| 1 | package integrations |
| 2 | |
| 3 | import ( |
| 4 | "bytes" |
| 5 | "encoding/json" |
| 6 | "errors" |
| 7 | "fmt" |
| 8 | "io" |
| 9 | "net/http" |
| 10 | "net/url" |
| 11 | "os" |
| 12 | "regexp" |
| 13 | "strings" |
| 14 | "time" |
| 15 | ) |
| 16 | |
| 17 | // The executor: runs one manifest op against the service's pinned host. The |
| 18 | // manifest is reviewed code; the args and token are member input — so every |
| 19 | // arg is bounded and escaped into the slot its template put it in, and the |
| 20 | // final URL is re-checked against auth.base's host before anything is sent. |
| 21 | // A value can steer WHAT an op says, never WHERE it goes. |
| 22 | |
| 23 | const ( |
| 24 | maxArgs = 16 |
| 25 | maxArgBytes = 2048 |
| 26 | maxRespBody = 256 << 10 // response read cap for un-projected ops |
| 27 | // A projected op may read much more: provider list responses run to |
| 28 | // hundreds of KB raw (a real /user/repos page is ~400KB), and the pick |
| 29 | // throws almost all of it away immediately — the projected result is what |
| 30 | // travels, not this buffer. |
| 31 | maxPickedRespBody = 4 << 20 |
| 32 | execTimeout = 20 * time.Second |
| 33 | ) |
| 34 | |
| 35 | // purePlaceholderRE matches a template that is EXACTLY one placeholder — the |
| 36 | // shape of an optional-arg slot like `"body": "{body}"`, whose key is dropped |
| 37 | // from the JSON when the arg is absent (so optional means absent, not ""). |
| 38 | var purePlaceholderRE = regexp.MustCompile(`^\{[a-z0-9_]+\}$`) |
| 39 | |
| 40 | // StatusError is a non-2xx provider response. The message keeps the exact |
| 41 | // shape errors always had; the code lets a caller distinguish "token expired" |
| 42 | // (401 → refresh and retry) from "the request was wrong". |
| 43 | type StatusError struct { |
| 44 | Service, Op string |
| 45 | Code int |
| 46 | Detail string |
| 47 | } |
| 48 | |
| 49 | func (e *StatusError) Error() string { |
| 50 | return fmt.Sprintf("%s.%s: status %d: %s", e.Service, e.Op, e.Code, e.Detail) |
| 51 | } |
| 52 | |
| 53 | // baseOverride returns the test-only base URL for a service, mirroring how |
| 54 | // LCHAT_DNSIMPLE_BASE points the dnsimple client at a loopback httptest |
| 55 | // server (internal/bot/dnsimple.go). Never set in production units. |
| 56 | func baseOverride(id string) string { |
| 57 | return os.Getenv("LCHAT_INTEGRATIONS_BASE_" + strings.ToUpper(strings.ReplaceAll(id, "-", "_"))) |
| 58 | } |
| 59 | |
| 60 | // Execute runs one op with the member's args and token, returning the |
| 61 | // response body as a string. Errors are values — the sidecar feeds them back |
| 62 | // into the thread as an honest failure, never a silent gap. |
| 63 | func Execute(m *Manifest, opID string, args map[string]string, token string) (string, error) { |
| 64 | return ExecuteAs(m, opID, args, token, "") |
| 65 | } |
| 66 | |
| 67 | // ExecuteAs is Execute plus the credential's kind ("pat" | "oauth"; "" for |
| 68 | // legacy credentials stored before kinds existed). An OAuth token is a Bearer |
| 69 | // token by definition (RFC 6750), whatever scheme the manifest picked for |
| 70 | // pasted keys — Linear's personal API keys go bare while its OAuth tokens |
| 71 | // carry Bearer. Any other kind keeps the manifest's scheme. |
| 72 | func ExecuteAs(m *Manifest, opID string, args map[string]string, token, tokenKind string) (string, error) { |
| 73 | op := m.Op(opID) |
| 74 | if op == nil { |
| 75 | // Absence is refusal: an op the manifest doesn't declare cannot run. |
| 76 | return "", fmt.Errorf("%s: unknown op %q", m.ID, opID) |
| 77 | } |
| 78 | if err := validateArgs(op, args); err != nil { |
| 79 | return "", fmt.Errorf("%s.%s: %w", m.ID, opID, err) |
| 80 | } |
| 81 | |
| 82 | base := m.Auth.Base |
| 83 | override := baseOverride(m.ID) |
| 84 | if override != "" { |
| 85 | base = override |
| 86 | } |
| 87 | baseURL, err := url.Parse(base) |
| 88 | if err != nil || baseURL.Host == "" { |
| 89 | return "", fmt.Errorf("%s.%s: bad base %q", m.ID, opID, base) |
| 90 | } |
| 91 | |
| 92 | final := strings.TrimRight(base, "/") + renderPath(op.Path, args) |
| 93 | u, err := url.Parse(final) |
| 94 | if err != nil { |
| 95 | return "", fmt.Errorf("%s.%s: bad url: %w", m.ID, opID, err) |
| 96 | } |
| 97 | // Pin the host: re-check the URL we actually built, not just the inputs — |
| 98 | // if escaping ever failed to contain a value, this is the backstop. |
| 99 | // Skipped only under the env override (loopback test servers are http). |
| 100 | if override == "" { |
| 101 | if u.Scheme != "https" || u.Host != baseURL.Host { |
| 102 | return "", fmt.Errorf("%s.%s: url escaped its pinned host", m.ID, opID) |
| 103 | } |
| 104 | } |
| 105 | |
| 106 | var body io.Reader |
| 107 | contentType := "" |
| 108 | switch { |
| 109 | case op.Encoding == "form": |
| 110 | vals := url.Values{} |
| 111 | for k, tmpl := range op.BodyForm { |
| 112 | vals.Set(k, substitute(tmpl, args)) |
| 113 | } |
| 114 | body = strings.NewReader(vals.Encode()) |
| 115 | contentType = "application/x-www-form-urlencoded" |
| 116 | case op.BodyJSON != nil: |
| 117 | rendered := substituteJSONMap(op.BodyJSON, args) |
| 118 | raw, err := json.Marshal(rendered) |
| 119 | if err != nil { |
| 120 | return "", fmt.Errorf("%s.%s: body: %w", m.ID, opID, err) |
| 121 | } |
| 122 | body = bytes.NewReader(raw) |
| 123 | contentType = "application/json" |
| 124 | } |
| 125 | |
| 126 | req, err := http.NewRequest(op.Method, u.String(), body) |
| 127 | if err != nil { |
| 128 | return "", fmt.Errorf("%s.%s: %w", m.ID, opID, err) |
| 129 | } |
| 130 | // Default Accept first, then the manifest's headers — a manifest Accept |
| 131 | // (GitHub's vnd header) wins over the default, never the reverse. |
| 132 | req.Header.Set("Accept", "application/json") |
| 133 | for k, v := range m.Auth.Headers { |
| 134 | req.Header.Set(k, v) |
| 135 | } |
| 136 | scheme := m.Auth.Authorization |
| 137 | if scheme == "" { |
| 138 | scheme = "Bearer" |
| 139 | } |
| 140 | if tokenKind == "oauth" { |
| 141 | scheme = "Bearer" |
| 142 | } |
| 143 | if scheme == "none" { |
| 144 | // Bare-token APIs: Linear rejects a Bearer prefix on personal API |
| 145 | // keys — the header value is the token alone. |
| 146 | req.Header.Set("Authorization", token) |
| 147 | } else { |
| 148 | req.Header.Set("Authorization", scheme+" "+token) |
| 149 | } |
| 150 | if contentType != "" { |
| 151 | req.Header.Set("Content-Type", contentType) |
| 152 | } |
| 153 | |
| 154 | client := &http.Client{ |
| 155 | Timeout: execTimeout, |
| 156 | CheckRedirect: func(req *http.Request, via []*http.Request) error { |
| 157 | if len(via) >= 5 { |
| 158 | return errors.New("too many redirects") |
| 159 | } |
| 160 | // A redirect is a new URL: hold it to the same pin. Off-host |
| 161 | // redirects are refused outright — an integration API that |
| 162 | // bounces us elsewhere is outside the manifest's contract. |
| 163 | if req.URL.Host != u.Host { |
| 164 | return errors.New("redirect off the pinned host blocked") |
| 165 | } |
| 166 | if override == "" && req.URL.Scheme != "https" { |
| 167 | return errors.New("redirect downgraded from https blocked") |
| 168 | } |
| 169 | return nil |
| 170 | }, |
| 171 | } |
| 172 | resp, err := client.Do(req) |
| 173 | if err != nil { |
| 174 | return "", fmt.Errorf("%s.%s: %w", m.ID, opID, err) |
| 175 | } |
| 176 | defer resp.Body.Close() |
| 177 | readCap := int64(maxRespBody) |
| 178 | if op.Pick != nil { |
| 179 | readCap = maxPickedRespBody |
| 180 | } |
| 181 | raw, err := io.ReadAll(io.LimitReader(resp.Body, readCap+1)) |
| 182 | if err != nil { |
| 183 | return "", fmt.Errorf("%s.%s: read: %w", m.ID, opID, err) |
| 184 | } |
| 185 | // Over the cap: refuse honestly — a silently truncated body would surface |
| 186 | // as a baffling JSON parse error (or worse, truncated data downstream). |
| 187 | if int64(len(raw)) > readCap { |
| 188 | return "", fmt.Errorf("%s.%s: response too large (over %d bytes)", m.ID, opID, readCap) |
| 189 | } |
| 190 | if resp.StatusCode/100 != 2 { |
| 191 | snippet := raw |
| 192 | if len(snippet) > 300 { |
| 193 | snippet = snippet[:300] |
| 194 | } |
| 195 | // Typed so a caller can react to the code (the sidecar refreshes an |
| 196 | // expired token on a 401) without parsing the message. |
| 197 | return "", &StatusError{Service: m.ID, Op: opID, Code: resp.StatusCode, Detail: string(snippet)} |
| 198 | } |
| 199 | if op.Pick != nil { |
| 200 | projected, err := project(raw, op.Pick) |
| 201 | if err != nil { |
| 202 | return "", fmt.Errorf("%s.%s: %w", m.ID, opID, err) |
| 203 | } |
| 204 | return projected, nil |
| 205 | } |
| 206 | return string(raw), nil |
| 207 | } |
| 208 | |
| 209 | // project applies an op's declared response projection (registry.go's Pick): |
| 210 | // descend the root path, then keep only the named fields of the object — or of |
| 211 | // each object of the array — found there. The manifest is the review surface, |
| 212 | // so a response that doesn't match its declared shape is an honest error, never |
| 213 | // a silent empty result. |
| 214 | func project(raw []byte, p *Pick) (string, error) { |
| 215 | var v any |
| 216 | if err := json.Unmarshal(raw, &v); err != nil { |
| 217 | return "", fmt.Errorf("response isn't JSON: %w", err) |
| 218 | } |
| 219 | for _, key := range splitPath(p.Root) { |
| 220 | obj, ok := v.(map[string]any) |
| 221 | if !ok { |
| 222 | return "", fmt.Errorf("response has no %q envelope", p.Root) |
| 223 | } |
| 224 | if v, ok = obj[key]; !ok { |
| 225 | return "", fmt.Errorf("response has no %q envelope", p.Root) |
| 226 | } |
| 227 | } |
| 228 | if len(p.Fields) > 0 { |
| 229 | switch t := v.(type) { |
| 230 | case []any: |
| 231 | out := make([]any, 0, len(t)) |
| 232 | for _, e := range t { |
| 233 | if obj, ok := e.(map[string]any); ok { |
| 234 | out = append(out, pickFields(obj, p.Fields)) |
| 235 | } |
| 236 | } |
| 237 | v = out |
| 238 | case map[string]any: |
| 239 | v = pickFields(t, p.Fields) |
| 240 | default: |
| 241 | return "", fmt.Errorf("response %q isn't an object or array", p.Root) |
| 242 | } |
| 243 | } |
| 244 | out, err := json.Marshal(v) |
| 245 | if err != nil { |
| 246 | return "", err |
| 247 | } |
| 248 | return string(out), nil |
| 249 | } |
| 250 | |
| 251 | // pickFields projects one object down to the named dot-paths. A field the |
| 252 | // response doesn't carry is omitted (provider APIs grow and shrink; the client |
| 253 | // treats absence as "—"). The output key is the full dot-path, flat — so |
| 254 | // "user.login" comes back as {"user.login": "octocat"}. |
| 255 | func pickFields(obj map[string]any, fields []string) map[string]any { |
| 256 | out := make(map[string]any, len(fields)) |
| 257 | for _, f := range fields { |
| 258 | v := any(obj) |
| 259 | found := true |
| 260 | for _, key := range splitPath(f) { |
| 261 | m, ok := v.(map[string]any) |
| 262 | if !ok { |
| 263 | found = false |
| 264 | break |
| 265 | } |
| 266 | if v, ok = m[key]; !ok { |
| 267 | found = false |
| 268 | break |
| 269 | } |
| 270 | } |
| 271 | if found { |
| 272 | out[f] = v |
| 273 | } |
| 274 | } |
| 275 | return out |
| 276 | } |
| 277 | |
| 278 | // splitPath splits a validated pick path into keys; "" is the empty path. |
| 279 | func splitPath(p string) []string { |
| 280 | if p == "" { |
| 281 | return nil |
| 282 | } |
| 283 | return strings.Split(p, ".") |
| 284 | } |
| 285 | |
| 286 | // validateArgs bounds member input before any of it touches a template: |
| 287 | // declared keys only, required keys present, values short and free of |
| 288 | // control characters. |
| 289 | func validateArgs(op *Op, args map[string]string) error { |
| 290 | if len(args) > maxArgs { |
| 291 | return fmt.Errorf("too many args (%d > %d)", len(args), maxArgs) |
| 292 | } |
| 293 | for k, v := range args { |
| 294 | if _, ok := op.Args[k]; !ok { |
| 295 | return fmt.Errorf("undeclared arg %q", k) |
| 296 | } |
| 297 | if len(v) > maxArgBytes { |
| 298 | return fmt.Errorf("arg %q too long (%d bytes)", k, len(v)) |
| 299 | } |
| 300 | for i := 0; i < len(v); i++ { |
| 301 | if v[i] < 0x20 || v[i] == 0x7f { |
| 302 | return fmt.Errorf("arg %q contains control characters", k) |
| 303 | } |
| 304 | } |
| 305 | } |
| 306 | for k, spec := range op.Args { |
| 307 | if spec.Required && args[k] == "" { |
| 308 | return fmt.Errorf("missing required arg %q", k) |
| 309 | } |
| 310 | } |
| 311 | return nil |
| 312 | } |
| 313 | |
| 314 | // renderPath substitutes `{arg}` into a path template, escaping each value |
| 315 | // for the slot it lands in: path-segment escaping before the first "?", |
| 316 | // query escaping after it. A value can never introduce an un-escaped |
| 317 | // `/`, `?`, `#`, or `&` — it stays inside its segment or query slot. |
| 318 | func renderPath(path string, args map[string]string) string { |
| 319 | pathPart, queryPart, hasQuery := strings.Cut(path, "?") |
| 320 | out := substituteWith(pathPart, args, escapePathValue) |
| 321 | if hasQuery { |
| 322 | out += "?" + substituteWith(queryPart, args, url.QueryEscape) |
| 323 | } |
| 324 | return out |
| 325 | } |
| 326 | |
| 327 | // escapePathValue is url.PathEscape plus the sub-delims PathEscape leaves |
| 328 | // alone (`&;=+,`). They're legal in a path segment, but our promise is |
| 329 | // stronger than legality: a substituted value must not be able to smuggle |
| 330 | // anything a server might parse as a delimiter. |
| 331 | func escapePathValue(v string) string { |
| 332 | return strings.NewReplacer( |
| 333 | "&", "%26", ";", "%3B", "=", "%3D", "+", "%2B", ",", "%2C", |
| 334 | ).Replace(url.PathEscape(v)) |
| 335 | } |
| 336 | |
| 337 | // substitute replaces `{arg}` with the raw arg value (bodies — no URL |
| 338 | // escaping; the encoder for the body format does its own). An absent |
| 339 | // optional arg substitutes to "". |
| 340 | func substitute(tmpl string, args map[string]string) string { |
| 341 | return substituteWith(tmpl, args, func(s string) string { return s }) |
| 342 | } |
| 343 | |
| 344 | func substituteWith(tmpl string, args map[string]string, escape func(string) string) string { |
| 345 | return placeholderRE.ReplaceAllStringFunc(tmpl, func(ph string) string { |
| 346 | return escape(args[ph[1:len(ph)-1]]) |
| 347 | }) |
| 348 | } |
| 349 | |
| 350 | // substituteJSONMap deep-copies a body_json template with `{arg}` substituted |
| 351 | // into string leaves. Keys whose template was exactly one placeholder and |
| 352 | // substituted to "" are dropped — an optional arg the member didn't give is |
| 353 | // omitted from the request, not sent as an empty string. |
| 354 | func substituteJSONMap(m map[string]any, args map[string]string) map[string]any { |
| 355 | out := make(map[string]any, len(m)) |
| 356 | for k, v := range m { |
| 357 | switch t := v.(type) { |
| 358 | case string: |
| 359 | s := substitute(t, args) |
| 360 | if s == "" && purePlaceholderRE.MatchString(t) { |
| 361 | continue |
| 362 | } |
| 363 | out[k] = s |
| 364 | case map[string]any: |
| 365 | out[k] = substituteJSONMap(t, args) |
| 366 | case []any: |
| 367 | out[k] = substituteJSONSlice(t, args) |
| 368 | default: |
| 369 | out[k] = v |
| 370 | } |
| 371 | } |
| 372 | return out |
| 373 | } |
| 374 | |
| 375 | func substituteJSONSlice(s []any, args map[string]string) []any { |
| 376 | out := make([]any, len(s)) |
| 377 | for i, v := range s { |
| 378 | switch t := v.(type) { |
| 379 | case string: |
| 380 | out[i] = substitute(t, args) |
| 381 | case map[string]any: |
| 382 | out[i] = substituteJSONMap(t, args) |
| 383 | case []any: |
| 384 | out[i] = substituteJSONSlice(t, args) |
| 385 | default: |
| 386 | out[i] = v |
| 387 | } |
| 388 | } |
| 389 | return out |
| 390 | } |
| 391 | |
amadan