import * as Context from "effect/Context";import * as Effect from "effect/Effect";import * as Layer from "effect/Layer";import * as Result from "effect/Result";import * as Schema from "effect/Schema";import type {  PullRequestAction,  PullRequestComment,  PullRequestInvolvement,  PullRequestListState,  PullRequestMergeMethod,} from "@t3tools/contracts"; import * as AzureDevOpsCli from "../sourceControl/AzureDevOpsCli.ts";import {  decodePullRequestJson,  decodePullRequestListJson,  decodeThreadsJson,  decodeViewerJson,  type AzureDevOpsPullRequest,} from "./azureDevOpsPullRequestJson.ts";import type { ProviderListCursor } from "./PullRequestProvider.ts"; /** * Names the read that produced unusable output, so a failure reports the call it came from * rather than borrowing another operation's message. */export class AzureDevOpsPullRequestReadError extends Schema.TaggedError<AzureDevOpsPullRequestReadError>()(  "AzureDevOpsPullRequestReadError",  {    command: Schema.Literal("az"),    cwd: Schema.String,    operation: Schema.String,    cause: Schema.Defect(),  },) {  get detail(): string {    return `Azure CLI returned an unreadable ${this.operation} response.`;  }   override get message(): string {    return `Azure CLI failed in ${this.operation}: ${this.detail}`;  }} /** Not a decode failure: az answered, the account it answered for just has no name. */export class AzureDevOpsViewerUnavailableError extends Schema.TaggedError<AzureDevOpsViewerUnavailableError>()(  "AzureDevOpsViewerUnavailableError",  {    command: Schema.Literal("az"),    cwd: Schema.String,  },) {  get detail(): string {    return "Azure CLI returned no account for the current sign-in.";  }   override get message(): string {    return `Azure CLI failed in getViewer: ${this.detail}`;  }} /** * Not a decode failure either: az answered with a well-formed pull request that simply carries * no branch or link, which is a response this cannot place rather than one it cannot read. */export class AzureDevOpsPullRequestIncompleteError extends Schema.TaggedError<AzureDevOpsPullRequestIncompleteError>()(  "AzureDevOpsPullRequestIncompleteError",  {    command: Schema.Literal("az"),    cwd: Schema.String,    number: Schema.Int,  },) {  get detail(): string {    return "Azure DevOps returned no branch or link for the pull request.";  }   override get message(): string {    return `Azure CLI failed in getPullRequest: ${this.detail}`;  }} /** * Not a decode failure: the reader named a reviewer `az` would read as a flag of its own. The * reviewers travel as argv rather than in a request body — `az repos pr reviewer` takes them no * other way — so anything that could leave the value position is refused rather than sent. */export class AzureDevOpsReviewerNameError extends Schema.TaggedError<AzureDevOpsReviewerNameError>()(  "AzureDevOpsReviewerNameError",  {    command: Schema.Literal("az"),    cwd: Schema.String,  },) {  get detail(): string {    return "A reviewer is named by an email address or an identity id.";  }   override get message(): string {    return `Azure CLI failed in setPullRequestReviewers: ${this.detail}`;  }} export type AzureDevOpsPullRequestCliError =  | AzureDevOpsCli.AzureDevOpsCliError  | AzureDevOpsPullRequestReadError  | AzureDevOpsPullRequestIncompleteError  | AzureDevOpsReviewerNameError  | AzureDevOpsViewerUnavailableError; /** The version every REST call below is pinned to, so a new default cannot reshape a response. */const REST_API_VERSION = "7.1";const PULL_REQUEST_LIST_MAX_OUTPUT_BYTES = 16 * 1024 * 1024; export class AzureDevOpsPullRequestCli extends Context.Service<  AzureDevOpsPullRequestCli,  {    readonly getViewer: (input: {      readonly cwd: string;    }) => Effect.Effect<string, AzureDevOpsPullRequestCliError>;     readonly listPullRequests: (input: {      readonly cwd: string;      readonly repository: string;      readonly state: PullRequestListState;      readonly involvement: PullRequestInvolvement;      readonly viewer: string;      readonly limit: number;      /**       * Where to carry on from. Azure has no date filter for a pull request listing, so the only       * part of a cursor it can use is how many rows have already been handed over.       */      readonly cursor?: ProviderListCursor | undefined;    }) => Effect.Effect<      {        readonly items: ReadonlyArray<AzureDevOpsPullRequest>;        readonly truncated: boolean;        /** Raw Azure rows consumed to produce this page, including malformed rows. */        readonly cursorAdvance: number;      },      AzureDevOpsPullRequestCliError    >;     readonly getPullRequest: (input: {      readonly cwd: string;      readonly number: number;    }) => Effect.Effect<AzureDevOpsPullRequest, AzureDevOpsPullRequestCliError>;     /** Threads are not reachable through `az repos pr`, so they come from the REST API. */    readonly listThreads: (input: {      readonly cwd: string;      readonly threadsUrl: string;    }) => Effect.Effect<ReadonlyArray<PullRequestComment>, AzureDevOpsPullRequestCliError>;     readonly runPullRequestAction: (input: {      readonly cwd: string;      readonly number: number;      readonly action: PullRequestAction;      readonly mergeMethod?: PullRequestMergeMethod;    }) => Effect.Effect<void, AzureDevOpsPullRequestCliError>;     /** Rewrites the pull request's own words, through the same command that moves it. */    readonly updatePullRequest: (input: {      readonly cwd: string;      readonly number: number;      readonly title?: string | undefined;      readonly body?: string | undefined;    }) => Effect.Effect<void, AzureDevOpsPullRequestCliError>;     /**     * Adds reviewers to a pull request, or takes them off it. `az repos pr reviewer` is the whole     * of what Azure offers here: it adds and removes named identities, and has no counterpart that     * says who could be named.     */    readonly setPullRequestReviewers: (input: {      readonly cwd: string;      readonly number: number;      readonly reviewers: ReadonlyArray<string>;      readonly requested: boolean;    }) => Effect.Effect<void, AzureDevOpsPullRequestCliError>;  }>()("t3/pullRequest/AzureDevOpsPullRequestCli") {} function statusArgs(state: PullRequestListState): ReadonlyArray<string> {  switch (state) {    case "open":      return ["--status", "active"];    case "merged":      return ["--status", "completed"];    case "closed":      return ["--status", "abandoned"];    case "all":      return ["--status", "all"];  }} function involvementArgs(input: {  readonly involvement: PullRequestInvolvement;  readonly viewer: string;}): ReadonlyArray<string> {  switch (input.involvement) {    case "authored":      return ["--creator", input.viewer];    case "reviewing":      return ["--reviewer", input.viewer];    case "all":      return [];  }} /** * Azure moves a pull request by setting its state rather than by named commands: completing it * is the merge, abandoning it is the close, and reactivating it is the reopen. Squashing is a * completion option rather than a strategy of its own. */function actionArgs(  action: PullRequestAction,  mergeMethod: PullRequestMergeMethod | undefined,): ReadonlyArray<string> {  switch (action) {    case "merge":      return ["--status", "completed", "--squash", mergeMethod === "squash" ? "true" : "false"];    // Auto-complete is Azure's own name for it: the pull request stays active and Azure completes    // it once its policies pass. The squash choice is stored with it, as it is for a merge now.    case "enable-auto-merge":      return [        "--auto-complete",        "true",        ...(mergeMethod === undefined          ? []          : ["--squash", mergeMethod === "squash" ? "true" : "false"]),      ];    case "disable-auto-merge":      return ["--auto-complete", "false"];    case "ready":      return ["--draft", "false"];    case "draft":      return ["--draft", "true"];    case "close":      return ["--status", "abandoned"];    // Never reached: this host does not declare the action, so nothing offers it.    case "update-branch":      return [];    case "reopen":      return ["--status", "active"];    // Never reached: this host does not declare the action, so the service refuses it first.    case "revert":    case "approve-workflows":      throw new Error(`Azure DevOps pull request action ${action} is unsupported`);  }} /** * A reviewer Azure could be given: an email address, a display name or an identity guid, and * nothing that starts with a dash. The dash is the whole point — these are argv, and a value that * looks like a flag stops being a value. */function isReviewerName(value: string): boolean {  const name = value.trim();  return name.length > 0 && !name.startsWith("-");} /** @public Service construction is part of the canonical Effect module API. */export const make = Effect.gen(function* () {  const azure = yield* AzureDevOpsCli.AzureDevOpsCli;   // Every command resolves the organization, project and repository from the checkout, which is  // what the rest of the Azure wrapper does. The remote takes three shapes and only `az` knows  // how to read all of them.  const detectArgs = ["--detect", "true"] as const;   const executeJson = (input: {    readonly cwd: string;    readonly args: ReadonlyArray<string>;    readonly maxOutputBytes?: number;  }) =>    azure.execute({      cwd: input.cwd,      args: [...input.args, "--only-show-errors", "--output", "json"],      ...(input.maxOutputBytes === undefined ? {} : { maxOutputBytes: input.maxOutputBytes }),    });   /**   * Azure pages by raw offset. Keep reading when malformed rows leave the decoded page short, and   * retain the raw count so the next public cursor skips every row this walk consumed.   */  const listPullRequestPage = (input: {    readonly cwd: string;    readonly repository: string;    readonly state: PullRequestListState;    readonly involvement: PullRequestInvolvement;    readonly viewer: string;    readonly limit: number;    readonly skip: number;    readonly cursorAdvance: number;    readonly items: ReadonlyArray<AzureDevOpsPullRequest>;  }): Effect.Effect<    {      readonly items: ReadonlyArray<AzureDevOpsPullRequest>;      readonly truncated: boolean;      readonly cursorAdvance: number;    },    AzureDevOpsPullRequestCliError  > => {    const remaining = input.limit - input.items.length;    const top = remaining + 1;    return executeJson({      cwd: input.cwd,      maxOutputBytes: PULL_REQUEST_LIST_MAX_OUTPUT_BYTES,      args: [        "repos",        "pr",        "list",        ...detectArgs,        "--repository",        input.repository,        ...statusArgs(input.state),        ...involvementArgs(input),        // A web link per row, which is the only url that needs no assembling.        "--include-links",        ...(input.skip === 0 ? [] : ["--skip", String(input.skip)]),        "--top",        String(top),      ],    }).pipe(      Effect.flatMap((result) => {        const raw = result.stdout.trim();        if (raw.length === 0) {          return Effect.succeed({            items: input.items,            truncated: false,            cursorAdvance: input.cursorAdvance,          });        }        const decoded = decodePullRequestListJson(raw);        if (!Result.isSuccess(decoded)) {          return Effect.fail(            new AzureDevOpsPullRequestReadError({              command: "az",              cwd: input.cwd,              operation: "listPullRequests",              cause: decoded.failure,            }),          );        }         const lastItemIndex = decoded.success.rawIndexes[remaining - 1];        if (lastItemIndex !== undefined) {          const consumed = lastItemIndex + 1;          return Effect.succeed({            items: [...input.items, ...decoded.success.items.slice(0, remaining)],            // A full raw response may have more rows even when malformed entries used the probe.            truncated: consumed < decoded.success.rawCount || decoded.success.rawCount === top,            cursorAdvance: input.cursorAdvance + consumed,          });        }         const items = [...input.items, ...decoded.success.items];        if (decoded.success.rawCount < top) {          return Effect.succeed({            items,            truncated: false,            cursorAdvance: input.cursorAdvance + decoded.success.rawCount,          });        }        return listPullRequestPage({          ...input,          skip: input.skip + decoded.success.rawCount,          cursorAdvance: input.cursorAdvance + decoded.success.rawCount,          items,        });      }),    );  };   return AzureDevOpsPullRequestCli.of({    getViewer: (input) =>      executeJson({ cwd: input.cwd, args: ["account", "show", "--query", "user"] }).pipe(        Effect.flatMap((result): Effect.Effect<string, AzureDevOpsPullRequestCliError> => {          // `--query user` narrows the payload to the account, so it is nested back under the          // key the decoder reads to keep one shape for the signed-in user.          const decoded = decodeViewerJson(`{"user":${result.stdout.trim() || "null"}}`);          if (!Result.isSuccess(decoded)) {            return Effect.fail(              new AzureDevOpsPullRequestReadError({                command: "az",                cwd: input.cwd,                operation: "getViewer",                cause: decoded.failure,              }),            );          }          return decoded.success === null            ? Effect.fail(new AzureDevOpsViewerUnavailableError({ command: "az", cwd: input.cwd }))            : Effect.succeed(decoded.success);        }),      ),     listPullRequests: (input) =>      listPullRequestPage({        cwd: input.cwd,        repository: input.repository,        state: input.state,        involvement: input.involvement,        viewer: input.viewer,        limit: input.limit,        // Azure counts rather than filters, so a slice carries on by stepping over every raw row        // the prior slice consumed. That is an offset into a list that can shift underneath it:        // a pull request opened between two slices moves everything down one, and the row on the        // seam is the one that pays for it.        skip: input.cursor?.delivered ?? 0,        cursorAdvance: 0,        items: [],      }),     getPullRequest: (input) =>      executeJson({        cwd: input.cwd,        args: ["repos", "pr", "show", ...detectArgs, "--id", String(input.number)],      }).pipe(        Effect.flatMap(          (result): Effect.Effect<AzureDevOpsPullRequest, AzureDevOpsPullRequestCliError> => {            const decoded = decodePullRequestJson(result.stdout.trim());            if (!Result.isSuccess(decoded)) {              return Effect.fail(                new AzureDevOpsPullRequestReadError({                  command: "az",                  cwd: input.cwd,                  operation: "getPullRequest",                  cause: decoded.failure,                }),              );            }            // Null means Azure answered with too little to place the pull request. Nothing            // failed underneath it, so it is its own outcome rather than a decode failure.            return decoded.success === null              ? Effect.fail(                  new AzureDevOpsPullRequestIncompleteError({                    command: "az",                    cwd: input.cwd,                    number: input.number,                  }),                )              : Effect.succeed(decoded.success);          },        ),      ),     listThreads: (input) =>      executeJson({        cwd: input.cwd,        args: [          "rest",          "--method",          "get",          "--url",          `${input.threadsUrl}?api-version=${REST_API_VERSION}`,        ],      }).pipe(        Effect.flatMap((result) => {          const decoded = decodeThreadsJson(result.stdout.trim());          return Result.isSuccess(decoded)            ? Effect.succeed(decoded.success)            : Effect.fail(                new AzureDevOpsPullRequestReadError({                  command: "az",                  cwd: input.cwd,                  operation: "listThreads",                  cause: decoded.failure,                }),              );        }),      ),     setPullRequestReviewers: (input) =>      input.reviewers.some((reviewer) => !isReviewerName(reviewer))        ? Effect.fail(new AzureDevOpsReviewerNameError({ command: "az", cwd: input.cwd }))        : azure            .execute({              cwd: input.cwd,              args: [                "repos",                "pr",                "reviewer",                input.requested ? "add" : "remove",                ...detectArgs,                "--id",                String(input.number),                // One `--reviewers` takes them all, because az reads the flag as a list and a                // second one would replace the first rather than add to it.                "--reviewers",                ...input.reviewers,                "--only-show-errors",                "--output",                "json",              ],            })            .pipe(Effect.asVoid), 

Lines 1–500 of 545. Load more