major cleanup, now handles nested multipart mime

3 files changed, 50 insertions, 41 deletions

diff --git a/tools/libskrewt.c b/tools/libskrewt.c
index dc131f9..595c3fe 100644
--- a/tools/libskrewt.c
+++ b/tools/libskrewt.c
@@ -6,8 +6,9 @@
 using namespace std;
 
 void parse_content(const string type_spec_line,
-        string &maintype, string &boundary) {
-  cerr << "parser called with: " << type_spec_line << endl;
+        string &maintype, string &boundary,
+        const string old_bdy) {
+  if (0) cerr << "parser called with: " << type_spec_line << endl;
   string get_type(type_spec_line);
 
   size_t where = get_type.find_first_of(" \t;\n");
@@ -46,7 +47,7 @@ void parse_content(const string type_spec_line,
       boundary = boundary.substr(0, where);
     }
   } else {
-    //xxxxxxx cerr << "boundary= not found in " << type_spec_line << endl;
+    boundary = old_bdy;
   }
 }
 
@@ -56,7 +57,7 @@ int skrewt::krunch_rfrom(){
   string word;
   parse >> word;
   if (word != "from") {
-    cerr << progid << " bad 'Received: from' line ... '"
+    cerr << progid << " bad 'Received: from' line; no from, instead '"
         << word << "'" << endl;
     return ex_syserr;
   }
@@ -136,8 +137,8 @@ xstr skrewt::getRecord(istream& xin){
           << " folded header rejection: bigger than " << maxsize << endl;
         exeunt(ex_spam);
       }
-      bigbuf.push_back(line.str);
-      headrec += "\n" + noCR(line.str);
+      bigbuf.push_back(more.str);
+      headrec += "\n" + noCR(more.str);
     }
     return {0, headrec};
 }
@@ -202,7 +203,7 @@ int skrewt::headers(istream& xin){
   return 0;
 }
 
-int skrewt::dump_headers(std::ostream& xout){
+int skrewt::dump_bigbuf(std::ostream& xout){
   for (vector<string>::const_iterator foo = bigbuf.begin();
         foo != bigbuf.end(); foo++){
     xout << *foo << endl;
@@ -394,11 +395,21 @@ public:
   string bdy;
 };
 
+void dump(const list<conner> sitch){
+  int depth = sitch.size();
+  cerr << "  depth: " << depth;
+  if (depth){
+    cerr << "  tp: " << sitch.front().tp
+         << "  bdy: " << sitch.front().bdy;
+  }
+  cerr << endl;
+}
+
 int skrewt::body(std::istream& xin, std::ostream& xout){
   list<conner> sitch;
   if (content_type.length()) {
     string tp, bdy;
-    parse_content(content_type, tp, bdy);
+    parse_content(content_type, tp, bdy, "");
     sitch.push_front({tp, bdy});
   } else {
     // assume single-part, all text if not otherwise specified:
@@ -409,10 +420,8 @@ int skrewt::body(std::istream& xin, std::ostream& xout){
 // Now spew the rest of the message
 // Note that multipart messages can have sub-headers.
 
-  cerr << "body begins: tp: " << sitch.front().tp
-       << "  bdy: " << sitch.front().bdy
-       << "  depth: " << sitch.size()
-       << endl;
+  cerr << "body begins: ";
+  dump(sitch);
   int in_subheads(0);
   int textlines(0);
 
@@ -420,29 +429,20 @@ int skrewt::body(std::istream& xin, std::ostream& xout){
   for (;;){             // outer loop over all lines in the body
     if (xin.eof()) break;
     if (xin.bad()) return 1;
-    string line;
-// on fail, go back to top of outer loop and check for eof versus bad
-    if (getline(xin, line).fail()) continue;
-    msgsize += line.length()+1;
-    if (msgsize > maxsize) {
-      cerr << progid << " rejection: bigger than " << maxsize << endl;
-      maybe_exeunt(ex_spam, error_exit);
-    }
-    bigbuf.push_back(line);
-    xout << line << endl;
+    xstr rec = getRecord(xin);
+    if (rec.err) break;
+    string line = rec.str;
     if (0) cerr << "+++" << line << endl;
     if (in_subheads){
       if (line == "" || line == "\r") {
         in_subheads = 0;
         cerr << "+++ end of subhead" << endl;
         string tp, bdy;
-        parse_content(sub_conntype, tp, bdy);
-        cerr << tp << "   '" << bdy << "'" << endl;
+        parse_content(sub_conntype, tp, bdy, sitch.front().bdy);
         sitch.push_front({tp, bdy});
-        cerr << "payload mode: tp: " << sitch.front().tp
-             << "  bdy: " << sitch.front().bdy
-             << "  depth: " << sitch.size()
-             << endl;
+
+        cerr << "payload mode: ";
+        dump(sitch);
       }
     }
     if (/* still */in_subheads){
@@ -463,20 +463,24 @@ int skrewt::body(std::istream& xin, std::ostream& xout){
           sub_conntype = rest;
         }
     } else {
-      if (sitch.front().tp == "multipart"
+      if (sitch.front().bdy != ""
            && line == "--" + sitch.front().bdy) {
-     cerr << "boundary: begin subhead block" << endl;
+        if (sitch.front().tp != "multipart")
+          sitch.pop_front();
+        cerr << "boundary: begin subhead block ";
+        dump(sitch);
         in_subheads = 1;
         sub_conntype = "";
         continue;
       }
-      if (sitch.front().tp == "multipart"
+      if (sitch.front().bdy != ""
            && line == "--" + sitch.front().bdy + "--") {
+        string dead_bdy = sitch.front().bdy;
         sitch.pop_front();
-        cerr << "found subhead termination --> "
-          << sitch.front().tp
-          << endl;
-        in_subheads = 1;
+        if (sitch.front().bdy == dead_bdy) sitch.pop_front();
+        cerr << "boundary: termination ";
+        dump(sitch);
+        in_subheads = 0;
         continue;
       }
       if (sitch.front().tp == "text") textlines++;
diff --git a/tools/libskrewt.h b/tools/libskrewt.h
index 5ae6ece..34cdfd3 100644
--- a/tools/libskrewt.h
+++ b/tools/libskrewt.h
@@ -31,30 +31,31 @@ public:
   std::string content_type;
   std::list<std::string> delivered_to;
   int msgsize;
-  std::vector<std::string> bigbuf;
   int saw_blank_line;
   int recno;
-
   int maxsize;
   int error_exit;
   int mid_required;
+  std::vector<std::string> headerbuf;
+  std::vector<std::string> bigbuf;
+  xstr lookahead;
 
   // constructor
   skrewt()
   : spf_result(SPF_RESULT_INVALID),
     boundary("x-xx-x"), msgsize(0), saw_blank_line(0), recno(0),
     maxsize(1000*1000), error_exit(0), mid_required(0),
+    headerbuf(0), bigbuf(0),
     lookahead(1, "")
   {}
 
   xstr getRecord(std::istream& xin);
   xstr getLine(std::istream& xin);
   int headers(std::istream& xin);
-  int dump_headers(std::ostream& xout);
+  int dump_bigbuf(std::ostream& xout);
   int interstage();
   int body(std::istream& xin, std::ostream& xout);
   int krunch_rfrom();
-  xstr lookahead;
 };
 
 void parse_content(const std::string type_spec_line,
diff --git a/tools/ward.c b/tools/ward.c
index b8bdee4..abea289 100644
--- a/tools/ward.c
+++ b/tools/ward.c
@@ -117,7 +117,9 @@ int main(int _argc, const char** _argv){
 
   int rslt = mysk.headers(cin);
   if (rslt) return rslt;
-  mysk.dump_headers(cout);
+  mysk.dump_bigbuf(cout);
+  mysk.headerbuf = mysk.bigbuf;
+  mysk.bigbuf = vector<string>(0);
 
 // Headers are done.
 // Do some early-stage thinking.
@@ -126,5 +128,7 @@ int main(int _argc, const char** _argv){
   if (rslt) return rslt;
 
   rslt = mysk.body(cin, cout);
-  return rslt;
+  if (rslt) return rslt;
+  mysk.dump_bigbuf(cout);
+  return 0;
 }