path: root/bbot/manifest.cxx

// file      : bbot/manifest.cxx -*- C++ -*-
// copyright : Copyright (c) 2014-2017 Code Synthesis Ltd
// license   : MIT; see accompanying LICENSE file

#include <bbot/manifest>

#include <vector>
#include <string>
#include <cctype>    // isxdigit()
#include <cassert>
#include <sstream>
#include <cstddef>   // size_t
#include <utility>   // move()
#include <cstdint>   // uint64_t
#include <stdexcept> // invalid_argument

#include <butl/utility>             // digit()
#include <butl/tab-parser>
#include <butl/string-parser>
#include <butl/manifest-parser>
#include <butl/manifest-serializer>

using namespace std;
using namespace butl;
using namespace bpkg;

namespace bbot
{
  using parser = manifest_parser;
  using parsing = manifest_parsing;
  using serializer = manifest_serializer;
  using serialization = manifest_serialization;
  using name_value = manifest_name_value;

  using strings = vector<string>;

  // result_status
  //
  string
  to_string (result_status s)
  {
    switch (s)
    {
    case result_status::success:  return "success";
    case result_status::warning:  return "warning";
    case result_status::error:    return "error";
    case result_status::abort:    return "abort";
    case result_status::abnormal: return "abnormal";
    }

    assert (false);
    return string ();
  }

  result_status
  to_result_status (const string& s)
  {
         if (s == "success")  return result_status::success;
    else if (s == "warning")  return result_status::warning;
    else if (s == "error")    return result_status::error;
    else if (s == "abort")    return result_status::abort;
    else if (s == "abnormal") return result_status::abnormal;
    else throw invalid_argument ("invalid result status '" + s + "'");
  }

  // Utility functions
  //
  inline static bool
  valid_sha256 (const string& s) noexcept
  {
    if (s.size () != 64)
      return false;

    for (const auto& c: s)
    {
      if ((c < 'a' || c > 'f' ) && !digit (c))
        return false;
    }

    return true;
  }

  inline static bool
  valid_fingerprint (const string& f) noexcept
  {
    size_t n (f.size ());
    if (n != 32 * 3 - 1)
      return false;

    for (size_t i (0); i < n; ++i)
    {
      char c (f[i]);
      if ((i + 1) % 3 == 0)
      {
        if (c != ':')
          return false;
      }
      else if (!isxdigit (c))
        return false;
    }

    return true;
  }

  // machine_header_manifest
  //
  machine_header_manifest::
  machine_header_manifest (parser& p, bool iu)
      : machine_header_manifest (p, p.next (), iu)
  {
    // Make sure this is the end.
    //
    name_value nv (p.next ());
    if (!nv.empty ())
      throw parsing (p.name (), nv.name_line, nv.name_column,
                     "single machine header manifest expected");
  }

  machine_header_manifest::
  machine_header_manifest (parser& p, name_value nv, bool iu)
  {
    auto bad_name = [&p, &nv] (const string& d)
    {
      throw parsing (p.name (), nv.name_line, nv.name_column, d);
    };

    auto bad_value = [&p, &nv] (const string& d)
    {
      throw parsing (p.name (), nv.value_line, nv.value_column, d);
    };

    // Make sure this is the start and we support the version.
    //
    if (!nv.name.empty ())
      bad_name ("start of machine header manifest expected");

    if (nv.value != "1")
      bad_value ("unsupported format version");

    for (nv = p.next (); !nv.empty (); nv = p.next ())
    {
      string& n (nv.name);
      string& v (nv.value);

      if (n == "id")
      {
        if (!id.empty ())
          bad_name ("machine id redefinition");

        if (v.empty ())
          bad_value ("empty machine id");

        id = move (v);
      }
      else if (n == "name")
      {
        if (!name.empty ())
          bad_name ("machine name redefinition");

        if (v.empty ())
          bad_value ("empty machine name");

        name = move (v);
      }
      else if (n == "summary")
      {
        if (!summary.empty ())
          bad_name ("machine summary redefinition");

        if (v.empty ())
          bad_value ("empty machine summary");

        summary = move (v);
      }
      else if (!iu)
        bad_name ("unknown name '" + n + "' in machine header manifest");
    }

    // Verify all non-optional values were specified.
    //
    if (id.empty ())
      bad_value ("no machine id specified");

    if (name.empty ())
      bad_value ("no machine name specified");

    if (summary.empty ())
      bad_value ("no machine summary specified");
  }

  void machine_header_manifest::
  serialize (serializer& s) const
  {
    // @@ Should we check that all non-optional values are specified and all
    //    values are valid?
    //
    s.next ("", "1"); // Start of manifest.
    s.next ("id", id);
    s.next ("name", name);
    s.next ("summary", summary);
    s.next ("", ""); // End of manifest.
  }

  // task_request_manifest
  //
  task_request_manifest::
  task_request_manifest (parser& p, bool iu)
  {
    name_value nv (p.next ());

    auto bad_name = [&p, &nv] (const string& d)
    {
      throw parsing (p.name (), nv.name_line, nv.name_column, d);
    };

    auto bad_value = [&p, &nv] (const string& d)
    {
      throw parsing (p.name (), nv.value_line, nv.value_column, d);
    };

    // Make sure this is the start and we support the version.
    //
    if (!nv.name.empty ())
      bad_name ("start of task request manifest expected");

    if (nv.value != "1")
      bad_value ("unsupported format version");

    // Parse the task request manifest.
    //
    for (nv = p.next (); !nv.empty (); nv = p.next ())
    {
      string& n (nv.name);
      string& v (nv.value);

      if (n == "agent")
      {
        if (!agent.empty ())
          bad_name ("task request agent redefinition");

        if (v.empty ())
          bad_value ("empty task request agent");

        agent = move (v);
      }
      else if (n == "fingerprint")
      {
        if (fingerprint)
          bad_name ("task request fingerprint redefinition");

        if (!valid_sha256 (v))
          bad_value ("invalid task request fingerprint");

        fingerprint = move (v);
      }
      else if (!iu)
        bad_name ("unknown name '" + n + "' in task request manifest");
    }

    // Verify all non-optional values were specified.
    //
    if (agent.empty ())
      bad_value ("no task request agent specified");

    // Parse machine header manifests.
    //
    for (nv = p.next (); !nv.empty (); nv = p.next ())
      machines.emplace_back (machine_header_manifest (p, nv, iu));

    if (machines.empty ())
      bad_value ("no task request machines specified");
  }

  void task_request_manifest::
  serialize (serializer& s) const
  {
    // @@ Should we check that all non-optional values are specified and all
    //    values are valid?
    //
    s.next ("", "1"); // Start of manifest.
    s.next ("agent", agent);

    if (fingerprint)
      s.next ("fingerprint", *fingerprint);

    s.next ("", "");  // End of manifest.

    for (const machine_header_manifest& m: machines)
      m.serialize (s);

    s.next ("", ""); // End of stream.
  }

  // task_manifest
  //
  task_manifest::
  task_manifest (parser& p, bool iu)
      : task_manifest (p, p.next (), iu)
  {
    // Make sure this is the end.
    //
    name_value nv (p.next ());
    if (!nv.empty ())
      throw parsing (p.name (), nv.name_line, nv.name_column,
                     "single task manifest expected");
  }

  task_manifest::
  task_manifest (parser& p, name_value nv, bool iu)
  {
    auto bad_name = [&p, &nv] (const string& d)
    {
      throw parsing (p.name (), nv.name_line, nv.name_column, d);
    };

    // Offsets are used to tie an error to the specific position inside a
    // manifest value (possibly a multiline one).
    //
    auto bad_value = [&p, &nv] (
      const string& d, uint64_t column_offset = 0, uint64_t line_offset = 0)
    {
      throw parsing (p.name (),
                     nv.value_line + line_offset,
                     (line_offset == 0 ? nv.value_column : 1) + column_offset,
                     d);
    };

    // Make sure this is the start and we support the version.
    //
    if (!nv.name.empty ())
      bad_name ("start of task manifest expected");

    if (nv.value != "1")
      bad_value ("unsupported format version");

    // Parse the task manifest.
    //
    for (nv = p.next (); !nv.empty (); nv = p.next ())
    {
      string& n (nv.name);
      string& v (nv.value);

      if (n == "name")
      {
        if (!name.empty ())
          bad_name ("task package name redefinition");

        if (v.empty ())
          bad_value ("empty task package name");

        name = move (v);
      }
      else if (n == "version")
      {
        if (!version.empty ())
          bad_name ("task package version redefinition");

        try
        {
          version = bpkg::version (move (v));
        }
        catch (const invalid_argument& e)
        {
          bad_value (string ("invalid task package version: ") + e.what ());
        }

        // Versions like 1.2.3- are forbidden in manifest as intended to be
        // used for version constrains rather than actual releases.
        //
        if (version.release && version.release->empty ())
          bad_value ("invalid task package version release");
      }
      else if (n == "repository")
      {
        if (!repository.empty ())
          bad_name ("task repository redefinition");

        if (v.empty ())
          bad_value ("empty task repository");

        try
        {
          // Call remote/absolute repository location constructor (throws
          // invalid_argument for relative location).
          //
          repository = repository_location (move (v));
        }
        catch (const invalid_argument& e)
        {
          bad_value (string ("invalid task repository: ") + e.what ());
        }
      }
      else if (n == "trust")
      {
        if (v != "yes" && !valid_fingerprint (v))
          bad_value ("invalid repository certificate fingerprint");

        trust.emplace_back (move (v));
      }
      else if (n == "machine")
      {
        if (!machine.empty ())
          bad_name ("task machine redefinition");

        if (v.empty ())
          bad_value ("empty task machine");

        machine = move (v);
      }
      else if (n == "target")
      {
        if (target)
          bad_name ("task target redefinition");

        try
        {
          target = target_triplet (v);
        }
        catch (const invalid_argument& e)
        {
          bad_value (string ("invalid task target: ") + e.what ());
        }
      }
      else if (n == "config")
      {
        if (!config.empty ())
          bad_name ("task configuration redefinition");

        // Note that when reporting errors we combine the manifest value
        // position with the respective field and error positions.
        //
        try
        {
          istringstream is (v);
          tab_parser parser (is, "");

          // Here we naturally support multiline config manifest.
          //
          tab_fields tl;
          while (!(tl = parser.next ()).empty ())
          {
            for (auto& tf: tl)
            {
              try
              {
                check_config (tf.value);
              }
              catch (const invalid_argument& e)
              {
                bad_value (string ("invalid task configuration: ") + e.what (),
                           tf.column - 1,
                           tl.line - 1);
              }

              config.emplace_back (move (tf.value));
            }
          }
        }
        catch (const tab_parsing& e)
        {
          bad_value ("invalid task configuration: " + e.description,
                     e.column - 1,
                     e.line - 1);
        }

        if (config.empty ())
          bad_value ("empty task configuration");
      }
      else if (!iu)
        bad_name ("unknown name '" + n + "' in task manifest");
    }

    // Verify all non-optional values were specified.
    //
    if (name.empty ())
      bad_value ("no task package name specified");

    if (version.empty ())
      bad_value ("no task package version specified");

    if (repository.empty ())
      bad_value ("no task repository specified");

    if (machine.empty ())
      bad_value ("no task machine specified");
  }

  void task_manifest::
  serialize (serializer& s) const
  {
    // @@ Should we check that all non-optional values are specified and all
    //    values are valid?
    //
    s.next ("", "1"); // Start of manifest.
    s.next ("name", name);
    s.next ("version", version.string ());
    s.next ("repository", repository.string ());

    for (const auto& v: trust)
      s.next ("trust", v);

    s.next ("machine", machine);

    if (target)
      s.next ("target", target->string ());

    // Recompose config string as a space-separated variable list,
    //
    if (!config.empty ())
    {
      string v;
      for (auto b (config.cbegin ()), i (b), e (config.cend ()); i != e; ++i)
      {
        if (i != b)
          v += ' ';

        v += *i;
      }

      s.next ("config", v);
    }

    s.next ("", ""); // End of manifest.
  }

  strings task_manifest::
  unquoted_config ()
  {
    return string_parser::unquote (config);
  }

  void task_manifest::
  check_config (const string& s)
  {
    auto i (s.begin ());
    auto e (s.end ());

    // Iterate until the variable name end and check that it contains no
    // whitespaces.
    //
    for (; i != e; ++i)
    {
      char c (*i);

      if (c == ' ' || c == '\t') // Whitespace in name.
        throw invalid_argument ("expected variable assignment");
      else if (c == '=')
        break;
    }

    if (i == e)
      throw invalid_argument ("no variable value");
  }

  // task_response_manifest
  //
  task_response_manifest::
  task_response_manifest (parser& p, bool iu)
  {
    name_value nv (p.next ());

    auto bad_name = [&p, &nv] (const string& d)
    {
      throw parsing (p.name (), nv.name_line, nv.name_column, d);
    };

    auto bad_value = [&p, &nv] (const string& d)
    {
      throw parsing (p.name (), nv.value_line, nv.value_column, d);
    };

    // Make sure this is the start and we support the version.
    //
    if (!nv.name.empty ())
      bad_name ("start of task response manifest expected");

    if (nv.value != "1")
      bad_value ("unsupported format version");

    // Parse the task response manifest.
    //
    // Note that we need to distinguish an empty and absent session.
    //
    optional<string> sess;
    for (nv = p.next (); !nv.empty (); nv = p.next ())
    {
      string& n (nv.name);
      string& v (nv.value);

      if (n == "session")
      {
        if (sess)
          bad_name ("task response session redefinition");

        sess = move (v);
      }
      else if (n == "challenge")
      {
        if (challenge)
          bad_name ("task response challenge redefinition");

        if (v.empty ())
          bad_value ("empty task response challenge");

        challenge = move (v);
      }
      else if (n == "result-url")
      {
        if (result_url)
          bad_name ("task response result url redefinition");

        if (v.empty ())
          bad_value ("empty task response result url");

        result_url = move (v);
      }
      else if (!iu)
        bad_name ("unknown name '" + n + "' in task response manifest");
    }

    // Verify all non-optional values were specified, and all values are
    // expected.
    //
    if (!sess)
      bad_value ("no task response session specified");

    session = move (*sess);

    // If session is not empty then the challenge may, and the result url
    // must, be present, otherwise they shouldn't.
    //
    if (!session.empty ())
    {
      if (!result_url)
        bad_value ("no task response result url specified");
    }
    else
    {
      if (challenge)
        bad_value ("unexpected task response challenge");

      if (result_url)
        bad_value ("unexpected task response result url");
    }

    // If session is not empty then the task manifest must follow, otherwise it
    // shouldn't.
    //
    nv = p.next ();

    if (!session.empty ())
    {
      if (nv.empty ())
        bad_value ("task manifest expected");

      task = task_manifest (p, nv, iu);

      nv = p.next ();
    }

    // Make sure this is the end.
    //
    if (!nv.empty ())
      throw parsing (p.name (), nv.name_line, nv.name_column,
                     "single task response manifest expected");
  }

  void task_response_manifest::
  serialize (serializer& s) const
  {
    // @@ Should we check that all non-optional values are specified and all
    //    values are valid?
    //
    s.next ("", "1"); // Start of manifest.
    s.next ("session", session);

    if (challenge)
      s.next ("challenge", *challenge);

    if (result_url)
      s.next ("result-url", *result_url);

    s.next ("", ""); // End of manifest.

    if (task)
      task->serialize (s);

    s.next ("", ""); // End of stream.
  }

  // result_manifest
  //
  result_manifest::
  result_manifest (parser& p, bool iu)
      : result_manifest (p, p.next (), iu)
  {
    // Make sure this is the end.
    //
    name_value nv (p.next ());
    if (!nv.empty ())
      throw parsing (p.name (), nv.name_line, nv.name_column,
                     "single result manifest expected");
  }

  result_manifest::
  result_manifest (parser& p, name_value nv, bool iu)
  {
    auto bad_name = [&p, &nv] (const string& d)
    {
      throw parsing (p.name (), nv.name_line, nv.name_column, d);
    };

    auto bad_value = [&p, &nv] (const string& d, size_t offset = 0)
    {
      throw parsing (p.name (), nv.value_line, nv.value_column + offset, d);
    };

    auto result_stat =
      [&bad_value] (const string& v, const string& what) -> result_status
    {
      try
      {
        return to_result_status (v);
      }
      catch (const invalid_argument&)
      {
        bad_value ("invalid " + what);
      }

      // Can't be here. Would be redundant if it were possible to declare
      // lambda with the [[noreturn]] attribute. Note that GCC (non-portably)
      // supports that.
      //
      return result_status::abnormal;
    };

    // Make sure this is the start and we support the version.
    //
    if (!nv.name.empty ())
      bad_name ("start of result manifest expected");

    if (nv.value != "1")
      bad_value ("unsupported format version");

    // Parse the result manifest.
    //
    optional<result_status> stat;

    // Number of parsed *-log values. Also denotes the next expected log type.
    //
    size_t nlog (0);

    for (nv = p.next (); !nv.empty (); nv = p.next ())
    {
      string& n (nv.name);
      string& v (nv.value);

      if (n == "name")
      {
        if (!name.empty ())
          bad_name ("result package name redefinition");

        if (v.empty ())
          bad_value ("empty result package name");

        name = move (v);
      }
      else if (n == "version")
      {
        if (!version.empty ())
          bad_name ("result package version redefinition");

        try
        {
          version = bpkg::version (move (v));
        }
        catch (const invalid_argument& e)
        {
          bad_value (string ("invalid result package version: ") + e.what ());
        }

        // Versions like 1.2.3- are forbidden in manifest as intended to be
        // used for version constrains rather than actual releases.
        //
        if (version.release && version.release->empty ())
          bad_value ("invalid result package version release");
      }
      else if (n == "status")
      {
        if (stat)
          bad_name ("result status redefinition");

        stat = result_stat (v, "result status");
      }
      else
      {
        size_t nn (n.size ()); // Name length.

        // Note: returns false if nothing preceeds a suffix.
        //
        auto suffix = [&n, nn] (const char* s, size_t sn) -> bool
        {
          return nn > sn && n.compare (nn - sn, sn, s) == 0;
        };

        size_t sn;
        if (suffix ("-status", sn = 7))
        {
          if (!stat)
            bad_name ("result status must appear first");

          if (nlog > 0) // Some logs have already been parsed.
            bad_name (n + " after operations logs");

          string op (n, 0, nn - sn);

          // Make sure the operation result status is not redefined.
          //
          for (const auto& r: results)
          {
            if (r.operation == op)
              bad_name ("result " + n + " redefinition");
          }

          // Add the operation result (log will come later).
          //
          results.push_back ({move (op), result_stat (v, n), string ()});
        }
        else if (suffix ("-log", sn = 4))
        {
          string op (n, 0, nn - sn);

          // Check that specifically this operation log is expected.
          //
          if (nlog >= results.size ())
            bad_name ("unexpected " + n);

          if (results[nlog].operation != op)
            bad_name (results[nlog].operation + "-log is expected");

          // Save operation log.
          //
          results[nlog++].log = move (v);
        }
        else if (!iu)
          bad_name ("unknown name '" + n + "' in result manifest");
      }
    }

    // Verify all non-optional values were specified.
    //
    if (name.empty ())
      bad_value ("no result package name specified");

    if (version.empty ())
      bad_value ("no result package version specified");

    if (!stat)
      bad_value ("no result status specified");

    // @@ Checking that the result status is consistent with operations
    //    statuses is a bit hairy, so let's postpone for now.
    //
    status = move (*stat);

    // Check that we have log for every operation result status.
    //
    if (nlog < results.size ())
      bad_name ("no result " + results[nlog].operation + "-log specified");
  }

  void result_manifest::
  serialize (serializer& s) const
  {
    // @@ Should we check that all non-optional values are specified and all
    //    values are valid?
    //
    s.next ("", "1"); // Start of manifest.
    s.next ("name", name);
    s.next ("version", version.string ());
    s.next ("status", to_string (status));

    // Serialize *-status values.
    //
    for (const auto& r: results)
      s.next (r.operation + "-status", to_string (r.status));

    // Serialize *-log values.
    //
    for (const auto& r: results)
      s.next (r.operation + "-log", r.log);

    s.next ("", ""); // End of manifest.
  }

  // result_request_manifest
  //
  result_request_manifest::
  result_request_manifest (parser& p, bool iu)
  {
    name_value nv (p.next ());

    auto bad_name = [&p, &nv] (const string& d)
    {
      throw parsing (p.name (), nv.name_line, nv.name_column, d);
    };

    auto bad_value = [&p, &nv] (const string& d)
    {
      throw parsing (p.name (), nv.value_line, nv.value_column, d);
    };

    // Make sure this is the start and we support the version.
    //
    if (!nv.name.empty ())
      bad_name ("start of result request manifest expected");

    if (nv.value != "1")
      bad_value ("unsupported format version");

    // Parse the result request manifest.
    //
    for (nv = p.next (); !nv.empty (); nv = p.next ())
    {
      string& n (nv.name);
      string& v (nv.value);

      if (n == "session")
      {
        if (!session.empty ())
          bad_name ("result request session redefinition");

        if (v.empty ())
          bad_value ("empty result request session");

        session = move (v);
      }
      else if (n == "challenge")
      {
        if (challenge)
          bad_name ("result request challenge redefinition");

        if (v.empty ())
          bad_value ("empty result request challenge");

        challenge = move (v);
      }
      else if (!iu)
        bad_name ("unknown name '" + n + "' in result request manifest");
    }

    // Verify all non-optional values were specified.
    //
    if (session.empty ())
      bad_value ("no result request session specified");

    nv = p.next ();
    if (nv.empty ())
      bad_value ("result manifest expected");

    result = result_manifest (p, nv, iu);

    // Make sure this is the end.
    //
    nv = p.next ();
    if (!nv.empty ())
      throw parsing (p.name (), nv.name_line, nv.name_column,
                     "single result request manifest expected");
  }

  void result_request_manifest::
  serialize (serializer& s) const
  {
    // @@ Should we check that all non-optional values are specified and all
    //    values are valid?
    //
    s.next ("", "1"); // Start of manifest.
    s.next ("session", session);

    if (challenge)
      s.next ("challenge", *challenge);

    s.next ("", ""); // End of manifest.

    result.serialize (s);
    s.next ("", ""); // End of stream.
  }
}