fix a bunch of DOS-CR bugs

3 files changed, 102 insertions, 33 deletions

diff --git a/tools/filters.conf b/tools/filters.conf
index bd8eb33..3cbd5bf 100644
--- a/tools/filters.conf
+++ b/tools/filters.conf
@@ -1,5 +1,5 @@
 # configuration file for hi-q
-series  /var/qmail/bin/skrewt
+series  /var/qmail/bin/skrewt -err
 stub    /var/qmail/bin/greylist -check -v
 sa      /usr/local/bin/spamc -Y 0 -s 1000000 -x
 qq      /var/qmail/bin/qmail-queue
diff --git a/tools/mail-scan.c b/tools/mail-scan.c
index 1ea245f..dc8aa5c 100644
--- a/tools/mail-scan.c
+++ b/tools/mail-scan.c
@@ -183,6 +183,16 @@ public:
   }
 };
 
+string noCR(const string bar){
+  string foo(bar);
+  int len = foo.length();
+  if (len){
+    if (foo[len-1] == '\r') {
+      foo.erase(len-1);
+    }
+  }
+  return foo;
+}
 
 ////////////////////////////////////////////////////////////
 int main(int _argc, const char** _argv){
@@ -268,6 +278,7 @@ int main(int _argc, const char** _argv){
             return 1;
           }
           if (getline(infile, line).fail()) continue;
+          line = noCR(line);
           Header.push_back(line);
           msgsize += line.length()+1;
           if (msgsize > maxsize) {
@@ -335,6 +346,7 @@ int main(int _argc, const char** _argv){
         break;
       }
     }   // end loop over matching records in this file
+
     if (vflag && !foundsome_infile) {
       cout << *file << endl;
       didprint++;
diff --git a/tools/skrewt.c b/tools/skrewt.c
index ad99e67..a43fd13 100644
--- a/tools/skrewt.c
+++ b/tools/skrewt.c
@@ -1,4 +1,4 @@
-///////////////////
+//////////////////
 // skrewt.c
 //
 // scrutinize email
@@ -13,6 +13,7 @@
 #include <stdio.h>              /* perror */
 #include <sstream>
 #include <vector>
+#include <list>
 
 using namespace std;
 
@@ -28,6 +29,7 @@ void usage(const int sts){
 "  Options\n"
 "    -help              print this msg (and exit immediately).\n"
 "    -maxsize ii        msg size in bytes; anything bigger will be rejected.\n"
+"    -error-exit        exit early if errors have been detected.\n"
 "\n"
 "  Messages containing the string '-please-bounce-this-' will be rejected.\n"
 "  Messages with no date will be rejected.\n"
@@ -36,7 +38,7 @@ void usage(const int sts){
 }
 
 //  error exit codes, mostly as stated in qmail.c
-#define bar \
+#define ErrorCodes \
 foo(good, 0) ;\
 foo(spam, 21) ;\
 foo(permerr, 31) ;\
@@ -46,7 +48,7 @@ foo(syserr, 71) ;\
 foo(comerr, 74) ;
 
 #define foo(name, num) const int ex_ ## name = num
-bar
+ErrorCodes
 #undef foo
 
 
@@ -104,12 +106,23 @@ string toLower(const std::string& a){
 }
 
 ////////////////
-string ltrim(string foo){
+string ltrim(const string foo){
   size_t where = foo.find_first_not_of(" \t\r\n");
   if (where == foo.npos) return foo;
   return foo.substr(where);
 }
 
+string noCR(const string bar){
+  string foo(bar);
+  int len = foo.length();
+  if (len){
+    if (foo[len-1] == '\r') {
+      foo.erase(len-1);
+    }
+  }
+  return foo;
+}
+
 ////////////////
 // little utility to help with argument parsing:
 //
@@ -117,7 +130,8 @@ int prefix(const string shorter, const string longer){
   return shorter == longer.substr(0, shorter.length());
 }
 
-void exeunt(const int sts){
+void maybe_exeunt(const int sts, const int really){
+  if (!really) return;
   if (sts == ex_good) exit(sts);
 
   const char* foo = getenv("HI_Q_GROUP");
@@ -136,6 +150,10 @@ void exeunt(const int sts){
   exit(sts);
 }
 
+void exeunt(const int sts){
+  maybe_exeunt(sts, 1);
+}
+
 string basename(const string path){
   size_t where = path.rfind("/");
   if (where != string::npos) return path.substr(1+where);
@@ -192,6 +210,15 @@ void parse_content(const string type_spec_line, string &maintype, string &bounda
   }
 }
 
+string join(const string sep, const list<string> stuff){
+  string rslt;
+  for (list<string>::const_iterator ptr = stuff.begin();
+        ptr != stuff.end(); ptr++){
+    if (rslt.length()) rslt += sep;
+    rslt += *ptr;
+  }
+  return rslt;
+}
 
 ////////////////////////////////////////////////////////////
 int main(int _argc, const char** _argv){
@@ -209,6 +236,8 @@ int main(int _argc, const char** _argv){
   }
 
   int maxsize(1000*1000);
+  int error_exit(0);
+  int mid_required(0);
 
   while (argc) {
     string arg(*argv); argv++; argc--;
@@ -216,14 +245,18 @@ int main(int _argc, const char** _argv){
     if (prefix(arg, "-help")) {
       usage(0);
     }
-    if (prefix(arg, "-maxsize")) {
+    if (0) {
+    } else if (prefix(arg, "-mid-required")) {
+      mid_required++;
+    } else if (prefix(arg, "-error-exit")) {
+      error_exit++;
+    } else if (prefix(arg, "-maxsize")) {
       if (!argc) {
         cerr << "Option -maxsize requires an argument" << endl;
         exit(ex_usage);
       }
       maxsize = atoi(*argv); argv++; argc--;
-    }
-    if (arg.substr(0,1) == "-") {
+    } else if (arg.substr(0,1) == "-") {
       cerr << "Unrecognized option '" << arg << "'" << endl;
       cerr << "For help, try:  " << progname << " -help" << endl;
       exit(ex_usage);
@@ -236,28 +269,32 @@ int main(int _argc, const char** _argv){
 
   int saw_blank_line(0);
   string boundary("x-xx-x");
-  string date;
+  string to;
+  string from;
   string subject;
-  string content_type;
+  string date;
   string message_id;
+  string content_type;
   int msgsize(0);
   vector<string> bigbuf;
-  cerr <<  "hi there" << endl;
+  int recno(0);
 
+  //xxxx cerr << progid << " begins" << endl;
   for (;;){             // outer loop over all records in the header
     if (cin.eof()) break;
     if (cin.bad()) return 1;
 
-    string headrec;
+    string line;
 // on fail, go back to top of outer loop and check for eof versus bad
-    if (getline(cin, headrec).fail()) continue;
-    msgsize += headrec.length()+1;
+    if (getline(cin, line).fail()) continue;
+    msgsize += line.length()+1;
     if (msgsize > maxsize) {
       cerr << progid << " rejection: bigger than " << maxsize << endl;
       exeunt(ex_spam);
     }
-    cout << headrec << endl;
-    bigbuf.push_back(headrec);  // for a folded record, this is the first line
+    cout << line << endl;
+    bigbuf.push_back(line);
+    string headrec = noCR(line);       // for a folded record, this is the first line
 
     for (;;) {        // inner loop to build a multi-line record e.g. folded record:
       if (cin.eof()) break;
@@ -276,12 +313,7 @@ int main(int _argc, const char** _argv){
       }
       cout << line << endl;
       bigbuf.push_back(line);
-      string cooked(line);
-      if (cooked.length()){
-        string::iterator ptr = cooked.end()-1;
-        if (*ptr == '\r') cooked.erase(ptr);
-      }
-      headrec += "\n" + cooked;
+      headrec += "\n" + noCR(line);
     }
 // here with a fully assembled header record
     int len = headrec.length();
@@ -301,6 +333,12 @@ int main(int _argc, const char** _argv){
     }
     headword = toLower(headword);
     if (0){
+    } else if (headword == "from") {
+      from = rest;
+    } else if (headword == "to") {
+      to = rest;
+    } else if (headword == "message-id") {
+      message_id = rest;
     } else if (headword == "date") {
       date = rest;
     } else if (headword == "subject") {
@@ -309,30 +347,49 @@ int main(int _argc, const char** _argv){
       content_type = rest;
     }
     //xxxx  cout << headrec.length() << " ... ";
+    recno++;
+    if (0) if (recno <= 6) cerr << progid << "#" << recno
+        << " " << headrec << endl;
   }
-  //xxx cerr << "headers are done. Delimited: " << saw_blank_line << endl;
+  cerr << progid <<" Mid '" << message_id << "'" << endl;
 
 // Headers are done.
 // Do some early-stage thinking.
 
+  list<string> badnews;
+
   if (subject.find("-please-bounce-this-") != string::npos) {
-    cerr << progid << " rejection: by request" << endl;
-    exeunt(ex_spam);
+    badnews.push_back("by request");
   }
 
   if (!date.length()) {
-    cerr << progid << " rejection: no date" << endl;
-    exeunt(ex_spam);              // disallow mail with no date
+    badnews.push_back("no date");
+  }
+
+  if (mid_required && !message_id.length()) {
+    badnews.push_back("no message-id");
+  }
+
+  if (badnews.size()){
+    cerr << progid << " " << join(", ", badnews) << endl;
+    if (error_exit){
+      cerr << progid << " '" << from
+             << "' to '" << to
+             << "'" << endl;
+      exeunt(ex_spam);
+    }
   }
 
   string main_contype;
-  parse_content(content_type, main_contype, boundary);
+  if (content_type.length())
+    parse_content(content_type, main_contype, boundary);
+// some slightly-useful booleans:
   int currently_text = main_contype == "text";
   int main_multipart = main_contype == "multipart";
 
 // early-stage thinking has been done.
 // Now spew the rest of the message
-  cerr << "body begins: " << main_contype << " " << currently_text << " " << boundary << endl;
+  //xxxx cerr << "body begins: " << main_contype << " " << currently_text << " " << boundary << endl;
   int in_subheads(0);
   int textlines(0);
 
@@ -345,7 +402,7 @@ int main(int _argc, const char** _argv){
     msgsize += line.length()+1;
     if (msgsize > maxsize) {
       cerr << progid << " rejection: bigger than " << maxsize << endl;
-      exeunt(ex_spam);
+      maybe_exeunt(ex_spam, error_exit);
     }
     bigbuf.push_back(line);
     cout << line << endl;
@@ -380,10 +437,10 @@ int main(int _argc, const char** _argv){
     }
   }
 
-  if (1) cerr << "textlines: " << textlines << endl;
+  if (0) cerr << "textlines: " << textlines << endl;
   if (1 && !textlines) {
     cerr << progid << " rejection: no text" << endl;
-//    exeunt(ex_spam);
+//    maybe_exeunt(ex_spam, error_exit);
   }
   cerr << progid << " normal completion" << endl;
   exit(ex_good);