progress toward cleaning up skrewt

1 files changed, 163 insertions, 102 deletions

diff --git a/tools/skrewt.c b/tools/skrewt.c
index 6de3dd9..ed0e627 100644
--- a/tools/skrewt.c
+++ b/tools/skrewt.c
@@ -12,6 +12,7 @@
 
 #include <stdio.h>              /* perror */
 #include <sstream>
+#include <vector>
 
 using namespace std;
 
@@ -144,6 +145,58 @@ string basename(const string path){
 string progname, progid;
 int mypid;
 
+
+/* Content-Type: text/plain; charset="us-ascii"                                         */
+/* Content-Type: multipart/mixed; boundary="1170861315-1262462055-1341954763=:92165"    */
+void parse_content(const string type_spec_line, string &maintype, string &boundary) {
+      string mainline(type_spec_line);
+
+      string get_type(toLower(mainline));
+      size_t where = get_type.find_first_of(" \t;\n");
+      if (where == string::npos) {
+        // keep whole string
+      }
+      else {
+        get_type = get_type.substr(0,where);
+      }
+      where = get_type.find("/");
+      if (where == string::npos){
+        maintype = "";
+      } else {
+        maintype = get_type.substr(0, where);
+      }
+}
+
+#ifdef xxxxxxxxxxxxxxxxxxxx
+      if
+      if (0) cerr << "type (" << get_type << ")  "
+            << (text_type?"text":"nope") << endl;
+
+
+      string srch = "boundary=";
+      where = headrec.find(srch);
+      if (where != string::npos) {
+        where += srch.length();
+        boundary = headrec.substr(where);
+        if (boundary[0] == '"') {
+          boundary = boundary.substr(1);
+          where = boundary.find_first_of("\"");
+        } else {
+          where = boundary.find_first_of(" \t;\n");
+        }
+        if (where == string::npos) {
+          /* do nothing, boundary=boundary as a whole */
+        } else {
+          boundary = boundary.substr(0, where);
+        }
+      }
+    }
+#endif
+
+
+
+
+
 ////////////////////////////////////////////////////////////
 int main(int _argc, const char** _argv){
 ////  pid_t pid = getpid();
@@ -159,8 +212,7 @@ int main(int _argc, const char** _argv){
     progid = binder.str();
   }
 
-
-  int maxsize(1000000);
+  int maxsize(1000*1000);
 
   while (argc) {
     string arg(*argv); argv++; argc--;
@@ -186,121 +238,130 @@ int main(int _argc, const char** _argv){
     }
   }
 
-  int inheads(1);
+  int saw_blank_line(0);
   string boundary("x-xx-x");
-  int text_type(1);
-  int textlines(0);
-  int gotdate(0);
+  string date;
+  string subject;
+  string content_type;
+  string message_id;
   int msgsize(0);
-  for (;;){
+  vector<string> bigbuf;
+  cerr <<  "hi there" << endl;
+
+  for (;;){             // outer loop over all records in the header
     if (cin.eof()) break;
     if (cin.bad()) return 1;
-    if (inheads) {
-      string header;
-      if (getline(cin, header).fail()) continue;
-      msgsize += header.length()+1;
+
+    string headrec;
+// 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 (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
+
+    for (;;) {        // inner loop to build a multi-line record e.g. folded record:
+      if (cin.eof()) break;
+      if (cin.bad()) return 1;
+      char ch;
+      if (cin.get(ch).fail()) continue;
+      cin.putback(ch);
+      if (ch != ' ' && ch != '\t') break;
+      string line;
+// on fail, go back to top of inner loop and check for eof versus bad
+      if (getline(cin, line).fail()) continue;
+      msgsize += line.length()+1;
       if (msgsize > maxsize) {
         cerr << progid << " rejection: bigger than " << maxsize << endl;
         exeunt(ex_spam);
       }
-      for (;;) {
-        if (cin.eof()) break;
-        if (cin.bad()) return 1;
-        char ch;
-        if (cin.get(ch).fail()) continue;
-        cin.putback(ch);
-        if (ch != ' ' && ch != '\t') break;
-        string line;
-        if (getline(cin, line).fail()) continue;
-        msgsize += line.length()+1;
-        if (msgsize > maxsize) {
-          cerr << progid << " rejection: bigger than " << maxsize << endl;
-          exeunt(ex_spam);
-        }
-        header += "\n" + line;
-      }
-      int len = header.length();
-      if (len && header[len-1] == '\r') len--;  // reduced length, not counting <cr>
-      if (len == 0) {
-        if (!gotdate) {
-          cerr << progid << " rejection: no date" << endl;
-          exeunt(ex_spam);              // disallow mail with no date
-        }
-        inheads = 0;
-        //cerr << "end of headers" << endl;
+      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);
       }
-      else {
+      headrec += "\n" + cooked;
+    }
+// here with a fully assembled header record
+    int len = headrec.length();
+    if (len && headrec[len-1] == '\r') len--;  // reduced length, not counting <cr>
+    if (len == 0) {
+      saw_blank_line = 1;
+      break;            // no more headers in this message
+    }
+
 // here if it's a header line
-        string headword;
-        string rest;
-        size_t where = header.find(":");
-        if (where != string::npos) {
-          headword = header.substr(0, where);
-          rest = ltrim(header.substr(1+where));
-        }
-        headword = toLower(headword);
-        if (headword == "content-type") {
-          string the_type = rest;
-          size_t where = the_type.find_first_of(" \t;\n");
-          if (where == string::npos) {
-            /* do nothing */
-          }
-          else {
-            the_type = the_type.substr(0,where);
-          }
-          the_type = toLower(the_type);
-          text_type = (the_type.find("text/") == 0);
-          if (0) cerr << "type (" << the_type << ")  "
-                << (text_type?"text":"nope") << endl;
-          string srch = "boundary=";
-          where = header.find(srch);
-          if (where != string::npos) {
-            where += srch.length();
-            boundary = header.substr(where);
-            if (boundary[0] == '"') {
-              boundary = boundary.substr(1);
-              where = boundary.find_first_of("\"");
-            } else {
-              where = boundary.find_first_of(" \t;\n");
-            }
-            if (where == string::npos) {
-              /* do nothing, boundary=boundary as a whole */
-            } else {
-              boundary = boundary.substr(0, where);
-            }
-          }
-        } else if (headword == "date") {
-          gotdate++;
-        } else if (headword == "subject") {
-          if (rest.find("-please-bounce-this-") != string::npos) {
-            cerr << progid << " rejection: by request" << endl;
-            exeunt(ex_spam);
-          }
-        }
-      }
-      //xxxx  cout << header.length() << " ... ";
-      cout << header << endl;
+    string headword;
+    string rest;
+    size_t where = headrec.find(":");
+    if (where != string::npos) {
+      headword = headrec.substr(0, where);
+      rest = ltrim(headrec.substr(1+where));
+    }
+    headword = toLower(headword);
+    if (0){
+    } else if (headword == "date") {
+      date = rest;
+    } else if (headword == "subject") {
+      subject = rest;
+    } else if (headword == "content-type") {
+      content_type = rest;
+    }
+    //xxxx  cout << headrec.length() << " ... ";
+  }
+  cerr << "headers are done. Delimited: " << saw_blank_line << endl;
+
+// Headers are done.
+// Do some early-stage thinking.
+
+  if (subject.find("-please-bounce-this-") != string::npos) {
+    cerr << progid << " rejection: by request" << endl;
+    exeunt(ex_spam);
+  }
+
+  if (!date.length()) {
+    cerr << progid << " rejection: no date" << endl;
+    exeunt(ex_spam);              // disallow mail with no date
+  }
+
+  string contype;
+  int textlines(0);
+  parse_content(content_type, contype, boundary);
+  int is_text = contype == "text";
+
+// early-stage thinking has been done.
+// Now spew the rest of the message
+  cerr << "body begins: " << contype << " " << is_text << endl;
+  int inheads(0);
+  for (;;){             // outer loop over all lines in the body
+    if (cin.eof()) break;
+    if (cin.bad()) return 1;
+    string line;
+// on fail, go back to top of outer loop and check for eof versus bad
+    if (getline(cin, line).fail()) continue;
+    msgsize += line.length()+1;
+    if (msgsize > maxsize) {
+      cerr << progid << " rejection: bigger than " << maxsize << endl;
+      exeunt(ex_spam);
+    }
+    bigbuf.push_back(line);
+    cout << line << endl;
+    if (line == "--" + boundary) {
+      inheads = 1;
     } else {
-      string line;
-      if (!getline(cin, line).fail()) {
-        msgsize += line.length()+1;
-        if (msgsize > maxsize) {
-          cerr << progid << " rejection: bigger than " << maxsize << endl;
-          exeunt(ex_spam);
-        }
-        if (line == "--" + boundary) {
-          inheads = 1;
-        } else {
-          if (text_type) {
-            if (ltrim(line).length()) textlines++;
-          }
-        }
-        cout << line << endl;
+      if (is_text) {
+        if (ltrim(line).length()) textlines++;
       }
     }
   }
+
   if (0) cerr << "textlines: " << textlines << endl;
-  if (!textlines) {
+  if (000 && !textlines) {
     cerr << progid << " rejection: no text" << endl;
     exeunt(ex_spam);
   }