Provide more informative error message for failed SAM header parsing.

PiperOrigin-RevId: 315543585

nucleus/io/python/sam_reader_wrap_test.py

@@ -130,8 +130,8 @@ def test_context_manager(self):
         self.assertIsInstance(query_iterable2, clif_postproc.WrappedCppIterable)
 
   def test_from_file_raises_with_missing_bam(self):
-    with self.assertRaisesRegexp(ValueError,
-                                 'Not found: Could not open missing.bam'):
+    with self.assertRaisesRegex(ValueError,
+                                'Not found: Could not open missing.bam'):
       sam_reader.SamReader.from_file(
           reads_path='missing.bam', ref_path='', options=self.options)
 
@@ -141,9 +141,9 @@ def test_ops_on_closed_reader_raise(self):
     with reader:
       pass
     # At this point the reader is closed.
-    with self.assertRaisesRegexp(ValueError, 'Cannot Iterate a closed'):
+    with self.assertRaisesRegex(ValueError, 'Cannot Iterate a closed'):
       reader.iterate()
-    with self.assertRaisesRegexp(ValueError, 'Cannot Query a closed'):
+    with self.assertRaisesRegex(ValueError, 'Cannot Query a closed'):
       reader.query(ranges.parse_literal('chr20:10,000,000-10,000,100'))
 
   @parameterized.parameters('test.sam', 'unindexed.bam')
@@ -152,15 +152,15 @@ def test_query_without_index_raises(self, unindexed_file_name):
     window = ranges.parse_literal('chr20:10,000,000-10,000,100')
     with sam_reader.SamReader.from_file(
         reads_path=path, ref_path='', options=self.options) as reader:
-      with self.assertRaisesRegexp(ValueError, 'Cannot query without an index'):
+      with self.assertRaisesRegex(ValueError, 'Cannot query without an index'):
         reader.query(window)
 
   def test_query_raises_with_bad_range(self):
     with sam_reader.SamReader.from_file(
         reads_path=self.bam, ref_path='', options=self.options) as reader:
-      with self.assertRaisesRegexp(ValueError, 'Unknown reference_name'):
+      with self.assertRaisesRegex(ValueError, 'Unknown reference_name'):
         reader.query(ranges.parse_literal('XXX:1-10'))
-      with self.assertRaisesRegexp(ValueError, 'unknown reference interval'):
+      with self.assertRaisesRegex(ValueError, 'unknown reference interval'):
         reader.query(ranges.parse_literal('chr20:10-5'))
 
   def test_sam_iterate_raises_on_malformed_record(self):
@@ -174,9 +174,10 @@ def test_sam_iterate_raises_on_malformed_record(self):
 
   def test_headless_sam_raises(self):
     headerless = test_utils.genomics_core_testdata('headerless.sam')
-    with self.assertRaisesRegexp(ValueError, 'Could not parse SAM header'):
+    with self.assertRaisesRegex(ValueError,
+                                'Could not parse file with bad SAM header'):
       sam_reader.SamReader.from_file(
-        reads_path=headerless, ref_path='', options=self.options)
+          reads_path=headerless, ref_path='', options=self.options)
 
 
 if __name__ == '__main__':

nucleus/io/sam_reader.cc

@@ -20,12 +20,14 @@
 
 #include <errno.h>
 #include <stdint.h>
+
 #include <map>
 #include <utility>
 #include <vector>
 
 #include "google/protobuf/repeated_field.h"
 #include "absl/strings/numbers.h"
+#include "absl/strings/str_cat.h"
 #include "absl/strings/str_split.h"
 #include "absl/strings/string_view.h"
 #include "htslib/cram.h"
@@ -649,12 +651,13 @@ StatusOr<std::unique_ptr<SamReader>> SamReader::FromFile(
 
   bam_hdr_t* header = sam_hdr_read(fp);
   if (header == nullptr) {
+    string errmsg = absl::StrCat("bad SAM header: ", fp->fn);
     int retval = hts_close(fp);
     fp = nullptr;
     if (retval < 0) {
-      return tf::errors::Internal("hts_close() failed during nonexistent header ");
+      return tf::errors::Internal("hts_close() failed on file with ", errmsg);
     }
-    return tf::errors::Unknown("Could not parse SAM header ");
+    return tf::errors::Unknown("Could not parse file with ", errmsg);
   }
 
   hts_idx_t* idx = nullptr;