Merge branch 'dev'

src/algorithm/algo_search.c

@@ -31,6 +31,9 @@ Schloss-Wolfsbrunnenweg 35, D-69118 Heidelberg, Germany
 #include "pllmod_algorithm.h"
 #include "../pllmod_common.h"
 
+/* constraint tree debugging */
+//#define CONS_DEBUG
+
 /* if not defined, branch length optimization will use
 * the same starting set of branch lengths for every topology */
 #define PLLMOD_SEARCH_GREEDY_BLO
@@ -733,6 +736,7 @@ static int best_reinsert_edge(pllmod_treeinfo_t * treeinfo,
     /* do not re-insert if resulting tree would contradict the constraint */
     if (check_cons && !pllmod_treeinfo_constraint_check_spr(treeinfo, p_edge, r_edge))
     {
+      DBG("SKIP incompatible: %u %u\n", j, r_edge->clv_index);
       ++j;
       continue;
     }
@@ -749,6 +753,18 @@ static int best_reinsert_edge(pllmod_treeinfo_t * treeinfo,
     retval = algo_utree_regraft(treeinfo, params, p_edge, r_edge);
     assert(retval == PLL_SUCCESS);
 
+#ifdef CONS_DEBUG
+      if (!pllmod_treeinfo_constraint_check_current(treeinfo))
+      {
+        pll_utree_show_ascii(treeinfo->root, PLL_UTREE_SHOW_LABEL | PLL_UTREE_SHOW_BRANCH_LENGTH |
+                                             PLL_UTREE_SHOW_CLV_INDEX );
+        printf("Constraint check failed after REGRAFT: %u %u\n", p_edge->clv_index, r_edge->clv_index);
+        pllmod_set_error(PLLMOD_TREE_ERROR_INVALID_TREE,
+                         "Constraint check failed after applying SPR!");
+        return PLL_FAILURE;
+      }
+#endif
+
     /* place root at the pruning branch and invalidate CLV at the new root */
     pllmod_treeinfo_set_root(treeinfo, p_edge);
     pllmod_treeinfo_invalidate_clv(treeinfo, p_edge);
@@ -947,12 +963,14 @@ static double reinsert_nodes(pllmod_treeinfo_t * treeinfo, pll_unode_t ** nodes,
       if (!retval)
         return PLL_FAILURE;
 
-#ifdef DEBUG
+#ifdef CONS_DEBUG
       if (!pllmod_treeinfo_constraint_check_current(treeinfo))
       {
-//        pll_utree_show_ascii(treeinfo->root, PLL_UTREE_SHOW_LABEL | PLL_UTREE_SHOW_BRANCH_LENGTH |
-//                                             PLL_UTREE_SHOW_CLV_INDEX );
+        pll_utree_show_ascii(treeinfo->root, PLL_UTREE_SHOW_LABEL | PLL_UTREE_SHOW_BRANCH_LENGTH |
+                                             PLL_UTREE_SHOW_CLV_INDEX );
         printf("Constraint check failed after applying SPR: %u %u\n", p_edge->clv_index, best_r_edge->clv_index);
+        pllmod_set_error(PLLMOD_TREE_ERROR_INVALID_TREE,
+                         "Constraint check failed after applying SPR!");
         return PLL_FAILURE;
       }
 #endif
@@ -1092,6 +1110,14 @@ PLL_EXPORT double pllmod_algo_spr_round(pllmod_treeinfo_t * treeinfo,
   /* reset error */
   pll_errno = 0;
 
+  /* make sure initial topology is compatible with constraint */
+  if (!pllmod_treeinfo_constraint_check_current(treeinfo))
+  {
+    pllmod_set_error(PLLMOD_TREE_ERROR_INVALID_TREE,
+                     "Constraint check failed before SPR round!");
+    return PLL_FAILURE;
+  }
+
   /* allocate brlen buffers */
   for (i = 0; i < BRLEN_BUF_COUNT; ++i)
   {
@@ -1155,6 +1181,14 @@ PLL_EXPORT double pllmod_algo_spr_round(pllmod_treeinfo_t * treeinfo,
     goto error_exit;
   }
 
+  /* make sure intermediate topology is compatible with constraint */
+  if (!pllmod_treeinfo_constraint_check_current(treeinfo))
+  {
+    pllmod_set_error(PLLMOD_TREE_ERROR_INVALID_TREE,
+                     "Constraint check failed after reinsert_nodes() in SPR round!");
+    return PLL_FAILURE;
+  }
+
   /* in FAST mode, we re-insert a subset of best-scoring subtrees with BLO
    * (i.e., in SLOW mode) */
   if (!params.thorough && bestnode_list->current > 0)

src/tree/utree_constraint.c

@@ -24,6 +24,36 @@
 
 #include "../pllmod_common.h"
 
+static inline unsigned int split_popcount(const pll_split_t bitv,
+                                          unsigned int bit_count,
+                                          unsigned int split_len)
+{
+  unsigned int split_size  = sizeof(pll_split_base_t) * 8;
+  unsigned int setb = 0;
+  unsigned int i;
+
+  if (!split_len)
+    split_len = bitv_length(bit_count);
+
+  for (i = 0; i < split_len; ++i)
+  {
+    setb += (unsigned int) PLL_POPCNT32(bitv[i]);
+  }
+
+  /* IMPORTANT: correct for padding bits in the last element! */
+  unsigned int split_offset = bit_count % split_size;
+  if (split_offset)
+  {
+    unsigned int mask = (1<<split_offset) - 1;
+    unsigned int last = bitv[split_len - 1];
+    /* count set bits in the padding part of the bit vector */
+    last &= ~mask;
+    setb -= (unsigned int) PLL_POPCNT32(last);
+  }
+
+  return setb;
+}
+
 static inline void invert_split(pll_split_t bitv, unsigned int bit_count)
 {
   unsigned int split_size  = sizeof(pll_split_base_t) * 8;
@@ -385,7 +415,10 @@ PLL_EXPORT int pllmod_utree_constraint_check_spr(pll_split_set_t * cons_splits,
     const pll_split_t prune_split = get_node_split(splits, p_edge->back);
     pll_split_t new_split = (pll_split_t) calloc(1, cons_split_len * sizeof(pll_split_base_t));
 
-    if (bitv_popcount(prune_split, cons_tip_count, cons_split_len) < cons_tip_count-1)
+    unsigned int pruned_count = split_popcount(prune_split, cons_tip_count, cons_split_len);
+    assert(pruned_count <= cons_tip_count);
+
+    if (pruned_count < cons_tip_count-1)
     {
       /* remaining subtree contains at least 2 constrained taxa -> traverse into regraft subtree */
       regraft_split = find_nonempty_regraft_split(splits, cons_split_len, prune_split, r_edge);
@@ -399,7 +432,7 @@ PLL_EXPORT int pllmod_utree_constraint_check_spr(pll_split_set_t * cons_splits,
     {
       /* remaining subtree contains just 1 constrained taxon  -> traverse into pruned subtree */
       copy_split(new_split, prune_split, cons_split_len);
-      invert_split(new_split, cons_split_len);
+      invert_split(new_split, cons_tip_count);
 
       regraft_split = find_nonempty_regraft_split(splits, cons_split_len, new_split, p_edge);
 

test/runtest.py

@@ -1,4 +1,4 @@
-#!/usr/bin/env python
+#!/usr/bin/env python3
 #
 #    Copyright (C) 2015 Diego Darriba
 #
@@ -40,9 +40,9 @@
 #####################
 #   Configuration   #
 #####################
-do_memtest       = 0           # Evaluate memory leaks
+do_memtest       = 1           # Evaluate memory leaks
 num_replicates   = 20           # Number of samples for the speed test
-all_args         = [6, 4, 0, 1, 2, 3, 4, 6]
+all_args         = [0, 1, 2, 3, 4, 6]
                                 # 0: No vector / No tip pattern
                                 # 1: No vector / Tip pattern
                                 # 2: AVX / No tip pattern
@@ -138,7 +138,7 @@ def testFAIL():
 def diffOutput(filename):
   p2 = Popen(["diff", "tmp", filename], stdout=PIPE)
   output = p2.communicate()[0]
-  return output
+  return output.decode()
 
 def runSpeedTest(files):
   speedtest_header()
@@ -216,15 +216,15 @@ def runSpeedTest(files):
           # Check the output
           skipOutput = diffOutput("out/skip.out")
 
-          if skipOutput == "":
+          if not skipOutput:
               fancyprint("orange", "\b\b\b\b\b")
               testSKIP()
               test_ok = -1
               break
           else:
               output = diffOutput("out/"+filename+".out")
 
-              if output != "":
+              if output:
                 fancyprint("red", "  Test failed\n")
                 call(["mv", "tmp", "result/"+filename+"_"+typestr+"_"+nowstr])
                 call(["mv", "tmperr", "result/"+filename+"_"+typestr+"_"+nowstr+".err"])
@@ -259,7 +259,7 @@ def runSpeedTest(files):
 
         success_count = success_count + result_ok*memory_ok
 
-        print
+        print()
 
   fancyprint("yellow", "{:<80}"
     .format(" "),True)
@@ -296,7 +296,7 @@ def runValidation(files):
         typestr   += "C"
 
       fancyprint("bluebg", "{:<80}"
-        .format(attribstr.rjust(40 + len(attribstr)/2)), True)
+        .format(attribstr.rjust(40 + int(len(attribstr)/2))), True)
 
       # Process each test case
       fancyprint("yellow", "{:<7}   {:<8} {:<18} {:>11}    Result    {:<17}"
@@ -338,21 +338,22 @@ def runValidation(files):
         # Check the output
         skipOutput = diffOutput("out/skip.out")
 
-        if skipOutput == "":
+        if not skipOutput:
           testSKIP()
-          print
+          print()
           result_ok = 1
           memory_ok = 1
         else:
           output = diffOutput("out/"+filename+".out")
-          if output == "":
+          if not output:
             result_ok = 1
             testOK()
             os.remove("tmp")
             os.remove("tmperr")
           else:
             testFAIL()
-            print
+            print(output)
+            print()
             call(["mv", "tmp", "result/"+filename+"_"+typestr+"_"+nowstr])
             call(["mv", "tmperr", "result/"+filename+"_"+typestr+"_"+nowstr+".err"])
             continue
@@ -362,7 +363,7 @@ def runValidation(files):
           if (do_memtest == 1 and not filename.endswith("exe")):
               # Check memory leaks
               p3 = Popen(["./eval_valgrind.sh", "obj/"+filename, nowstr, attrib], stdout=PIPE)
-              output = p3.communicate()[0]
+              output = p3.communicate()[0].decode()
               deflost        = int(output.split(' ')[0])
               indlost        = int(output.split(' ')[1])
               reachable      = int(output.split(' ')[2])
@@ -406,7 +407,7 @@ def runValidation(files):
   num_modules = len(test_modules)
 
   print("  %d modules" % num_modules)
-  print
+  print()
 
   files = []
   for cur_mod in test_modules:
@@ -444,7 +445,7 @@ def runValidation(files):
     success_count = runValidation(files)
 
   # Final summary
-  print
+  print()
   global_run_time = int(time.time()*1000) - global_start_time
   fancyprint("-", "Tests done... it took %d ms" % global_run_time, True)
   fancyprint("green", "      %d/%d (%3.2f%%) OK"
@@ -453,4 +454,4 @@ def runValidation(files):
     fancyprint("red", "      %d/%d (%3.2f%%) FAIL"
       % (num_tests-success_count, num_tests, 100 - 100*success_count/num_tests), True)
 
-  print
+  print()