Support fragment files with header; #609

stuart-lab · May 10, 2021 · 3d1bf00 · 3d1bf00
1 parent e5ab0c9
commit 3d1bf00
Show file tree

Hide file tree

Showing 4 changed files with 66 additions and 10 deletions.
diff --git a/src/filter.cpp b/src/filter.cpp
@@ -58,7 +58,9 @@ int filterCells(
   }
 
   if (!eof_check) {
-    Rcpp::Rcerr << "Incomplete: File has only headers";
+    Rcpp::Rcerr << "Error: fragment file contains header only\n" << std::flush;
+    gzclose(ifileHandler);
+    return 1;
   }
 
   // looping over the fragments file

diff --git a/src/group.cpp b/src/group.cpp
@@ -50,8 +50,9 @@ SEXP groupCommand(
   }
 
   // char * to string extraction
-  std::string cb_seq;
+  std::string cb_seq, line_seq;
   cb_seq.reserve(32);
+  line_seq.reserve(buffer_length);
 
   // metadata from fragments file
   size_t start, end, reads;
@@ -64,8 +65,25 @@ SEXP groupCommand(
     read_count.assign(num_whitelist_cells, 0);
   }
 
+  // skip header if present
+  bool eof_check;
+  while ((eof_check = gzgets(fileHandler, buffer, buffer_length)) !=0) {
+    line_seq.clear();
+    line_seq.append(buffer);
+
+    if (line_seq.at(0) != '#') {
+      break;
+    }
+  }
+
+  if (!eof_check) {
+    Rcpp::Rcerr << "Error: fragment file contains header only" << std::flush;
+    gzclose(fileHandler);
+    return (Rcpp::DataFrame::create());
+  }
+
   // looping over the fragments file
-  while(gzgets(fileHandler, buffer, buffer_length) !=0 ){
+  do {
     cb_char = strtok ( buffer, "\t" );
 
     for (auto i=1; i<=4; i++) {
@@ -136,7 +154,8 @@ SEXP groupCommand(
     if (line_counter % 2000000 == 0) {
       Rcpp::checkUserInterrupt();
     }
-  }
+    line_seq.clear();
+  } while(gzgets(fileHandler, buffer, buffer_length) !=0 );
 
   //Cleanup
   gzclose(fileHandler);

diff --git a/src/split.cpp b/src/split.cpp
@@ -44,7 +44,7 @@ int splitFragments(
     } else {
       out->open( fileName.c_str() );
     }
- 
+
     streams.push_back(out);
   }
 
@@ -80,8 +80,25 @@ int splitFragments(
   cb_seq.reserve(32);
   line_seq.reserve(buffer_length);
 
+  // skip header if present
+  bool eof_check;
+  while ((eof_check = gzgets(ifileHandler, buffer, buffer_length)) !=0) {
+    line_seq.clear();
+    line_seq.append(buffer);
+
+    if (line_seq.at(0) != '#') {
+      break;
+    }
+  }
+
+  if (!eof_check) {
+    Rcpp::Rcerr << "Error: fragment file contains header only\n" << std::flush;
+    gzclose(ifileHandler);
+    return 1;
+  }
+
   // looping over the fragments file
-  while(gzgets(ifileHandler, buffer, buffer_length) !=0 ){
+  do {
     line_seq.clear();
     line_seq.append(buffer);
 
@@ -124,7 +141,7 @@ int splitFragments(
     if (is_ten_mil) {
       Rcpp::checkUserInterrupt();
     }
-  }
+  } while(gzgets(ifileHandler, buffer, buffer_length) !=0 );
 
   // Cleanup
   gzclose(ifileHandler);

diff --git a/src/validate.cpp b/src/validate.cpp
@@ -46,11 +46,29 @@ bool validateCells(
   }
 
   // char * to string extraction
-  std::string cb_seq;
+  std::string cb_seq, line_seq;
   cb_seq.reserve(32);
+  line_seq.reserve(buffer_length);
+
+  // skip header if present
+  bool eof_check;
+  while ((eof_check = gzgets(fileHandler, buffer, buffer_length)) !=0) {
+    line_seq.clear();
+    line_seq.append(buffer);
+
+    if (line_seq.at(0) != '#') {
+      break;
+    }
+  }
+
+  if (!eof_check) {
+    Rcpp::Rcerr << "Error: fragment file contains header only\n" << std::flush;
+    gzclose(fileHandler);
+    return (false);
+  }
 
   // looping over the fragments file
-  while(gzgets(fileHandler, buffer, buffer_length) !=0 ){
+  do {
     cb_char = strtok ( buffer, "\t" );
 
     for (auto i=1; i<=3; i++) {
@@ -83,7 +101,7 @@ bool validateCells(
     if (line_counter % 2000000 == 0) {
       Rcpp::checkUserInterrupt();
     }
-  }
+  } while(gzgets(fileHandler, buffer, buffer_length) !=0 );
 
   //Cleanup
   gzclose(fileHandler);