dsh-remote 0.8.71 → 0.8.73

package/lib/ignore.js

@@ -1,202 +1,202 @@
-// dsh-remote — gitignore-style ignore matcher for mirror sync / remote search.
-import { readFileSync } from 'node:fs'
-//
-// Deliberately a SUBSET of gitignore semantics (documented, no surprises):
-//   - blank lines and `#` comments are skipped
-//   - `!` negation is NOT supported (later rules never un-ignore)
-//   - a trailing `/` matches directories only (and everything under them)
-//   - a pattern containing `/` (or starting with `/`) is anchored to the
-//     workspace root; otherwise it matches the basename at ANY depth
-//   - `*` matches any run of chars except `/`; `**` matches any run incl. `/`;
-//     `?` matches one char except `/`
-//   - character classes (`[abc]`) are not supported (treated literally)
-
-/** Convert one glob segment to a regex source (no `/` crossing). A leading
- * double-star-slash (`**` + `/`) matches zero or more directories (gitignore). */
-function globToRe(glob) {
-  let out = ''
-  let i = 0
-  const s = String(glob)
-  if (s.startsWith('**/')) { out = '(?:.*/)?'; i = 3 }
-  while (i < s.length) {
-    const c = s[i]
-    if (c === '*') {
-      if (s[i + 1] === '*') { out += '.*'; i += 2; continue }
-      out += '[^/]*'; i++; continue
-    }
-    if (c === '?') { out += '[^/]'; i++; continue }
-    if ('\\^$+{}[]()|.'.includes(c)) out += '\\' + c
-    else out += c
-    i++
-  }
-  return out
-}
-
-/**
- * Compile a list of ignore patterns into a matcher.
- * @param {string[]|string} patterns lines or an array of lines
- * @returns {(relPath: string, isDir?: boolean) => boolean} true = ignored.
- *   `relPath` is POSIX-style ('/' separated, no leading '/').
- */
-export function compileIgnore(patterns) {
-  const rules = []
-  const lines = Array.isArray(patterns) ? patterns : String(patterns || '').split('\n')
-  for (let raw of lines) {
-    let line = String(raw || '').trim()
-    if (!line || line.startsWith('#')) continue
-    let dirOnly = false
-    if (line.endsWith('/')) { dirOnly = true; line = line.slice(0, -1) }
-    let anchored = line.startsWith('/')
-    if (anchored) line = line.slice(1)
-    if (!line) continue
-    const hasSlash = line.includes('/')
-    const isAnchored = anchored || hasSlash
-    const segRe = globToRe(line)
-    let re
-    if (isAnchored) {
-      // Root-anchored: match the full path from the root.
-      re = dirOnly
-        ? new RegExp('^' + segRe + '(/.*)?$')
-        : new RegExp('^' + segRe + '$')
-    } else {
-      // Basename at any depth; a dir pattern also covers everything under it.
-      re = dirOnly
-        ? new RegExp('(^|/)' + segRe + '(/|$)')
-        : new RegExp('(^|/)' + segRe + '$')
-    }
-    rules.push({ re, dirOnly })
-  }
-  return function isIgnored(relPath, isDir) {
-    const rel = String(relPath || '').replace(/^\/+/, '').replace(/\\/g, '/')
-    if (!rel) return false
-    for (const r of rules) {
-      if (r.re.test(rel)) return true
-    }
-    return false
-  }
-}
-
-/** Built-in defaults: never pull/push these into/from a mirror. */
-export const DEFAULT_IGNORE = [
-  '.git/',
-  '.svn/',
-  '.hg/',
-  'node_modules/',
-  'target/',
-  'dist/',
-  'build/',
-  '.venv/',
-  'venv/',
-  '__pycache__/',
-  '.idea/',
-  '.vscode/',
-  '*.o',
-  '*.a',
-  '*.so',
-  '*.dylib',
-  '*.exe',
-  '*.dll',
-  '*.class',
-  '*.jar',
-  '*.log',
-  '*.tmp',
-  '.DS_Store',
-  'Thumbs.db',
-  '.dsh-remote-meta.json',
-  '.dsh-remote-sync-state.json',
-]
-
-/**
- * Extra directories skipped by `rw_search` (but NOT by mirror sync — a user may
- * legitimately want their cache tree mirrored, and silently excluding it from
- * sync would lose data).
- *
- * These are machine-local package/tool caches: enormous numbers of tiny files
- * that essentially never contain what a human is searching for, and that are
- * the main way a "search my home directory" turns into a multi-hour walk
- * (issue #44 reports `~/.npm/_cacache` specifically). Searching them is
- * available via an explicit `path`, which bypasses these excludes.
- */
-export const DEFAULT_SEARCH_IGNORE = [
-  ...DEFAULT_IGNORE,
-  '.npm/',
-  '.cache/',
-  '.local/',
-  '.cargo/',
-  '.rustup/',
-  '.gradle/',
-  '.m2/',
-  '.pnpm-store/',
-  '.yarn/',
-  '.conda/',
-  '.docker/',
-  '.nvm/',
-  '.pyenv/',
-  '.Trash/',
-  '.Trash-1000/',
-  'snap/',
-  'AppData/',
-  'Library/Caches/',
-]
-
-/**
- * Whether a normalized remote path is a FILESYSTEM ROOT: POSIX `/`, a Windows
- * drive root (`D:\`), or a UNC share root (`\\server\share`).
- *
- * Root workspaces are legal (issue #60) but dangerous to walk naively: `/proc`
- * and `/sys` are synthetic (new pids appear between readdirs, `/proc/kcore`
- * advertises 128 TB), `/dev` holds never-ending character devices, and
- * `/var/lib/docker` is tens of GB of image layers. ROOT_IGNORE exists so a
- * root workspace skips exactly those trees.
- */
-export function isFilesystemRoot(p) {
-  const s = String(p || '')
-  return s === '/' || /^[a-zA-Z]:\\?$/.test(s) || /^\\\\[^\\]+\\[^\\]+$/.test(s)
-}
-
-/**
- * Patterns merged in ONLY when the workspace is the filesystem root
- * (issue #60). Every entry is root-anchored AND directory-only, so even when
- * merged they can only match at the workspace top level: `/` skips `/proc`,
- * while a hypothetical `/home/u/proc` under a non-root workspace is ordinary
- * content (and non-root workspaces never receive these patterns at all, so
- * their behaviour is byte-identical to before).
- */
-export const ROOT_IGNORE = [
-  '/proc/',
-  '/sys/',
-  '/dev/',
-  '/run/',
-  '/boot/',
-  '/var/lib/docker/',
-]
-
-/**
- * Effective ignore patterns for a workspace rooted at `root`:
- * the defaults, plus ROOT_IGNORE only for a root workspace, plus the user's
- * own `.dsh-remote-ignore` lines (always last, they stack on top).
- *
- * @param {string} root - workspace root as passed to sync/push/search.
- * @param {{search?: boolean, fromFile?: string[]}} [opts]
- */
-export function ignorePatternsFor(root, { search = false, fromFile = [] } = {}) {
-  const base = search ? DEFAULT_SEARCH_IGNORE : DEFAULT_IGNORE
-  const extra = isFilesystemRoot(root) ? ROOT_IGNORE : []
-  return base.concat(extra, fromFile)
-}
-
-/**
- * Load the user ignore file (gitignore syntax) and merge with defaults.
- * Returns `{ patterns, fromFile }`.
- */
-export function loadIgnore(ignoreFile) {
-  let fromFile = []
-  if (ignoreFile) {
-    try {
-      const text = readFileSync(ignoreFile, 'utf8')
-      fromFile = text.split('\n').map((l) => l.trim()).filter((l) => l && !l.startsWith('#'))
-    } catch { /* missing file → defaults only */ }
-  }
-  const patterns = DEFAULT_IGNORE.concat(fromFile)
-  return { patterns, fromFile, matcher: compileIgnore(patterns) }
-}
+// dsh-remote — gitignore-style ignore matcher for mirror sync / remote search.
+import { readFileSync } from 'node:fs'
+//
+// Deliberately a SUBSET of gitignore semantics (documented, no surprises):
+//   - blank lines and `#` comments are skipped
+//   - `!` negation is NOT supported (later rules never un-ignore)
+//   - a trailing `/` matches directories only (and everything under them)
+//   - a pattern containing `/` (or starting with `/`) is anchored to the
+//     workspace root; otherwise it matches the basename at ANY depth
+//   - `*` matches any run of chars except `/`; `**` matches any run incl. `/`;
+//     `?` matches one char except `/`
+//   - character classes (`[abc]`) are not supported (treated literally)
+
+/** Convert one glob segment to a regex source (no `/` crossing). A leading
+ * double-star-slash (`**` + `/`) matches zero or more directories (gitignore). */
+function globToRe(glob) {
+  let out = ''
+  let i = 0
+  const s = String(glob)
+  if (s.startsWith('**/')) { out = '(?:.*/)?'; i = 3 }
+  while (i < s.length) {
+    const c = s[i]
+    if (c === '*') {
+      if (s[i + 1] === '*') { out += '.*'; i += 2; continue }
+      out += '[^/]*'; i++; continue
+    }
+    if (c === '?') { out += '[^/]'; i++; continue }
+    if ('\\^$+{}[]()|.'.includes(c)) out += '\\' + c
+    else out += c
+    i++
+  }
+  return out
+}
+
+/**
+ * Compile a list of ignore patterns into a matcher.
+ * @param {string[]|string} patterns lines or an array of lines
+ * @returns {(relPath: string, isDir?: boolean) => boolean} true = ignored.
+ *   `relPath` is POSIX-style ('/' separated, no leading '/').
+ */
+export function compileIgnore(patterns) {
+  const rules = []
+  const lines = Array.isArray(patterns) ? patterns : String(patterns || '').split('\n')
+  for (let raw of lines) {
+    let line = String(raw || '').trim()
+    if (!line || line.startsWith('#')) continue
+    let dirOnly = false
+    if (line.endsWith('/')) { dirOnly = true; line = line.slice(0, -1) }
+    let anchored = line.startsWith('/')
+    if (anchored) line = line.slice(1)
+    if (!line) continue
+    const hasSlash = line.includes('/')
+    const isAnchored = anchored || hasSlash
+    const segRe = globToRe(line)
+    let re
+    if (isAnchored) {
+      // Root-anchored: match the full path from the root.
+      re = dirOnly
+        ? new RegExp('^' + segRe + '(/.*)?$')
+        : new RegExp('^' + segRe + '$')
+    } else {
+      // Basename at any depth; a dir pattern also covers everything under it.
+      re = dirOnly
+        ? new RegExp('(^|/)' + segRe + '(/|$)')
+        : new RegExp('(^|/)' + segRe + '$')
+    }
+    rules.push({ re, dirOnly })
+  }
+  return function isIgnored(relPath, isDir) {
+    const rel = String(relPath || '').replace(/^\/+/, '').replace(/\\/g, '/')
+    if (!rel) return false
+    for (const r of rules) {
+      if (r.re.test(rel)) return true
+    }
+    return false
+  }
+}
+
+/** Built-in defaults: never pull/push these into/from a mirror. */
+export const DEFAULT_IGNORE = [
+  '.git/',
+  '.svn/',
+  '.hg/',
+  'node_modules/',
+  'target/',
+  'dist/',
+  'build/',
+  '.venv/',
+  'venv/',
+  '__pycache__/',
+  '.idea/',
+  '.vscode/',
+  '*.o',
+  '*.a',
+  '*.so',
+  '*.dylib',
+  '*.exe',
+  '*.dll',
+  '*.class',
+  '*.jar',
+  '*.log',
+  '*.tmp',
+  '.DS_Store',
+  'Thumbs.db',
+  '.dsh-remote-meta.json',
+  '.dsh-remote-sync-state.json',
+]
+
+/**
+ * Extra directories skipped by `rw_search` (but NOT by mirror sync — a user may
+ * legitimately want their cache tree mirrored, and silently excluding it from
+ * sync would lose data).
+ *
+ * These are machine-local package/tool caches: enormous numbers of tiny files
+ * that essentially never contain what a human is searching for, and that are
+ * the main way a "search my home directory" turns into a multi-hour walk
+ * (issue #44 reports `~/.npm/_cacache` specifically). Searching them is
+ * available via an explicit `path`, which bypasses these excludes.
+ */
+export const DEFAULT_SEARCH_IGNORE = [
+  ...DEFAULT_IGNORE,
+  '.npm/',
+  '.cache/',
+  '.local/',
+  '.cargo/',
+  '.rustup/',
+  '.gradle/',
+  '.m2/',
+  '.pnpm-store/',
+  '.yarn/',
+  '.conda/',
+  '.docker/',
+  '.nvm/',
+  '.pyenv/',
+  '.Trash/',
+  '.Trash-1000/',
+  'snap/',
+  'AppData/',
+  'Library/Caches/',
+]
+
+/**
+ * Whether a normalized remote path is a FILESYSTEM ROOT: POSIX `/`, a Windows
+ * drive root (`D:\`), or a UNC share root (`\\server\share`).
+ *
+ * Root workspaces are legal (issue #60) but dangerous to walk naively: `/proc`
+ * and `/sys` are synthetic (new pids appear between readdirs, `/proc/kcore`
+ * advertises 128 TB), `/dev` holds never-ending character devices, and
+ * `/var/lib/docker` is tens of GB of image layers. ROOT_IGNORE exists so a
+ * root workspace skips exactly those trees.
+ */
+export function isFilesystemRoot(p) {
+  const s = String(p || '')
+  return s === '/' || /^[a-zA-Z]:\\?$/.test(s) || /^\\\\[^\\]+\\[^\\]+$/.test(s)
+}
+
+/**
+ * Patterns merged in ONLY when the workspace is the filesystem root
+ * (issue #60). Every entry is root-anchored AND directory-only, so even when
+ * merged they can only match at the workspace top level: `/` skips `/proc`,
+ * while a hypothetical `/home/u/proc` under a non-root workspace is ordinary
+ * content (and non-root workspaces never receive these patterns at all, so
+ * their behaviour is byte-identical to before).
+ */
+export const ROOT_IGNORE = [
+  '/proc/',
+  '/sys/',
+  '/dev/',
+  '/run/',
+  '/boot/',
+  '/var/lib/docker/',
+]
+
+/**
+ * Effective ignore patterns for a workspace rooted at `root`:
+ * the defaults, plus ROOT_IGNORE only for a root workspace, plus the user's
+ * own `.dsh-remote-ignore` lines (always last, they stack on top).
+ *
+ * @param {string} root - workspace root as passed to sync/push/search.
+ * @param {{search?: boolean, fromFile?: string[]}} [opts]
+ */
+export function ignorePatternsFor(root, { search = false, fromFile = [] } = {}) {
+  const base = search ? DEFAULT_SEARCH_IGNORE : DEFAULT_IGNORE
+  const extra = isFilesystemRoot(root) ? ROOT_IGNORE : []
+  return base.concat(extra, fromFile)
+}
+
+/**
+ * Load the user ignore file (gitignore syntax) and merge with defaults.
+ * Returns `{ patterns, fromFile }`.
+ */
+export function loadIgnore(ignoreFile) {
+  let fromFile = []
+  if (ignoreFile) {
+    try {
+      const text = readFileSync(ignoreFile, 'utf8')
+      fromFile = text.split('\n').map((l) => l.trim()).filter((l) => l && !l.startsWith('#'))
+    } catch { /* missing file → defaults only */ }
+  }
+  const patterns = DEFAULT_IGNORE.concat(fromFile)
+  return { patterns, fromFile, matcher: compileIgnore(patterns) }
+}