+ new http req/res/url structs

8 years ago · 42b9b6426a
2 changed files with 510 additions and 0 deletions
--- a/HTTP.cpp
+++ b/HTTP.cpp
@ -0,0 +1,389 @@
				@@ -0,0 +1,389 @@
+/*
+* Copyright (c) 2013-2016, 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 "HTTP.h"
+
+namespace i2p {
+namespace http {
+  const char *HTTP_METHODS[] = {
+    "GET", "HEAD", "POST", "PUT", "PATCH",
+    "DELETE", "OPTIONS", "CONNECT",
+    NULL
+  };
+  const char *HTTP_VERSIONS[] = {
+    "HTTP/1.0", "HTTP/1.1", NULL
+  };
+
+  bool in_cstr_array(const char **haystack, const char *needle) {
+    for (const char *p = haystack[0]; p != NULL; p++) {
+      if (strcmp(p, needle) == 0)
+        return true;
+    }
+    return false;
+  }
+
+  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);
+    }
+  }
+
+  bool parse_header_line(const std::string & line, std::map<std::string, std::string> & headers) {
+    std::size_t pos = 0;
+    std::size_t len = 2; /* strlen(": ") */
+    if ((pos = line.find(": ", pos)) == std::string::npos)
+      return false;
+    while (isspace(line.at(pos + len)))
+      len++;
+    std::string name  = line.substr(0, pos);
+    std::string value = line.substr(pos + len);
+    headers[name] = value;
+    return true;
+  }
+
+  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) != '/') {
+      /* 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_c = url.find('@', pos_p);
+      if (pos_c != std::string::npos) {
+        std::size_t delim = url.find(':', pos_p);
+        if (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 {
+          user = url.substr(pos_p, pos_c - pos_p);
+        }
+        pos_p = pos_c + 1;
+      }
+      /* hostname[:port][/path] */
+      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 (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;
+  }
+
+  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 (!in_cstr_array(HTTP_METHODS, tokens[0].c_str()))
+          return -1;
+        if (!in_cstr_array(HTTP_VERSIONS, tokens[2].c_str()))
+          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);
+        if (!parse_header_line(line, headers))
+          return -1;
+      }
+      pos = eol + strlen(CRLF);
+      if (pos >= eoh)
+        break;
+    }
+    auto it = headers.find("Host");
+    if (it != headers.end ()) {
+      host = it->second;
+    } else if (version == "HTTP/1.1") {
+      return -1; /* 'Host' header required for HTTP/1.1 */
+    } else if (url.host != "") {
+      host = url.host;
+    }
+    return eoh + strlen(HTTP_EOH);
+  }
+
+  std::string HTTPReq::to_string() {
+    std::stringstream ss;
+    ss << method << " " << uri << " " << version << CRLF;
+    ss << "Host: " << host << CRLF;
+    for (auto & h : headers) {
+      ss << h.first << ": " << h.second << CRLF;
+    }
+    ss << CRLF;
+    return ss.str();
+  }
+
+  bool HTTPRes::is_chunked() {
+    auto it = headers.find("Transfer-Encoding");
+    if (it == headers.end())
+      return false;
+    if (it->second.find("chunked") == std::string::npos)
+      return true;
+    return false;
+  }
+
+  long int HTTPRes::length() {
+    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 (!in_cstr_array(HTTP_VERSIONS, tokens[0].c_str()))
+          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);
+        if (!parse_header_line(line, headers))
+          return -1;
+      }
+      pos = eol + strlen(CRLF);
+      if (pos >= eoh)
+        break;
+    }
+
+    return eoh + strlen(HTTP_EOH);
+  }
+
+  std::string HTTPRes::to_string() {
+    std::stringstream ss;
+    ss << version << " " << code << " " << status << CRLF;
+    for (auto & h : headers) {
+      ss << h.first << ": " << h.second << CRLF;
+    }
+    ss << CRLF;
+    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;
+    long int len;
+    while (!in.eof ()) {
+      std::getline (in, hexLen);
+      errno = 0;
+      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
--- a/HTTP.h
+++ b/HTTP.h
@ -0,0 +1,121 @@
				@@ -0,0 +1,121 @@
+/*
+* Copyright (c) 2013-2016, 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
+*/
+
+#ifndef HTTP_H__
+#define HTTP_H__
+
+#include <cstring>
+#include <map>
+#include <sstream>
+#include <string>
+#include <vector>
+
+namespace i2p {
+namespace http {
+  const char CRLF[] = "\r\n";         /**< HTTP line terminator */
+  const char HTTP_EOH[] = "\r\n\r\n"; /**< HTTP end-of-headers mark */
+  extern const char *HTTP_METHODS[];  /**< list of valid HTTP methods */
+  extern const char *HTTP_VERSIONS[]; /**< list of valid HTTP versions */
+
+  struct URL {
+    std::string schema;
+    std::string user;
+    std::string pass;
+    std::string host;
+    unsigned short int port;
+    std::string path;
+    std::string query;
+    std::string frag;
+
+    URL(): schema(""), user(""), pass(""), host(""), port(0), path(""), query(""), frag("") {};
+
+    /**
+     * @brief Tries to parse url from string
+     * @return true on success, false on invalid url
+     */
+    bool parse (const char *str, size_t len = 0);
+    bool parse (const std::string& url);
+
+    /**
+     * @brief Parse query part of url to key/value map
+     * @note Honestly, this should be implemented with std::multimap
+     */
+    bool parse_query(std::map<std::string, std::string> & params);
+
+    /**
+     * @brief Serialize URL structure to url
+     * @note Returns relative url if schema if empty, absolute url otherwise
+     */
+    std::string to_string ();
+  };
+
+  struct HTTPReq {
+    std::map<std::string, std::string> headers;
+    std::string version;
+    std::string method;
+    std::string uri;
+    std::string host;
+
+    HTTPReq (): version("HTTP/1.0"), method("GET"), uri("/") {};
+
+    /**
+     * @brief Tries to parse HTTP request from string
+     * @return -1 on error, 0 on incomplete query, >0 on success
+     * @note Positive return value is a size of header
+     */
+    int parse(const char *buf, size_t len);
+    int parse(const std::string& buf);
+
+    /** @brief Serialize HTTP request to string */
+    std::string to_string();
+  };
+
+  struct HTTPRes {
+    std::map<std::string, std::string> headers;
+    std::string version;
+    std::string status;
+    unsigned short int code;
+
+    HTTPRes (): version("HTTP/1.1"), status("OK"), code(200) {}
+
+    /**
+     * @brief Tries to parse HTTP response from string
+     * @return -1 on error, 0 on incomplete query, >0 on success
+     * @note Positive return value is a size of header
+     */
+    int parse(const char *buf, size_t len);
+    int parse(const std::string& buf);
+
+    /** @brief Serialize HTTP response to string */
+    std::string to_string();
+
+    /** @brief Checks that response declared as chunked data */
+    bool is_chunked();
+
+    /** @brief Returns declared response length or -1 if unknown */
+    long int length();
+  };
+
+  /**
+   * @brief returns HTTP status string by integer code
+   * @param code HTTP code [100, 599]
+   * @return Immutable string with status
+   */
+  const char * HTTPCodeToStatus(int code);
+
+  /**
+   * @brief Replaces %-encoded characters in string with their values
+   * @param data Source string
+   * @param null If set to true - decode also %00 sequence, otherwise - skip
+   * @return Decoded string
+   */
+  std::string UrlDecode(const std::string& data, bool null = false);
+} // http
+} // i2p
+
+#endif /* HTTP_H__ */