merge: close silent-drop gaps in attachment detection

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

worker/src/archive/detect.test.ts

@@ -1,7 +1,7 @@
 import { describe, it, expect } from "vitest";
 import { detectArchive, isArchiveAttachment } from "./detect.js";
 
-describe("detectArchive — 7z numbered multipart (pack.7z.001, pack.7z.002, ...)", () => {
+describe("detectArchive — numbered volumes (pack.EXT.001, pack.EXT.002, ...)", () => {
   it("recognizes a 7z multipart part as an archive attachment", () => {
     expect(isArchiveAttachment("Lost Adventures Vol2.7z.001")).toBe(true);
   });
@@ -12,7 +12,7 @@ describe("detectArchive — 7z numbered multipart (pack.7z.001, pack.7z.002, ...
       baseName: "Lost Adventures Vol2.7z",
       partNumber: 1,
       format: "7Z",
-      pattern: "SEVENZ_NUMBERED",
+      pattern: "ARCHIVE_NUMBERED",
     });
   });
 
@@ -36,4 +36,249 @@ describe("detectArchive — 7z numbered multipart (pack.7z.001, pack.7z.002, ...
       pattern: "SINGLE",
     });
   });
+
+  it("recognizes numbered ZIP volumes", () => {
+    expect(detectArchive("Big Pack.zip.001")).toEqual({
+      baseName: "Big Pack.zip",
+      partNumber: 1,
+      format: "ZIP",
+      pattern: "ARCHIVE_NUMBERED",
+    });
+  });
+
+  it("recognizes numbered RAR volumes (previously dropped silently)", () => {
+    expect(detectArchive("Big Pack.rar.001")).toEqual({
+      baseName: "Big Pack.rar",
+      partNumber: 1,
+      format: "RAR",
+      pattern: "ARCHIVE_NUMBERED",
+    });
+  });
+
+  it("derives the format from the archive extension, not a hardcoded value", () => {
+    expect(detectArchive("A.zip.004")?.format).toBe("ZIP");
+    expect(detectArchive("A.RAR.004")?.format).toBe("RAR");
+    expect(detectArchive("A.7z.004")?.format).toBe("7Z");
+  });
+
+  it("accepts hand-renamed two-digit volumes for every format", () => {
+    expect(detectArchive("Pack.zip.01")).toEqual({
+      baseName: "Pack.zip",
+      partNumber: 1,
+      format: "ZIP",
+      pattern: "ARCHIVE_NUMBERED",
+    });
+    expect(detectArchive("Pack.rar.02")?.partNumber).toBe(2);
+    expect(detectArchive("Pack.7z.03")?.partNumber).toBe(3);
+  });
+
+  it("still accepts four-or-more-digit volumes", () => {
+    expect(detectArchive("Pack.7z.0001")).toEqual({
+      baseName: "Pack.7z",
+      partNumber: 1,
+      format: "7Z",
+      pattern: "ARCHIVE_NUMBERED",
+    });
+    expect(detectArchive("Pack.zip.10001")?.partNumber).toBe(10001);
+  });
+
+  it("does not match a single-digit suffix (too ambiguous with real extensions)", () => {
+    expect(detectArchive("Pack.zip.1")).toBeNull();
+  });
+});
+
+describe("detectArchive — pattern ordering safety", () => {
+  it("does not let ZIP_LEGACY swallow a .7z.NNN name", () => {
+    expect(detectArchive("Pack.7z.001")?.pattern).toBe("ARCHIVE_NUMBERED");
+    expect(detectArchive("Pack.7z.001")?.format).toBe("7Z");
+  });
+
+  it("does not let ARCHIVE_NUMBERED swallow legacy .zNN names", () => {
+    expect(detectArchive("Pack.z01")).toEqual({
+      baseName: "Pack",
+      partNumber: 1,
+      format: "ZIP",
+      pattern: "ZIP_LEGACY",
+    });
+  });
+
+  it("does not let ARCHIVE_NUMBERED swallow legacy .rNN names", () => {
+    expect(detectArchive("Pack.r00")).toEqual({
+      baseName: "Pack",
+      partNumber: 0,
+      format: "RAR",
+      pattern: "RAR_LEGACY",
+    });
+    expect(detectArchive("Pack.r01")?.pattern).toBe("RAR_LEGACY");
+  });
+
+  it("keeps .partN.rar on the RAR_PART pattern", () => {
+    expect(detectArchive("Pack.part2.rar")).toEqual({
+      baseName: "Pack",
+      partNumber: 2,
+      format: "RAR",
+      pattern: "RAR_PART",
+    });
+  });
+});
+
+describe("detectArchive — filename normalization", () => {
+  it("recognizes a name with a trailing space and reports the trimmed baseName", () => {
+    expect(detectArchive("Pack.zip ")).toEqual({
+      baseName: "Pack",
+      partNumber: -1,
+      format: "ZIP",
+      pattern: "SINGLE",
+    });
+  });
+
+  it("recognizes a name with a leading space", () => {
+    expect(detectArchive("  Pack.rar")).toEqual({
+      baseName: "Pack",
+      partNumber: -1,
+      format: "RAR",
+      pattern: "SINGLE",
+    });
+  });
+
+  it("recognizes a multipart name with surrounding whitespace", () => {
+    expect(detectArchive("\tPack.rar.002 \n")).toEqual({
+      baseName: "Pack.rar",
+      partNumber: 2,
+      format: "RAR",
+      pattern: "ARCHIVE_NUMBERED",
+    });
+  });
+
+  it("recognizes a name with a trailing dot", () => {
+    expect(detectArchive("Pack.zip.")).toEqual({
+      baseName: "Pack",
+      partNumber: -1,
+      format: "ZIP",
+      pattern: "SINGLE",
+    });
+  });
+
+  it("recognizes a name with a trailing dot followed by a space", () => {
+    expect(detectArchive("Pack.part3.rar. ")?.pattern).toBe("RAR_PART");
+  });
+
+  it("still returns null for a whitespace-only or empty name", () => {
+    expect(detectArchive("   ")).toBeNull();
+    expect(detectArchive("")).toBeNull();
+  });
+});
+
+describe("detectArchive — self-extracting RAR first volume (.partN.exe)", () => {
+  it("recognizes Pack.part1.exe as the first volume of a RAR_PART set", () => {
+    expect(detectArchive("Pack.part1.exe")).toEqual({
+      baseName: "Pack",
+      partNumber: 1,
+      format: "RAR",
+      pattern: "RAR_PART",
+    });
+  });
+
+  it("groups the SFX first volume with its .rar continuation volumes", () => {
+    const sfx = detectArchive("Pack.part1.exe");
+    const cont = detectArchive("Pack.part2.rar");
+    expect(sfx?.baseName).toBe(cont?.baseName);
+    expect(sfx?.format).toBe(cont?.format);
+  });
+
+  it("does not open the door to arbitrary .exe attachments", () => {
+    expect(detectArchive("Installer.exe")).toBeNull();
+    expect(detectArchive("Pack.exe")).toBeNull();
+  });
+});
+
+describe("detectArchive — standalone documents", () => {
+  it("keeps recognizing the pre-existing document extensions", () => {
+    expect(detectArchive("Model.stl")).toEqual({
+      baseName: "Model",
+      partNumber: -1,
+      format: "DOCUMENT",
+      pattern: "SINGLE",
+    });
+    expect(detectArchive("Sheet.pdf")?.format).toBe("DOCUMENT");
+  });
+
+  it("recognizes slicer-project formats", () => {
+    for (const name of [
+      "Bust.lys",
+      "Bust.lyt",
+      "Bust.lymesh",
+      "Bust.chitubox",
+      "Bust.ctp",
+      "Bust.ctb",
+      "Bust.cbddlp",
+      "Bust.photon",
+      "Bust.pwmx",
+      "Bust.pwmo",
+      "Bust.pws",
+      "Bust.sl1",
+      "Bust.goo",
+      "Bust.phz",
+      "Bust.pm3",
+      "Bust.form",
+    ]) {
+      expect(detectArchive(name), name).toEqual({
+        baseName: "Bust",
+        partNumber: -1,
+        format: "DOCUMENT",
+        pattern: "SINGLE",
+      });
+    }
+  });
+
+  it("recognizes 3D-model and CAD formats", () => {
+    for (const name of [
+      "Bust.fbx",
+      "Bust.ply",
+      "Bust.glb",
+      "Bust.gltf",
+      "Bust.3ds",
+      "Bust.max",
+      "Bust.c4d",
+      "Bust.ztl",
+      "Bust.zpr",
+      "Bust.mtl",
+      "Bust.f3d",
+      "Bust.scad",
+      "Bust.igs",
+      "Bust.iges",
+      "Bust.sldprt",
+      "Bust.skp",
+      "Bust.wrl",
+    ]) {
+      expect(detectArchive(name), name).toEqual({
+        baseName: "Bust",
+        partNumber: -1,
+        format: "DOCUMENT",
+        pattern: "SINGLE",
+      });
+    }
+  });
+
+  it("recognizes the .blend1 autosave sibling of .blend", () => {
+    expect(detectArchive("Scene.blend")?.format).toBe("DOCUMENT");
+    expect(detectArchive("Scene.blend1")).toEqual({
+      baseName: "Scene",
+      partNumber: -1,
+      format: "DOCUMENT",
+      pattern: "SINGLE",
+    });
+  });
+
+  it("does NOT recognize image attachments (they must not become packages)", () => {
+    for (const name of ["Preview.jpg", "Preview.jpeg", "Preview.png", "Preview.webp", "Preview.gif"]) {
+      expect(detectArchive(name), name).toBeNull();
+    }
+  });
+
+  it("returns null for unrelated files", () => {
+    expect(detectArchive("notes.txt")).toBeNull();
+    expect(detectArchive("song.mp3")).toBeNull();
+    expect(detectArchive("noextension")).toBeNull();
+  });
 });

worker/src/archive/detect.ts

@@ -4,25 +4,29 @@ export interface MultipartInfo {
   baseName: string;
   partNumber: number;
   format: ArchiveFormat;
-  pattern: "ZIP_NUMBERED" | "ZIP_LEGACY" | "RAR_PART" | "RAR_LEGACY" | "SEVENZ_NUMBERED" | "SINGLE";
+  pattern: "ARCHIVE_NUMBERED" | "ZIP_LEGACY" | "RAR_PART" | "RAR_LEGACY" | "SINGLE";
 }
 
 const patterns: {
   regex: RegExp;
-  format: ArchiveFormat;
+  /** A fixed format, or one derived from the match for patterns spanning several formats. */
+  format: ArchiveFormat | ((match: RegExpMatchArray) => ArchiveFormat);
   pattern: MultipartInfo["pattern"];
   getBaseName: (match: RegExpMatchArray) => string;
   getPartNumber: (match: RegExpMatchArray) => number;
 }[] = [
-  // pack.zip.001, pack.zip.002
+  // pack.zip.001, pack.rar.001, pack.7z.001 (numbered volume split — one pattern for
+  // every format, so a new format can never be silently dropped for lack of its own entry).
+  // {2,} digits also picks up hand-renamed sets like pack.rar.01.
   {
-    regex: /^(.+\.zip)\.(\d{3,})$/i,
-    format: "ZIP",
-    pattern: "ZIP_NUMBERED",
-    getBaseName: (m) => m[1],
-    getPartNumber: (m) => parseInt(m[2], 10),
+    regex: /^(.+\.(zip|7z|rar))\.(\d{2,})$/i,
+    // The regex only ever captures zip/7z/rar, so uppercasing yields a valid ArchiveFormat.
+    format: (m) => m[2].toUpperCase() as ArchiveFormat,
+    pattern: "ARCHIVE_NUMBERED",
+    getBaseName: (m) => m[1], // includes the archive extension
+    getPartNumber: (m) => parseInt(m[3], 10),
   },
-  // pack.z01, pack.z02 (legacy split — final part is pack.zip)
+  // pack.z01, pack.z02 (legacy split — pack.zip is the FINAL disk of the set)
   {
     regex: /^(.+)\.z(\d{2,})$/i,
     format: "ZIP",
@@ -30,15 +34,16 @@ const patterns: {
     getBaseName: (m) => m[1],
     getPartNumber: (m) => parseInt(m[2], 10),
   },
-  // pack.part1.rar, pack.part2.rar
+  // pack.part1.rar, pack.part2.rar — .exe covers a self-extracting first volume
+  // (pack.part1.exe + pack.part2.rar + ...), which is still a RAR volume set.
   {
-    regex: /^(.+)\.part(\d+)\.rar$/i,
+    regex: /^(.+)\.part(\d+)\.(rar|exe)$/i,
     format: "RAR",
     pattern: "RAR_PART",
     getBaseName: (m) => m[1],
     getPartNumber: (m) => parseInt(m[2], 10),
   },
-  // pack.r00, pack.r01 (legacy split — final part is pack.rar)
+  // pack.r00, pack.r01 (legacy split — pack.rar is the FIRST volume, .r00 onwards follow it)
   {
     regex: /^(.+)\.r(\d{2,})$/i,
     format: "RAR",
@@ -46,50 +51,53 @@ const patterns: {
     getBaseName: (m) => m[1],
     getPartNumber: (m) => parseInt(m[2], 10),
   },
-  // pack.7z.001, pack.7z.002 (native 7z volume split)
-  {
-    regex: /^(.+\.7z)\.(\d{3,})$/i,
-    format: "7Z",
-    pattern: "SEVENZ_NUMBERED",
-    getBaseName: (m) => m[1],
-    getPartNumber: (m) => parseInt(m[2], 10),
-  },
 ];
 
-/** Extensions we recognize as fetchable documents (archives + standalone files) */
-const DOCUMENT_EXTENSIONS = /\.(pdf|stl|obj|3mf|step|stp|blend|gcode|svg|dxf|ai|eps|psd)$/i;
+/** Extensions we recognize as fetchable documents (archives + standalone files).
+ *  Deliberately excludes image formats — previews posted as uncompressed documents
+ *  must go through the photo-matching path, not become packages of their own. */
+const DOCUMENT_EXTENSIONS =
+  /\.(pdf|stl|obj|3mf|step|stp|blend1|blend|gcode|svg|dxf|ai|eps|psd|lys|lyt|lymesh|chitubox|ctp|ctb|cbddlp|photon|pwmx|pwmo|pws|sl1|goo|phz|pm3|fbx|ply|glb|gltf|3ds|max|c4d|ztl|zpr|mtl|f3d|scad|igs|iges|sldprt|form|skp|wrl)$/i;
 
 /**
  * Detect if a filename is an archive and extract multipart info.
  */
 export function detectArchive(fileName: string): MultipartInfo | null {
+  // TDLib hands us `document.file_name` verbatim and every pattern below is `$`-anchored,
+  // so "Pack.zip " or "Pack.zip." would otherwise be dropped without a trace. Trailing dots
+  // are stripped too: no filesystem or archiver can produce a meaningful one (Windows
+  // silently drops them), so a trailing dot is always cosmetic damage from a re-upload,
+  // never part of the real name. The normalized value is used for matching AND baseName.
+  const name = fileName.trim().replace(/[.\s]+$/, "");
+  if (!name) return null;
+
   // Check multipart patterns first
   for (const p of patterns) {
-    const match = fileName.match(p.regex);
+    const match = name.match(p.regex);
     if (match) {
       return {
         baseName: p.getBaseName(match),
         partNumber: p.getPartNumber(match),
-        format: p.format,
+        format: typeof p.format === "function" ? p.format(match) : p.format,
         pattern: p.pattern,
       };
     }
   }
 
   // Single .zip file — could be a standalone or the final part of a ZIP_LEGACY set
-  if (/\.zip$/i.test(fileName)) {
+  if (/\.zip$/i.test(name)) {
     return {
-      baseName: fileName.replace(/\.zip$/i, ""),
+      baseName: name.replace(/\.zip$/i, ""),
       partNumber: -1, // -1 signals "could be single or final legacy part"
       format: "ZIP",
       pattern: "SINGLE",
     };
   }
 
-  // Single .rar file — could be standalone or final part of RAR_LEGACY set
-  if (/\.rar$/i.test(fileName)) {
+  // Single .rar file — could be standalone or the FIRST part of a RAR_LEGACY set
+  if (/\.rar$/i.test(name)) {
     return {
-      baseName: fileName.replace(/\.rar$/i, ""),
+      baseName: name.replace(/\.rar$/i, ""),
       partNumber: -1,
       format: "RAR",
       pattern: "SINGLE",
@@ -97,20 +105,19 @@ export function detectArchive(fileName: string): MultipartInfo | null {
   }
 
   // Single .7z file
-  if (/\.7z$/i.test(fileName)) {
+  if (/\.7z$/i.test(name)) {
     return {
-      baseName: fileName.replace(/\.7z$/i, ""),
+      baseName: name.replace(/\.7z$/i, ""),
       partNumber: -1,
       format: "7Z",
       pattern: "SINGLE",
     };
   }
 
-  // Standalone documents (PDFs, STLs, 3D files, etc.)
-  if (DOCUMENT_EXTENSIONS.test(fileName)) {
-    const ext = fileName.match(DOCUMENT_EXTENSIONS)![0];
+  // Standalone documents (PDFs, STLs, 3D files, slicer projects, etc.)
+  if (DOCUMENT_EXTENSIONS.test(name)) {
     return {
-      baseName: fileName.replace(DOCUMENT_EXTENSIONS, ""),
+      baseName: name.replace(DOCUMENT_EXTENSIONS, ""),
       partNumber: -1,
       format: "DOCUMENT",
       pattern: "SINGLE",

worker/src/archive/multipart.test.ts

@@ -0,0 +1,72 @@
+import { describe, it, expect } from "vitest";
+import { groupArchiveSets, type TelegramMessage } from "./multipart.js";
+
+let nextId = 1000n;
+
+function msg(fileName: string): TelegramMessage {
+  const id = nextId++;
+  return {
+    id,
+    fileName,
+    fileId: `file-${id}`,
+    fileSize: 1024n,
+    date: new Date("2026-01-01T00:00:00Z"),
+  };
+}
+
+function names(files: string[]): string[] {
+  const sets = groupArchiveSets(files.map(msg));
+  expect(sets).toHaveLength(1);
+  return sets[0].parts.map((p) => p.fileName);
+}
+
+describe("groupArchiveSets — legacy split part ordering", () => {
+  it("puts the bare .rar FIRST in a RAR_LEGACY set (it is volume 1)", () => {
+    expect(names(["Pack.r01", "Pack.rar", "Pack.r00"])).toEqual([
+      "Pack.rar",
+      "Pack.r00",
+      "Pack.r01",
+    ]);
+  });
+
+  it("puts the bare .zip LAST in a ZIP_LEGACY set (it is the final disk)", () => {
+    expect(names(["Pack.z02", "Pack.zip", "Pack.z01"])).toEqual([
+      "Pack.z01",
+      "Pack.z02",
+      "Pack.zip",
+    ]);
+  });
+
+  it("marks both legacy sets as multipart with the right format", () => {
+    const rar = groupArchiveSets([msg("Pack.rar"), msg("Pack.r00")])[0];
+    expect(rar.isMultipart).toBe(true);
+    expect(rar.type).toBe("RAR");
+
+    const zip = groupArchiveSets([msg("Pack.zip"), msg("Pack.z01")])[0];
+    expect(zip.isMultipart).toBe(true);
+    expect(zip.type).toBe("ZIP");
+  });
+
+  it("orders numbered volume sets by part number", () => {
+    expect(names(["Pack.rar.003", "Pack.rar.001", "Pack.rar.002"])).toEqual([
+      "Pack.rar.001",
+      "Pack.rar.002",
+      "Pack.rar.003",
+    ]);
+  });
+
+  it("orders .partN sets by part number with an SFX first volume", () => {
+    expect(names(["Pack.part3.rar", "Pack.part1.exe", "Pack.part2.rar"])).toEqual([
+      "Pack.part1.exe",
+      "Pack.part2.rar",
+      "Pack.part3.rar",
+    ]);
+  });
+
+  it("treats unrelated singles as their own non-multipart sets", () => {
+    const sets = groupArchiveSets([msg("A.zip"), msg("B.rar")]);
+    expect(sets).toHaveLength(2);
+    expect(sets.every((s) => !s.isMultipart)).toBe(true);
+    expect(sets.every((s) => s.parts.length === 1)).toBe(true);
+  });
+});

worker/src/archive/multipart.ts

@@ -78,10 +78,17 @@ export function groupArchiveSets(messages: TelegramMessage[]): ArchiveSet[] {
         }
       }
 
-      // Sort by part number (singles get a very high number so they come last — they're the final part)
+      // Sort by part number. A bare single (partNumber -1) sits at a different end of the
+      // set depending on the legacy scheme: in a .zip/.z01/.z02 set the bare pack.zip is the
+      // FINAL disk, but in a .rar/.r00/.r01 set the bare pack.rar is volume 1 and .r00
+      // onwards follow it. Getting this backwards makes parts[0] a headerless continuation
+      // volume, which breaks listing and mislabels the package.
+      const singleRank = multipartEntries.some((e) => e.info.pattern === "RAR_LEGACY")
+        ? -1 // before .r00
+        : 999999;
       allEntries.sort((a, b) => {
-        const aNum = a.info.partNumber === -1 ? 999999 : a.info.partNumber;
-        const bNum = b.info.partNumber === -1 ? 999999 : b.info.partNumber;
+        const aNum = a.info.partNumber === -1 ? singleRank : a.info.partNumber;
+        const bNum = b.info.partNumber === -1 ? singleRank : b.info.partNumber;
         return aNum - bNum;
       });
 

worker/src/rebuild.ts

@@ -333,6 +333,14 @@ async function scanDestinationChannel(
           fileSize: BigInt(doc.document.size),
           date: new Date(msg.date * 1000),
         });
+      } else if (doc?.file_name) {
+        // Not matched by any pattern in archive/detect.ts, so it is dropped without a
+        // packages/skipped_packages row. Grep "unrecognized attachment" to find naming
+        // schemes we do not handle yet.
+        log.debug(
+          { chatId: chatId.toString(), messageId: msg.id, fileName: doc.file_name },
+          "Skipping unrecognized attachment (no archive/document pattern matched)"
+        );
       }
     }
 

worker/src/tdlib/download.ts

@@ -267,6 +267,15 @@ export async function getChannelMessages(
           });
           continue;
         }
+        if (doc?.file_name) {
+          // Not matched by any pattern in archive/detect.ts, so it is dropped without a
+          // packages/skipped_packages row. Grep "unrecognized attachment" to find naming
+          // schemes we do not handle yet.
+          log.debug(
+            { chatId: chatId.toString(), messageId: msg.id, fileName: doc.file_name },
+            "Skipping unrecognized attachment (no archive/document pattern matched)"
+          );
+        }
 
         // Check for photo messages (potential previews)
         const photo = msg.content?.photo;

worker/src/tdlib/topics.ts

@@ -280,6 +280,15 @@ export async function getTopicMessages(
         });
         continue;
       }
+      if (doc?.file_name) {
+        // Not matched by any pattern in archive/detect.ts, so it is dropped without a
+        // packages/skipped_packages row. Grep "unrecognized attachment" to find naming
+        // schemes we do not handle yet.
+        log.debug(
+          { chatId: chatId.toString(), topicId: topicId.toString(), messageId: msg.id, fileName: doc.file_name },
+          "Skipping unrecognized attachment (no archive/document pattern matched)"
+        );
+      }
 
       // Check for photo messages (potential previews)
       const photo = msg.content?.photo;