Add a minimal FASTA parser to run tests without Biopython or scikit-bio

.github/workflows/package.yml

@@ -72,9 +72,8 @@ jobs:
           path: dist
     - name: Install built wheel
       run: python -m pip install --no-index --find-links=dist pyfastani
-    - name: Install numpy
-      run: python -m pip install numpy
-      if: ${{ !startsWith(matrix.python-version, 'pypy') }}
+    - name: Get FastANI test files
+      run: git clone https://github.com/ParBLiSS/FastANI vendor/FastANI
     - name: Run tests without coverage
       run: python -m unittest pyfastani.tests -vv
 
@@ -136,9 +135,8 @@ jobs:
           path: dist
     - name: Install built wheel
       run: python -m pip install --no-index --find-links=dist pyfastani
-    - name: Install numpy
-      run: python -m pip install numpy
-      if: ${{ !startsWith(matrix.python-version, 'pypy') }}
+    - name: Get FastANI test files
+      run: git clone https://github.com/ParBLiSS/FastANI vendor/FastANI
     - name: Run tests without coverage
       run: python -m unittest pyfastani.tests -vv
 

pyfastani/tests/minifasta.py

+import collections
+
+
+class Record(collections.namedtuple("Record", ["id", "seq", "description"])):
+    pass
+
+
+def parse(path):
+
+    if isinstance(path, str):
+        file = open(path)
+    else:
+        file = path
+
+    with file:
+
+        id_ = None
+        for line in file:
+            if line.startswith(">"):
+                if id_ is not None:
+                    yield Record(id_, "".join(seq), desc)
+                id_ = line[1:].split()[0].strip()
+                desc = " ".join(line[1:].split(maxsplit=1))
+                seq = []
+            else:
+                seq.append(line.strip())
+        if id_ is not None:
+            yield Record(id_, "".join(seq), desc)

pyfastani/tests/requirements.txt

-biopython
+biopython   ; implementation_name == "cpython"
 scikit-bio  ; implementation_name == "cpython"

pyfastani/tests/test_mapper.py

@@ -3,11 +3,12 @@ import unittest
 
 import pyfastani
 
+from . import minifasta
+
 PROJECT_PATH = os.path.realpath(os.path.join(__file__, "..", "..", ".."))
 FASTANI_PATH = os.path.join(PROJECT_PATH, "vendor", "FastANI")
 
 
-
 class _TestMapper(object):
 
     def _load_fasta():
@@ -42,6 +43,15 @@ class _TestMapper(object):
         self.assertAlmostEqual(hits[0].identity, 97.664, places=4)
 
 
+class TestMapper(_TestMapper, unittest.TestCase):
+
+    def _load_fasta(self, path):
+        return list(minifasta.parse(path))
+
+    def _get_sequence(self, record):
+        return record.seq.encode()
+
+
 try:
     from skbio import io as skbio_io
 except ImportError: