From 2dff81806a7c1064ed59d5ef0ce569f2e2c0039b Mon Sep 17 00:00:00 2001
From: Thor Harald Johansen <thj@thj.no>
Date: Sat, 6 Jun 2020 02:43:52 +0200
Subject: [PATCH] Initial commit

---
 .gitignore                    |   4 ++
 Makefile                      |  30 +++++++++
 include/mail.hpp              |  20 ++++++
 include/protocol/protocol.hpp |  69 ++++++++++++++++++++
 include/protocol/token.hpp    |  33 ++++++++++
 include/util.hpp              |  12 ++++
 src/bogofilter-smtpd.cpp      |  50 +++++++++++++++
 src/mail.cpp                  |   2 +
 src/protocol/protocol.cpp     |  77 +++++++++++++++++++++++
 src/protocol/token.cpp        | 115 ++++++++++++++++++++++++++++++++++
 src/util.cpp                  |  36 +++++++++++
 11 files changed, 448 insertions(+)
 create mode 100644 .gitignore
 create mode 100644 Makefile
 create mode 100644 include/mail.hpp
 create mode 100644 include/protocol/protocol.hpp
 create mode 100644 include/protocol/token.hpp
 create mode 100644 include/util.hpp
 create mode 100644 src/bogofilter-smtpd.cpp
 create mode 100644 src/mail.cpp
 create mode 100644 src/protocol/protocol.cpp
 create mode 100644 src/protocol/token.cpp
 create mode 100644 src/util.cpp

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..4ed279e
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,4 @@
+build/
+.vscode/
+
+*.swp
diff --git a/Makefile b/Makefile
new file mode 100644
index 0000000..f934c2d
--- /dev/null
+++ b/Makefile
@@ -0,0 +1,30 @@
+# Makefile inspired by:
+# https://spin.atomicobject.com/2016/08/26/makefile-c-projects/
+
+EXEC := bogofilter-smtpd
+
+CXX=/usr/local/Cellar/gcc/9.3.0_1/bin/g++-9
+CXXFLAGS=-std=gnu++17 -Iinclude
+CPPFLAGS=-MMD -MP
+
+SRCDIR := src
+BUILDDIR := build
+
+SRCS := $(shell find src -name '*.cpp')
+OBJS := $(SRCS:%=$(BUILDDIR)/%.o)
+DEPS := $(OBJS:.o=.d)
+
+$(BUILDDIR)/$(EXEC) : $(OBJS)
+	$(CXX) -o $@ $^
+
+run : $(BUILDDIR)/$(EXEC)
+	$<
+
+clean :
+	rm -R $(BUILDDIR)
+
+$(BUILDDIR)/%.cpp.o: %.cpp
+	mkdir -p $(@D)
+	$(CXX) $(CPPFLAGS) $(CXXFLAGS) -c $< -o $@
+
+-include $(DEPS)
\ No newline at end of file
diff --git a/include/mail.hpp b/include/mail.hpp
new file mode 100644
index 0000000..5cfd067
--- /dev/null
+++ b/include/mail.hpp
@@ -0,0 +1,20 @@
+#ifndef _MAIL_
+#define _MAIL_
+
+#include <protocol/protocol.hpp>
+
+#include <functional>
+#include <vector>
+#include <string>
+#include <unordered_map>
+
+namespace mail {
+    struct message_t {
+        std::vector<std::string> header;
+        std::vector<std::string> body;
+    };
+
+    using session_map_t = std::unordered_map<protocol::session_t, message_t>;
+}
+
+#endif
\ No newline at end of file
diff --git a/include/protocol/protocol.hpp b/include/protocol/protocol.hpp
new file mode 100644
index 0000000..c13fc8c
--- /dev/null
+++ b/include/protocol/protocol.hpp
@@ -0,0 +1,69 @@
+#ifndef _PROTOCOL_
+#define _PROTOCOL_
+
+#include <protocol/token.hpp>
+
+#include <string>
+#include <map>
+#include <functional>
+#include <iostream>
+
+namespace protocol {
+    struct message_t {
+        std::string input;
+        token::pattern_t pattern;
+    };
+
+    class protocol_t;
+    using handler_t = std::function<void(protocol_t&, const message_t&)>;
+    using handler_map_t = std::map<token::pattern_t, handler_t>;
+
+    class protocol_t {
+    public:
+        void set_handlers(const handler_map_t handlers);
+        void absorb(const std::string& input);
+        void send(const token::list_t& tokens);
+
+    protected:
+        virtual void emit(const std::string& output) { };
+
+    private:
+        handler_map_t handlers;
+        size_t max_pattern_length;
+    };
+
+    class ios_protocol_t : public protocol_t {
+    public:
+        ios_protocol_t(std::istream& in, std::ostream& out);
+        void run();
+
+    protected:
+        virtual void emit(const std::string& output);
+
+    private:
+        std::istream& in;
+        std::ostream& out;
+    };
+
+    struct session_t {
+        std::string session;
+        std::string token;
+    };
+
+    bool operator==(const session_t& l, const session_t& r);
+
+    session_t session_from(token::list_t tokens);
+    session_t session_from(message_t message);
+}
+
+namespace std {
+    template<> struct hash<protocol::session_t> {
+        std::size_t operator()(protocol::session_t const& session) const noexcept {
+            auto session_hash = std::hash<std::string>{}(session.session);
+            auto token_hash = std::hash<std::string>{}(session.token);
+            return session_hash ^ (token_hash << 1);
+        }
+    };
+}
+
+#endif
\ No newline at end of file
diff --git a/include/protocol/token.hpp b/include/protocol/token.hpp
new file mode 100644
index 0000000..6a39133
--- /dev/null
+++ b/include/protocol/token.hpp
@@ -0,0 +1,33 @@
+#ifndef _TOKEN_
+#define _TOKEN_
+
+#include <vector>
+#include <map>
+#include <string>
+#include <iostream>
+
+namespace protocol::token {
+    using string_t = std::string;
+    using list_t = std::vector<string_t>;
+
+    std::string compose(const list_t& tokens);
+    list_t decompose(const std::string& input, size_t max_tokens = -1);
+
+    class pattern_t : public std::map<size_t, string_t> {
+    private:
+        bool wildcard;
+
+    public:
+        pattern_t();
+        pattern_t(list_t tokens, bool wildcard = false);
+        pattern_t(std::string input, bool wildcard = true);
+
+        list_t token_list() const;
+
+        friend std::ostream& operator<<(typeof(std::cout)& out, const pattern_t& pattern);
+        friend bool operator<(const pattern_t& l, const pattern_t& r);
+        friend bool operator==(const pattern_t& l, const pattern_t& r);
+    };
+}
+
+#endif
\ No newline at end of file
diff --git a/include/util.hpp b/include/util.hpp
new file mode 100644
index 0000000..887ee12
--- /dev/null
+++ b/include/util.hpp
@@ -0,0 +1,12 @@
+#ifndef _UTIL_
+#define _UTIL_
+
+#include <vector>
+#include <string>
+
+namespace util {
+    std::vector<std::string> split(const std::string input, const std::string::value_type separator, size_t max_tokens = -1);
+    std::string join(const std::vector<std::string> strings, const std::string::value_type separator);
+};
+
+#endif
\ No newline at end of file
diff --git a/src/bogofilter-smtpd.cpp b/src/bogofilter-smtpd.cpp
new file mode 100644
index 0000000..6f06a5d
--- /dev/null
+++ b/src/bogofilter-smtpd.cpp
@@ -0,0 +1,50 @@
+#include <protocol/protocol.hpp>
+#include <mail.hpp>
+
+#include <string>
+#include <vector>
+#include <map>
+#include <unordered_map>
+#include <iostream>
+#include <functional>
+
+const bool debug = true;
+
+using namespace std;
+
+mail::session_map_t mail_messages;
+
+void on_ready(protocol::protocol_t& protocol, protocol::message_t message) {
+    debug && cerr << "Ready" << endl;
+    protocol.send({"register", "filter", "smtp-in", "data"});
+    protocol.send({"register", "filter", "smtp-in", "data-line"});
+    protocol.send({"register", "filter", "smtp-in", "commit"});
+    protocol.send({"register", "ready"});
+}
+
+void on_data(protocol::protocol_t& protocol, protocol::message_t message) {
+    mail_messages.emplace(protocol::session_from(message), mail::message_t{});
+}
+
+void on_dataline(protocol::protocol_t& protocol, protocol::message_t message) {
+    debug && cerr << "Dataline" << endl;
+}
+
+void on_commit(protocol::protocol_t& protocol, protocol::message_t message) {
+
+}
+
+int main(int argc, char** argv) {
+    protocol::ios_protocol_t protocol(cin, cout);
+
+    protocol::handler_map_t handlers;
+    handlers.emplace("config|ready", on_ready);
+    handlers.emplace("filter|*|*|smtp-in|data", on_data);
+    handlers.emplace("filter|*|*|smtp-in|data-line", on_dataline);
+    handlers.emplace("filter|*|*|smtp-in|commit", on_commit);
+    protocol.set_handlers(handlers);
+
+    protocol.run();
+
+    return 0;
+}
\ No newline at end of file
diff --git a/src/mail.cpp b/src/mail.cpp
new file mode 100644
index 0000000..b3ebfdb
--- /dev/null
+++ b/src/mail.cpp
@@ -0,0 +1,2 @@
+#include <mail.hpp>
+
diff --git a/src/protocol/protocol.cpp b/src/protocol/protocol.cpp
new file mode 100644
index 0000000..e017748
--- /dev/null
+++ b/src/protocol/protocol.cpp
@@ -0,0 +1,77 @@
+#include <util.hpp>
+
+#include <protocol/protocol.hpp>
+
+#include <iostream>
+#include <string>
+
+using namespace std;
+using namespace util;
+
+    //config|smtpd-version|6.6.1
+    //config|smtp-session-timeout|300
+    //config|subsystem|smtp-in
+
+const bool debug = true;
+
+namespace protocol {
+    void protocol_t::set_handlers(const handler_map_t handlers) {
+        size_t length = 0;
+        for(auto&& [pattern, handler] : handlers) {
+            for(auto&& [index, token] : pattern) {
+                if(index + 1 > length) {
+                    length = index + 1;
+                }
+            }
+        }
+        debug && cerr << "Maximum handler pattern length is " << length << endl;
+
+        this->handlers = handlers;
+        max_pattern_length = length;
+    }
+
+    void protocol_t::send(const token::list_t& tokens) {
+        this->emit(token::compose(tokens));
+    }
+
+    void protocol_t::absorb(const std::string& input) {
+        auto tokens = token::decompose(input, max_pattern_length);
+        auto pattern = token::pattern_t(tokens, false);
+        debug && cout << "Received pattern " << pattern << endl;
+
+        auto element = handlers.find(pattern);
+        if(element != handlers.end()) {
+            element->second(*this, {input, element->first});
+        }
+    }
+
+    ios_protocol_t::ios_protocol_t(std::istream& in, std::ostream& out) :
+        in(in), out(out) { }
+
+    void ios_protocol_t::emit(const std::string& output) {
+        out << output << endl;
+    }
+
+    void ios_protocol_t::run() {
+        while(!in.eof()) {
+            string line;
+            std::getline(in, line);
+            absorb(line);
+        }
+    }
+
+    bool operator==(const session_t& l, const session_t& r) {
+        return l.session == r.session && l.token == r.token;
+    }
+
+    session_t session_from(token::list_t tokens) {
+        auto session_id = tokens.at(5);
+        auto session_token = tokens.at(6);
+        return {session_id, session_token};
+    }
+
+    session_t session_from(message_t message) {
+        auto tokens = token::decompose(message.input);
+        return session_from(tokens);
+    }
+}
diff --git a/src/protocol/token.cpp b/src/protocol/token.cpp
new file mode 100644
index 0000000..e870e93
--- /dev/null
+++ b/src/protocol/token.cpp
@@ -0,0 +1,115 @@
+#include <protocol/token.hpp>
+#include <util.hpp>
+
+using namespace std;
+using namespace util;
+
+const bool debug = false;
+
+namespace protocol::token {
+
+    std::string compose(const list_t& tokens) {
+        return join(tokens, '|');
+    }
+
+    list_t decompose(const std::string& input, size_t max_tokens) {
+        return split(input, '|', max_tokens);
+    }
+
+    pattern_t::pattern_t() {}
+
+    pattern_t::pattern_t(list_t tokens, bool wildcard) : wildcard(wildcard) {
+        for(auto index = 0; index < tokens.size(); ++index) {
+            auto token = tokens[index];
+            if(!wildcard || token != "*") {
+                this->emplace(index, tokens[index]);
+            }
+        }
+    }
+
+    pattern_t::pattern_t(string input, bool wildcard) : pattern_t(split(input, '|'), wildcard) {}
+
+    list_t pattern_t::token_list() const {
+        list_t output;
+        for(auto element = begin(); element != end(); ++element) {
+            output.push_back(element->second);
+        }
+        return output;
+    }
+
+    ostream& operator<<(typeof(cout)& out, const pattern_t& pattern) {
+        return out << join(pattern.token_list(), '|') << "<wildcard=" << pattern.wildcard << ">";
+    }
+
+    bool operator<(const pattern_t& l, const pattern_t& r) {
+        debug && cerr << "Checking if " << l << " < " << r << endl;
+
+        if(l == r) {
+            debug && cerr << "Not less because left equals right" << endl;
+            return false;
+        }
+
+        if(l.size() < r.size()) {
+            debug && cerr << "Less because right has more elements" << endl;
+            return true;
+        }
+
+        for(auto le = l.begin(); le != l.end(); ++le) {
+            if(r.count(le->first) == 0) {
+                debug && cerr << "Not less because right is missing " << le->first << " -> " << le->second << endl;
+                return false;
+            }
+
+            if(le->second < r.at(le->first)) {
+                debug && cerr << "Not less because " << le->second << " < " << r.at(le->first) << endl;
+                return true;
+            }
+        }
+
+        return false;
+    }
+
+    bool operator==(const pattern_t& l, const pattern_t& r) {
+        debug && cerr << "Checking if " << l << " == " << r << endl;
+
+        const pattern_t *subset, *superset;
+        if(r.wildcard == l.wildcard) {
+            debug && cerr << "Left and right are both (not) wildcards" << endl;
+            if(l.size() != r.size()) {
+                debug && cerr << "Not equal because sizes are unequal" << endl;
+                return false;
+            } else {
+                subset = &l;
+                superset = &r;
+            }
+        } else if(l.wildcard) {
+            debug && cerr << "Left is wildcard" << endl;
+            subset = &l;
+            superset = &r;
+        } else {
+            debug && cerr << "Right is wildcard" << endl;
+            subset = &r;
+            superset = &l;
+        }
+
+        for(auto le : *subset) {
+            if(superset->count(le.first) == 0) {
+                debug && cerr << "Not equal because " << le.first << " -> " << le.second << " is missing in superset" << endl;
+                return false;
+            }
+
+            auto rv = superset->at(le.first);
+
+            if(rv != le.second) {
+                debug && cerr << "Not equal because " << le.first << " -> " << le.second << " != " << rv << " in superset" << endl;
+                return false;
+            }
+
+            debug && cerr << le.first << " -> " << le.second << " == " << rv << " in superset" << endl;
+        }
+
+        debug && cerr << "Patterns match" << endl;
+
+        return true;
+    }
+}
\ No newline at end of file
diff --git a/src/util.cpp b/src/util.cpp
new file mode 100644
index 0000000..84e22fa
--- /dev/null
+++ b/src/util.cpp
@@ -0,0 +1,36 @@
+#include <util.hpp>
+
+std::vector<std::string> util::split(const std::string input, const std::string::value_type separator, size_t max_tokens) {
+    std::vector<std::string> tokens;
+
+    for(auto index = 0, last = -1; index != max_tokens; ++index) {
+        auto current = input.find(separator, last + 1);
+        auto eol = current == std::string::npos;
+
+        if(eol || index  == max_tokens) {
+            current = input.length();
+        }
+
+        auto length = current - last - 1;
+        auto token = input.substr(last + 1, length);
+        tokens.push_back(token);
+        last = current;
+
+        if(eol) {
+            break;
+        }
+    }
+
+    return tokens;
+}
+
+std::string util::join(const std::vector<std::string> strings, const std::string::value_type separator) {
+    std::string output;
+    for(size_t index = 0; index < strings.size(); ++index) {
+        output.append(strings.at(index));
+        if(index + 1 < strings.size()) {
+            output.push_back(separator);
+        }
+    }
+    return output;
+}
\ No newline at end of file