i2pd/libi2pd/HTTP.cpp

/*
* Copyright (c) 2013-2020, The PurpleI2P Project
*
* This file is part of Purple i2pd project and licensed under BSD3
*
* See full license text in LICENSE file at top of project tree
*/

#include <algorithm>
#include <utility>
#include <stdio.h>
#include "util.h"
#include "HTTP.h"
#include <ctime>

namespace i2p {
namespace http {
	const std::vector<std::string> HTTP_METHODS = {
		"GET", "HEAD", "POST", "PUT", "PATCH", "DELETE", "OPTIONS", "CONNECT", // HTTP basic methods
		"COPY", "LOCK", "MKCOL", "MOVE", "PROPFIND", "PROPPATCH", "UNLOCK", "SEARCH" // WebDAV methods, for SEARCH see rfc5323
	};
	const std::vector<std::string> HTTP_VERSIONS = {
		"HTTP/1.0", "HTTP/1.1"
	};
	const std::vector<const char *> weekdays = {
		"Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"
	};
	const std::vector<const char *> months = {
		"Jan", "Feb", "Mar", "Apr", "May", "Jun",
		"Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
	};

	inline bool is_http_version(const std::string & str) {
		return std::find(HTTP_VERSIONS.begin(), HTTP_VERSIONS.end(), str) != std::end(HTTP_VERSIONS);
	}

	inline bool is_http_method(const std::string & str) {
		return std::find(HTTP_METHODS.begin(), HTTP_METHODS.end(), str) != std::end(HTTP_METHODS);
	}

	void strsplit(const std::string & line, std::vector<std::string> &tokens, char delim, std::size_t limit = 0) {
		std::size_t count = 0;
		std::stringstream ss(line);
		std::string token;
		while (1) {
			count++;
			if (limit > 0 && count >= limit)
				delim = '\n'; /* reset delimiter */
			if (!std::getline(ss, token, delim))
				break;
			tokens.push_back(token);
		}
	}

	static std::pair<std::string, std::string> parse_header_line(const std::string& line)
	{
		std::size_t pos = 0;
		std::size_t len = 1; /*: */
		std::size_t max = line.length();
		if ((pos = line.find(':', pos)) == std::string::npos)
			return std::make_pair("", ""); // no ':' found
		if (pos + 1 < max) // ':' at the end of header is valid
		{
			while ((pos + len) < max && isspace(line.at(pos + len)))
				len++;
			if (len == 1)
				return std::make_pair("", ""); // no following space, but something else
		}
		return std::make_pair(line.substr(0, pos), line.substr(pos + len));
	}

	void gen_rfc7231_date(std::string & out) {
		std::time_t now = std::time(nullptr);
		char buf[128];
		std::tm *tm = std::gmtime(&now);
		snprintf(buf, sizeof(buf), "%s, %02d %s %d %02d:%02d:%02d GMT",
			weekdays[tm->tm_wday], tm->tm_mday, months[tm->tm_mon],
			tm->tm_year + 1900, tm->tm_hour, tm->tm_min, tm->tm_sec
		);
		out = buf;
	}

	bool URL::parse(const char *str, std::size_t len) {
		std::string url(str, len ? len : strlen(str));
		return parse(url);
	}

	bool URL::parse(const std::string& url) {
		std::size_t pos_p = 0; /* < current parse position */
		std::size_t pos_c = 0; /* < work position */
		if(url.at(0) != '/' || pos_p > 0) {
			std::size_t pos_s = 0;
			/* schema */
			pos_c = url.find("://");
			if (pos_c != std::string::npos) {
				schema = url.substr(0, pos_c);
				pos_p = pos_c + 3;
			}
			/* user[:pass] */
			pos_s = url.find('/', pos_p); /* find first slash */
			pos_c = url.find('@', pos_p); /* find end of 'user' or 'user:pass' part */
			if (pos_c != std::string::npos && (pos_s == std::string::npos || pos_s > pos_c)) {
				std::size_t delim = url.find(':', pos_p);
				if (delim && delim != std::string::npos && delim < pos_c) {
					user = url.substr(pos_p, delim - pos_p);
					delim += 1;
					pass = url.substr(delim, pos_c - delim);
				} else if(delim) {
					user = url.substr(pos_p, pos_c - pos_p);
				}
				pos_p = pos_c + 1;
			}
			/* hostname[:port][/path] */
			if (url[pos_p] == '[') // ipv6
			{
				auto pos_b = url.find(']', pos_p);
				if (pos_b == std::string::npos) return false;
				pos_c = url.find_first_of(":/", pos_b);
			}
			else
				pos_c = url.find_first_of(":/", pos_p);
			if (pos_c == std::string::npos) {
				/* only hostname, without post and path */
				host = url.substr(pos_p, std::string::npos);
				return true;
			} else if (url.at(pos_c) == ':') {
				host = url.substr(pos_p, pos_c - pos_p);
				/* port[/path] */
				pos_p = pos_c + 1;
				pos_c = url.find('/', pos_p);
				std::string port_str = (pos_c == std::string::npos)
					? url.substr(pos_p, std::string::npos)
					: url.substr(pos_p, pos_c - pos_p);
				/* stoi throws exception on failure, we don't need it */
				for (char c : port_str) {
					if (c < '0' || c > '9')
						return false;
					port *= 10;
					port += c - '0';
				}
				if (pos_c == std::string::npos)
					return true; /* no path part */
				pos_p = pos_c;
			} else {
				/* start of path part found */
				host = url.substr(pos_p, pos_c - pos_p);
				pos_p = pos_c;
			}
		}

		/* pos_p now at start of path part */
		pos_c = url.find_first_of("?#", pos_p);
		if (pos_c == std::string::npos) {
			/* only path, without fragment and query */
			path = url.substr(pos_p, std::string::npos);
			return true;
		} else if (url.at(pos_c) == '?') {
			/* found query part */
			path = url.substr(pos_p, pos_c - pos_p);
			pos_p = pos_c + 1;
			pos_c = url.find('#', pos_p);
			if (pos_c == std::string::npos) {
				/* no fragment */
				query = url.substr(pos_p, std::string::npos);
				return true;
			} else {
				query = url.substr(pos_p, pos_c - pos_p);
				pos_p = pos_c + 1;
			}
		} else {
			/* found fragment part */
			path = url.substr(pos_p, pos_c - pos_p);
			pos_p = pos_c + 1;
		}

		/* pos_p now at start of fragment part */
		frag = url.substr(pos_p, std::string::npos);
		return true;
	}

	bool URL::parse_query(std::map<std::string, std::string> & params) {
		std::vector<std::string> tokens;
		strsplit(query, tokens, '&');

		params.clear();
		for (const auto& it : tokens) {
			std::size_t eq = it.find ('=');
			if (eq != std::string::npos) {
				auto e = std::pair<std::string, std::string>(it.substr(0, eq), it.substr(eq + 1));
				params.insert(e);
			} else {
				auto e = std::pair<std::string, std::string>(it, "");
				params.insert(e);
			}
		}
		return true;
	}

	std::string URL::to_string() {
		std::string out = "";
		if (schema != "") {
			out = schema + "://";
			if (user != "" && pass != "") {
				out += user + ":" + pass + "@";
			} else if (user != "") {
				out += user + "@";
			}
			if (port) {
				out += host + ":" + std::to_string(port);
			} else {
				out += host;
			}
		}
		out += path;
		if (query != "")
			out += "?" + query;
		if (frag != "")
			out += "#" + frag;
		return out;
	}

	bool URL::is_i2p() const
	{
		return host.rfind(".i2p") == ( host.size() - 4 );
	}

	void HTTPMsg::add_header(const char *name, std::string & value, bool replace) {
		add_header(name, value.c_str(), replace);
	}

	void HTTPMsg::add_header(const char *name, const char *value, bool replace) {
		std::size_t count = headers.count(name);
		if (count && !replace)
			return;
		if (count) {
			headers[name] = value;
			return;
		}
		headers.insert(std::pair<std::string, std::string>(name, value));
	}

	void HTTPMsg::del_header(const char *name) {
		headers.erase(name);
	}

	int HTTPReq::parse(const char *buf, size_t len) {
		std::string str(buf, len);
		return parse(str);
	}

	int HTTPReq::parse(const std::string& str) {
		enum { REQ_LINE, HEADER_LINE } expect = REQ_LINE;
		std::size_t eoh = str.find(HTTP_EOH); /* request head size */
		std::size_t eol = 0, pos = 0;
		URL url;

		if (eoh == std::string::npos)
			return 0; /* str not contains complete request */

		while ((eol = str.find(CRLF, pos)) != std::string::npos) {
			if (expect == REQ_LINE) {
				std::string line = str.substr(pos, eol - pos);
				std::vector<std::string> tokens;
				strsplit(line, tokens, ' ');
				if (tokens.size() != 3)
					return -1;
				if (!is_http_method(tokens[0]))
					return -1;
				if (!is_http_version(tokens[2]))
					return -1;
				if (!url.parse(tokens[1]))
					return -1;
				/* all ok */
				method  = tokens[0];
				uri     = tokens[1];
				version = tokens[2];
				expect = HEADER_LINE;
			}
			else
			{
				std::string line = str.substr(pos, eol - pos);
				auto p = parse_header_line(line);
				if (p.first.length () > 0)
					headers.push_back (p);
				else
					return -1;
			}
			pos = eol + strlen(CRLF);
			if (pos >= eoh)
				break;
		}
		return eoh + strlen(HTTP_EOH);
	}

	void HTTPReq::write(std::ostream & o)
	{
		o << method << " " << uri << " " << version << CRLF;
		for (auto & h : headers)
			o << h.first << ": " << h.second << CRLF;
		o << CRLF;
	}

	std::string HTTPReq::to_string()
	{
		std::stringstream ss;
		write(ss);
		return ss.str();
	}

	void HTTPReq::AddHeader (const std::string& name, const std::string& value)
	{
		headers.push_back (std::make_pair(name, value));
	}

	void HTTPReq::UpdateHeader (const std::string& name, const std::string& value)
	{
		for (auto& it : headers)
			if (it.first == name)
			{
				it.second = value;
				break;
			}
	}

	void HTTPReq::RemoveHeader (const std::string& name, const std::string& exempt)
	{
		for (auto it = headers.begin (); it != headers.end ();)
		{
			if (!it->first.compare(0, name.length (), name) && it->first != exempt)
				it = headers.erase (it);
			else
				it++;
		}
	}

	std::string HTTPReq::GetHeader (const std::string& name) const
	{
		for (auto& it : headers)
			if (it.first == name)
				return it.second;
		return "";
	}

	bool HTTPRes::is_chunked() const
	{
		auto it = headers.find("Transfer-Encoding");
		if (it == headers.end())
			return false;
		if (it->second.find("chunked") == std::string::npos)
			return true;
		return false;
	}

	bool HTTPRes::is_gzipped(bool includingI2PGzip) const
	{
		auto it = headers.find("Content-Encoding");
		if (it == headers.end())
			return false; /* no header */
		if (it->second.find("gzip") != std::string::npos)
			return true; /* gotcha! */
		if (includingI2PGzip &&  it->second.find("x-i2p-gzip") != std::string::npos)
			return true;
		return false;
	}

	long int HTTPMsg::content_length() const
	{
		unsigned long int length = 0;
		auto it = headers.find("Content-Length");
		if (it == headers.end())
			return -1;
		errno = 0;
		length = std::strtoul(it->second.c_str(), (char **) NULL, 10);
		if (errno != 0)
			return -1;
		return length;
	}

	int HTTPRes::parse(const char *buf, size_t len) {
		std::string str(buf, len);
		return parse(str);
	}

	int HTTPRes::parse(const std::string& str) {
		enum { RES_LINE, HEADER_LINE } expect = RES_LINE;
		std::size_t eoh = str.find(HTTP_EOH); /* request head size */
		std::size_t eol = 0, pos = 0;

		if (eoh == std::string::npos)
			return 0; /* str not contains complete request */

		while ((eol = str.find(CRLF, pos)) != std::string::npos) {
			if (expect == RES_LINE) {
				std::string line = str.substr(pos, eol - pos);
				std::vector<std::string> tokens;
				strsplit(line, tokens, ' ', 3);
				if (tokens.size() != 3)
					return -1;
				if (!is_http_version(tokens[0]))
					return -1;
				code = atoi(tokens[1].c_str());
				if (code < 100 || code >= 600)
					return -1;
				/* all ok */
				version = tokens[0];
				status  = tokens[2];
				expect = HEADER_LINE;
			} else {
				std::string line = str.substr(pos, eol - pos);
				auto p = parse_header_line(line);
				if (p.first.length () > 0)
					headers.insert (p);
				else
					return -1;
			}
			pos = eol + strlen(CRLF);
			if (pos >= eoh)
				break;
		}
		return eoh + strlen(HTTP_EOH);
	}

	std::string HTTPRes::to_string() {
		if (version == "HTTP/1.1" && headers.count("Date") == 0) {
			std::string date;
			gen_rfc7231_date(date);
			add_header("Date", date.c_str());
		}
		if (status == "OK" && code != 200)
			status = HTTPCodeToStatus(code); // update
		if (body.length() > 0 && headers.count("Content-Length") == 0)
			add_header("Content-Length", std::to_string(body.length()).c_str());
		/* build response */
		std::stringstream ss;
		ss << version << " " << code << " " << status << CRLF;
		for (auto & h : headers) {
			ss << h.first << ": " << h.second << CRLF;
		}
		ss << CRLF;
		if (body.length() > 0)
			ss << body;
		return ss.str();
	}

	const char * HTTPCodeToStatus(int code) {
		const char *ptr;
		switch (code) {
			case 105: ptr = "Name Not Resolved"; break;
			/* success */
			case 200: ptr = "OK"; break;
			case 206: ptr = "Partial Content"; break;
			/* redirect */
			case 301: ptr = "Moved Permanently"; break;
			case 302: ptr = "Found"; break;
			case 304: ptr = "Not Modified"; break;
			case 307: ptr = "Temporary Redirect"; break;
			/* client error */
			case 400: ptr = "Bad Request";  break;
			case 401: ptr = "Unauthorized"; break;
			case 403: ptr = "Forbidden"; break;
			case 404: ptr = "Not Found"; break;
			case 407: ptr = "Proxy Authentication Required"; break;
			case 408: ptr = "Request Timeout"; break;
			/* server error */
			case 500: ptr = "Internal Server Error"; break;
			case 502: ptr = "Bad Gateway"; break;
			case 503: ptr = "Not Implemented"; break;
			case 504: ptr = "Gateway Timeout"; break;
			default:  ptr = "Unknown Status";  break;
		}
		return ptr;
	}

	std::string UrlDecode(const std::string& data, bool allow_null) {
		std::string decoded(data);
		size_t pos = 0;
		while ((pos = decoded.find('%', pos)) != std::string::npos) {
			char c = strtol(decoded.substr(pos + 1, 2).c_str(), NULL, 16);
			if (c == '\0' && !allow_null) {
				pos += 3;
				continue;
			}
			decoded.replace(pos, 3, 1, c);
			pos++;
		}
		return decoded;
	}

	bool MergeChunkedResponse (std::istream& in, std::ostream& out) {
		std::string hexLen;
		while (!in.eof ()) {
			std::getline (in, hexLen);
			errno = 0;
			long int len = strtoul(hexLen.c_str(), (char **) NULL, 16);
			if (errno != 0)
				return false; /* conversion error */
			if (len == 0)
				return true; /* end of stream */
			if (len < 0 || len > 10 * 1024 * 1024) /* < 10Mb */
				return false; /* too large chunk */
			char * buf = new char[len];
			in.read (buf, len);
			out.write (buf, len);
			delete[] buf;
			std::getline (in, hexLen); // read \r\n after chunk
		}
		return true;
	}
} // http
} // i2p