Add test_scripts/pdb2mega.bash, prepping for new release

src/Makefile

@@ -1,20 +1,20 @@
-########################################################
+######################################################
 # Makefile is generated by ./build_linux.py
-# Don't edit the Makefile -- update the generator script
-########################################################
+# Don't edit the Makefile -- update the python script
+######################################################
 
 BINDIR := ../bin
 OBJDIR := o
 BINPATH := $(BINDIR)/reseek
 
 CC = gcc
-CFLAGS := $(CFLAGS) -O3 -ffast-math -march=native
+CFLAGS := $(CFLAGS)  -ffast-math -march=native --std=c++17 -O3 -DNDEBUG -fopenmp
 
 CXX = g++
-CXXFLAGS := $(CFLAGS) -DNDEBUG -pthread -O3 -ffast-math -march=native --std=c++17
+CXXFLAGS := $(CFLAGS)  -ffast-math -march=native --std=c++17 -O3 -DNDEBUG -fopenmp
 
 UNAME_S := $(shell uname -s)
-LDFLAGS := $(LDFLAGS) -pthread -lpthread
+LDFLAGS := $(LDFLAGS)  -ffast-math -march=native -O3 -fopenmp
 ifeq ($(UNAME_S),Linux)
     LDFLAGS += -static
 endif
@@ -257,6 +257,3 @@ $(OBJDIR)/%.o : %.c $(HDRS)
 
 $(OBJDIR)/%.o : %.cpp $(HDRS)
 	$(CXX) $(CPPFLAGS) $(CXXFLAGS) -c -o $@ $<
-
-clean:
-	rm -rf $(OBJDIR)/ $(BINPATH)

src/build_linux.bash

@@ -0,0 +1 @@
+./build_linux.py --openmp --std 17

src/build_linux.py

@@ -2,25 +2,31 @@
 
 import os
 import sys
+import argparse
 
-DebugBuild = len(sys.argv) > 1 and sys.argv[1] == "d"
+Usage = \
+(
+"Convert Visual Studio .vcxproj file in current directory to Makefile and run make."
+)
 
-rc = os.system('test -z $(git status --porcelain) 2> /dev/null')
-if rc != 0:
-    sys.stderr.write("\n\nWARNING -- Uncommited changes\n\n")
-    ## sys.exit(1)
+AP = argparse.ArgumentParser(description = Usage)
 
-rc = os.system(r'echo \"$(git log --oneline | head -n1 | cut "-d " -f1)\" | tee gitver.txt')
-if rc != 0:
-    sys.stderr.write("\n\nERROR -- failed to generate gitver.txt\n\n")
-    sys.exit(1)
+# Value opts
+AP.add_argument("--std", required=False, help="Don't strip symbols (default if --debug or --symbols)")
 
-OBJDIR = "o"
-BINDIR = "../bin"
+# Flag opts
+AP.add_argument("--debug", required=False, action='store_true', help="Debug build")
+AP.add_argument("--openmp", required=False, action='store_true', help="Requires OMP")
+AP.add_argument("--pthread", required=False, action='store_true', help="Requires pthread")
+AP.add_argument("--lrt", required=False, action='store_true', help="Requires lrt")
+AP.add_argument("--symbols", required=False, action='store_true', help="Debug symbols (default if --debug)")
+AP.add_argument("--nostrip", required=False, action='store_true', help="Don't strip symbols (default if --debug or --symbols)")
 
-LRT = False
-DEBUG = False
-SVNVER = False
+Args = AP.parse_args()
+debug = Args.debug
+std = Args.std
+nostrip = debug or Args.symbols
+symbols = debug or Args.symbols
 
 ProjFileName = None
 HdrNames = []
@@ -30,13 +36,62 @@
     elif FileName.endswith(".h"):
         HdrNames.append(FileName)
 if ProjFileName is None:
-    print("Project file not found in current directory", file=sys.stderr)
+    sys.stderr.write("\nProject file not found in current directory\n")
+    sys.exit(1)
+
+binary = ProjFileName.replace(".vcxproj", "")
+sys.stderr.write("binary=" + binary + "\n")
+
+compiler_opts = " -ffast-math -march=native"
+linker_opts = " -ffast-math -march=native"
+
+if std:
+    assert std == "11" or std == "17"
+    compiler_opts += " --std=c++" + std
+
+if debug:
+    compiler_opts += " -O0 -DDEBUG"
+    linker_opts += " -O0"
+else:
+    compiler_opts += " -O3 -DNDEBUG"
+    linker_opts += " -O3"
+
+if symbols:
+    compiler_opts += " -g3"
+    linker_opts += " -g3"
+
+if Args.openmp:
+    compiler_opts += " -fopenmp"
+    linker_opts += " -fopenmp"
+
+if Args.pthread:
+    compiler_opts += " -pthread"
+    linker_opts += " -lpthread"
+
+rc = os.system('test -z $(git status --porcelain) 2> /dev/null')
+if rc != 0:
+    sys.stderr.write("\n\nWarning -- Uncommited changes\n\n")
+
+rc = os.system(r'echo \"$(git log --oneline | head -n1 | cut "-d " -f1)\" | tee gitver.txt')
+if rc != 0:
+    sys.stderr.write("\n\nERROR -- failed to generate gitver.txt\n\n")
+    sys.exit(1)
+sys.stderr.write("gitver.txt done.\n")
+
+rc = os.system(r'rm -rf o/ ../bin/%s*' % binary)
+if rc != 0:
+    sys.stderr.write("\n\nERROR -- failed to clean\n\n")
+    sys.exit(1)
+sys.stderr.write("clean done.\n")
+
+OBJDIR = "o"
+BINDIR = "../bin"
 
 Fields = ProjFileName.split("/")
 n = len(Fields)
 Name = Fields[n-1]
 Fields = Name.split(".")
-progname = Fields[0]
+binary = Fields[0]
 
 CXXNames = []
 CNames = []
@@ -58,60 +113,37 @@
             elif FileName.endswith(".c"):
                 FileName = FileName.replace(".c", "")
                 CNames.append(FileName)
-        #     <ProjectName>usearch</ProjectName>
-        elif Line.startswith("<ProjectName"):
-            Line = Line.replace("<ProjectName>", "")
-            Line = Line.replace("</ProjectName>", "")
-            progname = Line
 
 assert len(CXXNames) > 0 or len(CNames) > 0
 
-Makefile = "Makefile"
-if DebugBuild:
-    Makefile = "Makefiled"
-with open(Makefile, "w") as f:
+with open("Makefile", "w") as f:
     def Out(s):
         print(s, file=f)
 
-    if DebugBuild:
-        BINPATH = "$(BINDIR)/%s" % (progname) + "d"
-    else:
-        BINPATH = "$(BINDIR)/%s" % (progname) 
+    BINPATH = "$(BINDIR)/%s" % (binary) 
 
-    Out("########################################################")
-    Out("# Makefile is generated by %s" % sys.argv[0])
-    Out("# Don't edit the Makefile -- update the generator script")
-    Out("########################################################")
+    Out("######################################################")
+    Out("# Makefile is generated by " + sys.argv[0])
+    Out("# Don't edit the Makefile -- update the python script")
+    Out("######################################################")
     Out("")
     Out("BINDIR := %s" % BINDIR)
     Out("OBJDIR := %s" % OBJDIR)
     Out("BINPATH := %s" % BINPATH)
 
-    # Out("")
-    # Out("CPPFLAGS := $(CPPFLAGS) -DNDEBUG -pthread")
-
     if CNames:
         Out("")
         Out("CC = gcc")
-        if DebugBuild:
-            Out("CFLAGS := $(CFLAGS) -ffast-math -march=native")
-        else:
-            Out("CFLAGS := $(CFLAGS) -O3 -ffast-math -march=native")
+        Out("CFLAGS := $(CFLAGS) " + compiler_opts)
 
     if CXXNames:
         Out("")
         Out("CXX = g++")
-        if DebugBuild:
-            Out("CXXFLAGS := $(CFLAGS) -DDEBUG -pthread -march=native --std=c++17")
-        else:
-            Out("CXXFLAGS := $(CFLAGS) -DNDEBUG -pthread -O3 -ffast-math -march=native --std=c++17")
+        Out("CXXFLAGS := $(CFLAGS) " + compiler_opts)
 
     Out("")
     Out("UNAME_S := $(shell uname -s)")
-    if DebugBuild:
-        Out("LDFLAGS := $(LDFLAGS) -O3 -pthread -lpthread")
-    else:
-        Out("LDFLAGS := $(LDFLAGS) -pthread -lpthread")
+    Out("LDFLAGS := $(LDFLAGS) " + linker_opts)
     Out("ifeq ($(UNAME_S),Linux)")
     Out("    LDFLAGS += -static")
     Out("endif")
@@ -140,11 +172,11 @@ def Out(s):
     else:
         Cmd = "\t%(CC) $(LDFLAGS) $(OBJS) -o $(BINPATH)"
 
-    if LRT:
+    if Args.lrt:
         Cmd += " -lrt"
     Out(Cmd)
 
-    if not DebugBuild:
+    if not nostrip:
         Out("	strip $(BINPATH)")
 
     Out("")
@@ -165,17 +197,12 @@ def Out(s):
         Out("$(OBJDIR)/%.o : %.cpp $(HDRS)")
         Out("	$(CXX) $(CPPFLAGS) $(CXXFLAGS) -c -o $@ $<")
 
-    Out("")
-    Out("clean:")
-    Out("	rm -rf $(OBJDIR)/ $(BINPATH)")
-
-if DebugBuild:
-    rc = os.system("rm -f o/myutils.o ../bin/%sd" % progname)
-else:    
-    rc = os.system("rm -f o/myutils.o ../bin/%s" % progname)
+sys.stderr.write("Makefile done.\n")
 
-rc = os.system("make -f %s > make.stdout 2> make.stderr" % Makefile)
+rc = os.system("make 2> make.stderr | tee make.stdout")
 if rc != 0:
     os.system("tail make.stderr")
     sys.stderr.write("\n\nERROR -- make failed, see make.stderr\n\n")
     sys.exit(1)
+sys.stderr.write("make done.\n")
+os.system("ls -lh ../bin/" + binary + "\n")

src/daliscorer.cpp

@@ -95,6 +95,14 @@ void DALIScorer::LoadChains(const string &FN)
 		m_SeqToChainIdx[ptrChain->m_Seq] = Idx;
 		}
 	Progress("Reading chains done (%u)\n", SIZE(m_Chains));
+	{//@@
+	for (map<string, uint>::const_iterator iter = m_SeqToChainIdx.begin();
+		iter != m_SeqToChainIdx.end(); ++iter)
+			{
+			Log("[%u] %s\n", iter->second, iter->first);
+			}
+
+	}//@@
 	}
 
 void DALIScorer::SetCore()
@@ -129,6 +137,8 @@ bool DALIScorer::SetSeqIdxToChainIdx(bool MissingSeqOk)
 		const string &Label = string(m_MSA->GetLabel(SeqIdx));
 		string Seq;
 		m_MSA->GetSeq_StripGaps(SeqIdx, Seq, true);
+		Log(">%s\n", Label.c_str());//@@
+		Log("%s\n", Seq.c_str());//@@
 		map<string, uint>::const_iterator iter = m_SeqToChainIdx.find(Seq);
 		if (iter == m_SeqToChainIdx.end())
 			{

src/prettyaln.cpp

@@ -100,7 +100,7 @@ void PrettyAln(FILE *f, const PDBChain &A, const PDBChain &B,
 		if (ColHi == ColCount)
 			break;
 		asserta(ColHi < ColCount);
-		ColLo = ColHi + 1;
+		ColLo = ColHi;
 		}
 	double PctId = GetPct(Ids, ColCount);
 	double PctGaps = GetPct(Gaps, ColCount);

src/usage.h

@@ -56,6 +56,9 @@ const char *usage_txt[] =
 "           -bca FILENAME    # .bca format, binary .cal, recommended for DBs\n"
 "           -fasta FILENAME  # FASTA format\n"
 "\n"
+"Create input for Muscle-3D multiple structure alignment:\n"
+"    reseek -pdb2mega STRUCTS -output structs.mega\n"
+"\n"
 "STRUCTS argument is one of:\n"
 "   NAME.cif or NAME.mmcif     # PDBx/mmCIF file\n"
 "   NAME.pdb                   # Legacy format PDB file\n"

test_scripts/_run_tests.bash

@@ -1,6 +1,7 @@
 #!/bin/bash -e
 
 cd ../src
+rm -rf o/ ../bin/reseek*
 chmod +x ./build_linux.py
 python3 ./build_linux.py
 cd ../test_scripts
@@ -23,13 +24,14 @@ echo STARTED `date` >> $log
 ./columns.bash
 ./search.bash
 ./scop40.bash
+./pdb2mega.bash
 
 python3 ./check_logs.py >> $log
 python3 ./check_convert.py >> $log
 python3 ./check_columns.py >> $log
 python3 ./check_scop40.py >> $log
 
-./update_success_list.py $ver $date
+python3 ./update_success_list.py $ver $date
 
 echo COMPLETED $date >> $log
 

test_scripts/pdb2mega.bash

@@ -0,0 +1,8 @@
+#!/bin/bash -e
+
+cd ../test_output
+
+../bin/reseek \
+	-pdb2mega ../test_structures/4v40.cif \
+	-output ../test_output/4v40.mega \
+	-log pdb2mega.log