Fix FASTA parsing with Windows newlines

When iterating lines of a file, Python includes the trailing `\n`, such that
Windows lines ends with `\r\n`. Regex considers `$` to mean the end of the line,
not end of the string. Vamb's header regex check did not take the trailing
newline into account, but this happened to not matter since
* Only the identifier was used in Vamb's internals, so the newline in the
  description was not noticed, and
* It still passed the regex header check, as the regex matching stopped at the
  `\n` character.
However, `\r\n` threw a wrench in all that.
Fix it by simply removing the trailing `\r?\n` in the FASTA parser.

test/test_parsecontigs.py

@@ -108,3 +108,26 @@ def test_save_load(self):
         self.assertTrue(np.all(md1.lengths == md2.lengths))
         self.assertTrue(np.all(md1.refhash == md2.refhash))
         self.assertTrue(np.all(md1.minlength == md2.minlength))
+
+    def test_windows_newlines(self):
+        rng = random.Random()
+        buf1 = io.BytesIO()
+        buf2 = io.BytesIO()
+        for i in range(10):
+            record = testtools.make_randseq(rng, 10, 20)
+            buf1.write(b">" + record.header.encode())
+            buf2.write(b">" + record.header.encode())
+            buf1.write(b"\r\n")
+            buf2.write(b"\n")
+            buf1.write(record.sequence)
+            buf2.write(record.sequence)
+            buf1.write(b"\r\n")
+            buf2.write(b"\n")
+
+        buf1.seek(0)
+        buf2.seek(0)
+        comp1 = Composition.from_file(buf1)
+        comp2 = Composition.from_file(buf2)
+
+        self.assertEqual(comp1.metadata.refhash, comp2.metadata.refhash)
+        self.assertTrue(np.all(comp1.matrix == comp2.matrix))

vamb/vambtools.py

@@ -309,7 +309,8 @@ def byte_iterfasta(
         )
         raise TypeError(errormsg) from None
 
-    header = probeline[1:]
+    # 13 is the byte value of \r, meaning we remove either \r\n or \n
+    header = probeline[1 : -(1 + (probeline[-2] == 13))]
     buffer: list[bytes] = list()
 
     # Iterate over lines
@@ -320,7 +321,7 @@ def byte_iterfasta(
         elif line.startswith(b">"):
             yield FastaEntry(header, bytearray().join(buffer))
             buffer.clear()
-            header = line[1:]
+            header = line[1 : -(1 + (line[-2] == 13))]
 
         else:
             buffer.append(line)