a6.0: --check-sex/--impute-sex

2.0/include/plink2_string.cc

@@ -2871,7 +2871,7 @@ char* lntoa_g(double ln_val, char* start) {
 // makes sense to just promote to double like printf does.
 
 
-CXXCONST_CP ScanForDuplicateIds(const char* sorted_ids, uintptr_t id_ct, uintptr_t max_id_blen) {
+CXXCONST_CP FindSortedStrboxDuplicate(const char* sorted_ids, uintptr_t id_ct, uintptr_t max_id_blen) {
   --id_ct;
   for (uintptr_t id_idx = 0; id_idx != id_ct; ++id_idx) {
     if (strequal_overread(&(sorted_ids[id_idx * max_id_blen]), &(sorted_ids[(id_idx + 1) * max_id_blen]))) {
@@ -2881,49 +2881,6 @@ CXXCONST_CP ScanForDuplicateIds(const char* sorted_ids, uintptr_t id_ct, uintptr
   return nullptr;
 }
 
-uint32_t CollapseDuplicateIds(uintptr_t id_ct, uintptr_t max_id_blen, char* sorted_ids, uint32_t* id_starts) {
-  // Collapses array of sorted IDs to remove duplicates, and writes
-  // pre-collapse positions to id_starts (so e.g. duplication count of any
-  // sample ID can be determined via subtraction) if it isn't nullptr.
-  // Returns id_ct of collapsed array.
-  if (!id_ct) {
-    return 0;
-  }
-  uintptr_t read_idx = 1;
-  uintptr_t write_idx;
-  if (id_starts) {
-    id_starts[0] = 0;
-    for (; read_idx != id_ct; ++read_idx) {
-      if (strequal_overread(&(sorted_ids[(read_idx - 1) * max_id_blen]), &(sorted_ids[read_idx * max_id_blen]))) {
-        break;
-      }
-      id_starts[read_idx] = read_idx;
-    }
-    write_idx = read_idx;
-    while (++read_idx < id_ct) {
-      // this loop can probably be improved with string-length tracking...
-      if (!strequal_overread(&(sorted_ids[(write_idx - 1) * max_id_blen]), &(sorted_ids[read_idx * max_id_blen]))) {
-        strcpy(&(sorted_ids[write_idx * max_id_blen]), &(sorted_ids[read_idx * max_id_blen]));
-        id_starts[write_idx++] = read_idx;
-      }
-    }
-  } else {
-    for (; read_idx != id_ct; ++read_idx) {
-      if (strequal_overread(&(sorted_ids[(read_idx - 1) * max_id_blen]), &(sorted_ids[read_idx * max_id_blen]))) {
-        break;
-      }
-    }
-    write_idx = read_idx;
-    while (++read_idx < id_ct) {
-      if (!strequal_overread(&(sorted_ids[(write_idx - 1) * max_id_blen]), &(sorted_ids[read_idx * max_id_blen]))) {
-        strcpy(&(sorted_ids[write_idx * max_id_blen]), &(sorted_ids[read_idx * max_id_blen]));
-        ++write_idx;
-      }
-    }
-  }
-  return write_idx;
-}
-
 
 int32_t bsearch_strbox(const char* idbuf, const char* sorted_strbox, uintptr_t cur_id_slen, uintptr_t max_id_blen, uintptr_t end_idx) {
   // does not assume null-terminated idbuf, or nonempty array.

2.0/include/plink2_string.h

@@ -635,6 +635,10 @@ void SortStrbox64bFinish(uintptr_t str_ct, uintptr_t max_str_blen, uint32_t use_
 // Must be ok to overread.
 void SortStrboxIndexed2(uintptr_t str_ct, uintptr_t max_str_blen, uint32_t use_nsort, char* strbox, uint32_t* id_map, void* sort_wkspace);
 #else  // !__cplusplus
+HEADER_INLINE uint32_t strcmp_overread_lt(const char* s1, const char* s2) {
+  return strcmp_overread(s1, s2) < 0;
+}
+
 HEADER_INLINE void StrptrArrSort(uintptr_t ct, const char** strptr_arr) {
   qsort(strptr_arr, ct, sizeof(intptr_t), strcmp_deref);
 }
@@ -1513,11 +1517,11 @@ HEADER_INLINE char* i32toa_x(int32_t ii, char extra_char, char* start) {
 
 
 // overread must be ok.
-CXXCONST_CP ScanForDuplicateIds(const char* sorted_ids, uintptr_t id_ct, uintptr_t max_id_blen);
+CXXCONST_CP FindSortedStrboxDuplicate(const char* sorted_ids, uintptr_t id_ct, uintptr_t max_id_blen);
 
 #ifdef __cplusplus
-HEADER_INLINE char* ScanForDuplicateIds(char* sorted_ids, uintptr_t id_ct, uintptr_t max_id_blen) {
-  return const_cast<char*>(ScanForDuplicateIds(const_cast<const char*>(sorted_ids), id_ct, max_id_blen));
+HEADER_INLINE char* FindSortedStrboxDuplicate(char* sorted_ids, uintptr_t id_ct, uintptr_t max_id_blen) {
+  return const_cast<char*>(FindSortedStrboxDuplicate(const_cast<const char*>(sorted_ids), id_ct, max_id_blen));
 }
 #endif
 
@@ -1526,7 +1530,7 @@ HEADER_INLINE char* ScanForDuplicateIds(char* sorted_ids, uintptr_t id_ct, uintp
 // determined via subtraction) if it isn't nullptr.
 // Overread must be ok.
 // Returns id_ct of collapsed array.
-uint32_t CollapseDuplicateIds(uintptr_t id_ct, uintptr_t max_id_blen, char* sorted_ids, uint32_t* id_starts);
+// uint32_t CollapseSortedStrbox(uintptr_t id_ct, uintptr_t max_id_blen, char* sorted_ids, uint32_t* id_starts);
 
 
 // returns position of string, or -1 if not found.

2.0/include/plink2_thread.cc

@@ -395,6 +395,15 @@ void UpdateU64IfSmaller(uint64_t newval, uint64_t* oldval_ptr) {
   }
 }
 
+void UpdateU32IfSmaller(uint32_t newval, uint32_t* oldval_ptr) {
+  uint32_t oldval = *oldval_ptr;
+  while (oldval > newval) {
+    if (ATOMIC_COMPARE_EXCHANGE_N_U32(oldval_ptr, &oldval, newval, 1, __ATOMIC_RELAXED, __ATOMIC_RELAXED)) {
+      break;
+    }
+  }
+}
+
 #ifdef __cplusplus
 }  // namespace plink2
 #endif

2.0/include/plink2_thread.h

@@ -318,6 +318,8 @@ HEADER_INLINE void JoinThreads0(ThreadGroup* tg_ptr) {
 // deterministic behavior is desired.
 void UpdateU64IfSmaller(uint64_t newval, uint64_t* oldval_ptr);
 
+void UpdateU32IfSmaller(uint32_t newval, uint32_t* oldval_ptr);
+
 #ifdef __cplusplus
 }  // namespace plink2
 #endif