Updates

packages/bare-os-coreutils/src/grep.js

@@ -26,6 +26,12 @@ async function run(ctx, argv) {
   /** @type {'never' | 'always' | 'auto'} */
   let colorMode = 'never'
   let recursive = false
+  /** @type {string[]} */
+  const includeGlobs = []
+  /** @type {string[]} */
+  const excludeGlobs = []
+  /** @type {string[]} */
+  const excludeDirGlobs = []
 
   const args = argv.slice(1)
   let i = 0
@@ -93,6 +99,36 @@ async function run(ctx, argv) {
         i++
         continue
       }
+      if (a === '--include' && args[i + 1]) {
+        includeGlobs.push(args[++i])
+        i++
+        continue
+      }
+      if (a.startsWith('--include=')) {
+        includeGlobs.push(a.slice('--include='.length))
+        i++
+        continue
+      }
+      if (a === '--exclude' && args[i + 1]) {
+        excludeGlobs.push(args[++i])
+        i++
+        continue
+      }
+      if (a.startsWith('--exclude=')) {
+        excludeGlobs.push(a.slice('--exclude='.length))
+        i++
+        continue
+      }
+      if (a === '--exclude-dir' && args[i + 1]) {
+        excludeDirGlobs.push(args[++i])
+        i++
+        continue
+      }
+      if (a.startsWith('--exclude-dir=')) {
+        excludeDirGlobs.push(a.slice('--exclude-dir='.length))
+        i++
+        continue
+      }
       ctx.console.error('grep: unknown option ' + a)
       ctx.exitCode = 2
       return
@@ -304,7 +340,15 @@ async function run(ctx, argv) {
           continue
         }
         if (st.type === 'directory') {
-          await grepWalkFiles(ctx, p, acc, suppressErrors)
+          await grepWalkFiles(ctx, p, acc, suppressErrors, {
+            includeGlobs,
+            excludeGlobs,
+            excludeDirGlobs,
+            filterCap: Number.parseInt(
+              String(ctx.vfs?.env?.BARE_OS_GREP_FILTER_MAX || '32'),
+              10
+            ) || 32
+          })
         } else {
           acc.push(p)
         }
@@ -500,8 +544,31 @@ const GREP_RECURSE_MAX_DEPTH = 64
  * @param {string[]} acc
  * @param {boolean} suppressErrors
  */
-async function grepWalkFiles(ctx, dir, acc, suppressErrors) {
+/**
+ * @param {string} name
+ * @param {string} pat
+ */
+function grepSimpleGlobMatch(name, pat) {
+  if (!pat || pat === '*') return true
+  if (pat.includes('/')) return name === pat
+  if (pat.startsWith('*') && pat.length > 1 && pat.endsWith('*')) {
+    const mid = pat.slice(1, -1)
+    return mid !== '' && name.includes(mid)
+  }
+  if (pat.startsWith('*')) return name.endsWith(pat.slice(1))
+  if (pat.endsWith('*')) return name.startsWith(pat.slice(0, -1))
+  return name === pat
+}
+
+/**
+ * @param {{ includeGlobs?: string[], excludeGlobs?: string[], excludeDirGlobs?: string[], filterCap?: number }} [opts]
+ */
+async function grepWalkFiles(ctx, dir, acc, suppressErrors, opts = {}) {
   const vfs = ctx.vfs
+  const cap = opts.filterCap && opts.filterCap > 0 ? opts.filterCap : 32
+  const inc = (opts.includeGlobs || []).slice(0, cap)
+  const exc = (opts.excludeGlobs || []).slice(0, cap)
+  const excd = (opts.excludeDirGlobs || []).slice(0, cap)
   /** @param {string} d @param {number} depth */
   const walk = async (d, depth) => {
     if (depth > GREP_RECURSE_MAX_DEPTH) return
@@ -523,12 +590,21 @@ async function grepWalkFiles(ctx, dir, acc, suppressErrors) {
         continue
       }
       if (!st) continue
-      if (st.type === 'directory') await walk(sub, depth + 1)
-      else if (st.type === 'file') acc.push(sub)
-      else if (st.type === 'symlink') {
+      if (st.type === 'directory') {
+        if (excd.some((p) => grepSimpleGlobMatch(n, p))) continue
+        await walk(sub, depth + 1)
+      } else if (st.type === 'file') {
+        if (inc.length && !inc.some((p) => grepSimpleGlobMatch(n, p))) continue
+        if (exc.some((p) => grepSimpleGlobMatch(n, p))) continue
+        acc.push(sub)
+      } else if (st.type === 'symlink') {
         try {
           const ft = await vfs.stat(sub)
-          if (ft && ft.type === 'file') acc.push(sub)
+          if (ft && ft.type === 'file') {
+            if (inc.length && !inc.some((p) => grepSimpleGlobMatch(n, p))) continue
+            if (exc.some((p) => grepSimpleGlobMatch(n, p))) continue
+            acc.push(sub)
+          }
         } catch {
           /* skip */
         }

packages/bare-os-coreutils/src/sed.js

@@ -1,6 +1,7 @@
 async function run(ctx, argv) {
   let silent = false
   let extended = false
+  let nullData = false
   /** @type {string[]} */
   const scripts = []
   /** @type {string[]} */
@@ -11,6 +12,10 @@ async function run(ctx, argv) {
       silent = true
       continue
     }
+    if (a === '-z' || a === '--null-data') {
+      nullData = true
+      continue
+    }
     if (a === '-E' || a === '-r') {
       extended = true
       continue
@@ -89,6 +94,12 @@ async function run(ctx, argv) {
     readCache[rp] = buf ? ctx.b4a.toString(buf) : ''
   }
 
+  const maxNull =
+    Number.parseInt(
+      String(ctx.vfs?.env?.BARE_OS_SED_NULL_MAX_RECORDS || '100000'),
+      10
+    ) || 100000
+
   /** @type {string[]} */
   const lines = []
   async function pushFile(path) {
@@ -99,16 +110,22 @@ async function run(ctx, argv) {
       return false
     }
     const t = ctx.b4a.toString(buf)
-    const ls = t.split(/\r?\n/)
-    if (ls.length && ls[ls.length - 1] === '') ls.pop()
-    lines.push(...ls)
+    if (nullData) {
+      const rec = t.split('\0')
+      const room = maxNull - lines.length
+      lines.push(...rec.slice(0, Math.max(0, room)))
+    } else {
+      const ls = t.split(/\r?\n/)
+      if (ls.length && ls[ls.length - 1] === '') ls.pop()
+      lines.push(...ls)
+    }
     return true
   }
 
   if (!files.length) {
     const s = bareStdin(ctx)
-    const ls = s.split(/\r?\n/)
-    if (ls.length && ls[ls.length - 1] === '') ls.pop()
+    const ls = nullData ? s.split('\0').slice(0, maxNull) : s.split(/\r?\n/)
+    if (!nullData && ls.length && ls[ls.length - 1] === '') ls.pop()
     lines.push(...ls)
   } else {
     for (const f of files) {
@@ -121,6 +138,7 @@ async function run(ctx, argv) {
   const out = bareSedRun(lines, scripts, {
     silent,
     extended,
+    nullData,
     readFile: (p) => readCache[p] ?? null,
     writeFile: (p, chunk) => {
       wAccum[p] = (wAccum[p] || '') + chunk
@@ -141,6 +159,7 @@ async function run(ctx, argv) {
     }
   }
 
-  const t = out.replace(/\n$/, '')
+  const trail = nullData ? /\0$/ : /\n$/
+  const t = out.replace(trail, '')
   ctx.console.log(t)
 }