{AH} set MAX_POS to (1 << 31) - 1, fixes #741 and #732

pysam/libcalignmentfile.pyx

@@ -1,4 +1,3 @@
-
 # cython: embedsignature=True
 # cython: profile=True
 ########################################################
@@ -62,6 +61,8 @@ import warnings
 import array
 from libc.errno  cimport errno, EPIPE
 from libc.string cimport strcmp, strpbrk, strerror
+from libc.stdint cimport INT32_MAX
+
 from cpython cimport array as c_array
 from cpython.version cimport PY_MAJOR_VERSION
 
@@ -94,7 +95,8 @@ IndexStats = collections.namedtuple("IndexStats",
 ########################################################
 ## global variables
 # maximum genomic coordinace
-cdef uint32_t MAX_POS = 2 << 31
+# for some reason, using 'int' causes overlflow
+cdef int MAX_POS = (1 << 31) - 1
 
 # valid types for SAM headers
 VALID_HEADER_TYPES = {"HD" : collections.Mapping,
@@ -1314,8 +1316,9 @@ cdef class AlignmentFile(HTSFile):
         an iterator over genomic positions.
 
         """
-        cdef int rtid, rstart, rstop, has_coord
-
+        cdef int rtid, has_coord
+        cdef int32_t rstart, rstop
+
         if not self.is_open:
             raise ValueError("I/O operation on closed file")
 
@@ -2054,7 +2057,6 @@ cdef class IteratorRowRegion(IteratorRow):
 
         IteratorRow.__init__(self, samfile,
                              multiple_iterators=multiple_iterators)
-
         with nogil:
             self.iter = sam_itr_queryi(
                 self.index,

pysam/libcbcf.pyx

@@ -111,7 +111,7 @@ __all__ = ['VariantFile',
 ## Constants
 ########################################################################
 
-cdef uint32_t MAX_POS = 2 << 31
+cdef int MAX_POS = (1 << 31) - 1
 cdef tuple VALUE_TYPES = ('Flag', 'Integer', 'Float', 'String')
 cdef tuple METADATA_TYPES = ('FILTER', 'INFO', 'FORMAT', 'CONTIG', 'STRUCTURED', 'GENERIC')
 cdef tuple METADATA_LENGTHS = ('FIXED', 'VARIABLE', 'A', 'G', 'R')

pysam/libchtslib.pyx

@@ -9,10 +9,9 @@
 
 from posix.unistd cimport dup
 from libc.errno  cimport errno
+from libc.stdint cimport INT32_MAX
 from cpython cimport PyBytes_FromStringAndSize
-
 from pysam.libchtslib cimport *
-
 from pysam.libcutils cimport force_bytes, force_str, charptr_to_str, charptr_to_str_w_len
 from pysam.libcutils cimport encode_filename, from_string_and_size
 
@@ -25,7 +24,7 @@ from pysam.libcutils cimport encode_filename, from_string_and_size
 import os
 import io
 import re
-from warnings         import warn
+from warnings import warn
 
 
 ########################################################################
@@ -41,7 +40,7 @@ DEF SEEK_CUR = 1
 DEF SEEK_END = 2
 
 # maximum genomic coordinace
-cdef uint32_t MAX_POS = 2 << 31
+cdef int MAX_POS = (1 << 31) - 1
 
 cdef tuple FORMAT_CATEGORIES = ('UNKNOWN', 'ALIGNMENTS', 'VARIANTS', 'INDEX', 'REGIONS')
 cdef tuple FORMATS = ('UNKNOWN', 'BINARY_FORMAT', 'TEXT_FORMAT', 'SAM', 'BAM', 'BAI', 'CRAM', 'CRAI',
@@ -630,8 +629,8 @@ cdef class HTSFile(object):
 
         """
         cdef int rtid
-        cdef long long rstart
-        cdef long long rstop
+        cdef int32_t rstart
+        cdef int32_t rstop
 
         if reference is not None:
             if contig is not None:
@@ -644,7 +643,7 @@ cdef class HTSFile(object):
             stop = end
 
         if contig is None and tid is None and region is None:
-            return 0, 0, 0, 0
+            return 0, 0, 0, MAX_POS
 
         rtid = -1
         rstart = 0

pysam/libcutils.pyx

@@ -12,6 +12,7 @@ from cpython cimport PyBytes_Check, PyUnicode_Check
 from cpython cimport array as c_array
 from libc.stdlib cimport calloc, free
 from libc.string cimport strncpy
+from libc.stdint cimport INT32_MAX, int32_t
 from libc.stdio cimport fprintf, stderr, fflush
 from libc.stdio cimport stdout as c_stdout
 from posix.fcntl cimport open as c_open, O_WRONLY
@@ -24,7 +25,7 @@ from libcbcftools cimport bcftools_main, bcftools_set_stdout, bcftools_set_stder
 
 #####################################################################
 # hard-coded constants
-cdef uint32_t MAX_POS = 2 << 31
+cdef int MAX_POS = (1 << 31) - 1
 
 #################################################################
 # Utility functions for quality string conversions

tests/AlignmentFileHeader_test.py

@@ -121,6 +121,7 @@ def test_header_constructed_from_references_without_text(self):
         self.compare_headers(header, self.header_without_text)
         self.check_name_mapping(header)
 
+
 class TestHeaderSAM(unittest.TestCase):
     """testing header manipulation"""
 
@@ -287,6 +288,7 @@ def compare_headers(self, a, header_b):
     def check_read_write(self, flag_write, header):
 
         fn = get_temp_filename()
+        print(fn)
         with pysam.AlignmentFile(
                 fn,
                 flag_write,
@@ -299,17 +301,30 @@ def check_read_write(self, flag_write, header):
         with pysam.AlignmentFile(fn) as inf:
             read_header = inf.header
 
-        os.unlink(fn)
+        # os.unlink(fn)
         self.compare_headers(header, read_header)
+        expected_lengths = dict([(x["SN"], x["LN"]) for x in header["SQ"]])
+        self.assertEqual(expected_lengths,
+                         dict(zip(read_header.references,
+                                  read_header.lengths)))
 
     def test_SAM(self):
         self.check_read_write("wh", self.header)
 
     def test_BAM(self):
         self.check_read_write("wb", self.header)
-
+        
     def test_CRAM(self):
         header = copy.copy(self.header)
-        # for CRAM, \t needs to be quoted:
-        header['PG'][1]['CL'] = re.sub(r"\t", r"\\\\t", header['PG'][1]['CL'])
+        if "PG" in header:
+            # for CRAM, \t needs to be quoted:
+            header['PG'][1]['CL'] = re.sub(r"\t", r"\\\\t", header['PG'][1]['CL'])
         self.check_read_write("wc", header)
+
+
+class TestHeaderLargeContigs(TestHeaderWriteRead):
+    """see issue 741"""
+
+    header = {'SQ': [{'LN': 2147483647, 'SN': 'chr1'},
+                     {'LN': 1584, 'SN': 'chr2'}],
+              'HD': {'VN': '1.0'}}