitea/gitea/pull/22098#discussion_r1045234918 3. Remove deprecated and dirty `CmdArgCheck` function, hide the `CmdArg` type 4. Simplify code when using git command ## The main idea of this PR * Move the `git.CmdArg` to the `internal` package, then no other package except `git` could use it. Then developers could never do `AddArguments(git.CmdArg(userInput))` any more. * Introduce `git.ToTrustedCmdArgs`, it's for user-provided and already trusted arguments. It's only used in a few cases, for example: use git arguments from config file, help unit test with some arguments. * Introduce `AddOptionValues` and `AddOptionFormat`, they make code more clear and simple: * Before: `AddArguments("-m").AddDynamicArguments(message)` * After: `AddOptionValues("-m", message)` * - * Before: `AddArguments(git.CmdArg(fmt.Sprintf("--author='%s <%s>'", sig.Name, sig.Email)))` * After: `AddOptionFormat("--author='%s <%s>'", sig.Name, sig.Email)` ## FAQ ### Why these changes were not done in #21535 ? #21535 is mainly a search&replace, it did its best to not change too much logic. Making the framework better needs a lot of changes, so this separate PR is needed as the second step. ### The naming of `AddOptionXxx` According to git's manual, the `--xxx` part is called `option`. ### How can it guarantee that `internal.CmdArg` won't be not misused? Go's specification guarantees that. Trying to access other package's internal package causes compilation error. And, `golangci-lint` also denies the git/internal package. Only the `git/command.go` can use it carefully. ### There is still a `ToTrustedCmdArgs`, will it still allow developers to make mistakes and pass untrusted arguments? Generally speaking, no. Because when using `ToTrustedCmdArgs`, the code will be very complex (see the changes for examples). Then developers and reviewers can know that something might be unreasonable. ### Why there was a `CmdArgCheck` and why it's removed? At the moment of #21535, to reduce unnecessary changes, `CmdArgCheck` was introduced as a hacky patch. Now, almost all code could be written as `cmd := NewCommand(); cmd.AddXxx(...)`, then there is no need for `CmdArgCheck` anymore. ### Why many codes for `signArg == ""` is deleted? Because in the old code, `signArg` could never be empty string, it's either `-S[key-id]` or `--no-gpg-sign`. So the `signArg == ""` is just dead code. --------- Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
2023-02-04 11:30:43 +09:00
func NewCommandContextNoGlobals(ctx context.Context, args ...internal.CmdArg) *Command {
cargs := make([]string, 0, len(args))
for _, arg := range args {
cargs = append(cargs, string(arg))
}
return &Command{
name: GitExecutable,
args: cargs,
parentContext: ctx,
}
}
// SetParentContext sets the parent context for this command
func (c *Command) SetParentContext(ctx context.Context) *Command {
c.parentContext = ctx
return c
}
Refactor git command package to improve security and maintainability (#22678) This PR follows #21535 (and replace #22592) ## Review without space diff https://github.com/go-gitea/gitea/pull/22678/files?diff=split&w=1 ## Purpose of this PR 1. Make git module command completely safe (risky user inputs won't be passed as argument option anymore) 2. Avoid low-level mistakes like https://github.com/go-gitea/gitea/pull/22098#discussion_r1045234918 3. Remove deprecated and dirty `CmdArgCheck` function, hide the `CmdArg` type 4. Simplify code when using git command ## The main idea of this PR * Move the `git.CmdArg` to the `internal` package, then no other package except `git` could use it. Then developers could never do `AddArguments(git.CmdArg(userInput))` any more. * Introduce `git.ToTrustedCmdArgs`, it's for user-provided and already trusted arguments. It's only used in a few cases, for example: use git arguments from config file, help unit test with some arguments. * Introduce `AddOptionValues` and `AddOptionFormat`, they make code more clear and simple: * Before: `AddArguments("-m").AddDynamicArguments(message)` * After: `AddOptionValues("-m", message)` * - * Before: `AddArguments(git.CmdArg(fmt.Sprintf("--author='%s <%s>'", sig.Name, sig.Email)))` * After: `AddOptionFormat("--author='%s <%s>'", sig.Name, sig.Email)` ## FAQ ### Why these changes were not done in #21535 ? #21535 is mainly a search&replace, it did its best to not change too much logic. Making the framework better needs a lot of changes, so this separate PR is needed as the second step. ### The naming of `AddOptionXxx` According to git's manual, the `--xxx` part is called `option`. ### How can it guarantee that `internal.CmdArg` won't be not misused? Go's specification guarantees that. Trying to access other package's internal package causes compilation error. And, `golangci-lint` also denies the git/internal package. Only the `git/command.go` can use it carefully. ### There is still a `ToTrustedCmdArgs`, will it still allow developers to make mistakes and pass untrusted arguments? Generally speaking, no. Because when using `ToTrustedCmdArgs`, the code will be very complex (see the changes for examples). Then developers and reviewers can know that something might be unreasonable. ### Why there was a `CmdArgCheck` and why it's removed? At the moment of #21535, to reduce unnecessary changes, `CmdArgCheck` was introduced as a hacky patch. Now, almost all code could be written as `cmd := NewCommand(); cmd.AddXxx(...)`, then there is no need for `CmdArgCheck` anymore. ### Why many codes for `signArg == ""` is deleted? Because in the old code, `signArg` could never be empty string, it's either `-S[key-id]` or `--no-gpg-sign`. So the `signArg == ""` is just dead code. --------- Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
2023-02-04 11:30:43 +09:00
// SetDescription sets the description for this command which be returned on c.String()
func (c *Command) SetDescription(desc string) *Command {
c.desc = desc
return c
}
Refactor git command package to improve security and maintainability (#22678) This PR follows #21535 (and replace #22592) ## Review without space diff https://github.com/go-gitea/gitea/pull/22678/files?diff=split&w=1 ## Purpose of this PR 1. Make git module command completely safe (risky user inputs won't be passed as argument option anymore) 2. Avoid low-level mistakes like https://github.com/go-gitea/gitea/pull/22098#discussion_r1045234918 3. Remove deprecated and dirty `CmdArgCheck` function, hide the `CmdArg` type 4. Simplify code when using git command ## The main idea of this PR * Move the `git.CmdArg` to the `internal` package, then no other package except `git` could use it. Then developers could never do `AddArguments(git.CmdArg(userInput))` any more. * Introduce `git.ToTrustedCmdArgs`, it's for user-provided and already trusted arguments. It's only used in a few cases, for example: use git arguments from config file, help unit test with some arguments. * Introduce `AddOptionValues` and `AddOptionFormat`, they make code more clear and simple: * Before: `AddArguments("-m").AddDynamicArguments(message)` * After: `AddOptionValues("-m", message)` * - * Before: `AddArguments(git.CmdArg(fmt.Sprintf("--author='%s <%s>'", sig.Name, sig.Email)))` * After: `AddOptionFormat("--author='%s <%s>'", sig.Name, sig.Email)` ## FAQ ### Why these changes were not done in #21535 ? #21535 is mainly a search&replace, it did its best to not change too much logic. Making the framework better needs a lot of changes, so this separate PR is needed as the second step. ### The naming of `AddOptionXxx` According to git's manual, the `--xxx` part is called `option`. ### How can it guarantee that `internal.CmdArg` won't be not misused? Go's specification guarantees that. Trying to access other package's internal package causes compilation error. And, `golangci-lint` also denies the git/internal package. Only the `git/command.go` can use it carefully. ### There is still a `ToTrustedCmdArgs`, will it still allow developers to make mistakes and pass untrusted arguments? Generally speaking, no. Because when using `ToTrustedCmdArgs`, the code will be very complex (see the changes for examples). Then developers and reviewers can know that something might be unreasonable. ### Why there was a `CmdArgCheck` and why it's removed? At the moment of #21535, to reduce unnecessary changes, `CmdArgCheck` was introduced as a hacky patch. Now, almost all code could be written as `cmd := NewCommand(); cmd.AddXxx(...)`, then there is no need for `CmdArgCheck` anymore. ### Why many codes for `signArg == ""` is deleted? Because in the old code, `signArg` could never be empty string, it's either `-S[key-id]` or `--no-gpg-sign`. So the `signArg == ""` is just dead code. --------- Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
2023-02-04 11:30:43 +09:00
// isSafeArgumentValue checks if the argument is safe to be used as a value (not an option)
func isSafeArgumentValue(s string) bool {
return s == "" || s[0] != '-'
}
// isValidArgumentOption checks if the argument is a valid option (starting with '-').
// It doesn't check whether the option is supported or not
func isValidArgumentOption(s string) bool {
return s != "" && s[0] == '-'
}
// AddArguments adds new git arguments (option/value) to the command. It only accepts string literals, or trusted CmdArg.
// Type CmdArg is in the internal package, so it can not be used outside of this package directly,
// it makes sure that user-provided arguments won't cause RCE risks.
// User-provided arguments should be passed by other AddXxx functions
func (c *Command) AddArguments(args ...internal.CmdArg) *Command {
for _, arg := range args {
c.args = append(c.args, string(arg))
}
2016-11-04 07:16:01 +09:00
return c
}
Refactor git command package to improve security and maintainability (#22678) This PR follows #21535 (and replace #22592) ## Review without space diff https://github.com/go-gitea/gitea/pull/22678/files?diff=split&w=1 ## Purpose of this PR 1. Make git module command completely safe (risky user inputs won't be passed as argument option anymore) 2. Avoid low-level mistakes like https://github.com/go-gitea/gitea/pull/22098#discussion_r1045234918 3. Remove deprecated and dirty `CmdArgCheck` function, hide the `CmdArg` type 4. Simplify code when using git command ## The main idea of this PR * Move the `git.CmdArg` to the `internal` package, then no other package except `git` could use it. Then developers could never do `AddArguments(git.CmdArg(userInput))` any more. * Introduce `git.ToTrustedCmdArgs`, it's for user-provided and already trusted arguments. It's only used in a few cases, for example: use git arguments from config file, help unit test with some arguments. * Introduce `AddOptionValues` and `AddOptionFormat`, they make code more clear and simple: * Before: `AddArguments("-m").AddDynamicArguments(message)` * After: `AddOptionValues("-m", message)` * - * Before: `AddArguments(git.CmdArg(fmt.Sprintf("--author='%s <%s>'", sig.Name, sig.Email)))` * After: `AddOptionFormat("--author='%s <%s>'", sig.Name, sig.Email)` ## FAQ ### Why these changes were not done in #21535 ? #21535 is mainly a search&replace, it did its best to not change too much logic. Making the framework better needs a lot of changes, so this separate PR is needed as the second step. ### The naming of `AddOptionXxx` According to git's manual, the `--xxx` part is called `option`. ### How can it guarantee that `internal.CmdArg` won't be not misused? Go's specification guarantees that. Trying to access other package's internal package causes compilation error. And, `golangci-lint` also denies the git/internal package. Only the `git/command.go` can use it carefully. ### There is still a `ToTrustedCmdArgs`, will it still allow developers to make mistakes and pass untrusted arguments? Generally speaking, no. Because when using `ToTrustedCmdArgs`, the code will be very complex (see the changes for examples). Then developers and reviewers can know that something might be unreasonable. ### Why there was a `CmdArgCheck` and why it's removed? At the moment of #21535, to reduce unnecessary changes, `CmdArgCheck` was introduced as a hacky patch. Now, almost all code could be written as `cmd := NewCommand(); cmd.AddXxx(...)`, then there is no need for `CmdArgCheck` anymore. ### Why many codes for `signArg == ""` is deleted? Because in the old code, `signArg` could never be empty string, it's either `-S[key-id]` or `--no-gpg-sign`. So the `signArg == ""` is just dead code. --------- Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
2023-02-04 11:30:43 +09:00
// AddOptionValues adds a new option with a list of non-option values
// For example: AddOptionValues("--opt", val) means 2 arguments: {"--opt", val}.
// The values are treated as dynamic argument values. It equals to: AddArguments("--opt") then AddDynamicArguments(val).
func (c *Command) AddOptionValues(opt internal.CmdArg, args ...string) *Command {
if !isValidArgumentOption(string(opt)) {
c.brokenArgs = append(c.brokenArgs, string(opt))
return c
}
c.args = append(c.args, string(opt))
c.AddDynamicArguments(args...)
return c
}
// AddOptionFormat adds a new option with a format string and arguments
// For example: AddOptionFormat("--opt=%s %s", val1, val2) means 1 argument: {"--opt=val1 val2"}.
func (c *Command) AddOptionFormat(opt string, args ...any) *Command {
if !isValidArgumentOption(opt) {
c.brokenArgs = append(c.brokenArgs, opt)
return c
}
// a quick check to make sure the format string matches the number of arguments, to find low-level mistakes ASAP
if strings.Count(strings.ReplaceAll(opt, "%%", ""), "%") != len(args) {
c.brokenArgs = append(c.brokenArgs, opt)
return c
}
s := fmt.Sprintf(opt, args...)
c.args = append(c.args, s)
return c
}
// AddDynamicArguments adds new dynamic argument values to the command.
// The arguments may come from user input and can not be trusted, so no leading '-' is allowed to avoid passing options.
// TODO: in the future, this function can be renamed to AddArgumentValues
func (c *Command) AddDynamicArguments(args ...string) *Command {
for _, arg := range args {
Refactor git command package to improve security and maintainability (#22678) This PR follows #21535 (and replace #22592) ## Review without space diff https://github.com/go-gitea/gitea/pull/22678/files?diff=split&w=1 ## Purpose of this PR 1. Make git module command completely safe (risky user inputs won't be passed as argument option anymore) 2. Avoid low-level mistakes like https://github.com/go-gitea/gitea/pull/22098#discussion_r1045234918 3. Remove deprecated and dirty `CmdArgCheck` function, hide the `CmdArg` type 4. Simplify code when using git command ## The main idea of this PR * Move the `git.CmdArg` to the `internal` package, then no other package except `git` could use it. Then developers could never do `AddArguments(git.CmdArg(userInput))` any more. * Introduce `git.ToTrustedCmdArgs`, it's for user-provided and already trusted arguments. It's only used in a few cases, for example: use git arguments from config file, help unit test with some arguments. * Introduce `AddOptionValues` and `AddOptionFormat`, they make code more clear and simple: * Before: `AddArguments("-m").AddDynamicArguments(message)` * After: `AddOptionValues("-m", message)` * - * Before: `AddArguments(git.CmdArg(fmt.Sprintf("--author='%s <%s>'", sig.Name, sig.Email)))` * After: `AddOptionFormat("--author='%s <%s>'", sig.Name, sig.Email)` ## FAQ ### Why these changes were not done in #21535 ? #21535 is mainly a search&replace, it did its best to not change too much logic. Making the framework better needs a lot of changes, so this separate PR is needed as the second step. ### The naming of `AddOptionXxx` According to git's manual, the `--xxx` part is called `option`. ### How can it guarantee that `internal.CmdArg` won't be not misused? Go's specification guarantees that. Trying to access other package's internal package causes compilation error. And, `golangci-lint` also denies the git/internal package. Only the `git/command.go` can use it carefully. ### There is still a `ToTrustedCmdArgs`, will it still allow developers to make mistakes and pass untrusted arguments? Generally speaking, no. Because when using `ToTrustedCmdArgs`, the code will be very complex (see the changes for examples). Then developers and reviewers can know that something might be unreasonable. ### Why there was a `CmdArgCheck` and why it's removed? At the moment of #21535, to reduce unnecessary changes, `CmdArgCheck` was introduced as a hacky patch. Now, almost all code could be written as `cmd := NewCommand(); cmd.AddXxx(...)`, then there is no need for `CmdArgCheck` anymore. ### Why many codes for `signArg == ""` is deleted? Because in the old code, `signArg` could never be empty string, it's either `-S[key-id]` or `--no-gpg-sign`. So the `signArg == ""` is just dead code. --------- Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
2023-02-04 11:30:43 +09:00
if !isSafeArgumentValue(arg) {
c.brokenArgs = append(c.brokenArgs, arg)
}
}
if len(c.brokenArgs) != 0 {
return c
}
c.args = append(c.args, args...)
return c
}
// AddDashesAndList adds the "--" and then add the list as arguments, it's usually for adding file list
// At the moment, this function can be only called once, maybe in future it can be refactored to support multiple calls (if necessary)
func (c *Command) AddDashesAndList(list ...string) *Command {
c.args = append(c.args, "--")
// Some old code also checks `arg != ""`, IMO it's not necessary.
// If the check is needed, the list should be prepared before the call to this function
c.args = append(c.args, list...)
return c
}
Refactor git command package to improve security and maintainability (#22678) This PR follows #21535 (and replace #22592) ## Review without space diff https://github.com/go-gitea/gitea/pull/22678/files?diff=split&w=1 ## Purpose of this PR 1. Make git module command completely safe (risky user inputs won't be passed as argument option anymore) 2. Avoid low-level mistakes like https://github.com/go-gitea/gitea/pull/22098#discussion_r1045234918 3. Remove deprecated and dirty `CmdArgCheck` function, hide the `CmdArg` type 4. Simplify code when using git command ## The main idea of this PR * Move the `git.CmdArg` to the `internal` package, then no other package except `git` could use it. Then developers could never do `AddArguments(git.CmdArg(userInput))` any more. * Introduce `git.ToTrustedCmdArgs`, it's for user-provided and already trusted arguments. It's only used in a few cases, for example: use git arguments from config file, help unit test with some arguments. * Introduce `AddOptionValues` and `AddOptionFormat`, they make code more clear and simple: * Before: `AddArguments("-m").AddDynamicArguments(message)` * After: `AddOptionValues("-m", message)` * - * Before: `AddArguments(git.CmdArg(fmt.Sprintf("--author='%s <%s>'", sig.Name, sig.Email)))` * After: `AddOptionFormat("--author='%s <%s>'", sig.Name, sig.Email)` ## FAQ ### Why these changes were not done in #21535 ? #21535 is mainly a search&replace, it did its best to not change too much logic. Making the framework better needs a lot of changes, so this separate PR is needed as the second step. ### The naming of `AddOptionXxx` According to git's manual, the `--xxx` part is called `option`. ### How can it guarantee that `internal.CmdArg` won't be not misused? Go's specification guarantees that. Trying to access other package's internal package causes compilation error. And, `golangci-lint` also denies the git/internal package. Only the `git/command.go` can use it carefully. ### There is still a `ToTrustedCmdArgs`, will it still allow developers to make mistakes and pass untrusted arguments? Generally speaking, no. Because when using `ToTrustedCmdArgs`, the code will be very complex (see the changes for examples). Then developers and reviewers can know that something might be unreasonable. ### Why there was a `CmdArgCheck` and why it's removed? At the moment of #21535, to reduce unnecessary changes, `CmdArgCheck` was introduced as a hacky patch. Now, almost all code could be written as `cmd := NewCommand(); cmd.AddXxx(...)`, then there is no need for `CmdArgCheck` anymore. ### Why many codes for `signArg == ""` is deleted? Because in the old code, `signArg` could never be empty string, it's either `-S[key-id]` or `--no-gpg-sign`. So the `signArg == ""` is just dead code. --------- Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
2023-02-04 11:30:43 +09:00
// ToTrustedCmdArgs converts a list of strings (trusted as argument) to TrustedCmdArgs
// In most cases, it shouldn't be used. Use AddXxx function instead
func ToTrustedCmdArgs(args []string) TrustedCmdArgs {
ret := make(TrustedCmdArgs, len(args))
for i, arg := range args {
ret[i] = internal.CmdArg(arg)
}
Refactor git command package to improve security and maintainability (#22678) This PR follows #21535 (and replace #22592) ## Review without space diff https://github.com/go-gitea/gitea/pull/22678/files?diff=split&w=1 ## Purpose of this PR 1. Make git module command completely safe (risky user inputs won't be passed as argument option anymore) 2. Avoid low-level mistakes like https://github.com/go-gitea/gitea/pull/22098#discussion_r1045234918 3. Remove deprecated and dirty `CmdArgCheck` function, hide the `CmdArg` type 4. Simplify code when using git command ## The main idea of this PR * Move the `git.CmdArg` to the `internal` package, then no other package except `git` could use it. Then developers could never do `AddArguments(git.CmdArg(userInput))` any more. * Introduce `git.ToTrustedCmdArgs`, it's for user-provided and already trusted arguments. It's only used in a few cases, for example: use git arguments from config file, help unit test with some arguments. * Introduce `AddOptionValues` and `AddOptionFormat`, they make code more clear and simple: * Before: `AddArguments("-m").AddDynamicArguments(message)` * After: `AddOptionValues("-m", message)` * - * Before: `AddArguments(git.CmdArg(fmt.Sprintf("--author='%s <%s>'", sig.Name, sig.Email)))` * After: `AddOptionFormat("--author='%s <%s>'", sig.Name, sig.Email)` ## FAQ ### Why these changes were not done in #21535 ? #21535 is mainly a search&replace, it did its best to not change too much logic. Making the framework better needs a lot of changes, so this separate PR is needed as the second step. ### The naming of `AddOptionXxx` According to git's manual, the `--xxx` part is called `option`. ### How can it guarantee that `internal.CmdArg` won't be not misused? Go's specification guarantees that. Trying to access other package's internal package causes compilation error. And, `golangci-lint` also denies the git/internal package. Only the `git/command.go` can use it carefully. ### There is still a `ToTrustedCmdArgs`, will it still allow developers to make mistakes and pass untrusted arguments? Generally speaking, no. Because when using `ToTrustedCmdArgs`, the code will be very complex (see the changes for examples). Then developers and reviewers can know that something might be unreasonable. ### Why there was a `CmdArgCheck` and why it's removed? At the moment of #21535, to reduce unnecessary changes, `CmdArgCheck` was introduced as a hacky patch. Now, almost all code could be written as `cmd := NewCommand(); cmd.AddXxx(...)`, then there is no need for `CmdArgCheck` anymore. ### Why many codes for `signArg == ""` is deleted? Because in the old code, `signArg` could never be empty string, it's either `-S[key-id]` or `--no-gpg-sign`. So the `signArg == ""` is just dead code. --------- Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
2023-02-04 11:30:43 +09:00
return ret
}
// RunOpts represents parameters to run the command. If UseContextTimeout is specified, then Timeout is ignored.
type RunOpts struct {
Env []string
Timeout time.Duration
UseContextTimeout bool
Dir string
Stdout, Stderr io.Writer
Stdin io.Reader
PipelineFunc func(context.Context, context.CancelFunc) error
}
func commonBaseEnvs() []string {
// at the moment, do not set "GIT_CONFIG_NOSYSTEM", users may have put some configs like "receive.certNonceSeed" in it
envs := []string{
"HOME=" + HomeDir(), // make Gitea use internal git config only, to prevent conflicts with user's git config
"GIT_NO_REPLACE_OBJECTS=1", // ignore replace references (https://git-scm.com/docs/git-replace)
}
// some environment variables should be passed to git command
passThroughEnvKeys := []string{
"GNUPGHOME", // git may call gnupg to do commit signing
}
for _, key := range passThroughEnvKeys {
if val, ok := os.LookupEnv(key); ok {
envs = append(envs, key+"="+val)
}
}
return envs
}
// CommonGitCmdEnvs returns the common environment variables for a "git" command.
func CommonGitCmdEnvs() []string {
return append(commonBaseEnvs(), []string{
"LC_ALL=" + DefaultLocale,
"GIT_TERMINAL_PROMPT=0", // avoid prompting for credentials interactively, supported since git v2.3
}...)
}
// CommonCmdServEnvs is like CommonGitCmdEnvs, but it only returns minimal required environment variables for the "gitea serv" command
func CommonCmdServEnvs() []string {
return commonBaseEnvs()
}
var ErrBrokenCommand = errors.New("git command is broken")
// Run runs the command with the RunOpts
func (c *Command) Run(opts *RunOpts) error {
if len(c.brokenArgs) != 0 {
log.Error("git command is broken: %s, broken args: %s", c.String(), strings.Join(c.brokenArgs, " "))
return ErrBrokenCommand
}
if opts == nil {
opts = &RunOpts{}
}
// We must not change the provided options
timeout := opts.Timeout
if timeout <= 0 {
timeout = defaultCommandExecutionTimeout
2016-11-04 07:16:01 +09:00
}
if len(opts.Dir) == 0 {
log.Debug("%s", c)
2016-11-04 07:16:01 +09:00
} else {
log.Debug("%s: %v", opts.Dir, c)
2016-11-04 07:16:01 +09:00
}
desc := c.desc
if desc == "" {
args := c.args[c.globalArgsLength:]
var argSensitiveURLIndexes []int
for i, arg := range c.args {
if strings.Contains(arg, "://") && strings.Contains(arg, "@") {
argSensitiveURLIndexes = append(argSensitiveURLIndexes, i)
}
}
if len(argSensitiveURLIndexes) > 0 {
args = make([]string, len(c.args))
copy(args, c.args)
for _, urlArgIndex := range argSensitiveURLIndexes {
args[urlArgIndex] = util.SanitizeCredentialURLs(args[urlArgIndex])
}
}
desc = fmt.Sprintf("%s %s [repo_path: %s]", c.name, strings.Join(args, " "), opts.Dir)
}
var ctx context.Context
var cancel context.CancelFunc
var finished context.CancelFunc
if opts.UseContextTimeout {
ctx, cancel, finished = process.GetManager().AddContext(c.parentContext, desc)
} else {
ctx, cancel, finished = process.GetManager().AddContextTimeout(c.parentContext, timeout, desc)
}
defer finished()
cmd := exec.CommandContext(ctx, c.name, c.args...)
if opts.Env == nil {
cmd.Env = os.Environ()
} else {
cmd.Env = opts.Env
}
Disable new signal-based asynchronous goroutine preemption from GO 1.14 in git env (#11237) As seen in trouble shooting #11032 the new feature of Go 1.14 is causing several second delays in startup in certain situations. Debugging shows it spending several seconds handling SIGURG commands during init: ``` 6922:04:51.984234 trace init() ./modules/queue/unique_queue_wrapped.go remote: ) = 69 <0.000012> remote: [pid 15984] 22:04:51 write(1, "\ttime taken: 236.761\302\265s\n\n", 25 time taken: 236.761µs remote: remote: ) = 25 <0.000011> remote: [pid 15984] 22:04:51 --- SIGURG {si_signo=SIGURG, si_code=SI_TKILL, si_pid=15984, si_uid=0} --- remote: [pid 15984] 22:04:52 --- SIGURG {si_signo=SIGURG, si_code=SI_TKILL, si_pid=15984, si_uid=0} --- remote: [pid 15984] 22:04:52 --- SIGURG {si_signo=SIGURG, si_code=SI_TKILL, si_pid=15984, si_uid=0} --- ``` This causes up to 20 seconds added to a push in some cases as it happens for each call of the gitea hook command. This is likely the cause of #10661 as well and would start to effect users once we release 1.12 which would be the first release compiled with Go 1.14. I suspect this is just a slight issue with the upstream implementatation as there have been a few very similar bugs fixed and reported: https://github.com/golang/go/issues/37741 https://github.com/golang/go/issues/37942 We should revisit this in the future and see if a newer version of Go has solved it, but for now disable this option in the environment that gitea hook runs in to avoid it.
2020-04-29 00:45:32 +09:00
process.SetSysProcAttribute(cmd)
cmd.Env = append(cmd.Env, CommonGitCmdEnvs()...)
cmd.Dir = opts.Dir
cmd.Stdout = opts.Stdout
cmd.Stderr = opts.Stderr
cmd.Stdin = opts.Stdin
2016-11-04 07:16:01 +09:00
if err := cmd.Start(); err != nil {
return err
}
if opts.PipelineFunc != nil {
err := opts.PipelineFunc(ctx, cancel)
if err != nil {
cancel()
_ = cmd.Wait()
return err
}
}
if err := cmd.Wait(); err != nil && ctx.Err() != context.DeadlineExceeded {
return err
}
return ctx.Err()
2016-11-04 07:16:01 +09:00
}
type RunStdError interface {
error
Unwrap() error
Stderr() string
IsExitCode(code int) bool
2019-05-12 00:29:17 +09:00
}
type runStdError struct {
err error
stderr string
errMsg string
2019-05-12 00:29:17 +09:00
}
func (r *runStdError) Error() string {
// the stderr must be in the returned error text, some code only checks `strings.Contains(err.Error(), "git error")`
if r.errMsg == "" {
r.errMsg = ConcatenateError(r.err, r.stderr).Error()
}
return r.errMsg
2019-05-12 00:29:17 +09:00
}
func (r *runStdError) Unwrap() error {
return r.err
}
func (r *runStdError) Stderr() string {
return r.stderr
2016-11-04 07:16:01 +09:00
}
func (r *runStdError) IsExitCode(code int) bool {
var exitError *exec.ExitError
if errors.As(r.err, &exitError) {
return exitError.ExitCode() == code
}
return false
}
func bytesToString(b []byte) string {
return *(*string)(unsafe.Pointer(&b)) // that's what Golang's strings.Builder.String() does (go/src/strings/builder.go)
2019-05-12 00:29:17 +09:00
}
// RunStdString runs the command with options and returns stdout/stderr as string. and store stderr to returned error (err combined with stderr).
func (c *Command) RunStdString(opts *RunOpts) (stdout, stderr string, runErr RunStdError) {
stdoutBytes, stderrBytes, err := c.RunStdBytes(opts)
stdout = bytesToString(stdoutBytes)
stderr = bytesToString(stderrBytes)
if err != nil {
return stdout, stderr, &runStdError{err: err, stderr: stderr}
}
// even if there is no err, there could still be some stderr output, so we just return stdout/stderr as they are
return stdout, stderr, nil
}
// RunStdBytes runs the command with options and returns stdout/stderr as bytes. and store stderr to returned error (err combined with stderr).
func (c *Command) RunStdBytes(opts *RunOpts) (stdout, stderr []byte, runErr RunStdError) {
if opts == nil {
opts = &RunOpts{}
}
if opts.Stdout != nil || opts.Stderr != nil {
// we must panic here, otherwise there would be bugs if developers set Stdin/Stderr by mistake, and it would be very difficult to debug
panic("stdout and stderr field must be nil when using RunStdBytes")
}
stdoutBuf := &bytes.Buffer{}
stderrBuf := &bytes.Buffer{}
// We must not change the provided options as it could break future calls - therefore make a copy.
newOpts := &RunOpts{
Env: opts.Env,
Timeout: opts.Timeout,
UseContextTimeout: opts.UseContextTimeout,
Dir: opts.Dir,
Stdout: stdoutBuf,
Stderr: stderrBuf,
Stdin: opts.Stdin,
PipelineFunc: opts.PipelineFunc,
}
err := c.Run(newOpts)
stderr = stderrBuf.Bytes()
if err != nil {
return nil, stderr, &runStdError{err: err, stderr: bytesToString(stderr)}
}
// even if there is no err, there could still be some stderr output
return stdoutBuf.Bytes(), stderr, nil
2016-11-04 07:16:01 +09:00
}
// AllowLFSFiltersArgs return globalCommandArgs with lfs filter, it should only be used for tests
Refactor git command package to improve security and maintainability (#22678) This PR follows #21535 (and replace #22592) ## Review without space diff https://github.com/go-gitea/gitea/pull/22678/files?diff=split&w=1 ## Purpose of this PR 1. Make git module command completely safe (risky user inputs won't be passed as argument option anymore) 2. Avoid low-level mistakes like https://github.com/go-gitea/gitea/pull/22098#discussion_r1045234918 3. Remove deprecated and dirty `CmdArgCheck` function, hide the `CmdArg` type 4. Simplify code when using git command ## The main idea of this PR * Move the `git.CmdArg` to the `internal` package, then no other package except `git` could use it. Then developers could never do `AddArguments(git.CmdArg(userInput))` any more. * Introduce `git.ToTrustedCmdArgs`, it's for user-provided and already trusted arguments. It's only used in a few cases, for example: use git arguments from config file, help unit test with some arguments. * Introduce `AddOptionValues` and `AddOptionFormat`, they make code more clear and simple: * Before: `AddArguments("-m").AddDynamicArguments(message)` * After: `AddOptionValues("-m", message)` * - * Before: `AddArguments(git.CmdArg(fmt.Sprintf("--author='%s <%s>'", sig.Name, sig.Email)))` * After: `AddOptionFormat("--author='%s <%s>'", sig.Name, sig.Email)` ## FAQ ### Why these changes were not done in #21535 ? #21535 is mainly a search&replace, it did its best to not change too much logic. Making the framework better needs a lot of changes, so this separate PR is needed as the second step. ### The naming of `AddOptionXxx` According to git's manual, the `--xxx` part is called `option`. ### How can it guarantee that `internal.CmdArg` won't be not misused? Go's specification guarantees that. Trying to access other package's internal package causes compilation error. And, `golangci-lint` also denies the git/internal package. Only the `git/command.go` can use it carefully. ### There is still a `ToTrustedCmdArgs`, will it still allow developers to make mistakes and pass untrusted arguments? Generally speaking, no. Because when using `ToTrustedCmdArgs`, the code will be very complex (see the changes for examples). Then developers and reviewers can know that something might be unreasonable. ### Why there was a `CmdArgCheck` and why it's removed? At the moment of #21535, to reduce unnecessary changes, `CmdArgCheck` was introduced as a hacky patch. Now, almost all code could be written as `cmd := NewCommand(); cmd.AddXxx(...)`, then there is no need for `CmdArgCheck` anymore. ### Why many codes for `signArg == ""` is deleted? Because in the old code, `signArg` could never be empty string, it's either `-S[key-id]` or `--no-gpg-sign`. So the `signArg == ""` is just dead code. --------- Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
2023-02-04 11:30:43 +09:00
func AllowLFSFiltersArgs() TrustedCmdArgs {
// Now here we should explicitly allow lfs filters to run
Refactor git command package to improve security and maintainability (#22678) This PR follows #21535 (and replace #22592) ## Review without space diff https://github.com/go-gitea/gitea/pull/22678/files?diff=split&w=1 ## Purpose of this PR 1. Make git module command completely safe (risky user inputs won't be passed as argument option anymore) 2. Avoid low-level mistakes like https://github.com/go-gitea/gitea/pull/22098#discussion_r1045234918 3. Remove deprecated and dirty `CmdArgCheck` function, hide the `CmdArg` type 4. Simplify code when using git command ## The main idea of this PR * Move the `git.CmdArg` to the `internal` package, then no other package except `git` could use it. Then developers could never do `AddArguments(git.CmdArg(userInput))` any more. * Introduce `git.ToTrustedCmdArgs`, it's for user-provided and already trusted arguments. It's only used in a few cases, for example: use git arguments from config file, help unit test with some arguments. * Introduce `AddOptionValues` and `AddOptionFormat`, they make code more clear and simple: * Before: `AddArguments("-m").AddDynamicArguments(message)` * After: `AddOptionValues("-m", message)` * - * Before: `AddArguments(git.CmdArg(fmt.Sprintf("--author='%s <%s>'", sig.Name, sig.Email)))` * After: `AddOptionFormat("--author='%s <%s>'", sig.Name, sig.Email)` ## FAQ ### Why these changes were not done in #21535 ? #21535 is mainly a search&replace, it did its best to not change too much logic. Making the framework better needs a lot of changes, so this separate PR is needed as the second step. ### The naming of `AddOptionXxx` According to git's manual, the `--xxx` part is called `option`. ### How can it guarantee that `internal.CmdArg` won't be not misused? Go's specification guarantees that. Trying to access other package's internal package causes compilation error. And, `golangci-lint` also denies the git/internal package. Only the `git/command.go` can use it carefully. ### There is still a `ToTrustedCmdArgs`, will it still allow developers to make mistakes and pass untrusted arguments? Generally speaking, no. Because when using `ToTrustedCmdArgs`, the code will be very complex (see the changes for examples). Then developers and reviewers can know that something might be unreasonable. ### Why there was a `CmdArgCheck` and why it's removed? At the moment of #21535, to reduce unnecessary changes, `CmdArgCheck` was introduced as a hacky patch. Now, almost all code could be written as `cmd := NewCommand(); cmd.AddXxx(...)`, then there is no need for `CmdArgCheck` anymore. ### Why many codes for `signArg == ""` is deleted? Because in the old code, `signArg` could never be empty string, it's either `-S[key-id]` or `--no-gpg-sign`. So the `signArg == ""` is just dead code. --------- Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
2023-02-04 11:30:43 +09:00
filteredLFSGlobalArgs := make(TrustedCmdArgs, len(globalCommandArgs))
j := 0
for _, arg := range globalCommandArgs {
if strings.Contains(string(arg), "lfs") {
j--
} else {
filteredLFSGlobalArgs[j] = arg
j++
}
}
return filteredLFSGlobalArgs[:j]
}