ddarriba · astanabe · Nov 27, 2018 · Aug 8, 2020
diff --git a/.gitignore b/.gitignore
diff --git a/.gitmodules b/.gitmodules
diff --git a/CMakeLists.txt b/CMakeLists.txt
diff --git a/ChangeLog.md b/ChangeLog.md
diff --git a/LICENSE.txt b/LICENSE.txt
diff --git a/Makefile.am b/Makefile.am
diff --git a/README.md b/README.md
diff --git a/configure.ac b/configure.ac
diff --git a/doxy.cfg b/doxy.cfg
diff --git a/examples/consensus/Makefile b/examples/consensus/Makefile
diff --git a/examples/consensus/consensus.c b/examples/consensus/consensus.c
diff --git a/examples/rf-distance/Makefile b/examples/rf-distance/Makefile
diff --git a/examples/rf-distance/rf-distance.c b/examples/rf-distance/rf-distance.c
diff --git a/examples/spr-round/Makefile b/examples/spr-round/Makefile
diff --git a/examples/spr-round/spr-round.c b/examples/spr-round/spr-round.c
diff --git a/examples/weight-consensus/Makefile b/examples/weight-consensus/Makefile
diff --git a/examples/weight-consensus/weight-consensus.c b/examples/weight-consensus/weight-consensus.c
diff --git a/m4/ax_check_compile_flag.m4 b/m4/ax_check_compile_flag.m4
diff --git a/m4/ax_ext.m4 b/m4/ax_ext.m4
diff --git a/m4/ax_gcc_x86_avx_xgetbv.m4 b/m4/ax_gcc_x86_avx_xgetbv.m4
diff --git a/m4/ax_gcc_x86_cpuid.m4 b/m4/ax_gcc_x86_cpuid.m4
diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
diff --git a/src/Makefile.am b/src/Makefile.am
diff --git a/src/algorithm/CMakeLists.txt b/src/algorithm/CMakeLists.txt
diff --git a/src/algorithm/Makefile.am b/src/algorithm/Makefile.am
diff --git a/src/algorithm/README.md b/src/algorithm/README.md
diff --git a/src/algorithm/algo_callback.c b/src/algorithm/algo_callback.c
diff --git a/src/algorithm/algo_callback.h b/src/algorithm/algo_callback.h
diff --git a/src/algorithm/algo_search.c b/src/algorithm/algo_search.c
diff --git a/src/algorithm/pllmod_algorithm.c b/src/algorithm/pllmod_algorithm.c
diff --git a/src/algorithm/pllmod_algorithm.h b/src/algorithm/pllmod_algorithm.h
diff --git a/src/binary/CMakeLists.txt b/src/binary/CMakeLists.txt
diff --git a/src/binary/Makefile.am b/src/binary/Makefile.am
diff --git a/src/binary/README.md b/src/binary/README.md
diff --git a/src/binary/binary_io_operations.c b/src/binary/binary_io_operations.c
diff --git a/src/binary/binary_io_operations.h b/src/binary/binary_io_operations.h
diff --git a/src/binary/pll_binary.c b/src/binary/pll_binary.c
diff --git a/src/binary/pll_binary.h b/src/binary/pll_binary.h
diff --git a/src/msa/CMakeLists.txt b/src/msa/CMakeLists.txt
diff --git a/src/msa/Makefile.am b/src/msa/Makefile.am
diff --git a/src/msa/README.md b/src/msa/README.md
diff --git a/src/msa/pll_msa.c b/src/msa/pll_msa.c
@@ -28,7 +28,9 @@
   * @author Alexey Kozlov
   */
 
+#ifndef __MINGW32__
 #include <search.h>
+#endif
 
 #include "pll_msa.h"
 #include "../util/pllmod_util.h"
@@ -305,6 +307,97 @@ PLL_EXPORT double pllmod_msa_empirical_invariant_sites(pll_partition_t *partitio
   return empirical_pinv;
 }
 
+#if defined __MINGW32__
+static int find_duplicate_strings_htable(char ** const strings,
+                                  unsigned long string_count,
+                                  unsigned long ** duplicates,
+                                  unsigned long * duplicate_count)
+{
+  if (!strings)
+    return PLL_FAILURE;
+
+  unsigned long * tmpdup = (unsigned long *)malloc(string_count * 2 *
+                                                sizeof(unsigned long));
+
+  unsigned long * hash = (unsigned long *)calloc(string_count,
+                                                 sizeof(unsigned long));
+
+  unsigned char * dupflag = (unsigned char *)calloc(string_count,
+                                                 sizeof(unsigned char));
+
+  if (!hash || !tmpdup || !dupflag)
+  {
+    if (hash)
+      free(hash);
+    if (tmpdup)
+      free(tmpdup);
+    if (dupflag)
+      free(dupflag);
+    pllmod_set_error(PLL_ERROR_MEM_ALLOC,
+                     "Cannot allocate memory for duplicates array");
+    return PLL_FAILURE;
+  }
+
+  *duplicate_count = 0;
+
+  unsigned long i;
+  unsigned long j;
+
+  for (i = 0; i < string_count; ++i)
+  {
+    for (j = 0; j < strlen(strings[i]); ++j)
+      hash[i] = hash[i] + (j+1) * (unsigned long) strings[i][j];
+  }
+
+  int coll = 0;
+
+  for (i = 0; i < string_count; ++i)
+  {
+    if (dupflag[i])
+      continue;
+    for (j = i+1; j < string_count; ++j)
+    {
+      if (hash[i] == hash[j])
+      {
+        coll++;
+        if (strcmp(strings[i], strings[j]) == 0)
+        {
+          /* duplicate found, save it */
+          tmpdup[(*duplicate_count)*2] = i;
+          tmpdup[(*duplicate_count)*2+1] = j;
+          (*duplicate_count)++;
+          dupflag[j] = 1;
+        }
+      }
+    }
+  }
+
+//  printf("Collisions: %d, duplicates: %lu\n", coll, *duplicate_count);
+
+  free(hash);
+  free(dupflag);
+
+  if (*duplicate_count > 0)
+  {
+    *duplicates = (unsigned long *) realloc(tmpdup, (*duplicate_count) * 2 *
+                                                  sizeof(unsigned long));
+
+    if (!(*duplicates))
+    {
+      pllmod_set_error(PLL_ERROR_MEM_ALLOC,
+                       "Cannot allocate memory for duplicates array");
+      return PLL_FAILURE;
+    }
+  }
+  else
+  {
+    free(tmpdup);
+    *duplicates = NULL;
+  }
+
+  return PLL_SUCCESS;
+}
+#else
 /* Find duplicates using hash table from search.h. This works best for short
  * strings, so we use this method for checking taxa names */
 static int find_duplicate_strings_htable(char ** const strings,
@@ -376,6 +469,7 @@ static int find_duplicate_strings_htable(char ** const strings,
 
   return PLL_SUCCESS;
 }
+#endif
 
 /* Find duplicates using custom hash function optimized for long low-variance
  * strings - this method is used for detecting identical sequences */

diff --git a/src/msa/pll_msa.h b/src/msa/pll_msa.h
diff --git a/src/optimize/CMakeLists.txt b/src/optimize/CMakeLists.txt
diff --git a/src/optimize/Makefile.am b/src/optimize/Makefile.am
diff --git a/src/optimize/README.md b/src/optimize/README.md
diff --git a/src/optimize/lbfgsb/License.txt b/src/optimize/lbfgsb/License.txt
diff --git a/src/optimize/lbfgsb/lbfgsb.c b/src/optimize/lbfgsb/lbfgsb.c
diff --git a/src/optimize/lbfgsb/lbfgsb.h b/src/optimize/lbfgsb/lbfgsb.h
diff --git a/src/optimize/lbfgsb/linesearch.c b/src/optimize/lbfgsb/linesearch.c
diff --git a/src/optimize/lbfgsb/linpack.c b/src/optimize/lbfgsb/linpack.c
diff --git a/src/optimize/lbfgsb/miniCBLAS.c b/src/optimize/lbfgsb/miniCBLAS.c
diff --git a/src/optimize/lbfgsb/subalgorithms.c b/src/optimize/lbfgsb/subalgorithms.c
diff --git a/src/optimize/opt_algorithms.c b/src/optimize/opt_algorithms.c
diff --git a/src/optimize/pll_optimize.c b/src/optimize/pll_optimize.c
diff --git a/src/optimize/pll_optimize.h b/src/optimize/pll_optimize.h
diff --git a/src/pllmod_common.c b/src/pllmod_common.c
diff --git a/src/pllmod_common.h b/src/pllmod_common.h
diff --git a/src/tree/CMakeLists.txt b/src/tree/CMakeLists.txt
diff --git a/src/tree/Makefile.am b/src/tree/Makefile.am
diff --git a/src/tree/README.md b/src/tree/README.md
diff --git a/src/tree/consensus.c b/src/tree/consensus.c
diff --git a/src/tree/lex_split.l b/src/tree/lex_split.l
diff --git a/src/tree/pll_tree.c b/src/tree/pll_tree.c
diff --git a/src/tree/pll_tree.h b/src/tree/pll_tree.h
diff --git a/src/tree/rtree_operations.c b/src/tree/rtree_operations.c
diff --git a/src/tree/split_utree.y b/src/tree/split_utree.y
diff --git a/src/tree/tree_hashtable.c b/src/tree/tree_hashtable.c
diff --git a/src/tree/tree_hashtable.h b/src/tree/tree_hashtable.h
diff --git a/src/tree/treeinfo.c b/src/tree/treeinfo.c
diff --git a/src/tree/utree_distances.c b/src/tree/utree_distances.c
diff --git a/src/tree/utree_operations.c b/src/tree/utree_operations.c
diff --git a/src/util/CMakeLists.txt b/src/util/CMakeLists.txt
diff --git a/src/util/Makefile.am b/src/util/Makefile.am
diff --git a/src/util/README.md b/src/util/README.md
diff --git a/src/util/models.c b/src/util/models.c
diff --git a/src/util/models_aa.c b/src/util/models_aa.c
diff --git a/src/util/models_dna.c b/src/util/models_dna.c
diff --git a/src/util/models_gt.c b/src/util/models_gt.c
diff --git a/src/util/models_mult.c b/src/util/models_mult.c
diff --git a/src/util/pllmod_util.h b/src/util/pllmod_util.h
diff --git a/test/.gitignore b/test/.gitignore
diff --git a/test/Makefile b/test/Makefile
diff --git a/test/Makefile.w64 b/test/Makefile.w64
diff --git a/test/README.md b/test/README.md
diff --git a/test/out/README.md b/test/out/README.md
diff --git a/test/out/binary/binary-random.out b/test/out/binary/binary-random.out
diff --git a/test/out/binary/binary-sequential.out b/test/out/binary/binary-sequential.out
diff --git a/test/out/binary/binary-skeleton.out b/test/out/binary/binary-skeleton.out
diff --git a/test/out/optimize/blopt-5states.out b/test/out/optimize/blopt-5states.out
diff --git a/test/out/optimize/blopt-minimal.out b/test/out/optimize/blopt-minimal.out
diff --git a/test/out/skip.out b/test/out/skip.out
diff --git a/test/out/tree/parsimony-tree.out b/test/out/tree/parsimony-tree.out
diff --git a/test/out/tree/random-tree.out b/test/out/tree/random-tree.out
diff --git a/test/out/tree/rtreemove-spr.out b/test/out/tree/rtreemove-spr.out
diff --git a/test/out/tree/serialize.out b/test/out/tree/serialize.out
diff --git a/test/out/tree/split-reconstruct.out b/test/out/tree/split-reconstruct.out
diff --git a/test/out/tree/split-tbe.out b/test/out/tree/split-tbe.out
diff --git a/test/out/tree/treemove-nni.out b/test/out/tree/treemove-nni.out
diff --git a/test/out/tree/treemove-spr.out b/test/out/tree/treemove-spr.out
diff --git a/test/out/tree/treemove-tbr.out b/test/out/tree/treemove-tbr.out
diff --git a/test/src/README.md b/test/src/README.md
diff --git a/test/src/binary/binary-random.c b/test/src/binary/binary-random.c
diff --git a/test/src/binary/binary-sequential.c b/test/src/binary/binary-sequential.c
diff --git a/test/src/binary/binary-skeleton.c b/test/src/binary/binary-skeleton.c
diff --git a/test/src/common.c b/test/src/common.c
diff --git a/test/src/common.h b/test/src/common.h
diff --git a/test/src/optimize/blopt-5states.c b/test/src/optimize/blopt-5states.c
diff --git a/test/src/optimize/blopt-minimal.c b/test/src/optimize/blopt-minimal.c
diff --git a/test/src/rng.h b/test/src/rng.h
diff --git a/test/src/tree/parsimony-tree.c b/test/src/tree/parsimony-tree.c
diff --git a/test/src/tree/random-tree.c b/test/src/tree/random-tree.c
diff --git a/test/src/tree/serialize.c b/test/src/tree/serialize.c
diff --git a/test/src/tree/split-reconstruct.c b/test/src/tree/split-reconstruct.c
diff --git a/test/src/tree/split-tbe.c b/test/src/tree/split-tbe.c
diff --git a/test/src/tree/treemove-nni.c b/test/src/tree/treemove-nni.c
diff --git a/test/src/tree/treemove-spr.c b/test/src/tree/treemove-spr.c
diff --git a/test/src/tree/treemove-tbr.c b/test/src/tree/treemove-tbr.c