p-ranav / oystr

oystr recursively searches directories for a substring.

oystr

oystr is a command-line tool that recursively searches directories for a substring.

Quick Start

Build oystr using CMake. For more details, see BUILDING.md.

git clone https://github.com/p-ranav/oystr
cd oystr

# Build
cmake -S . -B build -D CMAKE_BUILD_TYPE=Release
cmake --build build

# Install
sudo cmake --install build

oystr's Issues

Use AVX2 needle search for case sensitive scenarios

#include <immintrin.h>

const char* find_avx2_more(const char* b, const char* e, char c)
{
  const char* i = b;

  __m256i q = _mm256_set1_epi8(c);

  for (; i + 32 < e; i += 32) {
    __m256i x = _mm256_lddqu_si256(reinterpret_cast<const __m256i*>(i));
    __m256i r = _mm256_cmpeq_epi8(x, q);
    int z = _mm256_movemask_epi8(r);
    if (z)
      return i + __builtin_ffs(z) - 1;
  }
  if (i < e) {
    if (e - b < 32) {
      for (; i < e; ++i)
        if (*i == c)
          return i;
    } else {
      i = e - 32;
      __m256i x = _mm256_lddqu_si256(reinterpret_cast<const __m256i*>(i));
      __m256i r = _mm256_cmpeq_epi8(x, q);
      int z = _mm256_movemask_epi8(r);
      if (z)
        return i + __builtin_ffs(z) - 1;
    }
  }
  return e;
}

auto needle_search_avx2(std::string_view needle,
                        std::string_view::const_iterator haystack_begin,
                        std::string_view::const_iterator haystack_end,
                        bool ignore_case)
{
  if (needle.empty()) {
    return haystack_end;
  }
  const char c = needle[0];

  // quickly find c in haystack
  auto it = haystack_begin;
  while (it != haystack_end) {
    const char* ptr = find_avx2_more(it, haystack_end, c);
    bool result = true;

    const char* i = ptr;
    for (auto& n : needle) {
      if (n != *i) {
        result = false;
        break;
      }
      i++;
    }

    if (result) {
      return ptr;
    } else {
      it = i;
    }
  }

  return haystack_end;
}

Look into Boost range adaptors for directory iteration

https://stackoverflow.com/questions/1257721/can-i-use-a-mask-to-iterate-files-in-a-directory-with-boost

Search fails in small file

bob@LAPTOP-JDRTARVO:~/dev/search$ time grep -irn 'qmi' ../quectel-CM/NOTICE
3:The APIs of QMI WWAMN interfaces are defined by Qualcomm. And this program complies with Qualcomm QMI WWAN interfaces specification.

real    0m0.002s
user    0m0.002s
sys     0m0.000s
bob@LAPTOP-JDRTARVO:~/dev/search$ time ./build/search -irn 'qmi' ../quectel-CM/NOTICE

real    0m0.002s
user    0m0.002s
sys     0m0.001s