From fbaaf6ea3d04403181de0ae79d916abbc99e7556 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Fran=C3=A7ois?= <francois@kaz.bzh>
Date: Sat, 17 Dec 2022 08:47:22 +0100
Subject: [PATCH] fix unexpected boundary not in multipart

---
 src/cpp/Attachment.cpp | 61 +++++++++++++++++++-----------------------
 src/cpp/kazMisc.cpp    |  4 +--
 2 files changed, 29 insertions(+), 36 deletions(-)

diff --git a/src/cpp/Attachment.cpp b/src/cpp/Attachment.cpp
index 9d2c11f..069ac81 100644
--- a/src/cpp/Attachment.cpp
+++ b/src/cpp/Attachment.cpp
@@ -73,7 +73,7 @@ const regex Attachment::nameRegEx			(".*name=[ \t]*((\"(\\\\.|[^\\\\\r])*\")|[^\
 // boundary="_000_PAVPR10MB6792713B313048E3A259B215B2079PAVPR10MB6792EURP_"
 // boundary=--boundary_1351_64006126-2b0e-4a3b-98ac-4797d1634188
 // boundary=--boundary_1352_7e294c9a-cfab-44a0-bfb3-7310380ac7cb;
-const regex Attachment::boundaryRegEx			(".*boundary=[ \t]*((\"(\\\\.|[^\\\\\r])*\")|[^\r; ]*);?.*");
+const regex Attachment::boundaryRegEx			(".*boundary=[ \t]*((\"(\\\\.|[^\\\\])*\")|[^; ]*);?.*");
 const regex Attachment::cidDefRegEx			(".*<([^>]*)>.*");
 const regex Attachment::textRegEx			(".*text/("+PLAIN+"|"+HTML+").*");
 const regex Attachment::multiRegEx			("\\s*multipart/(mixed|"+RELATED+"|"+ALTERNATIVE+").*");
@@ -85,6 +85,7 @@ const string Attachment::IMG_END			(">");
 
 static const string SRC_BEGIN ("SRC=\"");
 static const string RFC822 ("message/rfc822");
+static const string MULTIPART ("multipart/");
 
 // ================================================================================
 string
@@ -182,37 +183,26 @@ Attachment::getContentType () const {
 
 const string
 Attachment::getAttachName () const {
+  static string tokens [] = {contentTypeToken, contentDispositionToken};
   DEF_LOG ("Attachment::getAttachName", "");
-  string result = getProp (contentTypeToken, nameRegEx);
-  removeQuote (result);
-  if (result.length ()) {
-    LOG ("name=: " << result);
-    encodedWord (result);
-    return result;
-  }
-  result = getProp (contentTypeToken, nameCharsetRegEx);
-  removeQuote (result);
-  if (result.length ()) {
-    LOG ("name*=: " << result);
-    charsetValue (result);
-    return result;
+  for (string token : tokens) {
+    string result = getProp (token, nameRegEx);
+    removeQuote (result);
+    if (result.length ()) {
+      LOG ("name=: " << result);
+      encodedWord (result);
+      return result;
+    }
+    result = getProp (token, nameCharsetRegEx);
+    removeQuote (result);
+    if (result.length ()) {
+      LOG ("name*=: " << result);
+      charsetValue (result);
+      return result;
+    }
   }
   // XXX il faut composer s'il y a plusieurs ligne filename*x=
-  result = getProp (contentDispositionToken, nameRegEx);
-  removeQuote (result);
-  if (result.length ()) {
-    LOG ("filename=: " << result);
-    encodedWord (result);
-    return result;
-  }
   // XXX il faut composer s'il y a plusieurs ligne filename*x*=
-  result = getProp (contentDispositionToken, nameCharsetRegEx);
-  removeQuote (result);
-  if (result.length ()) {
-    LOG ("filename*=: " << result);
-    charsetValue (result);
-    return result;
-  }
   return getUnknown (getContentType ());
 }
 
@@ -328,12 +318,15 @@ Attachment::readMime (ifstream &mbox, streamoff &curPos) {
 
   contentPos = curPos;
   cid = getProp (contentIDToken, cidDefRegEx);
-  boundary = getProp (contentTypeToken, boundaryRegEx);
-  removeQuote (boundary);
-  LOG ("boundary: " << boundary);
-  if (boundary.length ()) {
-    boundary = "--"+boundary+"--";
-    boundaryMiddleSize = boundary.length () - 2;
+
+  if (caseInsensitiveFind (getContentType (), MULTIPART) != string::npos) {
+    boundary = getProp (contentTypeToken, boundaryRegEx);
+    removeQuote (boundary);
+    LOG ("boundary: " << boundary);
+    if (boundary.length ()) {
+      boundary = "--"+boundary+"--";
+      boundaryMiddleSize = boundary.length () - 2;
+    }
   }
   LOG ("readMime contentPos: " << contentPos << " cid: " << cid << " boundary: " << boundary);
 }
diff --git a/src/cpp/kazMisc.cpp b/src/cpp/kazMisc.cpp
index 9982afa..ae31a52 100644
--- a/src/cpp/kazMisc.cpp
+++ b/src/cpp/kazMisc.cpp
@@ -186,13 +186,13 @@ kaz::quotedDecode (string &content) {
     }
     if (p+1 < content.end () && *(p+1) == '\n') {
 
-      LOG_BUG (q == content.begin (), ++p;continue, "kazMisc::quotedDecode bug: bad quoted-printable format. (start with '=', content: " << content << ")");
+      LOG_BUG (q == content.begin (), ++p;continue, "kazMisc::quotedDecode bug: bad quoted-printable format. (start with '=', delim: " << int (delim) << " content: " << content << ")");
       ++p;
       --q;
       continue;
     }
 
-    LOG_BUG (p+3 > content.end () || !isxdigit (p[1]) || !isxdigit (p[2]), return, "kazMisc::quotedDecode bug: bad quoted-printable format. (content: " << content << ")");
+    LOG_BUG (p+3 > content.end () || !isxdigit (p[1]) || !isxdigit (p[2]), return, "kazMisc::quotedDecode bug: bad quoted-printable format. (delim: " << int (delim)  << " content: " << content << ")");
     *q = (char) ((getHexaVal (p[1]) << 4) + getHexaVal (p[2]));
     p += 2;
   }