log_parser.cpp

#include "log_parser.h"

namespace {

// Advance p past any spaces.
inline void skip_spaces(const char*& p, const char* end) {
  while (p < end && *p == ' ') ++p;
}

// Return the next space-delimited token as [tok_begin, p) and advance p to the
// delimiter. Returns an empty view if none remain.
inline std::string_view next_token(const char*& p, const char* end) {
  skip_spaces(p, end);
  const char* tok = p;
  while (p < end && *p != ' ') ++p;
  return std::string_view(tok, static_cast<std::size_t>(p - tok));
}

inline bool starts_with(std::string_view s, std::string_view prefix) {
  return s.size() >= prefix.size() && s.substr(0, prefix.size()) == prefix;
}

// Parse an unsigned integer from a string_view. Returns false on empty input or
// any non-digit. Caps at uint64 without overflow checks (log values fit easily).
template <typename T>
inline bool parse_uint(std::string_view s, T& out) {
  if (s.empty()) return false;
  T v = 0;
  for (char c : s) {
    if (c < '0' || c > '9') return false;
    v = v * 10 + static_cast<T>(c - '0');
  }
  out = v;
  return true;
}

}  // namespace

bool parse_line(const char* begin, const char* end, LogRecord& out) {
  const char* p = begin;

  // Timestamp: "YYYY-MM-DD" then "HH:MM:SS". They are contiguous in the buffer
  // separated by one space, so the 16-byte minute label "YYYY-MM-DD HH:MM" is a
  // single substring starting at the date token.
  std::string_view date = next_token(p, end);
  if (date.size() != 10) return false;
  std::string_view time = next_token(p, end);
  if (time.size() < 5) return false;
  // Require the single-space layout so the 16-byte slice is exactly the minute.
  if (time.data() != date.data() + 11) return false;
  out.minute = std::string_view(date.data(), 16);

  // Level.
  out.level = next_token(p, end);
  if (out.level.empty()) return false;
  out.is_error = (out.level == "ERROR");

  // user=<int>
  std::string_view user_tok = next_token(p, end);
  if (!starts_with(user_tok, "user=")) return false;
  if (!parse_uint(user_tok.substr(5), out.user_id)) return false;

  // endpoint=</path>
  std::string_view ep_tok = next_token(p, end);
  if (!starts_with(ep_tok, "endpoint=")) return false;
  out.endpoint = ep_tok.substr(9);
  if (out.endpoint.empty()) return false;

  // latency=<int>
  std::string_view lat_tok = next_token(p, end);
  if (!starts_with(lat_tok, "latency=")) return false;
  if (!parse_uint(lat_tok.substr(8), out.latency)) return false;

  return true;
}