fix: histogram

tlemane · tlemane · commit 25bf19e59263 · 2021-11-03T11:48:13.000+01:00
diff --git a/include/kmtricks/cmd.hpp b/include/kmtricks/cmd.hpp
@@ -49,7 +49,7 @@
 #ifdef WITH_HOWDE
 #include <cmd_cluster.h>
 #include <cmd_build_sbt.h>
-#include <cmd_query_km.h>
+#include <cmd_query.h>
 #endif
 
 namespace km {
@@ -178,14 +178,14 @@ struct main_count
           spdlog::debug("[push] - CountTask - S={}, P={}", opt->id, i);
           pool.add_task(std::make_shared<CountTask<MAX_K, DMAX_C, SuperKStorageReader>>(
             path, config, superk_storage, pinfo, i, KmDir::get().m_fof.get_i(opt->id),
-            config._kmerSize, opt->c_ab_min, opt->lz4, hist->clone(), opt->clear));
+            config._kmerSize, opt->c_ab_min, opt->lz4, get_hist_clone(hist), opt->clear));
         }
         else if (opt->format == "kff")
         {
           spdlog::debug("[push] - KffCountTask - S={}, P={}", opt->id, i);
           pool.add_task(std::make_shared<KffCountTask<MAX_K, DMAX_C, SuperKStorageReader>>(
             path, config, superk_storage, pinfo, i, KmDir::get().m_fof.get_i(opt->id),
-            config._kmerSize, opt->c_ab_min, hist->clone(), opt->clear));
+            config._kmerSize, opt->c_ab_min, get_hist_clone(hist), opt->clear));
         }
       }
       else if (opt->format == "hash" || opt->format == "vector")
@@ -199,14 +199,14 @@ struct main_count
           pool.add_task(std::make_shared<HashCountTask<MAX_K, DMAX_C, SuperKStorageReader>>(
                 path, config, superk_storage, pinfo, i, KmDir::get().m_fof.get_i(opt->id),
                 hw.get_window_size_bits(), config._kmerSize, opt->c_ab_min, opt->lz4,
-                hist->clone(), opt->clear));
+                get_hist_clone(hist), opt->clear));
         }
         else
         {
           spdlog::debug("[push] - HashVecCountTask - S={}, P={}", opt->id, i);
           pool.add_task(std::make_shared<HashVecCountTask<MAX_K, DMAX_C, SuperKStorageReader>>(
             path, config, superk_storage, pinfo, i, KmDir::get().m_fof.get_i(opt->id),
-            hw.get_window_size_bits(), config._kmerSize, opt->c_ab_min, opt->lz4, hist->clone(), opt->clear));
+            hw.get_window_size_bits(), config._kmerSize, opt->c_ab_min, opt->lz4, get_hist_clone(hist), opt->clear));
         }
       }
     }
@@ -700,7 +700,7 @@ struct main_query
     for (size_t i=0; i<howde_query.size(); i++)
       arr[i] = strdup(howde_query.at(i).c_str());
 
-    QueryCommandKm query_cmd("queryKm");
+    QueryCommand query_cmd("queryKm");
     query_cmd.parse(howde_query.size(), arr);
     auto path = fs::current_path();
 
diff --git a/include/kmtricks/histogram.hpp b/include/kmtricks/histogram.hpp
@@ -208,6 +208,13 @@ class KHist
 
 using hist_t = std::shared_ptr<KHist>;
 
+inline hist_t get_hist_clone(hist_t hist)
+{
+  if (hist)
+    return hist->clone();
+  return nullptr;
+}
+
 inline std::vector<uint32_t> compute_merge_thresholds(std::vector<hist_t>& histograms,
                                                       double p,
                                                       const std::string& path)
diff --git a/include/kmtricks/task_scheduler.hpp b/include/kmtricks/task_scheduler.hpp
@@ -202,7 +202,7 @@ class TaskScheduler
               sid, p, m_opt->lz4, KM_FILE::KMER);
             task = std::make_shared<CountTask<MAX_K, MAX_C, SuperKStorageReader>>(
               path, m_config, sk_storage, pinfos, p, iid, m_config._kmerSize,
-              a_min, m_opt->lz4, m_hists[iid]->clone(), !m_opt->keep_tmp);
+              a_min, m_opt->lz4, get_hist_clone(m_hists[iid]), !m_opt->keep_tmp);
           }
           else if (m_opt->kff)
           {
@@ -211,7 +211,7 @@ class TaskScheduler
               sid, p, m_opt->lz4, KM_FILE::KFF);
             task = std::make_shared<KffCountTask<MAX_K, MAX_C, SuperKStorageReader>>(
               path, m_config, sk_storage, pinfos, p, iid,
-              m_config._kmerSize, a_min, m_hists[iid]->clone(), !m_opt->keep_tmp);
+              m_config._kmerSize, a_min, get_hist_clone(m_hists[iid]), !m_opt->keep_tmp);
           }
         }
         else
@@ -224,7 +224,7 @@ class TaskScheduler
             task = std::make_shared<HashCountTask<MAX_K, MAX_C, SuperKStorageReader>>(
               path, m_config, sk_storage, pinfos, p, iid,
               m_hw.get_window_size_bits(), m_config._kmerSize, a_min, m_opt->lz4,
-              m_hists[iid]->clone(), !m_opt->keep_tmp);
+              get_hist_clone(m_hists[iid]), !m_opt->keep_tmp);
           }
           else
           {
@@ -234,7 +234,7 @@ class TaskScheduler
             task = std::make_shared<HashVecCountTask<MAX_K, MAX_C, SuperKStorageReader>>(
               path, m_config, sk_storage, pinfos, p, iid,
               m_hw.get_window_size_bits(), m_config._kmerSize, a_min, m_opt->lz4,
-              m_hists[iid]->clone(), !m_opt->keep_tmp);
+              get_hist_clone(m_hists[iid]), !m_opt->keep_tmp);
           }
         }
         if (m_is_info) task->set_callback([this](){ this->m_dyn[1].tick(); });
@@ -290,7 +290,7 @@ class TaskScheduler
                 sid, p, this->m_opt->lz4, KM_FILE::KMER);
               task = std::make_shared<CountTask<MAX_K, MAX_C, SuperKStorageReader>>(
                 path, this->m_config, sk_storage, pinfos, p, iid,
-                this->m_config._kmerSize, a_min, m_opt->lz4, this->m_hists[iid]->clone(),
+                this->m_config._kmerSize, a_min, m_opt->lz4, get_hist_clone(this->m_hists[iid]),
                 !this->m_opt->keep_tmp);
             }
             else if (m_opt->kff)
@@ -300,7 +300,7 @@ class TaskScheduler
                 sid, p, this->m_opt->lz4, KM_FILE::KFF);
               task = std::make_shared<KffCountTask<MAX_K, MAX_C, SuperKStorageReader>>(
                 path, this->m_config, sk_storage, pinfos, p, iid,
-                this->m_config._kmerSize, a_min, this->m_hists[iid]->clone(), !this->m_opt->keep_tmp);
+                this->m_config._kmerSize, a_min, get_hist_clone(this->m_hists[iid]), !this->m_opt->keep_tmp);
             }
           }
           else
@@ -313,7 +313,7 @@ class TaskScheduler
               task = std::make_shared<HashCountTask<MAX_K, MAX_C, SuperKStorageReader>>(
                 path, m_config, sk_storage, pinfos, p, iid,
                 m_hw.get_window_size_bits(), m_config._kmerSize, a_min, m_opt->lz4,
-                m_hists[iid]->clone(), !this->m_opt->keep_tmp);
+                get_hist_clone(this->m_hists[iid]), !this->m_opt->keep_tmp);
             }
             else
             {
@@ -323,7 +323,7 @@ class TaskScheduler
               task = std::make_shared<HashVecCountTask<MAX_K, MAX_C, SuperKStorageReader>>(
                 path, this->m_config, sk_storage, pinfos, p, iid,
                 this->m_hw.get_window_size_bits(), this->m_config._kmerSize, a_min, false,
-                this->m_hists[iid]->clone(), !this->m_opt->keep_tmp);
+                get_hist_clone(this->m_hists[iid]), !this->m_opt->keep_tmp);
             }
           }
           if (m_is_info)

Original file line number	Diff line number	Diff line change
`@@ -49,7 +49,7 @@`
`49`	`49`	`#ifdef WITH_HOWDE`
`50`	`50`	`#include <cmd_cluster.h>`
`51`	`51`	`#include <cmd_build_sbt.h>`
`52`		`-#include <cmd_query_km.h>`
	`52`	`+#include <cmd_query.h>`
`53`	`53`	`#endif`
`54`	`54`
`55`	`55`	`namespace km {`
`@@ -178,14 +178,14 @@ struct main_count`
`178`	`178`	`spdlog::debug("[push] - CountTask - S={}, P={}", opt->id, i);`
`179`	`179`	`pool.add_task(std::make_shared<CountTask<MAX_K, DMAX_C, SuperKStorageReader>>(`
`180`	`180`	`path, config, superk_storage, pinfo, i, KmDir::get().m_fof.get_i(opt->id),`
`181`		`- config._kmerSize, opt->c_ab_min, opt->lz4, hist->clone(), opt->clear));`
	`181`	`+ config._kmerSize, opt->c_ab_min, opt->lz4, get_hist_clone(hist), opt->clear));`
`182`	`182`	`}`
`183`	`183`	`else if (opt->format == "kff")`
`184`	`184`	`{`
`185`	`185`	`spdlog::debug("[push] - KffCountTask - S={}, P={}", opt->id, i);`
`186`	`186`	`pool.add_task(std::make_shared<KffCountTask<MAX_K, DMAX_C, SuperKStorageReader>>(`
`187`	`187`	`path, config, superk_storage, pinfo, i, KmDir::get().m_fof.get_i(opt->id),`
`188`		`- config._kmerSize, opt->c_ab_min, hist->clone(), opt->clear));`
	`188`	`+ config._kmerSize, opt->c_ab_min, get_hist_clone(hist), opt->clear));`
`189`	`189`	`}`
`190`	`190`	`}`
`191`	`191`	`else if (opt->format == "hash" \|\| opt->format == "vector")`
`@@ -199,14 +199,14 @@ struct main_count`
`199`	`199`	`pool.add_task(std::make_shared<HashCountTask<MAX_K, DMAX_C, SuperKStorageReader>>(`
`200`	`200`	`path, config, superk_storage, pinfo, i, KmDir::get().m_fof.get_i(opt->id),`
`201`	`201`	`hw.get_window_size_bits(), config._kmerSize, opt->c_ab_min, opt->lz4,`
`202`		`- hist->clone(), opt->clear));`
	`202`	`+ get_hist_clone(hist), opt->clear));`
`203`	`203`	`}`
`204`	`204`	`else`
`205`	`205`	`{`
`206`	`206`	`spdlog::debug("[push] - HashVecCountTask - S={}, P={}", opt->id, i);`
`207`	`207`	`pool.add_task(std::make_shared<HashVecCountTask<MAX_K, DMAX_C, SuperKStorageReader>>(`
`208`	`208`	`path, config, superk_storage, pinfo, i, KmDir::get().m_fof.get_i(opt->id),`
`209`		`- hw.get_window_size_bits(), config._kmerSize, opt->c_ab_min, opt->lz4, hist->clone(), opt->clear));`
	`209`	`+ hw.get_window_size_bits(), config._kmerSize, opt->c_ab_min, opt->lz4, get_hist_clone(hist), opt->clear));`
`210`	`210`	`}`
`211`	`211`	`}`
`212`	`212`	`}`
`@@ -700,7 +700,7 @@ struct main_query`
`700`	`700`	`for (size_t i=0; i<howde_query.size(); i++)`
`701`	`701`	`arr[i] = strdup(howde_query.at(i).c_str());`
`702`	`702`
`703`		`- QueryCommandKm query_cmd("queryKm");`
	`703`	`+ QueryCommand query_cmd("queryKm");`
`704`	`704`	`query_cmd.parse(howde_query.size(), arr);`
`705`	`705`	`auto path = fs::current_path();`
`706`	`706`
Original file line number	Diff line number	Diff line change
`@@ -202,7 +202,7 @@ class TaskScheduler`
`202`	`202`	`sid, p, m_opt->lz4, KM_FILE::KMER);`
`203`	`203`	`task = std::make_shared<CountTask<MAX_K, MAX_C, SuperKStorageReader>>(`
`204`	`204`	`path, m_config, sk_storage, pinfos, p, iid, m_config._kmerSize,`
`205`		`- a_min, m_opt->lz4, m_hists[iid]->clone(), !m_opt->keep_tmp);`
	`205`	`+ a_min, m_opt->lz4, get_hist_clone(m_hists[iid]), !m_opt->keep_tmp);`
`206`	`206`	`}`
`207`	`207`	`else if (m_opt->kff)`
`208`	`208`	`{`
`@@ -211,7 +211,7 @@ class TaskScheduler`
`211`	`211`	`sid, p, m_opt->lz4, KM_FILE::KFF);`
`212`	`212`	`task = std::make_shared<KffCountTask<MAX_K, MAX_C, SuperKStorageReader>>(`
`213`	`213`	`path, m_config, sk_storage, pinfos, p, iid,`
`214`		`- m_config._kmerSize, a_min, m_hists[iid]->clone(), !m_opt->keep_tmp);`
	`214`	`+ m_config._kmerSize, a_min, get_hist_clone(m_hists[iid]), !m_opt->keep_tmp);`
`215`	`215`	`}`
`216`	`216`	`}`
`217`	`217`	`else`
`@@ -224,7 +224,7 @@ class TaskScheduler`
`224`	`224`	`task = std::make_shared<HashCountTask<MAX_K, MAX_C, SuperKStorageReader>>(`
`225`	`225`	`path, m_config, sk_storage, pinfos, p, iid,`
`226`	`226`	`m_hw.get_window_size_bits(), m_config._kmerSize, a_min, m_opt->lz4,`
`227`		`- m_hists[iid]->clone(), !m_opt->keep_tmp);`
	`227`	`+ get_hist_clone(m_hists[iid]), !m_opt->keep_tmp);`
`228`	`228`	`}`
`229`	`229`	`else`
`230`	`230`	`{`
`@@ -234,7 +234,7 @@ class TaskScheduler`
`234`	`234`	`task = std::make_shared<HashVecCountTask<MAX_K, MAX_C, SuperKStorageReader>>(`
`235`	`235`	`path, m_config, sk_storage, pinfos, p, iid,`
`236`	`236`	`m_hw.get_window_size_bits(), m_config._kmerSize, a_min, m_opt->lz4,`
`237`		`- m_hists[iid]->clone(), !m_opt->keep_tmp);`
	`237`	`+ get_hist_clone(m_hists[iid]), !m_opt->keep_tmp);`
`238`	`238`	`}`
`239`	`239`	`}`
`240`	`240`	`if (m_is_info) task->set_callback([this](){ this->m_dyn[1].tick(); });`
`@@ -290,7 +290,7 @@ class TaskScheduler`
`290`	`290`	`sid, p, this->m_opt->lz4, KM_FILE::KMER);`
`291`	`291`	`task = std::make_shared<CountTask<MAX_K, MAX_C, SuperKStorageReader>>(`
`292`	`292`	`path, this->m_config, sk_storage, pinfos, p, iid,`
`293`		`- this->m_config._kmerSize, a_min, m_opt->lz4, this->m_hists[iid]->clone(),`
	`293`	`+ this->m_config._kmerSize, a_min, m_opt->lz4, get_hist_clone(this->m_hists[iid]),`
`294`	`294`	`!this->m_opt->keep_tmp);`
`295`	`295`	`}`
`296`	`296`	`else if (m_opt->kff)`
`@@ -300,7 +300,7 @@ class TaskScheduler`
`300`	`300`	`sid, p, this->m_opt->lz4, KM_FILE::KFF);`
`301`	`301`	`task = std::make_shared<KffCountTask<MAX_K, MAX_C, SuperKStorageReader>>(`
`302`	`302`	`path, this->m_config, sk_storage, pinfos, p, iid,`
`303`		`- this->m_config._kmerSize, a_min, this->m_hists[iid]->clone(), !this->m_opt->keep_tmp);`
	`303`	`+ this->m_config._kmerSize, a_min, get_hist_clone(this->m_hists[iid]), !this->m_opt->keep_tmp);`
`304`	`304`	`}`
`305`	`305`	`}`
`306`	`306`	`else`
`@@ -313,7 +313,7 @@ class TaskScheduler`
`313`	`313`	`task = std::make_shared<HashCountTask<MAX_K, MAX_C, SuperKStorageReader>>(`
`314`	`314`	`path, m_config, sk_storage, pinfos, p, iid,`
`315`	`315`	`m_hw.get_window_size_bits(), m_config._kmerSize, a_min, m_opt->lz4,`
`316`		`- m_hists[iid]->clone(), !this->m_opt->keep_tmp);`
	`316`	`+ get_hist_clone(this->m_hists[iid]), !this->m_opt->keep_tmp);`
`317`	`317`	`}`
`318`	`318`	`else`
`319`	`319`	`{`
`@@ -323,7 +323,7 @@ class TaskScheduler`
`323`	`323`	`task = std::make_shared<HashVecCountTask<MAX_K, MAX_C, SuperKStorageReader>>(`
`324`	`324`	`path, this->m_config, sk_storage, pinfos, p, iid,`
`325`	`325`	`this->m_hw.get_window_size_bits(), this->m_config._kmerSize, a_min, false,`
`326`		`- this->m_hists[iid]->clone(), !this->m_opt->keep_tmp);`
	`326`	`+ get_hist_clone(this->m_hists[iid]), !this->m_opt->keep_tmp);`
`327`	`327`	`}`
`328`	`328`	`}`
`329`	`329`	`if (m_is_info)`