@cplieger/deadset-ts 8.0.0-dev.6 → 8.0.0-dev.8

package/src/workspace.ts

@@ -7,12 +7,23 @@
  * manifest, never one below `node_modules`.
  */
 
-import { DiscoveryError } from "./discover.ts";
+import { DiscoveryError, isConfigFile } from "./discover.ts";
 import type { EmitLayout } from "./emit-map.ts";
 import { commonDirectory } from "./emit-map.ts";
 import type { Host } from "./host.ts";
-import { dirnamePath, joinPath, relativePath } from "./paths.ts";
-import { pnpmPackages } from "./pnpm-workspace.ts";
+import { ancestors, dirnamePath, joinPath, relativePath } from "./paths.ts";
+import { bytewise } from "./text/compare.ts";
+import { isRecord } from "./json-document.ts";
+import { directoriesIn, filesIn, repositoryOf } from "./fs/tree.ts";
+import {
+  holdsManifest,
+  inInstalledPackages,
+  INSTALLED_PACKAGES,
+  MANIFEST,
+  manifestAt,
+} from "./fs/manifest.ts";
+import { braceExpanded, pathMatches, segmentMatches } from "./text/glob.ts";
+import { yamlKey } from "./text/yaml.ts";
 
 /** What one compiler configuration of a member says about where it reads and writes. */
 export interface ParsedLayout {
@@ -56,29 +67,7 @@ export interface Workspace {
   memberOf(path: string): WorkspaceMember | undefined;
 }
 
-const MANIFEST = "package.json";
 const PNPM_WORKSPACE = "pnpm-workspace.yaml";
-const CONFIG_FILE = /^tsconfig.*\.json$/u;
-
-function isRecord(value: unknown): value is Record<string, unknown> {
-  return typeof value === "object" && value !== null && !Array.isArray(value);
-}
-
-/** One manifest, decoded, or `undefined` where there is none to read. */
-function manifestAt(host: Host, dir: string): Record<string, unknown> | undefined {
-  let text: string;
-  try {
-    text = host.readFile(joinPath(dir, MANIFEST));
-  } catch {
-    return undefined;
-  }
-  try {
-    const value: unknown = JSON.parse(text);
-    return isRecord(value) ? value : undefined;
-  } catch {
-    return undefined;
-  }
-}
 
 /** The patterns one directory declares a workspace with, or `undefined` where it declares none. */
 function declarationAt(
@@ -87,14 +76,20 @@ function declarationAt(
 ): { readonly file: string; readonly patterns: readonly string[] } | undefined {
   const pnpm = joinPath(dir, PNPM_WORKSPACE);
   if (host.kindOf(pnpm) === "file") {
-    const read = pnpmPackages(host.readFile(pnpm));
-    if (read.kind === "refused") {
+    const read = yamlKey(host.readFile(pnpm), "packages");
+    if (read.kind === "absent") {
+      return { file: pnpm, patterns: [] };
+    }
+    const listed = read.kind === "value" ? read.value : undefined;
+    if (!Array.isArray(listed) || !listed.every((entry) => typeof entry === "string")) {
+      const reason =
+        read.kind === "refused" ? read.reason : "packages is not a sequence of strings";
       throw new DiscoveryError(
-        `${pnpm}: ${read.reason}; the workspace's packages are read from a sequence of strings`,
+        `${pnpm}: ${reason}; the workspace's packages are read from a sequence of strings`,
         [],
       );
     }
-    return { file: pnpm, patterns: read.kind === "listed" ? read.patterns : [] };
+    return { file: pnpm, patterns: listed };
   }
   const manifest = manifestAt(host, dir);
   if (manifest === undefined || !Object.hasOwn(manifest, "workspaces")) {
@@ -120,21 +115,6 @@ function declaresWorkspace(host: Host, dir: string): boolean {
   return manifest !== undefined && Object.hasOwn(manifest, "workspaces");
 }
 
-/** A pattern with every brace list expanded into its alternatives. */
-function braceExpanded(pattern: string): readonly string[] {
-  const open = pattern.indexOf("{");
-  const close = open < 0 ? -1 : pattern.indexOf("}", open);
-  if (close < 0) {
-    return [pattern];
-  }
-  const before = pattern.slice(0, open);
-  const after = pattern.slice(close + 1);
-  return pattern
-    .slice(open + 1, close)
-    .split(",")
-    .flatMap((member) => braceExpanded(before + member + after));
-}
-
 /**
  * The segments of one pattern below the workspace root, `.` and empty segments
  * dropped and `..` taken back, or `undefined` where it leaves the root.
@@ -156,56 +136,6 @@ function segmentsOf(pattern: string): readonly string[] | undefined {
   return segments;
 }
 
-/**
- * Whether one directory name matches one pattern segment: `*` stands for any run of
- * characters and `?` for one, and neither matches a leading dot.
- */
-function segmentMatches(segment: string, name: string): boolean {
-  if (name.startsWith(".") && !segment.startsWith(".")) {
-    return false;
-  }
-  const source = Array.from(segment)
-    .map((character) => {
-      if (character === "*") {
-        return ".*";
-      }
-      if (character === "?") {
-        return ".";
-      }
-      return character.replaceAll(/[.+^${}()|[\]\\/]/gu, (c) => `\\${c}`);
-    })
-    .join("");
-  return new RegExp(`^${source}$`, "u").test(name);
-}
-
-/** Whether one path's segments match one pattern's, `**` standing for any run of segments. */
-function pathMatches(pattern: readonly string[], path: readonly string[]): boolean {
-  const [head, ...rest] = pattern;
-  if (head === undefined) {
-    return path.length === 0;
-  }
-  if (head === "**") {
-    return (
-      pathMatches(rest, path) ||
-      (path.length > 0 && !(path[0] ?? "").startsWith(".") && pathMatches(pattern, path.slice(1)))
-    );
-  }
-  const [name, ...below] = path;
-  return name !== undefined && segmentMatches(head, name) && pathMatches(rest, below);
-}
-
-function subdirectories(host: Host, dir: string): readonly string[] {
-  try {
-    return host
-      .readDirectory(dir)
-      .filter((entry) => entry.directory && entry.name !== "node_modules" && entry.name !== ".git")
-      .map((entry) => entry.name)
-      .sort();
-  } catch {
-    return [];
-  }
-}
-
 /**
  * Every directory below `dir` one pattern names. `**` does not descend into a
  * directory that declares a workspace of its own, whose packages are that
@@ -219,7 +149,7 @@ function expand(host: Host, dir: string, pattern: readonly string[], into: Set<s
   }
   if (head === "**") {
     expand(host, dir, rest, into);
-    for (const name of subdirectories(host, dir)) {
+    for (const name of directoriesIn(host, dir)) {
       if (name.startsWith(".")) {
         continue;
       }
@@ -230,12 +160,12 @@ function expand(host: Host, dir: string, pattern: readonly string[], into: Set<s
   }
   if (!head.includes("*") && !head.includes("?")) {
     const child = joinPath(dir, head);
-    if (head !== "node_modules" && host.kindOf(child) === "directory") {
+    if (head !== INSTALLED_PACKAGES && host.kindOf(child) === "directory") {
       expand(host, child, rest, into);
     }
     return;
   }
-  for (const name of subdirectories(host, dir)) {
+  for (const name of directoriesIn(host, dir)) {
     if (segmentMatches(head, name)) {
       expand(host, joinPath(dir, name), rest, into);
     }
@@ -277,10 +207,10 @@ function namesDirectory(root: string, patterns: readonly string[], dir: string):
   if (below === undefined) {
     return false;
   }
-  const path = below.split("/");
-  if (path.includes("node_modules")) {
+  if (inInstalledPackages(below)) {
     return false;
   }
+  const path = below.split("/");
   let named = false;
   for (const written of patterns) {
     const negated = written.startsWith("!");
@@ -314,17 +244,7 @@ function configurationsOf(
   dir: string,
   parse: (configFile: string) => ParsedLayout | undefined,
 ): readonly MemberConfiguration[] {
-  let names: readonly string[];
-  try {
-    names = host
-      .readDirectory(dir)
-      .filter((entry) => !entry.directory && CONFIG_FILE.test(entry.name))
-      .map((entry) => entry.name)
-      .sort();
-  } catch {
-    return [];
-  }
-  return names.flatMap((name) => {
+  return filesIn(host, dir, isConfigFile).flatMap((name) => {
     const configFile = joinPath(dir, name);
     const parsed = parse(configFile);
     return parsed === undefined
@@ -353,22 +273,17 @@ function nearestDeclaration(
       readonly highestPackage: string | undefined;
     }
   | undefined {
-  let dir = targetRoot;
   let highestPackage: string | undefined;
-  for (;;) {
+  for (const dir of ancestors(targetRoot, repositoryOf(host, targetRoot)?.root)) {
     const declared = declarationAt(host, dir);
     if (declared !== undefined) {
       return { root: dir, ...declared, highestPackage };
     }
-    if (host.kindOf(joinPath(dir, MANIFEST)) === "file") {
+    if (holdsManifest(host, dir)) {
       highestPackage = dir;
     }
-    const up = dirnamePath(dir);
-    if (up === dir || host.kindOf(joinPath(dir, ".git")) !== "absent") {
-      return undefined;
-    }
-    dir = up;
   }
+  return undefined;
 }
 
 /**
@@ -414,7 +329,7 @@ export function findWorkspace(
     return undefined;
   }
   members.sort((a, b) =>
-    a.dir === declared.root ? -1 : b.dir === declared.root ? 1 : a.dir < b.dir ? -1 : 1,
+    a.dir === declared.root ? -1 : b.dir === declared.root ? 1 : bytewise(a.dir, b.dir),
   );
   const named = new Map<string, WorkspaceMember[]>();
   for (const member of members) {
@@ -427,8 +342,7 @@ export function findWorkspace(
       return scopes.get(dir);
     }
     const up = dirnamePath(dir);
-    const found =
-      host.kindOf(joinPath(dir, MANIFEST)) === "file" ? dir : up === dir ? undefined : scopeOf(up);
+    const found = holdsManifest(host, dir) ? dir : up === dir ? undefined : scopeOf(up);
     scopes.set(dir, found);
     return found;
   };

package/dist/src/glob.js

@@ -1,65 +0,0 @@
-/**
- * Glob patterns over the paths of the target's own files, as the configuration
- * writes them. Both `ts.test_files` and `ts.entry_files` name files this way, and
- * one grammar answers for both.
- *
- * A configured root is not written in this grammar. It names a symbol reference
- * rather than a path, `*` there spans the solidus and a brace list means nothing,
- * so that rule is written where it is read.
- */
-/**
- * One literal run of a pattern, with every character the expression syntax claims
- * escaped.
- *
- * The set is the syntax characters and the solidus, and it is closed: an escape of any
- * other character is refused outright by the expression syntax under the unicode flag,
- * so escaping a hyphen, which needs none outside a character class, would refuse every
- * pattern that names a hyphenated path.
- */
-function quoted(text) {
-    return text.replaceAll(/[.*+?^${}()|[\]\\/]/gu, (character) => `\\${character}`);
-}
-/**
- * The expression one glob pattern denotes. `**` spans path separators while `*` and `?`
- * do not, a brace list is an alternation of its members, and every other character
- * stands for itself.
- */
-export function globExpression(pattern) {
-    let source = "^";
-    let at = 0;
-    while (at < pattern.length) {
-        const rest = pattern.slice(at);
-        if (rest.startsWith("**/")) {
-            source += "(?:[^/]*/)*";
-            at += 3;
-            continue;
-        }
-        if (rest.startsWith("**")) {
-            source += ".*";
-            at += 2;
-            continue;
-        }
-        const character = pattern[at] ?? "";
-        at += 1;
-        if (character === "*") {
-            source += "[^/]*";
-            continue;
-        }
-        if (character === "?") {
-            source += "[^/]";
-            continue;
-        }
-        if (character === "{") {
-            const close = pattern.indexOf("}", at);
-            if (close < 0) {
-                source += "\\{";
-                continue;
-            }
-            source += `(?:${pattern.slice(at, close).split(",").map(quoted).join("|")})`;
-            at = close + 1;
-            continue;
-        }
-        source += quoted(character);
-    }
-    return new RegExp(`${source}$`, "u");
-}

package/dist/src/pnpm-workspace.js

@@ -1,120 +0,0 @@
-/**
- * The `packages` list of a `pnpm-workspace.yaml`. The document is read as far as that
- * one key needs and no further: a top-level `packages` key holding a block sequence or
- * a one-line flow sequence of plain or quoted strings. Any other value of the key is
- * refused rather than guessed at, because a list read wrong names the wrong packages.
- */
-/** The characters a plain scalar may not start with. */
-const INDICATORS = new Set(["[", "]", "{", "}", ",", "&", "*", "!", "|", ">", "%", "@", "`", "#"]);
-/** One line with its comment removed, a `#` inside quotes kept. */
-function uncommented(line) {
-    let quote;
-    for (let at = 0; at < line.length; at += 1) {
-        const character = line[at];
-        if (quote !== undefined) {
-            if (character === quote) {
-                quote = undefined;
-            }
-            continue;
-        }
-        if (character === "'" || character === '"') {
-            quote = character;
-        }
-        else if (character === "#" && (at === 0 || /\s/u.test(line[at - 1] ?? ""))) {
-            return line.slice(0, at).trimEnd();
-        }
-    }
-    return line.trimEnd();
-}
-/** One scalar as written, or `undefined` where it is not a string this reader reads. */
-function scalarOf(text) {
-    const value = text.trim();
-    if (value.length >= 2 && value.startsWith("'") && value.endsWith("'")) {
-        const inner = value.slice(1, -1);
-        return /'(?!')/u.test(inner.replaceAll("''", "")) ? undefined : inner.replaceAll("''", "'");
-    }
-    if (value.length >= 2 && value.startsWith('"') && value.endsWith('"')) {
-        const inner = value.slice(1, -1);
-        return inner.includes("\\") || inner.includes('"') ? undefined : inner;
-    }
-    if (value === "" ||
-        INDICATORS.has(value[0] ?? "") ||
-        value.startsWith("'") ||
-        value.startsWith('"')) {
-        return undefined;
-    }
-    if (value.startsWith("- ") || value.startsWith("? ") || /:(?:\s|$)/u.test(value)) {
-        return undefined;
-    }
-    return value;
-}
-/** The entries of a one-line flow sequence, or `undefined` where one does not read. */
-function flowOf(text) {
-    const inner = text.slice(1, -1).trim();
-    if (inner === "") {
-        return [];
-    }
-    const entries = inner.split(",");
-    if (entries.length > 1 && entries.at(-1)?.trim() === "") {
-        entries.pop();
-    }
-    const found = [];
-    for (const entry of entries) {
-        const value = scalarOf(entry);
-        if (value === undefined) {
-            return undefined;
-        }
-        found.push(value);
-    }
-    return found;
-}
-const KEY = /^(?:packages|'packages'|"packages")\s*:(?:\s(.*))?$/u;
-/** The `packages` list one `pnpm-workspace.yaml` text declares. */
-export function pnpmPackages(text) {
-    const lines = text.split(/\r?\n/u).map(uncommented);
-    const keys = lines.flatMap((line, at) => (KEY.test(line) ? [at] : []));
-    const [at] = keys;
-    if (at === undefined) {
-        return { kind: "absent" };
-    }
-    if (keys.length > 1) {
-        return { kind: "refused", reason: "the packages key is declared more than once" };
-    }
-    const rest = (KEY.exec(lines[at] ?? "")?.[1] ?? "").trim();
-    if (rest !== "") {
-        const listed = rest.startsWith("[") && rest.endsWith("]") ? flowOf(rest) : undefined;
-        return listed === undefined
-            ? {
-                kind: "refused",
-                reason: `packages holds ${JSON.stringify(rest)}, not a sequence of strings`,
-            }
-            : { kind: "listed", patterns: listed };
-    }
-    const patterns = [];
-    for (const line of lines.slice(at + 1)) {
-        if (line.trim() === "") {
-            continue;
-        }
-        const item = /^(\s*)-(?:\s+(.*))?$/u.exec(line);
-        if (item === null) {
-            if (/^\s/u.test(line)) {
-                return {
-                    kind: "refused",
-                    reason: `packages holds ${JSON.stringify(line.trim())}, not a sequence item`,
-                };
-            }
-            break;
-        }
-        const value = scalarOf(item[2] ?? "");
-        if (value === undefined) {
-            return {
-                kind: "refused",
-                reason: `the packages item ${JSON.stringify(line.trim())} is not a plain or quoted string`,
-            };
-        }
-        patterns.push(value);
-    }
-    return patterns.length === 0
-        ? { kind: "refused", reason: "packages holds no sequence" }
-        : { kind: "listed", patterns };
-}

package/src/glob.ts

@@ -1,67 +0,0 @@
-/**
- * Glob patterns over the paths of the target's own files, as the configuration
- * writes them. Both `ts.test_files` and `ts.entry_files` name files this way, and
- * one grammar answers for both.
- *
- * A configured root is not written in this grammar. It names a symbol reference
- * rather than a path, `*` there spans the solidus and a brace list means nothing,
- * so that rule is written where it is read.
- */
-
-/**
- * One literal run of a pattern, with every character the expression syntax claims
- * escaped.
- *
- * The set is the syntax characters and the solidus, and it is closed: an escape of any
- * other character is refused outright by the expression syntax under the unicode flag,
- * so escaping a hyphen, which needs none outside a character class, would refuse every
- * pattern that names a hyphenated path.
- */
-function quoted(text: string): string {
-  return text.replaceAll(/[.*+?^${}()|[\]\\/]/gu, (character) => `\\${character}`);
-}
-
-/**
- * The expression one glob pattern denotes. `**` spans path separators while `*` and `?`
- * do not, a brace list is an alternation of its members, and every other character
- * stands for itself.
- */
-export function globExpression(pattern: string): RegExp {
-  let source = "^";
-  let at = 0;
-  while (at < pattern.length) {
-    const rest = pattern.slice(at);
-    if (rest.startsWith("**/")) {
-      source += "(?:[^/]*/)*";
-      at += 3;
-      continue;
-    }
-    if (rest.startsWith("**")) {
-      source += ".*";
-      at += 2;
-      continue;
-    }
-    const character = pattern[at] ?? "";
-    at += 1;
-    if (character === "*") {
-      source += "[^/]*";
-      continue;
-    }
-    if (character === "?") {
-      source += "[^/]";
-      continue;
-    }
-    if (character === "{") {
-      const close = pattern.indexOf("}", at);
-      if (close < 0) {
-        source += "\\{";
-        continue;
-      }
-      source += `(?:${pattern.slice(at, close).split(",").map(quoted).join("|")})`;
-      at = close + 1;
-      continue;
-    }
-    source += quoted(character);
-  }
-  return new RegExp(`${source}$`, "u");
-}

package/src/pnpm-workspace.ts

@@ -1,135 +0,0 @@
-/**
- * The `packages` list of a `pnpm-workspace.yaml`. The document is read as far as that
- * one key needs and no further: a top-level `packages` key holding a block sequence or
- * a one-line flow sequence of plain or quoted strings. Any other value of the key is
- * refused rather than guessed at, because a list read wrong names the wrong packages.
- */
-
-/** What the document says about `packages`. */
-type PnpmPackages =
-  | { readonly kind: "listed"; readonly patterns: readonly string[] }
-  /** The document declares no `packages` key. */
-  | { readonly kind: "absent" }
-  /** The key holds something this reader does not read, with what it holds. */
-  | { readonly kind: "refused"; readonly reason: string };
-
-/** The characters a plain scalar may not start with. */
-const INDICATORS = new Set(["[", "]", "{", "}", ",", "&", "*", "!", "|", ">", "%", "@", "`", "#"]);
-
-/** One line with its comment removed, a `#` inside quotes kept. */
-function uncommented(line: string): string {
-  let quote: string | undefined;
-  for (let at = 0; at < line.length; at += 1) {
-    const character = line[at];
-    if (quote !== undefined) {
-      if (character === quote) {
-        quote = undefined;
-      }
-      continue;
-    }
-    if (character === "'" || character === '"') {
-      quote = character;
-    } else if (character === "#" && (at === 0 || /\s/u.test(line[at - 1] ?? ""))) {
-      return line.slice(0, at).trimEnd();
-    }
-  }
-  return line.trimEnd();
-}
-
-/** One scalar as written, or `undefined` where it is not a string this reader reads. */
-function scalarOf(text: string): string | undefined {
-  const value = text.trim();
-  if (value.length >= 2 && value.startsWith("'") && value.endsWith("'")) {
-    const inner = value.slice(1, -1);
-    return /'(?!')/u.test(inner.replaceAll("''", "")) ? undefined : inner.replaceAll("''", "'");
-  }
-  if (value.length >= 2 && value.startsWith('"') && value.endsWith('"')) {
-    const inner = value.slice(1, -1);
-    return inner.includes("\\") || inner.includes('"') ? undefined : inner;
-  }
-  if (
-    value === "" ||
-    INDICATORS.has(value[0] ?? "") ||
-    value.startsWith("'") ||
-    value.startsWith('"')
-  ) {
-    return undefined;
-  }
-  if (value.startsWith("- ") || value.startsWith("? ") || /:(?:\s|$)/u.test(value)) {
-    return undefined;
-  }
-  return value;
-}
-
-/** The entries of a one-line flow sequence, or `undefined` where one does not read. */
-function flowOf(text: string): readonly string[] | undefined {
-  const inner = text.slice(1, -1).trim();
-  if (inner === "") {
-    return [];
-  }
-  const entries = inner.split(",");
-  if (entries.length > 1 && entries.at(-1)?.trim() === "") {
-    entries.pop();
-  }
-  const found: string[] = [];
-  for (const entry of entries) {
-    const value = scalarOf(entry);
-    if (value === undefined) {
-      return undefined;
-    }
-    found.push(value);
-  }
-  return found;
-}
-
-const KEY = /^(?:packages|'packages'|"packages")\s*:(?:\s(.*))?$/u;
-
-/** The `packages` list one `pnpm-workspace.yaml` text declares. */
-export function pnpmPackages(text: string): PnpmPackages {
-  const lines = text.split(/\r?\n/u).map(uncommented);
-  const keys = lines.flatMap((line, at) => (KEY.test(line) ? [at] : []));
-  const [at] = keys;
-  if (at === undefined) {
-    return { kind: "absent" };
-  }
-  if (keys.length > 1) {
-    return { kind: "refused", reason: "the packages key is declared more than once" };
-  }
-  const rest = (KEY.exec(lines[at] ?? "")?.[1] ?? "").trim();
-  if (rest !== "") {
-    const listed = rest.startsWith("[") && rest.endsWith("]") ? flowOf(rest) : undefined;
-    return listed === undefined
-      ? {
-          kind: "refused",
-          reason: `packages holds ${JSON.stringify(rest)}, not a sequence of strings`,
-        }
-      : { kind: "listed", patterns: listed };
-  }
-  const patterns: string[] = [];
-  for (const line of lines.slice(at + 1)) {
-    if (line.trim() === "") {
-      continue;
-    }
-    const item = /^(\s*)-(?:\s+(.*))?$/u.exec(line);
-    if (item === null) {
-      if (/^\s/u.test(line)) {
-        return {
-          kind: "refused",
-          reason: `packages holds ${JSON.stringify(line.trim())}, not a sequence item`,
-        };
-      }
-      break;
-    }
-    const value = scalarOf(item[2] ?? "");
-    if (value === undefined) {
-      return {
-        kind: "refused",
-        reason: `the packages item ${JSON.stringify(line.trim())} is not a plain or quoted string`,
-      };
-    }
-    patterns.push(value);
-  }
-  return patterns.length === 0
-    ? { kind: "refused", reason: "packages holds no sequence" }
-    : { kind: "listed", patterns };
-}