Files
ThothII/tools/thothctl/internal/workspaceops/operations.go
T

454 lines
13 KiB
Go

// Package workspaceops defines the deliberately closed host workspace contract.
package workspaceops
import (
"bytes"
"context"
"crypto/sha256"
"encoding/hex"
"encoding/json"
"errors"
"fmt"
"io"
"regexp"
"strings"
"github.com/aritmolab/thothii/tools/thothctl/internal/compose"
"github.com/aritmolab/thothii/tools/thothctl/internal/config"
"github.com/aritmolab/thothii/tools/thothctl/internal/safeio"
)
const CodeRegistryBootstrapRecoveryConflict = "registry_bootstrap_recovery_conflict"
const (
maxSQLFile = 1 << 20
maxSQLTotal = 16 << 20
maxAssumptions = 256
maxAssumptionBytes = 256
maxResult = 1 << 20
maxCandidate = 700 << 10
)
var workspaceIDPattern = regexp.MustCompile(`^[a-z][a-z0-9-]{2,62}$`)
var runIDPattern = regexp.MustCompile(`^[0-9a-f]{32}$`)
var digestPattern = regexp.MustCompile(`^sha256:[0-9a-f]{64}$`)
type Command interface{ workspaceCommand() }
type InspectCommand struct {
WorkspaceID string
JSON bool
}
type DwhRequest struct {
WorkspaceID string
Resume string
JSON bool
}
type SuggestFksRequest struct {
WorkspaceID string
FromSQL []string
Assume []string
Output string
JSON bool
}
type CheckSchemaRequest struct {
WorkspaceID string
Resume string
Annotations string
ReviewedCandidates string
JSON bool
}
type IndexSchemaRequest struct {
WorkspaceID string
JSON bool
}
type EvidenceRequest struct {
WorkspaceID string
DryRun bool
Resume string
JSON bool
}
type RunRequest struct {
WorkspaceID string
Resume string
JSON bool
}
func (InspectCommand) workspaceCommand() {}
func (DwhRequest) workspaceCommand() {}
func (SuggestFksRequest) workspaceCommand() {}
func (CheckSchemaRequest) workspaceCommand() {}
func (IndexSchemaRequest) workspaceCommand() {}
func (EvidenceRequest) workspaceCommand() {}
func (RunRequest) workspaceCommand() {}
type Result struct {
SchemaVersion int `json:"schemaVersion"`
Status string `json:"status"`
Code string `json:"code"`
WorkspaceID string `json:"workspaceId"`
WorkspaceRevision string `json:"workspaceRevision"`
DescriptorBlob string `json:"descriptorBlob"`
Operation string `json:"operation"`
RunID string `json:"runId,omitempty"`
ChildRuns map[string]string `json:"childRuns,omitempty"`
CompletedStages []string `json:"completedStages"`
Counts map[string]int `json:"counts,omitempty"`
ArtifactIdentities []ArtifactIdentity `json:"artifactIdentities,omitempty"`
Warnings []string `json:"warnings,omitempty"`
}
type ArtifactIdentity struct {
Kind string `json:"kind"`
Digest string `json:"digest"`
}
func invalid(msg string) (Command, error) { return nil, errors.New(msg) }
func requireWorkspace(v string) error {
if !workspaceIDPattern.MatchString(v) {
return errors.New("--workspace must be a valid workspace ID")
}
return nil
}
func requireRun(v string) error {
if !runIDPattern.MatchString(v) {
return errors.New("--resume must be a lowercase 32-hex run ID")
}
return nil
}
func one(args []string, i *int, flag string) (string, error) {
if *i+1 >= len(args) || strings.HasPrefix(args[*i+1], "--") {
return "", fmt.Errorf("%s requires a value", flag)
}
*i = *i + 1
return args[*i], nil
}
func ParseWorkspaceCommand(args []string) (Command, error) {
if len(args) < 2 || args[0] != "workspace" {
return invalid("workspace command is required")
}
area, action := args[1], ""
rest := args[2:]
if area == "preprocess" || area == "schema" {
if len(rest) == 0 {
return invalid("workspace group requires an operation")
}
action = rest[0]
rest = rest[1:]
} else {
action = area
}
if action != "inspect" && action != "dwh" && action != "suggest-fks" && action != "check" && action != "index-schema" && action != "evidence" && action != "run" {
return invalid("unknown workspace command")
}
var ws, resume, annotations, reviewed, output string
var jsonOut, dry bool
var sql, assume []string
seen := map[string]bool{}
for i := 0; i < len(rest); i++ {
f := rest[i]
if f == "--json" {
if seen[f] {
return invalid("duplicate --json")
}
seen[f] = true
jsonOut = true
continue
}
switch f {
case "--workspace":
if seen[f] {
return invalid("duplicate --workspace")
}
seen[f] = true
v, e := one(rest, &i, f)
if e != nil {
return nil, e
}
ws = v
case "--resume":
if seen[f] {
return invalid("duplicate --resume")
}
seen[f] = true
v, e := one(rest, &i, f)
if e != nil {
return nil, e
}
resume = v
case "--from-sql":
if action != "suggest-fks" {
return invalid("--from-sql is valid only for schema suggest-fks")
}
v, e := one(rest, &i, f)
if e != nil {
return nil, e
}
if len(sql) >= 32 {
return invalid("too many --from-sql files")
}
sql = append(sql, v)
case "--assume":
if action != "suggest-fks" {
return invalid("--assume is valid only for schema suggest-fks")
}
v, e := one(rest, &i, f)
if e != nil {
return nil, e
}
if len(assume) >= maxAssumptions || len([]byte(v)) > maxAssumptionBytes {
return invalid("--assume exceeds limit")
}
assume = append(assume, v)
case "--output":
if action != "suggest-fks" {
return invalid("--output is valid only for schema suggest-fks")
}
if seen[f] {
return invalid("duplicate --output")
}
seen[f] = true
v, e := one(rest, &i, f)
if e != nil {
return nil, e
}
output = v
case "--annotations":
if action != "check" {
return invalid("--annotations is valid only for schema check")
}
if seen[f] {
return invalid("duplicate --annotations")
}
seen[f] = true
v, e := one(rest, &i, f)
if e != nil {
return nil, e
}
annotations = v
case "--reviewed-candidates":
if action != "check" {
return invalid("--reviewed-candidates is valid only for schema check")
}
if seen[f] {
return invalid("duplicate --reviewed-candidates")
}
seen[f] = true
v, e := one(rest, &i, f)
if e != nil {
return nil, e
}
reviewed = v
case "--dry-run":
if action != "evidence" {
return invalid("--dry-run is valid only for evidence")
}
if seen[f] {
return invalid("duplicate --dry-run")
}
seen[f] = true
dry = true
default:
return invalid("unknown workspace option")
}
}
if err := requireWorkspace(ws); err != nil {
return nil, err
}
if action != "inspect" && resume != "" {
if err := requireRun(resume); err != nil {
return nil, err
}
}
if resume != "" && action != "dwh" && action != "evidence" && action != "run" && action != "check" {
return invalid("--resume is not valid for this workspace command")
}
if action == "inspect" && resume != "" {
return invalid("inspect does not accept --resume")
}
if action == "check" {
if resume == "" {
return invalid("schema check requires --resume")
}
if (annotations == "") != (reviewed == "") {
return invalid("--annotations and --reviewed-candidates must be supplied together")
}
if reviewed != "" && !digestPattern.MatchString(reviewed) {
return invalid("--reviewed-candidates must be sha256:<hex>")
}
}
if action != "suggest-fks" && (len(sql) > 0 || len(assume) > 0 || output != "") {
return invalid("schema options are valid only for suggest-fks")
}
switch action {
case "inspect":
return InspectCommand{ws, jsonOut}, nil
case "dwh":
return DwhRequest{ws, resume, jsonOut}, nil
case "suggest-fks":
return SuggestFksRequest{ws, sql, assume, output, jsonOut}, nil
case "check":
return CheckSchemaRequest{ws, resume, annotations, reviewed, jsonOut}, nil
case "index-schema":
return IndexSchemaRequest{ws, jsonOut}, nil
case "evidence":
return EvidenceRequest{ws, dry, resume, jsonOut}, nil
default:
return RunRequest{ws, resume, jsonOut}, nil
}
}
type inputEnvelope struct {
SchemaVersion int `json:"schemaVersion"`
Operation string `json:"operation"`
WorkspaceID string `json:"workspaceId"`
Resume string `json:"resume,omitempty"`
SQL []sqlInput `json:"sql,omitempty"`
Assume []string `json:"assume,omitempty"`
Annotations string `json:"annotations,omitempty"`
ReviewedCandidates string `json:"reviewedCandidates,omitempty"`
DryRun bool `json:"dryRun,omitempty"`
}
type sqlInput struct {
Name string `json:"name"`
Content string `json:"content"`
}
func operationName(c Command) string {
switch c.(type) {
case InspectCommand:
return "inspect"
case DwhRequest:
return "dwh"
case SuggestFksRequest:
return "suggest-fks"
case CheckSchemaRequest:
return "check"
case IndexSchemaRequest:
return "index-schema"
case EvidenceRequest:
return "evidence"
default:
return "run"
}
}
func Run(ctx context.Context, installation config.Installation, runner compose.Runner, command Command, stdin io.Reader) (Result, error) {
env := inputEnvelope{SchemaVersion: 1, Operation: operationName(command)}
var outputPath string
switch c := command.(type) {
case InspectCommand:
env.WorkspaceID = c.WorkspaceID
case DwhRequest:
env.WorkspaceID, env.Resume = c.WorkspaceID, c.Resume
case SuggestFksRequest:
env.WorkspaceID, env.Assume, outputPath = c.WorkspaceID, c.Assume, c.Output
if outputPath != "" {
if err := safeio.ValidateCanonicalOutputPath(outputPath); err != nil {
return Result{}, errors.New("unsafe output file")
}
}
var total int64
for _, path := range c.FromSQL {
b, e := safeio.ReadCanonicalUTF8(path, maxSQLFile)
if e != nil {
return Result{}, errors.New("unsafe SQL input")
}
total += int64(len(b))
if total > maxSQLTotal {
return Result{}, errors.New("SQL input exceeds limit")
}
env.SQL = append(env.SQL, sqlInput{path, string(b)})
}
if len(env.SQL) > 0 {
var n int
for _, x := range env.SQL {
n += len(x.Content)
}
if n > maxCandidate {
return Result{}, errors.New("candidate input exceeds limit")
}
}
case CheckSchemaRequest:
env.WorkspaceID, env.Resume, env.Annotations, env.ReviewedCandidates = c.WorkspaceID, c.Resume, c.Annotations, c.ReviewedCandidates
if c.Annotations != "" {
b, e := safeio.ReadCanonicalUTF8(c.Annotations, 16<<20)
if e != nil {
return Result{}, errors.New("unsafe annotation input")
}
env.Annotations = string(b)
}
case IndexSchemaRequest:
env.WorkspaceID = c.WorkspaceID
case EvidenceRequest:
env.WorkspaceID, env.Resume, env.DryRun = c.WorkspaceID, c.Resume, c.DryRun
case RunRequest:
env.WorkspaceID, env.Resume = c.WorkspaceID, c.Resume
default:
return Result{}, errors.New("unsupported workspace command")
}
payload, e := json.Marshal(env)
if e != nil {
return Result{}, e
}
if stdin != nil {
payload, e = io.ReadAll(io.LimitReader(stdin, 1<<20+1))
if e != nil {
return Result{}, e
}
if len(payload) > 1<<20 {
return Result{}, errors.New("request exceeds limit")
}
var supplied inputEnvelope
d := json.NewDecoder(bytes.NewReader(payload))
d.DisallowUnknownFields()
if e = d.Decode(&supplied); e != nil || supplied.SchemaVersion != 1 || supplied.Operation != env.Operation || supplied.WorkspaceID != env.WorkspaceID {
return Result{}, errors.New("invalid workspace request")
}
}
args := installation.ComposeArgs("run", "--rm", "--no-deps", "workspace-maintenance", "--operation", operationName(command), "--workspace", env.WorkspaceID)
cr, runErr := runner.RunBounded(ctx, args, bytes.NewReader(payload), compose.CaptureLimits{StdoutBytes: 1 << 20, StderrBytes: 64 << 10})
if runErr != nil && (errors.Is(runErr, compose.ErrOutputLimit) || cr.Stdout == "") {
return Result{}, runErr
}
if len(cr.Stdout) > maxResult {
return Result{}, errors.New("invalid workspace result")
}
var result Result
dec := json.NewDecoder(strings.NewReader(cr.Stdout))
dec.DisallowUnknownFields()
if e := dec.Decode(&result); e != nil {
return Result{}, errors.New("invalid workspace result")
}
var extra any
if e := dec.Decode(&extra); e != io.EOF {
return Result{}, errors.New("invalid workspace result")
}
if result.SchemaVersion != 1 || result.WorkspaceID != env.WorkspaceID || result.Operation != operationName(command) || !validStatus(result.Status) || !validCode(result.Code) {
return Result{}, errors.New("invalid workspace result")
}
if outputPath != "" {
if len(cr.Stdout) > maxCandidate {
return Result{}, errors.New("candidate export exceeds limit")
}
if err := safeio.WriteCanonicalExclusive(outputPath, []byte(cr.Stdout), 0o600); err != nil {
return Result{}, errors.New("unsafe output file")
}
}
return result, nil
}
func DigestBytes(b []byte) string { s := sha256.Sum256(b); return "sha256:" + hex.EncodeToString(s[:]) }
func validStatus(v string) bool {
switch v {
case "succeeded", "unchanged", "dry_run", "blocked", "failed":
return true
}
return false
}
func validCode(v string) bool {
switch v {
case "ok", "workspace_not_found", "workspace_not_activatable", "binding_missing", "preprocessing_conflict", "preprocessing_resume_mismatch", "manual_review_required", "evidence_materialization_required", "effective_config_mismatch", "semantic_index_incompatible", "annotation_invalid", "egress_policy_refused", CodeRegistryBootstrapRecoveryConflict:
return true
}
return false
}