Move HTML→Markdown and Unicode-abuse sanitization into Post

Package.swift

@@ -23,7 +23,7 @@ let package = Package(
     dependencies: [
         .package(url: "https://github.com/Cocoanetics/SwiftMCP", .upToNextMajor(from: "1.4.3")),
         .package(url: "https://github.com/Cocoanetics/SwiftMail", .upToNextMajor(from: "1.4.0")),
-        .package(url: "https://github.com/Cocoanetics/SwiftText", .upToNextMajor(from: "1.1.4")),
+        .package(url: "https://github.com/Cocoanetics/SwiftText", .upToNextMajor(from: "1.1.7")),
         .package(url: "https://github.com/apple/swift-argument-parser", from: "1.3.0"),
         .package(url: "https://github.com/apple/swift-log.git", from: "1.0.0")
     ],
@@ -38,6 +38,7 @@ let package = Package(
                 .product(name: "SwiftMCP", package: "SwiftMCP"),
                 .product(name: "SwiftMail", package: "SwiftMail"),
                 .product(name: "SwiftTextHTML", package: "SwiftText"),
+                .product(name: "SwiftTextCore", package: "SwiftText"),
                 .product(name: "Logging", package: "swift-log")
             ],
             plugins: [

Sources/PostServer/Models/MessageDetail.swift

@@ -1,5 +1,6 @@
 import Foundation
 import SwiftMCP
+import SwiftTextCore
 import SwiftTextHTML
 
 @Schema
@@ -19,6 +20,8 @@ public struct MessageDetail: Codable, Sendable {
     public let messageId: String?
     /// The References header value (RFC 822, space-separated Message-IDs)
     public let references: String?
+    /// Optional description of Unicode abuse removed from subject and/or body.
+    public let unicodeAbuse: String?
 
     public init(
         uid: Int,
@@ -32,7 +35,8 @@ public struct MessageDetail: Codable, Sendable {
         attachments: [AttachmentInfo],
         additionalHeaders: [String: String]? = nil,
         messageId: String? = nil,
-        references: String? = nil
+        references: String? = nil,
+        unicodeAbuse: String? = nil
     ) {
         self.uid = uid
         self.from = from
@@ -46,22 +50,40 @@ public struct MessageDetail: Codable, Sendable {
         self.additionalHeaders = additionalHeaders
         self.messageId = messageId
         self.references = references
+        self.unicodeAbuse = unicodeAbuse
     }
 }
 
 extension MessageDetail {
+    public func sanitizedSubject() -> SanitizedText {
+        UnicodeAbuseSummary.sanitize(subject, field: "Subject")
+    }
+
+    public func sanitizedTextBody() -> SanitizedText {
+        UnicodeAbuseSummary.sanitize(textBody ?? "", field: "Body")
+    }
+
+    public func sanitizedHTMLBody() -> SanitizedText {
+        UnicodeAbuseSummary.sanitize(htmlBody ?? textBody ?? "", field: "Body")
+    }
+
     /// Returns the message body as markdown.
     /// Converts HTML to markdown when available, falls back to plain text.
     public func markdown() async throws -> String {
+        try await markdownSanitized().text
+    }
+
+    public func markdownSanitized() async throws -> SanitizedText {
         if let htmlBody, !htmlBody.isEmpty {
             let converter = HTMLToMarkdown(data: Data(htmlBody.utf8))
-            return try await converter.markdown()
+            let raw = try await converter.markdown()
+            return UnicodeAbuseSummary.sanitize(raw, field: "Body")
         }
 
         if let textBody, !textBody.isEmpty {
-            return textBody
+            return UnicodeAbuseSummary.sanitize(textBody, field: "Body")
         }
 
-        return ""
+        return SanitizedText(text: "")
     }
 }

Sources/PostServer/Models/MessageHeader+Sanitization.swift

@@ -0,0 +1,7 @@
+import Foundation
+
+public extension MessageHeader {
+    func sanitizedSubject() -> SanitizedText {
+        UnicodeAbuseSummary.sanitize(subject, field: "Subject")
+    }
+}

Sources/PostServer/Models/MessageHeader.swift

@@ -8,12 +8,22 @@ public struct MessageHeader: Codable, Sendable {
     public let subject: String
     public let date: String
     public let flags: MessageFlags
+    /// Optional description of Unicode abuse removed from subject.
+    public let unicodeAbuse: String?
 
-    public init(uid: Int, from: String, subject: String, date: String, flags: MessageFlags) {
+    public init(
+        uid: Int,
+        from: String,
+        subject: String,
+        date: String,
+        flags: MessageFlags,
+        unicodeAbuse: String? = nil
+    ) {
         self.uid = uid
         self.from = from
         self.subject = subject
         self.date = date
         self.flags = flags
+        self.unicodeAbuse = unicodeAbuse
     }
 }

Sources/PostServer/PostServer+IDLE.swift

@@ -347,7 +347,10 @@ extension PostServer {
         let replyTo = extractReplyTo(from: decodedAdditionalHeaders)
         let decodedFrom = decodeHeaderValue(messageInfo.from ?? header.from)
         let decodedTo = decodeRecipientList(messageInfo.to)
-        let decodedSubject = decodeHeaderValue(messageInfo.subject ?? header.subject)
+        let decodedSubject = UnicodeAbuseSummary.sanitize(
+            decodeHeaderValue(messageInfo.subject ?? header.subject),
+            field: "Subject"
+        )
         let attachmentParts = messageInfo.parts.filter(isAttachmentPart)
         let attachments: [HookAttachmentPayload] = attachmentParts.map { part in
             let filename = canonicalAttachmentFilename(part)
@@ -368,7 +371,7 @@ extension PostServer {
             from: decodedFrom,
             to: decodedTo,
             replyTo: replyTo,
-            subject: decodedSubject,
+            subject: decodedSubject.text,
             date: formatHookDate(resolvedDate)
         )
         return HookMessagePayload(
@@ -378,8 +381,9 @@ extension PostServer {
             to: decodedTo,
             replyTo: replyTo,
             date: resolvedDate,
-            subject: decodedSubject,
-            markdown: markdown,
+            subject: decodedSubject.text,
+            markdown: markdown?.text,
+            unicodeAbuse: UnicodeAbuseSummary.combine([decodedSubject.unicodeAbuse, markdown?.unicodeAbuse]),
             flags: messageInfo.flags.map(Self.flagToString),
             attachments: attachments,
             headers: headers
@@ -406,15 +410,17 @@ extension PostServer {
         header: MessageHeader
     ) async -> HookMessagePayload {
         guard (1...Int(UInt32.max)).contains(header.uid) else {
+            let subject = UnicodeAbuseSummary.sanitize(decodeHeaderValue(header.subject), field: "Subject")
             return HookMessagePayload(
                 uid: header.uid,
                 mailbox: mailbox,
                 from: decodeHeaderValue(header.from),
                 to: [],
                 replyTo: nil,
                 date: resolveHookDate(messageDate: nil, headerDate: header.date),
-                subject: decodeHeaderValue(header.subject),
+                subject: subject.text,
                 markdown: nil,
+                unicodeAbuse: subject.unicodeAbuse,
                 flags: [],
                 attachments: [],
                 headers: [:]
@@ -425,15 +431,17 @@ extension PostServer {
             _ = try await connection.selectMailbox(mailbox)
             let identifier = UID(UInt32(header.uid))
             guard let messageInfo = try await connection.fetchMessageInfo(for: identifier) else {
+                let subject = UnicodeAbuseSummary.sanitize(decodeHeaderValue(header.subject), field: "Subject")
                 return HookMessagePayload(
                     uid: header.uid,
                     mailbox: mailbox,
                     from: decodeHeaderValue(header.from),
                     to: [],
                     replyTo: nil,
                     date: resolveHookDate(messageDate: nil, headerDate: header.date),
-                    subject: decodeHeaderValue(header.subject),
+                    subject: subject.text,
                     markdown: nil,
+                    unicodeAbuse: subject.unicodeAbuse,
                     flags: [],
                     attachments: [],
                     headers: [:]
@@ -452,7 +460,10 @@ extension PostServer {
             let replyTo = extractReplyTo(from: decodedAdditionalHeaders)
             let decodedFrom = decodeHeaderValue(messageInfo.from ?? header.from)
             let decodedTo = decodeRecipientList(messageInfo.to)
-            let decodedSubject = decodeHeaderValue(messageInfo.subject ?? header.subject)
+            let decodedSubject = UnicodeAbuseSummary.sanitize(
+                decodeHeaderValue(messageInfo.subject ?? header.subject),
+                field: "Subject"
+            )
             let attachmentParts = messageInfo.parts.filter(isAttachmentPart)
             let attachments: [HookAttachmentPayload] = attachmentParts.map { part in
                 let filename = canonicalAttachmentFilename(part)
@@ -473,7 +484,7 @@ extension PostServer {
                 from: decodedFrom,
                 to: decodedTo,
                 replyTo: replyTo,
-                subject: decodedSubject,
+                subject: decodedSubject.text,
                 date: formatHookDate(resolvedDate)
             )
             return HookMessagePayload(
@@ -483,8 +494,9 @@ extension PostServer {
                 to: decodedTo,
                 replyTo: replyTo,
                 date: resolvedDate,
-                subject: decodedSubject,
-                markdown: markdown,
+                subject: decodedSubject.text,
+                markdown: markdown?.text,
+                unicodeAbuse: UnicodeAbuseSummary.combine([decodedSubject.unicodeAbuse, markdown?.unicodeAbuse]),
                 flags: messageInfo.flags.map(Self.flagToString),
                 attachments: attachments,
                 headers: headers
@@ -493,15 +505,17 @@ extension PostServer {
             Self.logDiagnostic("ERROR failed to fetch hook message details for \(mailbox) uid=\(header.uid): \(String(describing: error))")
         }
 
+        let subject = UnicodeAbuseSummary.sanitize(decodeHeaderValue(header.subject), field: "Subject")
         return HookMessagePayload(
             uid: header.uid,
             mailbox: mailbox,
             from: decodeHeaderValue(header.from),
             to: [],
             replyTo: nil,
             date: resolveHookDate(messageDate: nil, headerDate: header.date),
-            subject: decodeHeaderValue(header.subject),
+            subject: subject.text,
             markdown: nil,
+            unicodeAbuse: subject.unicodeAbuse,
             flags: [],
             attachments: [],
             headers: [:]
@@ -527,7 +541,7 @@ extension PostServer {
     /// Fetches ALL RFC 822 headers by fetching the raw message and parsing headers.
     /// This is a workaround for SwiftMail not populating MessageInfo.additionalFields.
     /// Produces markdown by fetching only text/html body parts (no attachment download).
-    fileprivate static func fetchHookMarkdown(using connection: IMAPNamedConnection, messageInfo: MessageInfo) async -> String? {
+    fileprivate static func fetchHookMarkdown(using connection: IMAPNamedConnection, messageInfo: MessageInfo) async -> SanitizedText? {
         let textPart = messageInfo.parts.first { part in
             part.contentType.lowercased().hasPrefix("text/plain")
                 && part.disposition?.lowercased() != "attachment"
@@ -585,7 +599,7 @@ extension PostServer {
         )
 
         do {
-            return try await detail.markdown()
+            return try await detail.markdownSanitized()
         } catch {
             let uid = messageInfo.uid?.value ?? 0
             Self.logDiagnostic("ERROR failed to convert body to markdown uid=\(uid): \(String(describing: error))")
@@ -609,7 +623,7 @@ extension PostServer {
                 }
             }
 
-            return textBody
+            return UnicodeAbuseSummary.sanitize(textBody ?? "", field: "Body")
         }
     }
 

Sources/PostServer/PostServer.swift

@@ -120,6 +120,7 @@ public actor PostServer {
         let date: Date
         let subject: String
         let markdown: String?
+        let unicodeAbuse: String?
         let flags: [String]
         let attachments: [HookAttachmentPayload]
         let headers: [String: String]
@@ -131,6 +132,7 @@ public actor PostServer {
             case date
             case subject
             case markdown
+            case unicodeAbuse
             case flags
             case attachments
             case headers
@@ -144,6 +146,7 @@ public actor PostServer {
             try container.encode(date, forKey: .date)
             try container.encode(subject, forKey: .subject)
             try container.encode(markdown ?? "", forKey: .markdown)
+            try container.encodeIfPresent(unicodeAbuse, forKey: .unicodeAbuse)
             try container.encode(flags, forKey: .flags)
             try container.encode(attachments, forKey: .attachments)
             try container.encode(headers, forKey: .headers)
@@ -1359,12 +1362,15 @@ public actor PostServer {
     }
 
     internal func messageHeader(from message: Message) -> MessageHeader {
-        MessageHeader(
+        let sanitizedSubject = UnicodeAbuseSummary.sanitize(message.subject ?? "(No Subject)", field: "Subject")
+
+        return MessageHeader(
             uid: messageUID(from: message),
             from: message.from ?? "Unknown",
-            subject: message.subject ?? "(No Subject)",
+            subject: sanitizedSubject.text,
             date: formatDate(message.date),
-            flags: MessageFlags(message.flags)
+            flags: MessageFlags(message.flags),
+            unicodeAbuse: sanitizedSubject.unicodeAbuse
         )
     }
 
@@ -1380,20 +1386,28 @@ public actor PostServer {
         let referencesString = message.header.references?.map { $0.description }.joined(separator: " ")
 
         let filteredHeaders = Self.filterNoiseHeaders(additionalHeaders ?? message.header.additionalFields ?? [:])
+        let sanitizedSubject = UnicodeAbuseSummary.sanitize(message.subject ?? "(No Subject)", field: "Subject")
+        let sanitizedTextBody = message.textBody.map { UnicodeAbuseSummary.sanitize($0, field: "Body") }
+        let sanitizedHTMLBody = message.htmlBody.map { UnicodeAbuseSummary.sanitize($0, field: "Body") }
 
         return MessageDetail(
             uid: messageUID(from: message),
             from: message.from ?? "Unknown",
             to: message.to,
             cc: message.cc.isEmpty ? nil : message.cc,
-            subject: message.subject ?? "(No Subject)",
+            subject: sanitizedSubject.text,
             date: formatDate(message.date),
-            textBody: message.textBody,
-            htmlBody: message.htmlBody,
+            textBody: sanitizedTextBody?.text,
+            htmlBody: sanitizedHTMLBody?.text,
             attachments: attachments,
             additionalHeaders: filteredHeaders.isEmpty ? nil : filteredHeaders,
             messageId: messageId,
-            references: referencesString
+            references: referencesString,
+            unicodeAbuse: UnicodeAbuseSummary.combine([
+                sanitizedSubject.unicodeAbuse,
+                sanitizedTextBody?.unicodeAbuse,
+                sanitizedHTMLBody?.unicodeAbuse
+            ])
         )
     }