Fail if CONTENT_TYPE is not application/x-www-form-urlencoded.

As part of the partial improvement to the POST reading
code I also added a max_post_size setting which defaults
to 1 MiB. POST inputs longer than that size get truncated.
This is separate to max_pastie_size, which is just the
size of one of the values in the POST data.

src/tawashi/main.cpp

@@ -84,6 +84,7 @@ namespace {
 		parSettings.add_default("resubmit_wait", "10");
 		parSettings.add_default("log_file", "-");
 		parSettings.add_default("highlight_css", "sh_darkness.css");
+		parSettings.add_default("max_post_size", "1048576");
 	}
 
 	void print_buildtime_info() {

src/tawashi_implem/cgi_env.cpp

@@ -18,6 +18,7 @@
 #include "cgi_env.hpp"
 #include "cgi_environment_vars.hpp"
 #include "duckhandy/lexical_cast.hpp"
+#include "tawashi_exception.hpp"
 #include <cassert>
 #include <ciso646>
 #include <boost/spirit/include/qi_core.hpp>
@@ -114,6 +115,18 @@ namespace cgi {
 		m_skip_path_info(calculate_skip_path_length(m_cgi_env[CGIVars::PATH_INFO], parBasePath)),
 		m_request_method_type(RequestMethodType::_from_string(m_cgi_env[CGIVars::REQUEST_METHOD].data()))
 	{
+		{
+			const std::string& content_type = m_cgi_env.at(CGIVars::CONTENT_TYPE);
+			int parsed_chars;
+			bool parse_ok;
+			m_split_mime = string_to_mime(&content_type, parse_ok, parsed_chars);
+			if (not parse_ok) {
+				std::string err_msg = "Parsing failed at position " +
+					std::to_string(parsed_chars) + " for input \"" +
+					content_type + "\"";
+				throw TawashiException(ErrorReasons::InvalidContentType, boost::string_ref(err_msg));
+			}
+		}
 	}
 
 	Env::~Env() noexcept = default;
@@ -217,6 +230,10 @@ namespace cgi {
 		return retval;
 	}
 
+	const SplitMime& Env::content_type_split() const {
+		return m_split_mime;
+	}
+
 	std::ostream& Env::print_all (std::ostream& parStream, const char* parNewline) const {
 		for (std::size_t z = 0; z < m_cgi_env.size(); ++z) {
 			parStream << CGIVars::_from_integral(z) <<

src/tawashi_implem/cgi_env.hpp

@@ -22,6 +22,7 @@
 #include "escapist.hpp"
 #include "kakoune/safe_ptr.hh"
 #include "request_method_type.hpp"
+#include "mime_split.hpp"
 #include <vector>
 #include <string>
 #include <boost/utility/string_ref.hpp>
@@ -67,6 +68,7 @@ namespace tawashi {
 			const std::string& server_software() const;
 
 			GetMapType query_string_split() const a_pure;
+			const SplitMime& content_type_split() const a_pure;
 
 			std::ostream& print_all (std::ostream& parStream, const char* parNewline) const;
 
@@ -75,6 +77,7 @@ namespace tawashi {
 			Escapist m_houdini;
 			std::size_t m_skip_path_info;
 			RequestMethodType m_request_method_type;
+			SplitMime m_split_mime;
 		};
 	} //namespace cgi
 } //namespace tawashi

src/tawashi_implem/cgi_post.cpp

@@ -28,11 +28,42 @@
 #include <ciso646>
 
 namespace tawashi {
+	UnsupportedContentTypeException::UnsupportedContentTypeException (const boost::string_ref& parMessage) :
+		TawashiException(ErrorReasons::UnsupportedContentType, parMessage)
+	{
+	}
+
 	namespace cgi {
 		namespace {
+			bool valid_content_type (const Env& parEnv) {
+				if (parEnv.content_type_split().type != "application" or
+					parEnv.content_type_split().subtype !=
+						"x-www-form-urlencoded") {
+					return false;
+				}
+				return true;
+			}
+
+			std::string read_n (std::istream& parSrc, std::size_t parSize) {
+				if (0 == parSize)
+					return std::string();
+
+				std::string original_data;
+				original_data.reserve(parSize);
+				std::copy_n(
+					std::istream_iterator<char>(parSrc),
+					parSize,
+					std::back_inserter(original_data)
+				);
+				return sanitized_utf8(original_data);
+			}
 		} //unnamed namespace
 
 		const PostMapType& read_post (std::istream& parSrc, const Env& parEnv) {
+			return read_post(parSrc, parEnv, parEnv.content_length());
+		}
+
+		const PostMapType& read_post (std::istream& parSrc, const Env& parEnv, std::size_t parMaxLen) {
 			static bool already_read = false;
 			static PostMapType map;
 			static std::string original_data;
@@ -41,22 +72,17 @@ namespace tawashi {
 				assert(original_data.empty());
 				assert(map.empty());
 
-				const auto input_len = parEnv.content_length();
-				if (input_len > 0) {
-					original_data.reserve(input_len);
-					std::copy_n(
-						std::istream_iterator<char>(parSrc),
-						input_len,
-						std::back_inserter(original_data)
-					);
-					original_data = sanitized_utf8(original_data);
+				if (not valid_content_type(parEnv)) {
+					throw UnsupportedContentTypeException(parEnv.content_type());
+				}
 
-					Escapist houdini;
-					for (auto& itm : split_env_vars(original_data)) {
-						std::string key(houdini.unescape_url(itm.first));
-						std::string val(houdini.unescape_url(itm.second));
-						map[std::move(key)] = std::move(val);
-					}
+				const auto input_len = std::min(parMaxLen, parEnv.content_length());
+				original_data = read_n(parSrc, input_len);
+				Escapist houdini;
+				for (auto& itm : split_env_vars(original_data)) {
+					std::string key(houdini.unescape_url(itm.first));
+					std::string val(houdini.unescape_url(itm.second));
+					map[std::move(key)] = std::move(val);
 				}
 
 				already_read = true;

src/tawashi_implem/cgi_post.hpp

@@ -17,11 +17,17 @@
 
 #pragma once
 
+#include "tawashi_exception.hpp"
 #include <boost/container/flat_map.hpp>
 #include <string>
 #include <istream>
+#include <cstddef>
 
 namespace tawashi {
+	class UnsupportedContentTypeException : public TawashiException {
+	public:
+		explicit UnsupportedContentTypeException (const boost::string_ref& parMessage);
+	};
 
 	namespace cgi {
 		class Env;
@@ -29,5 +35,6 @@ namespace tawashi {
 		typedef boost::container::flat_map<std::string, std::string> PostMapType;
 
 		const PostMapType& read_post (std::istream& parSrc, const Env& parEnv);
+		const PostMapType& read_post (std::istream& parSrc, const Env& parEnv, std::size_t parMaxLen);
 	} //namespace cgi
 } //namespace tawashi

src/tawashi_implem/error_reasons.hpp

@@ -27,6 +27,8 @@ namespace tawashi {
 		UnkownReason,
 		RedisDisconnected,
 		MissingPostVariable,
-		PastieNotFound
+		PastieNotFound,
+		InvalidContentType,
+		UnsupportedContentType
 	)
 } //namespace tawashi

src/tawashi_implem/error_response.cpp

@@ -50,7 +50,9 @@ namespace tawashi {
 			"An unknown error was raised.",
 			"Unable to connect to Redis.",
 			"Request is missing a POST variable.",
-			"Pastie not found."
+			"Pastie not found.",
+			"Invalid CONTENT_TYPE.",
+			"Unsupported CONTENT_TYPE."
 		};
 		constexpr const auto lengths = string_lengths(err_descs);
 		static_assert(err_descs.static_size == lengths.static_size, "Mismatching array sizes between strings and their lengths");

src/tawashi_implem/submit_paste_response.cpp

@@ -84,7 +84,6 @@ namespace tawashi {
 	}
 
 	HttpHeader SubmitPasteResponse::on_process() {
-		auto post = cgi::read_post(std::cin, cgi_env());
 		boost::string_ref pastie;
 		boost::string_ref lang;
 		boost::string_ref duration;
@@ -92,18 +91,25 @@ namespace tawashi {
 		auto statuslog = spdlog::get("statuslog");
 		assert(statuslog);
 
+		const SettingsBag& settings = this->settings();
 		try {
+			auto post = cgi::read_post(std::cin, cgi_env(), settings.as<uint32_t>("max_post_size"));
 			pastie = get_value_from_post(post, make_string_ref(g_post_key));
+			lang = get_value_from_post_log_failure(post, make_string_ref(g_language_key));
+			duration = get_value_from_post_log_failure(post, make_string_ref(g_duration_key));
+		}
+		catch (const UnsupportedContentTypeException& err) {
+			statuslog->info(
+				"Unsupported content type exception: \"{}\"",
+				err.what()
+			);
+			return make_error_redirect(ErrorReasons::UnsupportedContentType);
 		}
 		catch (const TawashiException& e) {
 			statuslog->error(e.what());
 			return make_error_redirect(e.reason());
 		}
 
-		lang = get_value_from_post_log_failure(post, make_string_ref(g_language_key));
-		duration = get_value_from_post_log_failure(post, make_string_ref(g_duration_key));
-
-		const SettingsBag& settings = this->settings();
 		const auto max_sz = settings.as<uint32_t>("max_pastie_size");
 		if (pastie.size() < settings.as<uint32_t>("min_pastie_size")) {
 			return make_error_redirect(ErrorReasons::PostLengthNotInRange);

test/unit/test_invalid_utf8_post.cpp

@@ -53,6 +53,7 @@ TEST_CASE ("Retrieve and sanitize invalid an invalid utf-8 text from POST data",
 		content_length.c_str(),
 		"PATH_INFO=/",
 		"REQUEST_METHOD=GET",
+		"CONTENT_TYPE=application/x-www-form-urlencoded",
 		nullptr
 	};