rtorrent/src/rpc/parse.cc

#include "config.h"

#include <cctype>
#include <charconv>
#include <cstring>
#include <cstdio>
#include <limits>
#include <locale>
#include <torrent/exceptions.h>

#include "globals.h"
#include "parse.h"

namespace rpc {

constexpr uint32_t max_parse_depth = 1024;

const char*
parse_skip_wspace(const char* first, const char* last) {
  while (first != last && parse_is_space(*first))
    first++;

  return first;
}

const char*
parse_skip_wspace(const char* first) {
  while (parse_is_space(*first))
    first++;

  return first;
}

const char*
parse_string(const char* first, const char* last, std::string* dest, bool (*delim)(const char)) {
  if (first == last)
    return first;

  if (parse_is_quote(*first)) {
    first++;

    while (first != last) {
      if (parse_is_quote(*first))
        return ++first;

      if (parse_is_escape(*first) && ++first == last)
        throw torrent::input_error("Escape character at end of input.");

      dest->push_back(*first++);
    }

    throw torrent::input_error("Missing closing quote.");

  } else {
    while (first != last) {
      if (delim(*first))
        return first;

      if (parse_is_escape(*first) && ++first == last)
        throw torrent::input_error("Escape character at end of input.");

      dest->push_back(*first++);
    }

    return first;
  }
}

void
parse_whole_string(const char* first, const char* last, std::string* dest) {
  first = parse_skip_wspace(first, last);
  first = parse_string(first, last, dest);
  first = parse_skip_wspace(first, last);

  if (first != last)
    throw torrent::input_error("Junk at end of input.");
}

const char*
parse_value(const char* src, int64_t* value, int base, int unit) {
  const char* last = parse_value_nothrow(src, value, base, unit);

  if (last == src)
    throw torrent::input_error("Could not convert string to value.");

  return last;
}

void
parse_whole_value(const char* src, int64_t* value, int base, int unit) {
  const char* last = parse_value_nothrow(src, value, base, unit);

  if (last == src || *parse_skip_wspace(last) != '\0')
    throw torrent::input_error("Could not convert string to value.");
}

bool
parse_whole_value_nothrow(const char* src, int64_t* value, int base, int unit) {
  const char* last = parse_value_nothrow(src, value, base, unit);

  if (last == src || *parse_skip_wspace(last) != '\0')
    return false;

  return true;
}

static bool
value_fits_shifted(int64_t value, int shift) {
  return value <= (std::numeric_limits<int64_t>::max() >> shift) &&
         value >= (std::numeric_limits<int64_t>::min() >> shift);
}

const char*
parse_value_nothrow(const char* src, int64_t* value, int base, int unit) {
  if (unit <= 0)
    throw torrent::input_error("Command::string_to_value_unit(...) received unit <= 0.");

  if (base != 0 && base != 8 && base != 10 && base != 16)
    throw torrent::input_error("Command::string_to_value_unit(...) received invalid base.");

  const char* first = src;

  while (parse_is_space(*src))
    src++;

  if (src[0] == '+')
    return first;

  if (src[0] == '-') {
    if (base == 8 || base == 16)
      return first;

    if (src[1] == '0')
      return first;
  }

  if (base == 10 && src[0] == '0' && (src[1] >= '0' && src[1] <= '9'))
    return first;

  char* last{};

  errno  = 0;
  *value = strtoll(src, &last, base);

  if (errno == ERANGE)
    return first;

  if (last == src) {
    *value = 0;

    if (strcasecmp(src, "no") == 0)    { *value = 0; return src + strlen("no"); }
    if (strcasecmp(src, "yes") == 0)   { *value = 1; return src + strlen("yes"); }
    if (strcasecmp(src, "true") == 0)  { *value = 1; return src + strlen("true"); }
    if (strcasecmp(src, "false") == 0) { *value = 0; return src + strlen("false"); }

    return first;
  }

  switch (*last) {
  case 'b':
  case 'B': ++last; break;
  case 'k':
  case 'K':
    if (!value_fits_shifted(*value, 10)) return first; // overflow guard
    *value = *value << 10; ++last; break;
  case 'm':
  case 'M':
    if (!value_fits_shifted(*value, 20)) return first; // overflow guard
    *value = *value << 20; ++last; break;
  case 'g':
  case 'G':
    if (!value_fits_shifted(*value, 30)) return first; // overflow guard
    *value = *value << 30; ++last; break;
//   case ' ':
//   case '\0': *value = *value * unit; break;
//   default: throw torrent::input_error("Could not parse value.");
  default:
    if (*value > std::numeric_limits<int64_t>::max() / unit ||
        *value < std::numeric_limits<int64_t>::min() / unit)
      return first; // overflow guard

    *value = *value * unit;
    break;
  }

  return last;
}

// Somewhat ugly...
const char*
parse_object(const char* first, const char* last, torrent::Object* dest, bool (*delim)(const char), uint32_t depth) {
  if (++depth >= max_parse_depth)
    throw torrent::input_error("Max parse depth reached.");

  if (first != last && *first == '{') {
    *dest = torrent::Object::create_list();
    first = parse_list(first + 1, last, dest, &parse_is_delim_block, depth);
    first = parse_skip_wspace(first, last);

    if (first == last || *first != '}')
      throw torrent::input_error("Could not find closing '}'.");

    return ++first;

  } else if (first != last && *first == '(') {
    int32_t parentheses = 1;

    while (first + 1 != last && *(first + 1) == '(') {
      first++;
      parentheses++;
    }

    if (parentheses > 3)
      throw torrent::input_error("Max 3 parentheses per object allowed.");

    *dest = torrent::Object::create_dict_key();
    dest->set_flags(torrent::Object::flag_function << (parentheses - 1));

    first = parse_string(first + 1, last, &dest->as_dict_key(), &parse_is_delim_func);
    first = parse_skip_wspace(first, last);

    if (first == last || !parse_is_delim_func(*first))
      throw torrent::input_error("Could not find closing ')'.");

    if (*first == ',') {
      // This will always create a list even for single argument functions...
      dest->as_dict_obj() = torrent::Object::create_list();
      first = parse_list(first + 1, last, &dest->as_dict_obj(), &parse_is_delim_func, depth);
      first = parse_skip_wspace(first, last);
    }

    while (parentheses != 0 && first != last && *first == ')') {
      first++;
      parentheses--;
    }

    if (parentheses != 0)
      throw torrent::input_error("Parentheses mismatch.");

    return first;

  } else {
    *dest = std::string();

    return parse_string(first, last, &dest->as_string(), delim);
  }
}

const char*
parse_list(const char* first, const char* last, torrent::Object* dest, bool (*delim)(const char), uint32_t depth) {
  if (!dest->is_list())
    throw torrent::internal_error("parse_list(...) !dest->is_list().");

  while (true) {
    torrent::Object tmp;

    first = parse_skip_wspace(first, last);
    first = parse_object(first, last, &tmp, delim, depth);
    first = parse_skip_wspace(first, last);

    dest->as_list().push_back(tmp);

    if (first == last || !parse_is_seperator(*first))
      break;

    first++;
  }

  return first;
}

const char*
parse_whole_list(const char* first, const char* last, torrent::Object* dest, bool (*delim)(const char), uint32_t depth) {
  first = parse_skip_wspace(first, last);
  first = parse_object(first, last, dest, delim, depth);
  first = parse_skip_wspace(first, last);

  if (first != last && parse_is_seperator(*first)) {
    torrent::Object tmp = torrent::Object::create_list();
    tmp.swap(*dest);

    dest->as_list().push_back(tmp);
    first = parse_list(++first, last, dest, delim, depth);
  }

  return first;
}

std::string
convert_to_string(const torrent::Object& rawSrc) {
  const torrent::Object& src = convert_to_single_argument(rawSrc);

  switch (src.type()) {
  case torrent::Object::TYPE_VALUE: {
    char buffer[64];
    snprintf(buffer, 64, "%lli", (long long int)src.as_value());
    return std::string(buffer);
  }
  case torrent::Object::TYPE_STRING: return src.as_string();
  case torrent::Object::TYPE_NONE:   return std::string();

  case torrent::Object::TYPE_RAW_BENCODE:
    if (src.as_raw_bencode().is_empty())
      return std::string();

    if (src.as_raw_bencode().is_raw_string())
      return src.as_raw_bencode().as_raw_string().as_string();

    if (src.as_raw_bencode().is_value())
      return src.as_raw_bencode().as_value_string();

  default: throw torrent::input_error("Not a string.");
  }
}

std::string
convert_list_to_string(const torrent::Object& src) {
  if (!src.is_list())
    throw torrent::internal_error("convert_list_to_string(...) !src->is_list().");

  return convert_list_to_string(src.as_list().begin(), src.as_list().end());
}

std::string
convert_list_to_string(torrent::Object::list_const_iterator first,
                       torrent::Object::list_const_iterator last) {
  std::string dest;

  while (first != last) {
    if (!first->is_string())
      throw torrent::input_error("Could not convert non-string list element to string.");

    // Meh.
    if (!dest.empty())
      dest += ",\"";
    else
      dest += '"';

    std::string::size_type quoteItr = dest.size();
    dest += first->as_string();

    // Finding a quote inside the string should be relatively rare, so
    // use something that is fast in the general case and ignore the
    // cost of the unusual one.
    while (quoteItr != dest.size()) {
      if (dest[quoteItr] == '"' || dest[quoteItr] == '\\')
        dest.insert(quoteItr++, 1, '\\');

      quoteItr++;
    }

    dest += '"';
    first++;
  }

  return dest;
}

std::string
convert_list_to_command(torrent::Object::list_const_iterator first,
                        torrent::Object::list_const_iterator last) {
  if (first == last)
    throw torrent::input_error("Too few arguments.");

  auto dest     = (first++)->as_string();
  auto quoteItr = dest.find('=');

  if (quoteItr == std::string::npos)
    throw torrent::input_error("Could not find '=' in command.");

  // We should only escape backslash, not quote here as the string
  // will start with the command name which isn't quoted.
  while ((quoteItr = dest.find('\\', quoteItr + 1)) != std::string::npos)
    dest.insert(quoteItr++, 1, '\\');

  while (first != last) {
    if (!first->is_string())
      throw torrent::input_error("Could not convert non-string list element to string.");

    dest += ",\"";

    std::string::size_type quoteItr = dest.size();
    dest += first->as_string();

    // Finding a quote inside the string should be relatively rare, so
    // use something that is fast in the general case and ignore the
    // cost of the unusual one.
    while (quoteItr != dest.size()) {
      if (dest[quoteItr] == '"' || dest[quoteItr] == '\\')
        dest.insert(quoteItr++, 1, '\\');

      quoteItr++;
    }

    dest += '"';
    first++;
  }

  return dest;
}

int64_t
convert_to_value(const torrent::Object& src, int base, int unit) {
  int64_t value;

  if (!convert_to_value_nothrow(src, &value, base, unit))
    throw torrent::input_error("Not convertible to a value.");

  return value;
}

bool
convert_to_value_nothrow(const torrent::Object& src, int64_t* value, int base, int unit) {
  const torrent::Object& unpacked = (src.is_list() && src.as_list().size() == 1) ? src.as_list().front() : src;

  switch (unpacked.type()) {
  case torrent::Object::TYPE_VALUE:
    *value = unpacked.as_value();
    break;

  case torrent::Object::TYPE_STRING:
    return parse_skip_wspace(parse_value(unpacked.as_string().c_str(), value, base, unit),
                             unpacked.as_string().c_str() + unpacked.as_string().size())
      == unpacked.as_string().c_str() + unpacked.as_string().size();

  case torrent::Object::TYPE_RAW_STRING: {
    const torrent::raw_string& str = unpacked.as_raw_string();

    auto buffer = std::make_unique<char[]>(str.size() + 1);
    std::memcpy(buffer.get(), str.data(), str.size());
    buffer[str.size()] = '\0';

    return parse_skip_wspace(parse_value(buffer.get(), value, base, unit), buffer.get() + str.size())
      == buffer.get() + str.size();
  }
  case torrent::Object::TYPE_NONE:
    *value = 0;
    break;

  default:
    return false;
  }

  return true;
}

char*
print_object(char* first, char* last, const torrent::Object* src, int flags) {
  switch (src->type()) {
  case torrent::Object::TYPE_STRING:
  {
    const std::string& str = src->as_string();

    if (first == last)
      return first;

    if ((flags & print_expand_tilde) && *str.c_str() == '~') {
      auto expanded = expand_path(str);

      size_t n = std::min<size_t>(expanded.size(), std::distance(first, last) - 1);
      std::memcpy(first, expanded.c_str(), n);
      *(first += n) = '\0';

    } else {
      size_t n = std::min<size_t>(str.size(), std::distance(first, last) - 1);
      std::memcpy(first, str.c_str(), n);
      *(first += n) = '\0';
    }

    return first;
  }

  case torrent::Object::TYPE_VALUE:
    return std::min(first + snprintf(first, std::distance(first, last), "%lli", (long long int)src->as_value()), last);

  case torrent::Object::TYPE_LIST:
    if (first != last)
      *first = '\0';

    for (const auto& itr : src->as_list()) {
      first = print_object(first, last, &itr, flags);

      // Don't expand tilde after the first element in the list.
      flags &= ~print_expand_tilde;
    }

    return first;

  case torrent::Object::TYPE_NONE:
    if (first != last)
      *first = '\0';

    return first;
  default:
    throw torrent::input_error("Invalid type.");
  }
}

void
print_object_std(std::string* dest, const torrent::Object* src, int flags) {
  switch (src->type()) {
  case torrent::Object::TYPE_STRING:
  {
    const std::string& str = src->as_string();

    if ((flags & print_expand_tilde) && *str.c_str() == '~')
      *dest += expand_path(str);
    else
      *dest += str;

    return;
  }
  case torrent::Object::TYPE_VALUE:
  {
    char buffer[64];
    snprintf(buffer, 64, "%lli", (long long int)src->as_value());

    *dest += buffer;
    return;
  }
  case torrent::Object::TYPE_LIST:
    for (const auto& itr : src->as_list()) {
      print_object_std(dest, &itr, flags);

      // Don't expand tilde after the first element in the list.
      flags &= ~print_expand_tilde;
    }

    return;

  case torrent::Object::TYPE_NONE:
    return;
  default:
    throw torrent::input_error("Invalid type.");
  }
}

}