Support native builds on Windows with MinGW/UCRT

Makefile

@@ -43,6 +43,13 @@ endif
 BT_LDFLAGS =
 BT_LIBS    = -lz -lm -lbz2 -llzma -lpthread
 
+# Support building on MSYS2
+UNAME := $(shell sh -c 'uname 2>/dev/null || echo Unknown')
+ifeq (MINGW64_NT,$(findstring MINGW64_NT,$(UNAME)))
+BT_CPPFLAGS += -D_GNU_SOURCE -D_DEFAULT_SOURCE
+BT_LIBS += -lwsock32
+endif
+
 prefix ?= /usr/local
 
 SUBDIRS = $(SRC_DIR)/annotateBed \

src/complementFile/complementFile.h

@@ -32,7 +32,7 @@ class ComplementFile : public ToolBase {
 
 
 protected:
-	FileRecordMergeMgr *_frm;
+	FileRecordMergeMgr *_frm{};
 	Bed3Interval _outRecord;
 	string _currChrom;
 	const NewGenomeFile *_genomeFile;

src/mergeFile/mergeFile.h

@@ -29,7 +29,7 @@ class MergeFile : public ToolBase {
 
 
 protected:
-	FileRecordMergeMgr *_frm;
+	FileRecordMergeMgr *_frm{};
 
 	virtual ContextMerge *upCast(ContextBase *context) { return static_cast<ContextMerge *>(context); }
 

src/randomBed/randomBed.h

@@ -46,7 +46,7 @@ class BedRandom {
     int _seed;
     bool _haveSeed;
 
-    GenomeFile *_genome;
+    GenomeFile *_genome{};
     CHRPOS _length;
     uint32_t _numToGenerate;
 

src/regressTest/RegressTest.cpp

@@ -68,7 +68,11 @@ bool RegressTest::init(int argc, char **argv)
 		}
 	}
 	_tmpDirname += timeStr;
+#if (defined(_WIN32) || defined(__WIN32__))
+	int mkdirRetval = _mkdir(_tmpDirname.c_str()); //mkdir directory.
+#else
 	int mkdirRetval = mkdir(_tmpDirname.c_str(), S_IRWXU | S_IRWXG | S_IRWXO ); //mkdir directory with all permissions allowed.
+#endif
 	if (mkdirRetval != 0) {
 		fprintf(stderr, "Error: Unable to create temporary output directory %s.\n", _tmpDirname.c_str());
 		return false;

src/regressTest/RegressTest.h

@@ -12,6 +12,10 @@
 #include <cstdio>
 #include <vector>
 #include <fstream>
+#if (defined(_WIN32) || defined(__WIN32__))
+#include <direct.h>
+#endif
+
 
 using namespace std;
 
@@ -42,7 +46,7 @@ class RegressTest {
 	int _filesPerRun;
 	vector<string> _filePrecessorOptions;
 
-	FILE *_fpReportFile;
+	FILE *_fpReportFile{};
 	ifstream _configFile;
 
 	typedef vector<pair<string, string> > fileListType;

src/reldist/reldist.h

@@ -56,7 +56,7 @@ class RelativeDistance {
     size_t _tot_queries;
 
     // instance of a bed file class.
-    BedFile *_bedA, *_bedB;
+    BedFile *_bedA{}, *_bedB{};
     bool _summary;
 
     //------------------------------------------------

src/spacingFile/spacingFile.h

@@ -30,7 +30,7 @@ class SpacingFile : public ToolBase {
     Record *_currRec;
     string _distance;
 
-    FileRecordMgr *_inputFile;
+    FileRecordMgr *_inputFile{};
     virtual ContextSpacing *upCast(ContextBase *context) { return static_cast<ContextSpacing *>(context); }
 
 

src/split/splitBed.h

@@ -29,7 +29,7 @@ class BedSplit {
     std::string bedFileName;
     std::string outfileprefix;
     unsigned int num_chuncks;
-    int bedType;
+    int bedType{};
     std::vector<BED> items;
 
     void usage(std::ostream& out);

src/summaryFile/summaryFile.h

@@ -31,15 +31,15 @@ class SummaryFile : public ToolBase {
 
 
 protected:
-    FileRecordMgr *_frm;
+    FileRecordMgr *_frm{};
     Record *_currRec;
     uint64_t _total_length;
     uint64_t _total_intervals;
     const NewGenomeFile *_genomeFile;
     const vector<string> &_chromList;
     map<string, vector<Interval>, std::less<string> > _chromData;
 
-    FileRecordMgr *_inputFile;
+    FileRecordMgr *_inputFile{};
     virtual ContextSummary *upCast(ContextBase *context) { return static_cast<ContextSummary *>(context); }
 
 };

src/utils/BamTools/include/api/BamAux.h

@@ -1,4 +1,5 @@
 #include <string>
+#include <cstdint>
 
 #ifndef BAMAUX_H
 #define BAMAUX_H

src/utils/Contexts/ContextBase.cpp

@@ -69,7 +69,8 @@ ContextBase::ContextBase()
   _allFileHaveLeadingZeroInChromNames(UNTESTED),
   _noEnforceCoordSort(false),
   _inheader(false),
-  _nameConventionWarningTripped(false)
+  _nameConventionWarningTripped(false),
+  _runToQueryEnd(false)
 
 {
 	_programNames["intersect"] = INTERSECT;

src/utils/Contexts/ContextBase.h

@@ -215,7 +215,7 @@ class ContextBase {
     int _bamHeaderAndRefIdx;
     int _maxNumDatabaseFields;
     bool _useFullBamTags;
-	bool _isCram;   // Used when a "BAM" type which is actually a CRAM
+	bool _isCram{};   // Used when a "BAM" type which is actually a CRAM
 
 	int _numOutputRecords;
 
@@ -247,9 +247,9 @@ class ContextBase {
 
 	//set cmd line params and counter, i, as members so code
 	//is more readable (as opposed to passing all 3 everywhere).
-	int _argc;
-	char **_argv;
-	int _i;
+	int _argc{};
+	char **_argv{};
+	int _i{};
 
 	//track whether each file has the letters chr in their chrom names.
 	//this is needed for enforcing consistent naming conventions across

src/utils/Contexts/ContextClosest.cpp

@@ -12,6 +12,8 @@ ContextClosest::ContextClosest()
 	_ignoreOverlaps(false),
 	_ignoreUpstream(false),
 	_ignoreDownstream(false),
+	_forceUpstream(false),
+	_forceDownstream(false),
 	_reportDistance(false),
 	_signDistance(false),
 	_haveStrandedDistMode(false),

src/utils/Contexts/ContextComplement.h

@@ -21,7 +21,7 @@ class ContextComplement : public ContextBase {
 
 private:
   bool handle_limit_chroms();
-  bool _onlyChromsWithBedRecords;
+  bool _onlyChromsWithBedRecords{};
 protected:
 
 };

src/utils/Contexts/ContextCoverage.h

@@ -26,7 +26,7 @@ class ContextCoverage : public ContextIntersect {
     bool _count;
     bool _perBase;
     bool _showHist;
-    bool _mean;
+    bool _mean{};
     coverageType _coverageType;
 
 	bool handle_c();

src/utils/Contexts/ContextIntersect.h

@@ -109,8 +109,8 @@ class ContextIntersect : public ContextBase {
 
 protected:
 
-	BlockMgr *_splitBlockMgr;
-	bool _shouldRunToDbEnd;
+	BlockMgr *_splitBlockMgr{};
+	bool _shouldRunToDbEnd{};
 
 	virtual bool handle_a();
 	virtual bool handle_abam();

src/utils/Fasta/Fasta.h

@@ -20,7 +20,9 @@
 #include "bedFile.h"
 #include "htslib/faidx.h"
 #include <sys/stat.h>
+#if (!defined(_WIN32) || !defined(__WIN32__))
 #include <sys/mman.h>
+#endif
 #include "split.h"
 #include <stdlib.h>
 #include <ctype.h>
@@ -34,7 +36,7 @@ class FastaIndexEntry {
         FastaIndexEntry(void);
         ~FastaIndexEntry(void);
         string name;  // sequence name
-        CHRPOS length;  // length of sequence
+        CHRPOS length{};  // length of sequence
         void clear(void);
 };
 

src/utils/Fasta/LargeFileSupport.h

@@ -2,7 +2,7 @@
 
 #define _FILE_OFFSET_BITS 64
 
-#ifdef WIN32
+#ifdef  _MSC_VER
 #define ftell64(a)     _ftelli64(a)
 #define fseek64(a,b,c) _fseeki64(a,b,c)
 typedef __int64_t off_type;

src/utils/FileRecordTools/FileReaders/InputStreamMgr.cpp

@@ -86,7 +86,7 @@ bool InputStreamMgr::init()
 			_isStdin = true;
 		}
 
-		_inputFileStream = new ifstream(_filename.c_str());
+		_inputFileStream = new ifstream(_filename.c_str(), std::ios::binary);
 		if (_inputFileStream->fail()) {
 			cerr << "Error: Unable to open file " << _filename << ". Exiting." << endl;
 			delete _inputFileStream;

src/utils/FileRecordTools/FileReaders/InputStreamMgr.h

@@ -51,7 +51,7 @@ class InputStreamMgr {
 	BTlist<int> _scanBuffer;
 	string _saveDataStr;
 	InflateStreamBuf *_infStreamBuf;
-	istream * _finalInputStream;
+	istream * _finalInputStream{};
 	istream *_oldInputStream;
 	bool _isStdin;
 	bool _isGzipped;

src/utils/FileRecordTools/Records/GffRecord.h

@@ -38,7 +38,7 @@ class GffRecord : public Bed6Interval {
 protected:
 	void printRemainingFields(string &outbuf) const;
 
-	int _numFields;
+	int _numFields{};
 	string _source;
 	string _frame;
 	string _group;

src/utils/FileRecordTools/Records/PlusFields.h

@@ -32,7 +32,7 @@ class PlusFields {
 
 protected:
 	vector<string> _fields;
-	int _numOffsetFields; //could be 3 for BedPlus, but GFF has 8 or 9
+	int _numOffsetFields{}; //could be 3 for BedPlus, but GFF has 8 or 9
 };
 
 

src/utils/FileRecordTools/Records/RecordList.h

@@ -48,7 +48,7 @@ class RecordListNode {
 
 private:
 
-	Record * _val;
+	Record * _val{};
 	RecordListNode *_next;
 };
 

src/utils/GenomeFile/GenomeFile.h

@@ -65,7 +65,7 @@ class GenomeFile {
     vector<string> _chromList;
 
     // projecting chroms into a single coordinate system
-    CHRPOS _genomeLength;
+    CHRPOS _genomeLength{};
     vector<CHRPOS> _startOffsets;
 
 };

src/utils/GenomeFile/NewGenomeFile.h

@@ -48,20 +48,20 @@ class NewGenomeFile {
 
 private:
     string  _genomeFileName;
-    istream   *_genomeFile;
+    istream   *_genomeFile{};
     typedef map<string, pair<CHRPOS, int> > lookupType;
     lookupType _chromSizeIds;
     vector<string> _chromList;
     int _maxId;
 
     // projecting chroms onto a single coordinate system
-    CHRPOS _genomeLength;
+    CHRPOS _genomeLength{};
     vector<CHRPOS> _startOffsets;
 
     //cache members for quick lookup
     string _currChromName;
-    CHRPOS _currChromSize;
-    int _currChromId;
+    CHRPOS _currChromSize{};
+    int _currChromId{};
 
 };
 

src/utils/KeyListOps/KeyListOpsMethods.h

@@ -105,9 +105,9 @@ class KeyListOpsMethods {
 
 	typedef enum { UNSORTED, ASC, DESC} SORT_TYPE;
 
-	bool _nonNumErrFlag;
+	bool _nonNumErrFlag{};
 	string _errMsg;
-	bool _isBam;
+	bool _isBam{};
 
 	typedef multimap<int, string, less<int> > histAscType;
 	typedef multimap<int, string, greater<int> > histDescType;

src/utils/bedFile/bedFile.h

@@ -175,8 +175,8 @@ struct BED {
 */
 struct MATE {
     BED bed;
-    int lineNum;
-    MATE *mate;
+    int lineNum{};
+    MATE *mate{};
 };
 
 
@@ -483,10 +483,10 @@ class BedFile {
 
     // the bedfile with which this instance is associated
     string bedFile;
-    unsigned int bedType;  // 3-6, 12 for BED
+    unsigned int bedType{};  // 3-6, 12 for BED
                            // 9 for GFF
-    bool isBed12;          // is it file of true blocked BED12 records?
-    bool isZeroBased;
+    bool isBed12{};          // is it file of true blocked BED12 records?
+    bool isZeroBased{};
 
     // Main data structures used by BEDTools
     masterBedCovMap      bedCovMap;
@@ -496,7 +496,7 @@ class BedFile {
     masterBedMapNoBin    bedMapNoBin;
 
     BedLineStatus _status;
-    int _lineNum;
+    int _lineNum{};
 
 private:
 
@@ -505,21 +505,21 @@ class BedFile {
     bool _isVcf;
     bool _typeIsKnown;        // do we know the type?   (i.e., BED, GFF, VCF)
     FileType   _fileType;     // what is the file type? (BED? GFF? VCF?)
-    istream   *_bedStream;
+    istream   *_bedStream{};
     string _bedLine;
 
     BED _nullBed;
     string _header;
-    bool _firstLine;
+    bool _firstLine{};
     vector<string> _bedFields;
-    unsigned int _numFields;
+    unsigned int _numFields{};
     CHRPOS _merged_start;
     CHRPOS _merged_end;
     string _merged_chrom;
     CHRPOS _prev_start;
     string _prev_chrom;
     unsigned long _total_length;
-    unsigned long _total_flattened_length;
+    unsigned long _total_flattened_length{};
 
     void setZeroBased(bool zeroBased);
     void setGff (bool isGff);

src/utils/bedFilePE/bedFilePE.h

@@ -80,13 +80,13 @@ class BedFilePE {
 
 
     string bedFile;
-    unsigned int bedType;
+    unsigned int bedType{};
 
     masterMateMap bedMapEnd1;
     masterMateMap bedMapEnd2;
 
 private:
-    istream *_bedStream;
+    istream *_bedStream{};
 
     // methods
     BedLineStatus parseLine (BEDPE &bedpe, const vector<string> &lineVector, int &lineNum);