Complete archive listings by walking RAR headers, at low priority

Inspection lists only the entries whose headers are in the first segment
of a first volume. A walk hops from header to header, fetching one segment
per entry and volume, in its own queue (one job per collection worker) on
NNTP connections inspection isn't waiting for. content:walk-archives queues
releases inspected earlier (CONCEPT D40).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

lib/rar/parse.ts

@@ -31,6 +31,12 @@ export interface RarInfo {
   entries: RarEntry[]
   /** The buffer ended before the end-of-archive header (normal for one segment). */
   truncated: boolean
+  /**
+   * Offset of the first block that was not read: where the next header
+   * starts (possibly beyond the buffer, after an entry's data), or the
+   * header the buffer cut off. After the end-of-archive header, its end.
+   */
+  next: number
 }
 
 const RAR4_MARKER = Buffer.from([0x52, 0x61, 0x72, 0x21, 0x1a, 0x07, 0x00])
@@ -45,11 +51,16 @@ export function rarFormat(data: Buffer): 'rar4' | 'rar5' | null {
 /**
  * Lists a RAR archive from its headers only, so the first segment of a first
  * volume is enough. Returns null when the data is not a RAR archive.
+ *
+ * With `continueAs`, `data` starts at a header in the middle of a volume (no
+ * signature), e.g. where a previous parse said the next header is.
  */
-export function parseRarHeaders(data: Buffer): RarInfo | null {
+export function parseRarHeaders(data: Buffer, continueAs?: 'rar4' | 'rar5'): RarInfo | null {
+  if (continueAs === 'rar4') return parseRar4(data, 0)
+  if (continueAs === 'rar5') return parseRar5(data, 0)
   const format = rarFormat(data)
-  if (format === 'rar4') return parseRar4(data)
-  if (format === 'rar5') return parseRar5(data)
+  if (format === 'rar4') return parseRar4(data, RAR4_MARKER.length)
+  if (format === 'rar5') return parseRar5(data, RAR5_MARKER.length)
   return null
 }
 
@@ -60,7 +71,7 @@ const R4_FILE = 0x74
 const R4_END = 0x7b
 const R4_LONG_BLOCK = 0x8000
 
-function parseRar4(data: Buffer): RarInfo {
+function parseRar4(data: Buffer, start: number): RarInfo {
   const info: RarInfo = {
     format: 'rar4',
     volume: false,
@@ -69,8 +80,9 @@ function parseRar4(data: Buffer): RarInfo {
     headersEncrypted: false,
     entries: [],
     truncated: true,
+    next: start,
   }
-  let p = RAR4_MARKER.length
+  let p = start
 
   while (p + 7 <= data.length) {
     const headCrc = data.readUInt16LE(p)
@@ -81,6 +93,7 @@ function parseRar4(data: Buffer): RarInfo {
 
     if (type === R4_END) {
       info.truncated = false
+      p += size
       break
     }
 
@@ -130,6 +143,7 @@ function parseRar4(data: Buffer): RarInfo {
   if (info.volume && !info.firstVolume && info.entries.length > 0 && !info.entries[0].splitBefore) {
     info.firstVolume = true
   }
+  info.next = p
   return info
 }
 
@@ -205,7 +219,7 @@ function vint(data: Buffer, offset: number): { value: number; next: number } | n
   return null
 }
 
-function parseRar5(data: Buffer): RarInfo {
+function parseRar5(data: Buffer, start: number): RarInfo {
   const info: RarInfo = {
     format: 'rar5',
     volume: false,
@@ -214,8 +228,9 @@ function parseRar5(data: Buffer): RarInfo {
     headersEncrypted: false,
     entries: [],
     truncated: true,
+    next: start,
   }
-  let p = RAR5_MARKER.length
+  let p = start
 
   while (p + 5 <= data.length) {
     const headCrc = data.readUInt32LE(p)
@@ -246,6 +261,7 @@ function parseRar5(data: Buffer): RarInfo {
 
     if (type.value === R5_END) {
       info.truncated = false
+      p = headerEnd
       break
     }
     if (type.value === R5_ENCRYPTION) {
@@ -270,6 +286,7 @@ function parseRar5(data: Buffer): RarInfo {
     if (next <= p) break
     p = next
   }
+  info.next = p
   return info
 }