Manually define `std::unique` with concrete types to fix issue with Clang

pyfastani/_fastani.pyx

@@ -6,15 +6,15 @@
 # --- C imports --------------------------------------------------------------
 
 cimport libcpp11.chrono
-from cython.operator cimport dereference, postincrement
+from cython.operator cimport dereference, preincrement, postincrement
 from libc.string cimport memcpy
 from libc.limits cimport INT_MAX
 from libc.stdint cimport int64_t, uint64_t
 from libc.stdlib cimport malloc, realloc, free
 from libcpp cimport bool, nullptr
-from libcpp.algorithm cimport sort, unique
+from libcpp.algorithm cimport sort
 from libcpp.deque cimport deque
-from libcpp.utility cimport pair
+from libcpp.utility cimport move, pair
 from libcpp.functional cimport function
 from libcpp.string cimport string
 from libcpp.unordered_map cimport unordered_map
@@ -46,7 +46,7 @@ from fastani.map.base_types cimport (
 # HACK: we need kseq_t* as a template argument, which is not supported by
 #       Cython at the moment, so we just `typedef kseq_t* kseq_ptr_t` in
 #       an external C++ header to make Cython happy
-from _utils cimport kseq_ptr_t, toupper, complement, distance
+from _utils cimport kseq_ptr_t, toupper, complement, distance, unique_minimizers
 from _unicode cimport *
 
 
@@ -580,7 +580,12 @@ cdef class Mapper(_Parameterized):
 
         # find the unique minimizers in thos that were just obtained
         sort(query.minimizerTableQuery.begin(), query.minimizerTableQuery.end(), MinimizerInfo_t.lessByHash)
-        uniq_end_iter = unique(query.minimizerTableQuery.begin(), query.minimizerTableQuery.end(), MinimizerInfo_t.equalityByHash)
+
+        # manually implement `unique` as template instantiation has issues on OSX
+        it = query.minimizerTableQuery.begin()
+        uniq_end_iter = unique_minimizers(query.minimizerTableQuery.begin(), query.minimizerTableQuery.end())
+
+        # early return if no minimizers were found
         query.sketchSize = distance(query.minimizerTableQuery.begin(), uniq_end_iter)
         if query.sketchSize == 0:
             return

pyfastani/_utils.cpp

@@ -24,3 +24,9 @@ ZEXTERN int ZEXPORT gzread(gzFile file, void* buf, unsigned int len) {
 ZEXTERN int ZEXPORT gzclose(gzFile file) {
     return 0;
 }
+
+// Implementation from: https://en.cppreference.com/w/cpp/algorithm/
+std::vector<skch::MinimizerInfo>::iterator unique_minimizers(std::vector<skch::MinimizerInfo>::iterator first, std::vector<skch::MinimizerInfo>::iterator last)
+{
+    return std::unique(first, last, skch::MinimizerInfo::equalityByHash);
+}

pyfastani/_utils.hpp

@@ -2,6 +2,7 @@
 #define __UTILS_HPP
 
 #include <stdint.h>
+#include <algorithm>
 #include <chrono>
 #include <limits>
 #include <iostream>
@@ -12,38 +13,39 @@
 #include "map/include/base_types.hpp"
 #include "map/include/winSketch.hpp"
 
-#ifdef __cplusplus
 extern "C" {
-#endif
-
-// compatibility layer for Cython
-typedef kseq_t* kseq_ptr_t;
-
-// efficient nucleotide complement with a lookup table
-static const char COMPLEMENT_LOOKUP[128] = {
-    '\x00', '\x01', '\x02', '\x03', '\x04', '\x05', '\x06', '\x07',
-    '\x08', '\t',   '\n',   '\x0', '\x0c', '\r',   '\x0e', '\x0f',
-    '\x10', '\x11', '\x12', '\x13', '\x14', '\x15', '\x16', '\x17',
-    '\x18', '\x19', '\x1a', '\x1', '\x1c', '\x1d', '\x1e', '\x1f',
-    ' ',    '!',    '"',    '#',    '$',    '%',    '&',    '\'',
-    '(',    ')',    '*',    '+',    ',',    '-',    '.',    '/',
-    '0',    '1',    '2',    '3',    '4',    '5',    '6',    '7',
-    '8',    '9',    ':',    ';',    '<',    '=',    '>',    '?',
-    '@',    'T',    'V',    'G',    'H',    'E',    'F',    'C',
-    'D',    'I',    'J',    'M',    'L',    'K',    'N',    'O',
-    'P',    'Q',    'Y',    'S',    'A',    'U',    'B',    'W',
-    'X',    'R',    'Z',    '[',    '\\',   ']',    '^',    '_',
-    '`',    't',    'v',    'g',    'h',    'e',    'f',    'c',
-    'd',    'i',    'j',    'm',    'l',    'k',    'n',    'o',
-    'p',    'q',    'y',    's',    'a',    'u',    'b',    'w',
-    'x',    'r',    'z',    '{',    '|',    '}',    '~',    '\x7f'
-};
-
-inline char complement(char base) {
-    return COMPLEMENT_LOOKUP[(size_t) (base & 0x7F)];
-}
 
-#ifdef __cplusplus
+    // compatibility layer for Cython
+    typedef kseq_t* kseq_ptr_t;
+
+    // efficient nucleotide complement with a lookup table
+    static const char COMPLEMENT_LOOKUP[128] = {
+        '\x00', '\x01', '\x02', '\x03', '\x04', '\x05', '\x06', '\x07',
+        '\x08', '\t',   '\n',   '\x0', '\x0c', '\r',   '\x0e', '\x0f',
+        '\x10', '\x11', '\x12', '\x13', '\x14', '\x15', '\x16', '\x17',
+        '\x18', '\x19', '\x1a', '\x1', '\x1c', '\x1d', '\x1e', '\x1f',
+        ' ',    '!',    '"',    '#',    '$',    '%',    '&',    '\'',
+        '(',    ')',    '*',    '+',    ',',    '-',    '.',    '/',
+        '0',    '1',    '2',    '3',    '4',    '5',    '6',    '7',
+        '8',    '9',    ':',    ';',    '<',    '=',    '>',    '?',
+        '@',    'T',    'V',    'G',    'H',    'E',    'F',    'C',
+        'D',    'I',    'J',    'M',    'L',    'K',    'N',    'O',
+        'P',    'Q',    'Y',    'S',    'A',    'U',    'B',    'W',
+        'X',    'R',    'Z',    '[',    '\\',   ']',    '^',    '_',
+        '`',    't',    'v',    'g',    'h',    'e',    'f',    'c',
+        'd',    'i',    'j',    'm',    'l',    'k',    'n',    'o',
+        'p',    'q',    'y',    's',    'a',    'u',    'b',    'w',
+        'x',    'r',    'z',    '{',    '|',    '}',    '~',    '\x7f'
+    };
+
+    inline char complement(char base) {
+        return COMPLEMENT_LOOKUP[(size_t) (base & 0x7F)];
+    }
+
 }
-#endif // ifdef __cplusplus
+
+// OSX seems to have trouble instantiating `std::unique` with our
+// custom type, so here is a dedicated implementation
+std::vector<skch::MinimizerInfo>::iterator unique_minimizers(std::vector<skch::MinimizerInfo>::iterator first, std::vector<skch::MinimizerInfo>::iterator last);
+
 #endif // ifdef __UTILS_HPP

pyfastani/_utils.pxd

 from libc.stdint cimport uint64_t
+from libcpp.vector cimport vector
 
 from kseq cimport kseq_t
-from fastani.map.base_types cimport MappingResultsVector_t
 from fastani.map.compute_map cimport Map
 from fastani.map.map_parameters cimport Parameters
 from fastani.map.win_sketch cimport Sketch
+from fastani.map.base_types cimport (
+    MappingResultsVector_t,
+    MinimizerInfo as MinimizerInfo_t
+)
 
 
 cdef extern from *:
@@ -36,4 +40,5 @@ cdef extern from "_utils.hpp" nogil:
 
     ctypedef kseq_t* kseq_ptr_t
 
+    cdef vector[MinimizerInfo_t].iterator unique_minimizers(vector[MinimizerInfo_t].iterator, vector[MinimizerInfo_t].iterator)
     int complement(int)