Merge pull request #1 from lckarssen/master

A cleaner autotools implementation of the gzip reading code

configure.ac

@@ -88,6 +88,37 @@ if test x$ac_cv_header_Eigen_Dense = xno; then
 fi
 
 
+# See if we can use the Boost IOstreams libraries for reading gzipped
+# files
+AC_ARG_WITH([boost-iostreams],
+ AS_HELP_STRING([--with-boost-iostreams], [Use the Boost Iostreams library to
+ allow reading from gzipped files. This is enabled by default]
+               )
+)
+
+if test "x$with_boost_iostreams" != "xno"; then
+    AC_MSG_NOTICE([building using the Boost IOstreams library enabled])
+
+    AC_ARG_WITH([boost-include-path],
+        [AS_HELP_STRING([--with-boost-include-path],
+          [location of the Boost headers, defaults to /usr/include/boost])],
+        [CPPFLAGS+=" -I${withval}"],
+        [CPPFLAGS+=' -I/usr/include/'])
+
+    # Check for the Boost IOstreams header files
+    AC_CHECK_HEADER([boost/iostreams/filter/gzip.hpp],
+        [],
+        [AC_MSG_ERROR([Could not find the Boost IOstreams header files. Did \
+you specify --with-boost-include-path correctly? Or use --without-boost \
+to disable the reading of gzipped files.])]
+    )
+else
+   AC_MSG_NOTICE([not using the Boost IOstreams libraries, so input \
+from gzipped files is not possible])
+fi
+AM_CONDITIONAL([WITH_BOOST_IOSTREAMS], test "x$with_boost_iostreams" != "xno")
+
+
 # Checks for typedefs, structures, and compiler characteristics.
 AC_HEADER_STDBOOL
 AC_C_INLINE

src/Makefile.am

@@ -37,8 +37,6 @@ RHEADERS = include/R.h include/Rmath.h include/R_ext/Arith.h		\
  include/R_ext/Print.h include/R_ext/Random.h include/R_ext/Utils.h	\
  include/R_ext/RS.h
 
-LDADD = -lboost_iostreams -lz 
-
 bin_PROGRAMS =
 if BUILD_palinear
 bin_PROGRAMS += palinear
@@ -62,19 +60,32 @@ endif
 
 palinear_SOURCES = $(REGFILES) $(FVSRC) $(FVHEADERS)
 palinear_CPPFLAGS = -DLINEAR $(AM_CPPFLAGS)
+palinear_LDADD =
 palinear_SOURCES += $(EIGENFILES)
 
 palogist_SOURCES = $(REGFILES) $(FVSRC) $(FVHEADERS)
 palogist_CPPFLAGS = -DLOGISTIC $(AM_CPPFLAGS)
+palogist_LDADD =
 palogist_SOURCES += $(EIGENFILES)
 
 pacoxph_SOURCES = $(COXSRC) $(REGFILES) $(FVSRC) $(FVHEADERS)	\
  $(RHEADERS) survS.h survproto.h coxph_data.h coxph_data.cpp
 pacoxph_CXXFLAGS = -I $(top_srcdir)/src/include $(AM_CXXFLAGS)
 pacoxph_CPPFLAGS = -DCOXPH $(AM_CPPFLAGS)
 pacoxph_CFLAGS = -I $(top_srcdir)/src/include $(AM_CFLAGS)
+pacoxph_LDADD =
 pacoxph_SOURCES += $(EIGENFILES)
 
+if WITH_BOOST_IOSTREAMS
+palinear_CPPFLAGS += -DWITH_BOOST_IOSTREAMS
+palinear_LDADD += -lboost_iostreams -lz
+palogist_CPPFLAGS += -DWITH_BOOST_IOSTREAMS
+palogist_LDADD += -lboost_iostreams -lz
+pacoxph_CPPFLAGS += -DWITH_BOOST_IOSTREAMS
+pacoxph_LDADD += -lboost_iostreams -lz
+endif
+
+
 extract_snp_SOURCES = extract-snp.cpp $(FVSRC) $(FVHEADERS)
 
 ## Install these scripts in the bin directory as well:

src/gendata.cpp

@@ -36,9 +36,11 @@
 #include "utilities.h"
 #include <iostream>
 #include <fstream>
+
+#if WITH_BOOST_IOSTREAMS
 #include <boost/iostreams/filtering_stream.hpp>
 #include <boost/iostreams/filter/gzip.hpp>
-
+#endif
 
 void gendata::mldose_line_to_matrix(const int k,
                                     const char *all_numbers,
@@ -251,17 +253,34 @@ void gendata::re_gendata(const char * fname,
 
     G.reinit(nids, (nsnps * ngpreds));
 
-	std::ifstream file(fname, std::ios_base::in | std::ios_base::binary);
-	boost::iostreams::filtering_istream infile;
-	std::string filename=fname;
-    if (filename.compare(filename.length()-2,2,"gz")== 0){
-    	infile.push(boost::iostreams::gzip_decompressor());
-    }else{
-    	std::cout << "no gziped:"<< ":\n";
+#if WITH_BOOST_IOSTREAMS
+    std::ifstream file(fname, std::ios_base::in | std::ios_base::binary);
+    boost::iostreams::filtering_istream infile;
+    std::string filename = fname;
+    // Note: a better check would be to read the first two bytes of
+    // the file and check for the gzip signature: 0x1f8b
+    // W.r.t. endianness and bytt width: compare each byte separately,
+    // see the comment to this SE answer:
+    // http://stackoverflow.com/a/6059342/881084
+    if (filename.compare(filename.length() - 2, 2, "gz") == 0)
+    {
+        infile.push(boost::iostreams::gzip_decompressor());
+    }
+    else
+    {
+        std::cout << "no gziped:" << ":\n";
     }
     infile.push(file);
+#else
+    std::ifstream infile;
+    infile.open(fname);
+#endif
 
+#if WITH_BOOST_IOSTREAMS
     if (!file)
+#else
+    if (!infile)
+#endif
     {
         std::cerr << "gendata: cannot open file " << fname << endl;
         exit(1);
@@ -294,7 +313,11 @@ void gendata::re_gendata(const char * fname,
                     cerr << "phenotype file and dose or probability file "
                             << "did not match at line " << i + 2 << " ("
                             << tmpid << " != " << idnames[k] << ")" << endl;
+#if WITH_BOOST_IOSTREAMS
                     file.close();
+#else
+                    infile.close();
+#endif
                     exit(1);
                 }
             }
@@ -324,7 +347,11 @@ void gendata::re_gendata(const char * fname,
         }
     }
 
+#if WITH_BOOST_IOSTREAMS
     file.close();
+#else
+    infile.close();
+#endif
 }
 
 // HERE NEED A NEW CONSTRUCTOR BASED ON DATABELBASECPP OBJECT

src/usage.cpp

@@ -31,6 +31,10 @@
 #include "config.h"
 #endif
 
+#if WITH_BOOST_IOSTREAMS
+#include <boost/version.hpp>
+#endif
+
 #include "eigen_mematrix.h"
 
 
@@ -103,6 +107,10 @@ void print_version(void) {
     cout << "Using EIGEN version " << EIGEN_WORLD_VERSION
          << "." << EIGEN_MAJOR_VERSION << "." << EIGEN_MINOR_VERSION
          << " for matrix operations\n";
+#if WITH_BOOST_IOSTREAMS
+    cout << "Using Boost libraries version " << BOOST_LIB_VERSION << endl;
+#endif
+
 }