Skip to content

Commit

Permalink
feat: Add support for tools from github enterprise.
Browse files Browse the repository at this point in the history
  • Loading branch information
johnrengelman authored and John Engelman committed Aug 9, 2024
1 parent 4fd8e8a commit 2e51afe
Show file tree
Hide file tree
Showing 4 changed files with 178 additions and 75 deletions.
65 changes: 33 additions & 32 deletions docs/docs/04-command-line-reference/gptscript.md
Original file line number Diff line number Diff line change
Expand Up @@ -12,38 +12,39 @@ gptscript [flags] PROGRAM_FILE [INPUT...]
### Options

```
--cache-dir string Directory to store cache (default: $XDG_CACHE_HOME/gptscript) ($GPTSCRIPT_CACHE_DIR)
--chat-state string The chat state to continue, or null to start a new chat and return the state ($GPTSCRIPT_CHAT_STATE)
-C, --chdir string Change current working directory ($GPTSCRIPT_CHDIR)
--color Use color in output (default true) ($GPTSCRIPT_COLOR)
--config string Path to GPTScript config file ($GPTSCRIPT_CONFIG)
--confirm Prompt before running potentially dangerous commands ($GPTSCRIPT_CONFIRM)
--credential-context string Context name in which to store credentials ($GPTSCRIPT_CREDENTIAL_CONTEXT) (default "default")
--credential-override strings Credentials to override (ex: --credential-override github.com/example/cred-tool:API_TOKEN=1234) ($GPTSCRIPT_CREDENTIAL_OVERRIDE)
--debug Enable debug logging ($GPTSCRIPT_DEBUG)
--debug-messages Enable logging of chat completion calls ($GPTSCRIPT_DEBUG_MESSAGES)
--default-model string Default LLM model to use ($GPTSCRIPT_DEFAULT_MODEL) (default "gpt-4o")
--default-model-provider string Default LLM model provider to use, this will override OpenAI settings ($GPTSCRIPT_DEFAULT_MODEL_PROVIDER)
--disable-cache Disable caching of LLM API responses ($GPTSCRIPT_DISABLE_CACHE)
--disable-tui Don't use chat TUI but instead verbose output ($GPTSCRIPT_DISABLE_TUI)
--dump-state string Dump the internal execution state to a file ($GPTSCRIPT_DUMP_STATE)
--events-stream-to string Stream events to this location, could be a file descriptor/handle (e.g. fd://2), filename, or named pipe (e.g. \\.\pipe\my-pipe) ($GPTSCRIPT_EVENTS_STREAM_TO)
--force-chat Force an interactive chat session if even the top level tool is not a chat tool ($GPTSCRIPT_FORCE_CHAT)
--force-sequential Force parallel calls to run sequentially ($GPTSCRIPT_FORCE_SEQUENTIAL)
-h, --help help for gptscript
-f, --input string Read input from a file ("-" for stdin) ($GPTSCRIPT_INPUT_FILE)
--list-models List the models available and exit ($GPTSCRIPT_LIST_MODELS)
--list-tools List built-in tools and exit ($GPTSCRIPT_LIST_TOOLS)
--no-trunc Do not truncate long log messages ($GPTSCRIPT_NO_TRUNC)
--openai-api-key string OpenAI API KEY ($OPENAI_API_KEY)
--openai-base-url string OpenAI base URL ($OPENAI_BASE_URL)
--openai-org-id string OpenAI organization ID ($OPENAI_ORG_ID)
-o, --output string Save output to a file, or - for stdout ($GPTSCRIPT_OUTPUT)
-q, --quiet No output logging (set --quiet=false to force on even when there is no TTY) ($GPTSCRIPT_QUIET)
--save-chat-state-file string A file to save the chat state to so that a conversation can be resumed with --chat-state ($GPTSCRIPT_SAVE_CHAT_STATE_FILE)
--sub-tool string Use tool of this name, not the first tool in file ($GPTSCRIPT_SUB_TOOL)
--ui Launch the UI ($GPTSCRIPT_UI)
--workspace string Directory to use for the workspace, if specified it will not be deleted on exit ($GPTSCRIPT_WORKSPACE)
--cache-dir string Directory to store cache (default: $XDG_CACHE_HOME/gptscript) ($GPTSCRIPT_CACHE_DIR)
--chat-state string The chat state to continue, or null to start a new chat and return the state ($GPTSCRIPT_CHAT_STATE)
-C, --chdir string Change current working directory ($GPTSCRIPT_CHDIR)
--color Use color in output (default true) ($GPTSCRIPT_COLOR)
--config string Path to GPTScript config file ($GPTSCRIPT_CONFIG)
--confirm Prompt before running potentially dangerous commands ($GPTSCRIPT_CONFIRM)
--credential-context string Context name in which to store credentials ($GPTSCRIPT_CREDENTIAL_CONTEXT) (default "default")
--credential-override strings Credentials to override (ex: --credential-override github.com/example/cred-tool:API_TOKEN=1234) ($GPTSCRIPT_CREDENTIAL_OVERRIDE)
--debug Enable debug logging ($GPTSCRIPT_DEBUG)
--debug-messages Enable logging of chat completion calls ($GPTSCRIPT_DEBUG_MESSAGES)
--default-model string Default LLM model to use ($GPTSCRIPT_DEFAULT_MODEL) (default "gpt-4o")
--default-model-provider string Default LLM model provider to use, this will override OpenAI settings ($GPTSCRIPT_DEFAULT_MODEL_PROVIDER)
--disable-cache Disable caching of LLM API responses ($GPTSCRIPT_DISABLE_CACHE)
--disable-tui Don't use chat TUI but instead verbose output ($GPTSCRIPT_DISABLE_TUI)
--dump-state string Dump the internal execution state to a file ($GPTSCRIPT_DUMP_STATE)
--events-stream-to string Stream events to this location, could be a file descriptor/handle (e.g. fd://2), filename, or named pipe (e.g. \\.\pipe\my-pipe) ($GPTSCRIPT_EVENTS_STREAM_TO)
--force-chat Force an interactive chat session if even the top level tool is not a chat tool ($GPTSCRIPT_FORCE_CHAT)
--force-sequential Force parallel calls to run sequentially ($GPTSCRIPT_FORCE_SEQUENTIAL)
--github-enterprise-hostname string The host name for a Github Enterprise instance to enable for remote loading ($GPTSCRIPT_GITHUB_ENTERPRISE_HOSTNAME)
-h, --help help for gptscript
-f, --input string Read input from a file ("-" for stdin) ($GPTSCRIPT_INPUT_FILE)
--list-models List the models available and exit ($GPTSCRIPT_LIST_MODELS)
--list-tools List built-in tools and exit ($GPTSCRIPT_LIST_TOOLS)
--no-trunc Do not truncate long log messages ($GPTSCRIPT_NO_TRUNC)
--openai-api-key string OpenAI API KEY ($OPENAI_API_KEY)
--openai-base-url string OpenAI base URL ($OPENAI_BASE_URL)
--openai-org-id string OpenAI organization ID ($OPENAI_ORG_ID)
-o, --output string Save output to a file, or - for stdout ($GPTSCRIPT_OUTPUT)
-q, --quiet No output logging (set --quiet=false to force on even when there is no TTY) ($GPTSCRIPT_QUIET)
--save-chat-state-file string A file to save the chat state to so that a conversation can be resumed with --chat-state ($GPTSCRIPT_SAVE_CHAT_STATE_FILE)
--sub-tool string Use tool of this name, not the first tool in file ($GPTSCRIPT_SUB_TOOL)
--ui Launch the UI ($GPTSCRIPT_UI)
--workspace string Directory to use for the workspace, if specified it will not be deleted on exit ($GPTSCRIPT_WORKSPACE)
```

### SEE ALSO
Expand Down
44 changes: 25 additions & 19 deletions pkg/cli/gptscript.go
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,7 @@ import (
"github.com/gptscript-ai/gptscript/pkg/gptscript"
"github.com/gptscript-ai/gptscript/pkg/input"
"github.com/gptscript-ai/gptscript/pkg/loader"
"github.com/gptscript-ai/gptscript/pkg/loader/github"
"github.com/gptscript-ai/gptscript/pkg/monitor"
"github.com/gptscript-ai/gptscript/pkg/mvl"
"github.com/gptscript-ai/gptscript/pkg/openai"
Expand Down Expand Up @@ -54,25 +55,26 @@ type GPTScript struct {
Output string `usage:"Save output to a file, or - for stdout" short:"o"`
EventsStreamTo string `usage:"Stream events to this location, could be a file descriptor/handle (e.g. fd://2), filename, or named pipe (e.g. \\\\.\\pipe\\my-pipe)" name:"events-stream-to"`
// Input should not be using GPTSCRIPT_INPUT env var because that is the same value that is set in tool executions
Input string `usage:"Read input from a file (\"-\" for stdin)" short:"f" env:"GPTSCRIPT_INPUT_FILE"`
SubTool string `usage:"Use tool of this name, not the first tool in file" local:"true"`
Assemble bool `usage:"Assemble tool to a single artifact, saved to --output" hidden:"true" local:"true"`
ListModels bool `usage:"List the models available and exit" local:"true"`
ListTools bool `usage:"List built-in tools and exit" local:"true"`
ListenAddress string `usage:"Server listen address" default:"127.0.0.1:0" hidden:"true"`
Chdir string `usage:"Change current working directory" short:"C"`
Daemon bool `usage:"Run tool as a daemon" local:"true" hidden:"true"`
Ports string `usage:"The port range to use for ephemeral daemon ports (ex: 11000-12000)" hidden:"true"`
CredentialContext string `usage:"Context name in which to store credentials" default:"default"`
CredentialOverride []string `usage:"Credentials to override (ex: --credential-override github.com/example/cred-tool:API_TOKEN=1234)"`
ChatState string `usage:"The chat state to continue, or null to start a new chat and return the state" local:"true"`
ForceChat bool `usage:"Force an interactive chat session if even the top level tool is not a chat tool" local:"true"`
ForceSequential bool `usage:"Force parallel calls to run sequentially" local:"true"`
Workspace string `usage:"Directory to use for the workspace, if specified it will not be deleted on exit"`
UI bool `usage:"Launch the UI" local:"true" name:"ui"`
DisableTUI bool `usage:"Don't use chat TUI but instead verbose output" local:"true" name:"disable-tui"`
SaveChatStateFile string `usage:"A file to save the chat state to so that a conversation can be resumed with --chat-state" local:"true"`
DefaultModelProvider string `usage:"Default LLM model provider to use, this will override OpenAI settings"`
Input string `usage:"Read input from a file (\"-\" for stdin)" short:"f" env:"GPTSCRIPT_INPUT_FILE"`
SubTool string `usage:"Use tool of this name, not the first tool in file" local:"true"`
Assemble bool `usage:"Assemble tool to a single artifact, saved to --output" hidden:"true" local:"true"`
ListModels bool `usage:"List the models available and exit" local:"true"`
ListTools bool `usage:"List built-in tools and exit" local:"true"`
ListenAddress string `usage:"Server listen address" default:"127.0.0.1:0" hidden:"true"`
Chdir string `usage:"Change current working directory" short:"C"`
Daemon bool `usage:"Run tool as a daemon" local:"true" hidden:"true"`
Ports string `usage:"The port range to use for ephemeral daemon ports (ex: 11000-12000)" hidden:"true"`
CredentialContext string `usage:"Context name in which to store credentials" default:"default"`
CredentialOverride []string `usage:"Credentials to override (ex: --credential-override github.com/example/cred-tool:API_TOKEN=1234)"`
ChatState string `usage:"The chat state to continue, or null to start a new chat and return the state" local:"true"`
ForceChat bool `usage:"Force an interactive chat session if even the top level tool is not a chat tool" local:"true"`
ForceSequential bool `usage:"Force parallel calls to run sequentially" local:"true"`
Workspace string `usage:"Directory to use for the workspace, if specified it will not be deleted on exit"`
UI bool `usage:"Launch the UI" local:"true" name:"ui"`
DisableTUI bool `usage:"Don't use chat TUI but instead verbose output" local:"true" name:"disable-tui"`
SaveChatStateFile string `usage:"A file to save the chat state to so that a conversation can be resumed with --chat-state" local:"true"`
DefaultModelProvider string `usage:"Default LLM model provider to use, this will override OpenAI settings"`
GithubEnterpriseHostname string `usage:"The host name for a Github Enterprise instance to enable for remote loading" local:"true"`

readData []byte
}
Expand Down Expand Up @@ -334,6 +336,10 @@ func (r *GPTScript) Run(cmd *cobra.Command, args []string) (retErr error) {
return err
}

if r.GithubEnterpriseHostname != "" {
loader.AddVSC(github.LoaderForPrefix(r.GithubEnterpriseHostname))
}

// If the user is trying to launch the chat-builder UI, then set up the tool and options here.
if r.UI {
if os.Getenv(system.BinEnvVar) == "" {
Expand Down
87 changes: 63 additions & 24 deletions pkg/loader/github/github.go
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,7 @@ package github

import (
"context"
"crypto/tls"
"encoding/json"
"fmt"
"io"
Expand All @@ -18,52 +19,63 @@ import (
"github.com/gptscript-ai/gptscript/pkg/types"
)

const (
GithubPrefix = "github.com/"
githubRepoURL = "https://github.com/%s/%s.git"
githubDownloadURL = "https://raw.githubusercontent.com/%s/%s/%s/%s"
githubCommitURL = "https://api.github.com/repos/%s/%s/commits/%s"
)
type Config struct {
Prefix string
RepoURL string
DownloadURL string
CommitURL string
AuthToken string
}

var (
githubAuthToken = os.Getenv("GITHUB_AUTH_TOKEN")
log = mvl.Package()
log = mvl.Package()
defaultGithubConfig = &Config{
Prefix: "github.com/",
RepoURL: "https://github.com/%s/%s.git",
DownloadURL: "https://raw.githubusercontent.com/%s/%s/%s/%s",
CommitURL: "https://api.github.com/repos/%s/%s/commits/%s",
AuthToken: os.Getenv("GITHUB_AUTH_TOKEN"),
}
)

func init() {
loader.AddVSC(Load)
}

func getCommitLsRemote(ctx context.Context, account, repo, ref string) (string, error) {
url := fmt.Sprintf(githubRepoURL, account, repo)
func getCommitLsRemote(ctx context.Context, account, repo, ref string, config *Config) (string, error) {
url := fmt.Sprintf(config.RepoURL, account, repo)
return git.LsRemote(ctx, url, ref)
}

// regexp to match a git commit id
var commitRegexp = regexp.MustCompile("^[a-f0-9]{40}$")

func getCommit(ctx context.Context, account, repo, ref string) (string, error) {
func getCommit(ctx context.Context, account, repo, ref string, config *Config) (string, error) {
if commitRegexp.MatchString(ref) {
return ref, nil
}

url := fmt.Sprintf(githubCommitURL, account, repo, ref)
url := fmt.Sprintf(config.CommitURL, account, repo, ref)
req, err := http.NewRequestWithContext(ctx, http.MethodGet, url, nil)
if err != nil {
return "", fmt.Errorf("failed to create request of %s/%s at %s: %w", account, repo, url, err)
}

if githubAuthToken != "" {
req.Header.Add("Authorization", "Bearer "+githubAuthToken)
if config.AuthToken != "" {
req.Header.Add("Authorization", "Bearer "+config.AuthToken)
}

resp, err := http.DefaultClient.Do(req)
client := http.DefaultClient
if req.Host == config.Prefix && strings.ToLower(os.Getenv("GH_ENTERPRISE_SKIP_VERIFY")) == "true" {
client = &http.Client{Transport: &http.Transport{TLSClientConfig: &tls.Config{InsecureSkipVerify: true}}}
}
resp, err := client.Do(req)
if err != nil {
return "", err
} else if resp.StatusCode != http.StatusOK {
c, _ := io.ReadAll(resp.Body)
resp.Body.Close()
commit, fallBackErr := getCommitLsRemote(ctx, account, repo, ref)
commit, fallBackErr := getCommitLsRemote(ctx, account, repo, ref, config)
if fallBackErr == nil {
return commit, nil
}
Expand All @@ -88,8 +100,28 @@ func getCommit(ctx context.Context, account, repo, ref string) (string, error) {
return commit.SHA, nil
}

func Load(ctx context.Context, _ *cache.Client, urlName string) (string, string, *types.Repo, bool, error) {
if !strings.HasPrefix(urlName, GithubPrefix) {
func LoaderForPrefix(prefix string) func(context.Context, *cache.Client, string) (string, string, *types.Repo, bool, error) {
return func(ctx context.Context, c *cache.Client, urlName string) (string, string, *types.Repo, bool, error) {
return LoadWithConfig(ctx, c, urlName, NewGithubEnterpriseConfig(prefix))
}
}

func Load(ctx context.Context, c *cache.Client, urlName string) (string, string, *types.Repo, bool, error) {
return LoadWithConfig(ctx, c, urlName, defaultGithubConfig)
}

func NewGithubEnterpriseConfig(prefix string) *Config {
return &Config{
Prefix: prefix,
RepoURL: fmt.Sprintf("https://%s/%%s/%%s.git", prefix),
DownloadURL: fmt.Sprintf("https://raw.%s/%%s/%%s/%%s/%%s", prefix),
CommitURL: fmt.Sprintf("https://%s/api/v3/repos/%%s/%%s/commits/%%s", prefix),
AuthToken: os.Getenv("GH_ENTERPRISE_TOKEN"),
}
}

func LoadWithConfig(ctx context.Context, _ *cache.Client, urlName string, config *Config) (string, string, *types.Repo, bool, error) {
if !strings.HasPrefix(urlName, config.Prefix) {
return "", "", nil, false, nil
}

Expand All @@ -107,12 +139,12 @@ func Load(ctx context.Context, _ *cache.Client, urlName string) (string, string,
account, repo := parts[1], parts[2]
path := strings.Join(parts[3:], "/")

ref, err := getCommit(ctx, account, repo, ref)
ref, err := getCommit(ctx, account, repo, ref, config)
if err != nil {
return "", "", nil, false, err
}

downloadURL := fmt.Sprintf(githubDownloadURL, account, repo, ref, path)
downloadURL := fmt.Sprintf(config.DownloadURL, account, repo, ref, path)
if path == "" || path == "/" || !strings.Contains(parts[len(parts)-1], ".") {
var (
testPath string
Expand All @@ -124,13 +156,20 @@ func Load(ctx context.Context, _ *cache.Client, urlName string) (string, string,
} else {
testPath = path + "/" + ext
}
testURL = fmt.Sprintf(githubDownloadURL, account, repo, ref, testPath)
testURL = fmt.Sprintf(config.DownloadURL, account, repo, ref, testPath)
if i == len(types.DefaultFiles)-1 {
// no reason to test the last one, we are just going to use it. Being that the default list is only
// two elements this loop could have been one check, but hey over-engineered code ftw.
break
}
if resp, err := http.Head(testURL); err == nil {
headReq, err := http.NewRequest("HEAD", testURL, nil)
if err != nil {
break
}
if config.AuthToken != "" {
headReq.Header.Add("Authorization", "Bearer "+config.AuthToken)
}
if resp, err := http.DefaultClient.Do(headReq); err == nil {
_ = resp.Body.Close()
if resp.StatusCode == 200 {
break
Expand All @@ -141,9 +180,9 @@ func Load(ctx context.Context, _ *cache.Client, urlName string) (string, string,
path = testPath
}

return downloadURL, githubAuthToken, &types.Repo{
return downloadURL, config.AuthToken, &types.Repo{
VCS: "git",
Root: fmt.Sprintf(githubRepoURL, account, repo),
Root: fmt.Sprintf(config.RepoURL, account, repo),
Path: gpath.Dir(path),
Name: gpath.Base(path),
Revision: ref,
Expand Down
Loading

0 comments on commit 2e51afe

Please sign in to comment.