detect several MODEL records without ENDMDL between them

Because there is nothing that doesn't exist at CASP, stop 
parsing when hitting a second MODEL tag as if there would 
have been and ENDMDL tag (only if SKIP_FAULTY_RECORDS 
flag is set).


git-svn-id: https://dng.biozentrum.unibas.ch/svn/openstructure/trunk@2102 5a81b35b-ba03-0410-adc8-b2c5c5119f08

modules/io/src/mol/pdb_reader.cc

@@ -81,6 +81,7 @@ PDBReader::PDBReader(const boost::filesystem::path& loc):
 
 void PDBReader::Init(const boost::filesystem::path& loc)
 {
+  num_model_records_=0;
   if (boost::iequals(".gz", boost::filesystem::extension(loc))) {
     in_.push(boost::iostreams::gzip_decompressor());
   }
@@ -134,7 +135,7 @@ void PDBReader::Import(mol::EntityHandle& ent,
   do {
     if (curr_line_.empty()) {
       continue;
-    }    
+    }
     StringRef curr_line(curr_line_.c_str(), curr_line_.size());
     switch(curr_line[0]) {
       case 'A':
@@ -166,6 +167,7 @@ void PDBReader::Import(mol::EntityHandle& ent,
         }
         if (IEquals(curr_line.substr(0, 6), StringRef("ENDMDL", 6))) {
           go_on=false;
+          num_model_records_=0;
           break;
         }
       case 'H':
@@ -182,6 +184,26 @@ void PDBReader::Import(mol::EntityHandle& ent,
         } else if (IEquals(curr_line.substr(0, 6), StringRef("HELIX ", 6))) {
           this->ParseHelixEntry(curr_line);
         }
+        break;
+      case 'M':
+      case 'm':
+        if (curr_line.size()<6) {
+          continue;
+        }
+        if (IEquals(curr_line.substr(0, 6), StringRef("MODEL ", 6))) {
+          ++num_model_records_;          
+          if (num_model_records_<2) {
+            continue;
+          }
+          if (PDB::Flags() & PDB::SKIP_FAULTY_RECORDS) {
+            go_on=false;
+            num_model_records_=1;
+            break;
+          }
+          String msg=str(format("MODEL record without matching ENDMDL on line %d")%line_num_);
+          throw IOException(msg);
+        }
+        break;
       case 'S':
       case 's':
         if (curr_line.size()<6) {
@@ -190,6 +212,7 @@ void PDBReader::Import(mol::EntityHandle& ent,
         if (IEquals(curr_line.substr(0, 6), StringRef("SHEET ", 6))) {
           this->ParseStrandEntry(curr_line);
         }
+        break;
       default:
         break;
     }

modules/io/src/mol/pdb_reader.hh

@@ -72,6 +72,7 @@ private:
   int atom_count_;
   int line_num_;
   bool hard_end_;
+  int num_model_records_;
   String restrict_chains_;
   HSList helix_list_;
   HSList strand_list_;

modules/io/tests/test_io_pdb.cc

@@ -19,6 +19,7 @@
 #include <ost/mol/mol.hh>
 #include <ost/io/mol/entity_io_pdb_handler.hh>
 #include <ost/io/pdb_reader.hh>
+#include <ost/io/io_exception.hh>
 #define BOOST_TEST_DYN_LINK
 #include <boost/test/unit_test.hpp>
 using boost::unit_test_framework::test_suite;
@@ -179,4 +180,31 @@ BOOST_AUTO_TEST_CASE(only_66_cols)
   reader.Import(ent);
 }
 
+BOOST_AUTO_TEST_CASE(no_endmdl_record)
+{
+  String fname("testfiles/pdb/model.pdb");
+  PDBReader reader(fname);
+  mol::EntityHandle ent=mol::CreateEntity();
+  BOOST_CHECK_THROW(reader.Import(ent), IOException);
+
+}
+
+BOOST_AUTO_TEST_CASE(no_endmdl_record_fault_tolerant)
+{
+  String fname("testfiles/pdb/model.pdb");
+  PDBReader reader(fname);
+  PDB::PushFlags(PDB::SKIP_FAULTY_RECORDS);  
+  mol::EntityHandle ent=mol::CreateEntity();
+  reader.Import(ent);
+  BOOST_CHECK_EQUAL(ent.GetChainCount(), 1);
+  BOOST_CHECK_EQUAL(ent.GetResidueCount(), 1);  
+  BOOST_CHECK_EQUAL(ent.GetAtomCount(), 3);
+  ent=mol::CreateEntity();
+  reader.Import(ent);
+  BOOST_CHECK_EQUAL(ent.GetChainCount(), 1);
+  BOOST_CHECK_EQUAL(ent.GetResidueCount(), 1);  
+  BOOST_CHECK_EQUAL(ent.GetAtomCount(), 2);  
+  PDB::PopFlags();
+}
+
 BOOST_AUTO_TEST_SUITE_END()

modules/io/tests/testfiles/pdb/model.pdb

+MODEL  1
+ATOM      1  N   SER     1     -84.454  -4.478 -71.554  1.00 29.08
+ATOM      2  CA  SER     1     -84.657  -3.972 -70.181  1.00 29.08
+ATOM      3  CB  SER     1     -86.156  -3.976 -69.839  1.00 29.08
+MODEL  2
+ATOM      1  N   SER     1     -84.454  -4.478 -71.554  1.00 29.08
+ATOM      2  CA  SER     1     -84.657  -3.972 -70.181  1.00 29.08
+MODEL  3
+ATOM      1  N   SER     1     -84.454  -4.478 -71.554  1.00 29.08
+ATOM      2  CA  SER     1     -84.657  -3.972 -70.181  1.00 29.08
+ATOM      3  CB  SER     1     -86.156  -3.976 -69.839  1.00 29.08
+