Kmc db input fixes (#14)

Fixes #178 and #179
refresh-bio · Jan 3, 2022 · 61ee8d8 · 61ee8d8
1 parent cc843b8
commit 61ee8d8
Show file tree

Hide file tree

Showing 4 changed files with 27 additions and 9 deletions.
diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
@@ -5,6 +5,8 @@ on:
     branches: [ master ]
     paths-ignore:
       - '**.md'
+  pull_request:
+    branches: [ master ]      
   workflow_dispatch:
 
 jobs:
@@ -27,6 +29,15 @@ jobs:
       run: |
         $EXE -k28 -ci1 -jstats.json $KMC_SINGLE_READ 28mers .        
         cmp <(cat stats.json | grep '#Total no. of k-mers' | cut -f 2 -d ':' | cut -f 2 -d ' ' | cut -f 1 -d ',') <(echo "70")
+    
+    - name: small k, no counters
+      run: |
+        $EXE -k4 -cs1 -v $KMC_SINGLE_READ 4mers .
+        
+    - name: count k-mers from KMC1 database when small-k-opt may be applied
+      run: |
+        $EXE -k10 -v  $KMC_SINGLE_READ 10mers .
+        $EXE -k9 -v -fkmc 10mers 9mers .
         
   macos-remote:
     name: macOS build (remote)

diff --git a/kmc_core/binary_reader.h b/kmc_core/binary_reader.h
@@ -339,13 +339,7 @@ class CBinaryFilesReader
 					std::ostringstream ostr;
 					ostr << "Cannot open KMC database: " << f_name;
 					CCriticalErrorHandler::Inst().HandleCriticalError(ostr.str());
-				}
-				if (!kmc_file.IsKMC2())
-				{
-					std::ostringstream ostr;
-					ostr << "Error: currently only KMC databases in version 2 can be readed. If needed to read other version please post an GitHub issue.";
-					CCriticalErrorHandler::Inst().HandleCriticalError(ostr.str());
-				}
+				}				
 				CKMCFileInfo info;
 				kmc_file.Info(info);
 				fsize = info.total_kmers;

diff --git a/kmc_core/kb_completer.h b/kmc_core/kb_completer.h
@@ -160,7 +160,11 @@ bool CSmallKCompleter::CompleteKMCFormat(CSmallKBuf<COUNTER_TYPE> result)
 	uint32 lut_buf_recs = (uint32)(MIN(lut_recs * sizeof(uint64), (uint64)mem_tot_small_k_completer / 2) / sizeof(uint64));
 	uint32 lut_buf_pos = 0;
 	uint32 suf_size = (uint32)(mem_tot_small_k_completer - lut_buf_recs * sizeof(uint64));
-	uint32 suf_recs = (uint32)(suf_size / (counter_size + kmer_suf_bytes) * (counter_size + kmer_suf_bytes));
+	uint32 suf_recs;
+	if (counter_size + kmer_suf_bytes == 0) //fixes #178
+		suf_recs = 0;
+	else
+		suf_recs = (uint32)(suf_size / (counter_size + kmer_suf_bytes) * (counter_size + kmer_suf_bytes));
 	uint32 suf_pos = 0;
 
 	uint64* lut = (uint64*)raw_buffer;

diff --git a/kmc_core/s_mapper.h b/kmc_core/s_mapper.h
@@ -44,7 +44,6 @@ class CSignatureMapper
 	};
 
 public:	
-
 	void InitKMC(const std::string& path)
 	{
 		std::string pre_file_name = path + ".kmc_pre";
@@ -58,6 +57,16 @@ class CSignatureMapper
 
 		my_fseek(file, 0, SEEK_END);
 
+		my_fseek(file, -12, SEEK_END);
+		uint32_t kmc_version;
+		fread(&kmc_version, sizeof(uint32), 1, file);
+		if (kmc_version != 0x200)
+		{
+			std::ostringstream ostr;
+			ostr << "currently only KMC databases in version 2 can be readed. If needed to read other version please post an GitHub issue.";
+			CCriticalErrorHandler::Inst().HandleCriticalError(ostr.str());
+		}
+
 		uint32_t header_offset;
 		my_fseek(file, -8, SEEK_END);
 		fread(&header_offset, sizeof(uint32_t), 1, file);