XrdHttpTpcState.cc

Go to the documentation of this file.

 
#include <algorithm>
#include <sstream>
#include <stdexcept>
 
#include "XrdVersion.hh"
#include "XrdHttp/XrdHttpExtHandler.hh"
#include "XrdSfs/XrdSfsInterface.hh"
#include "XrdOuc/XrdOucTUtils.hh"
 
#include <curl/curl.h>
 
#include "XrdHttpTpcState.hh"
#include "XrdHttpTpcStream.hh"
 
#include "XrdHttp/XrdHttpHeaderUtils.hh"
 
using namespace TPC;
 
 
State::~State() {
    if (m_headers) {
            curl_slist_free_all(m_headers);
            m_headers = NULL;
            if (m_curl) {curl_easy_setopt(m_curl, CURLOPT_HTTPHEADER, m_headers);}
    }
}
 
 
void State::Move(State &other)
{
    m_push = other.m_push;
    m_recv_status_line = other.m_recv_status_line;
    m_recv_all_headers = other.m_recv_all_headers;
    m_offset = other.m_offset;
    m_start_offset = other.m_start_offset;
    m_status_code = other.m_status_code;
    m_content_length = other.m_content_length;
    m_push_length = other.m_push_length;
    m_stream = other.m_stream;
    m_curl = other.m_curl;
    m_headers = other.m_headers;
    m_headers_copy = other.m_headers_copy;
    m_resp_protocol = other.m_resp_protocol;
    m_is_transfer_state = other.m_is_transfer_state;
    curl_easy_setopt(m_curl, CURLOPT_HEADERDATA, this);
    if (m_is_transfer_state) {
        if (m_push) {
            curl_easy_setopt(m_curl, CURLOPT_READDATA, this);
        } else {
            curl_easy_setopt(m_curl, CURLOPT_WRITEDATA, this);
        }
    }
    tpcForwardCreds = other.tpcForwardCreds;
    other.m_headers_copy.clear();
    other.m_curl = NULL;
    other.m_headers = NULL;
    other.m_stream = NULL;
    other.m_repr_digests = m_repr_digests;
}
 
 
bool State::InstallHandlers(CURL *curl) {
    curl_easy_setopt(curl, CURLOPT_USERAGENT, "xrootd-tpc/" XrdVERSION);
    curl_easy_setopt(curl, CURLOPT_HEADERFUNCTION, &State::HeaderCB);
    curl_easy_setopt(curl, CURLOPT_HEADERDATA, this);
    if(m_is_transfer_state) {
        if (m_push) {
            curl_easy_setopt(curl, CURLOPT_UPLOAD, 1);
            curl_easy_setopt(curl, CURLOPT_READFUNCTION, &State::ReadCB);
            curl_easy_setopt(curl, CURLOPT_READDATA, this);
            curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, &State::PushRespCB);
            curl_easy_setopt(curl, CURLOPT_WRITEDATA, this);
            struct stat buf;
            if (SFS_OK == m_stream->Stat(&buf)) {
                m_push_length = buf.st_size;
                curl_easy_setopt(curl, CURLOPT_INFILESIZE_LARGE, buf.st_size);
            }
        } else {
            curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, &State::WriteCB);
            curl_easy_setopt(curl, CURLOPT_WRITEDATA, this);
        }
    }
    curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1L);
    if(tpcForwardCreds) {
      curl_easy_setopt(curl,CURLOPT_UNRESTRICTED_AUTH,1L);
    }
 
    return true;
}
 
void State::SetupHeaders(XrdHttpExtReq &req) {
    struct curl_slist *list = NULL;
    for (const auto & [header,value]: req.headers) {
        if (!strncasecmp(header.c_str(),"copy-header", 11)) {
            list = curl_slist_append(list, value.c_str());
            m_headers_copy.emplace_back(value);
        }
        // Note: len("TransferHeader") == 14
        if (!strncasecmp(header.c_str(),"transferheader",14)) {
            std::stringstream ss;
            ss << header.substr(14) << ": " << value;
            list = curl_slist_append(list, ss.str().c_str());
            m_headers_copy.emplace_back(ss.str());
        }
    }
 
    if(m_is_transfer_state && !m_push && !req.mReprDigest.empty()) {
      size_t reprDigestSize = req.mReprDigest.size();
      std::stringstream ss;
      ss << "Want-Repr-Digest: ";
      size_t cpt = 1;
      for (const auto &kv: req.mReprDigest) {
        // We put the same weight for the digest names as we do not have any way, according to the specs,
        // to give priority to a digest name in particular
        ss << kv.first << '=' << 5;
        if(cpt < reprDigestSize) {
          ss << ',';
        }
        cpt++;
      }
      list = curl_slist_append(list, ss.str().c_str());
      m_headers_copy.emplace_back(ss.str());
    }
 
    if (m_is_transfer_state && m_push && m_push_length > 0) {
        // On libcurl 8.5.0 - 8.9.1, we've observed bugs causing failures whenever
        // `Expect: 100-continue` is not used.  Older versions of libcurl unconditionally
        // set `Expect` whenever PUT is used (likely an older bug).  To workaround the issue,
        // we force `Expect` to be set, triggering the older libcurl behavior.
        // See: https://github.com/xrootd/xrootd/issues/2470
        // See: https://github.com/curl/curl/issues/17004
        list = curl_slist_append(list, "Expect: 100-continue");
        // Add Repr-Digest header to PUT request (PUSH)
        auto reprDigest = XrdOucTUtils::caseInsensitiveFind(req.headers,"repr-digest");
        if(reprDigest != req.headers.end()) {
          std::string reprDigestHeader {"Repr-Digest: " + reprDigest->second};
          curl_slist_append(list,reprDigestHeader.c_str());
        }
    }
 
    if (list != nullptr) {
        curl_easy_setopt(m_curl, CURLOPT_HTTPHEADER, list);
        m_headers = list;
    }
}
 
void State::SetupHeadersForHEAD(XrdHttpExtReq &req) {
  struct curl_slist *list = NULL;
  for (const auto & [header,value]: req.headers) {
    if (!strncasecmp(header.c_str(),"copy-header", 11)) {
      list = curl_slist_append(list, value.c_str());
    }
    // Note: len("TransferHeader") == 14
    if (!strncasecmp(header.c_str(),"transferheader",14)) {
      std::stringstream ss;
      ss << header.substr(14) << ": " << value;
      list = curl_slist_append(list, ss.str().c_str());
    }
  }
  if(!req.mReprDigest.empty()) {
    size_t reprDigestSize = req.mReprDigest.size();
    std::stringstream ss;
    ss << "Want-Repr-Digest: ";
    size_t cpt = 1;
    for (const auto &kv: req.mReprDigest) {
      // We put the same weight for the digest names as we do not have any way, according to the specs,
      // to give priority to a digest name in particular
      ss << kv.first << '=' << 5;
      if(cpt < reprDigestSize) {
        ss << ',';
      }
      cpt++;
    }
    list = curl_slist_append(list, ss.str().c_str());
  }
 
  if (list != nullptr) {
    curl_easy_setopt(m_curl, CURLOPT_HTTPHEADER, list);
  }
}
 
void State::ResetAfterRequest() {
    m_offset = 0;
    m_status_code = -1;
    m_content_length = -1;
    m_push_length = -1;
    m_recv_all_headers = false;
    m_recv_status_line = false;
    m_repr_digests.clear();
}
 
size_t State::HeaderCB(char *buffer, size_t size, size_t nitems, void *userdata)
{
    State *obj = static_cast<State*>(userdata);
    std::string header(buffer, size*nitems);
    return obj->Header(header);
}
 
int State::Header(const std::string &header) {
    //printf("Received remote header (%d, %d): %s", m_recv_all_headers, m_recv_status_line, header.c_str());
    if (m_recv_all_headers) {  // This is the second request -- maybe processed a redirect?
        m_recv_all_headers = false;
        m_recv_status_line = false;
    }
    if (!m_recv_status_line) {
        std::stringstream ss(header);
        std::string item;
        if (!std::getline(ss, item, ' ')) return 0;
        m_resp_protocol = item;
        //printf("\n\nResponse protocol: %s\n", m_resp_protocol.c_str());
        if (!std::getline(ss, item, ' ')) return 0;
        try {
            m_status_code = std::stol(item);
        } catch (...) {
            return 0;
        }
        m_recv_status_line = true;
    } else if (header.size() == 0 || header == "\n" || header == "\r\n") {
        m_recv_all_headers = true;
    }
    else if (header != "\r\n") {
        // Parse the header
        std::size_t found = header.find(":");
        if (found != std::string::npos) {
            std::string header_name = header.substr(0, found);
            std::transform(header_name.begin(), header_name.end(), header_name.begin(), ::tolower);
            std::string header_value = header.substr(found+1);
            if (header_name == "content-length")
            {
                try {
                    m_content_length = std::stoll(header_value);
                } catch (...) {
                    // Header unparseable -- not a great sign, fail request.
                    //printf("Content-length header unparseable\n");
                    return 0;
                }
            }
            if(header_name == "repr-digest") {
              XrdHttpHeaderUtils::parseReprDigest(header_value,m_repr_digests);
            }
        } else {
            // Non-empty header that isn't the status line, but no ':' present --
            // malformed request?
            //printf("Malformed header: %s\n", header.c_str());
            return 0;
        }
    }
    return header.size();
}
 
size_t State::WriteCB(void *buffer, size_t size, size_t nitems, void *userdata) {
    State *obj = static_cast<State*>(userdata);
    if (obj->GetStatusCode() < 0) {
        return 0;
     }  // malformed request - got body before headers.
    if (obj->GetStatusCode() >= 400) {
        obj->m_error_buf += std::string(static_cast<char*>(buffer),
                                        std::min(static_cast<size_t>(1024), size*nitems));
        // Record error messages until we hit a KB; at that point, fail out.
        if (obj->m_error_buf.size() >= 1024)
            return 0;
        else
            return size*nitems;
    }  // Status indicates failure.
    return obj->Write(static_cast<char*>(buffer), size*nitems);
}
 
size_t State::PushRespCB(void *buffer, size_t size, size_t nitems, void *userdata) {
  State *obj = static_cast<State*>(userdata);
  // Note: The obj's status code is set by the HeaderCB once there's a reply from the passive server
  if (obj->GetStatusCode() < 0) {
    return 0;
  }  // malformed request - got body before headers.
  if (obj->GetStatusCode() >= 400) {
    obj->m_error_buf += std::string(static_cast<char*>(buffer),
                                    std::min(static_cast<size_t>(1024), size*nitems));
    // Record error messages until we hit a KB; at that point, fail out.
    if (obj->m_error_buf.size() >= 1024)
      return 0;
    else
      return size*nitems;
  }
  return size*nitems;
}
 
ssize_t State::Write(char *buffer, size_t size) {
    ssize_t retval = m_stream->Write(m_start_offset + m_offset, buffer, size, false);
    if (retval == SFS_ERROR) {
        m_error_buf = m_stream->GetErrorMessage();
        m_error_code = errWrite;
        return -1;
    }
    m_offset += retval;
    return retval;
}
 
void State::RecordFinalizeError(int error_code, const std::string &error_msg) {
    if (m_finalize_error_code) {
        return;
    }
    m_finalize_error_code = error_code;
    m_finalize_error_buf = error_msg;
}
 
int State::Flush() {
    if (m_push) {
        return 0;
    }
 
    if (m_stream->Flush() == SFS_ERROR) {
        RecordFinalizeError(errFlush, m_stream->GetErrorMessage());
        return -1;
    }
    return 0;
}
 
size_t State::ReadCB(void *buffer, size_t size, size_t nitems, void *userdata) {
    State *obj = static_cast<State*>(userdata);
    if (obj->GetStatusCode() < 0) {return 0;}  // malformed request - got body before headers.
    if (obj->GetStatusCode() >= 400) {return 0;}  // Status indicates failure.
    return obj->Read(static_cast<char*>(buffer), size*nitems);
}
 
int State::Read(char *buffer, size_t size) {
    int retval = m_stream->Read(m_start_offset + m_offset, buffer, size);
    if (retval == SFS_ERROR) {
        return -1;
    }
    m_offset += retval;
    //printf("Read a total of %ld bytes.\n", m_offset);
    return retval;
}
 
State *State::Duplicate() {
    CURL *curl = curl_easy_duphandle(m_curl);
    if (!curl) {
        throw std::runtime_error("Failed to duplicate existing curl handle.");
    }
 
    State *state = new State(0, *m_stream, curl, m_push, tpcForwardCreds);
 
    if (m_headers) {
        state->m_headers_copy.reserve(m_headers_copy.size());
        for (std::vector<std::string>::const_iterator header_iter = m_headers_copy.begin();
             header_iter != m_headers_copy.end();
             header_iter++) {
            state->m_headers = curl_slist_append(state->m_headers, header_iter->c_str());
            state->m_headers_copy.push_back(*header_iter);
        }
        curl_easy_setopt(curl, CURLOPT_HTTPHEADER, NULL);
        curl_easy_setopt(curl, CURLOPT_HTTPHEADER, state->m_headers);
    }
 
    return state;
}
 
void State::SetTransferParameters(off_t offset, size_t size) {
    m_start_offset = offset;
    m_offset = 0;
    m_content_length = size;
    std::stringstream ss;
    ss << offset << "-" << (offset+size-1);
    curl_easy_setopt(m_curl, CURLOPT_RANGE, ss.str().c_str());
}
 
int State::AvailableBuffers() const
{
    return m_stream->AvailableBuffers();
}
 
void State::DumpBuffers() const
{
    m_stream->DumpBuffers();
}
 
bool State::Finalize()
{
    if (!m_stream->Finalize()) {
        RecordFinalizeError(errClose, m_stream->GetErrorMessage());
        return false;
    }
    return true;
}
 
std::string State::GetConnectionDescription()
{
    // CURLINFO_PRIMARY_PORT is only defined for 7.21.0 or later; on older
    // library versions, simply omit this information.
#if LIBCURL_VERSION_NUM >= 0x071500
    char *curl_ip = NULL;
    CURLcode rc = curl_easy_getinfo(m_curl, CURLINFO_PRIMARY_IP, &curl_ip);
    if ((rc != CURLE_OK) || !curl_ip) {
        return "";
    }
    long curl_port = 0;
    rc = curl_easy_getinfo(m_curl, CURLINFO_PRIMARY_PORT, &curl_port);
    if ((rc != CURLE_OK) || !curl_port) {
        return "";
    }
    std::stringstream ss;
    // libcurl returns IPv6 addresses of the form:
    //    2600:900:6:1301:5054:ff:fe0b:9cba:8000
    // However the HTTP-TPC spec says to use the form
    //   [2600:900:6:1301:5054:ff:fe0b:9cba]:8000
    // Hence, we add '[' and ']' whenever a ':' is seen.
    if (NULL == strchr(curl_ip, ':'))
        ss << "tcp:" << curl_ip << ":" << curl_port;
    else
        ss << "tcp:[" << curl_ip << "]:" << curl_port;
    return ss.str();
#else
    return "";
#endif
}