Merge pull request #86 from nileshpatra/switch-to-pygtrie

Replace Biopython trie with pygtrie

.travis.yml

@@ -3,7 +3,7 @@ python:
   # - "2.7"
   # - "pypy"
   # - "3.4"
-  - "3.5"
+  # - "3.5"
   - "3.6"
   - "3.7"
   - "3.8"
@@ -12,7 +12,7 @@ python:
 # BioPython doesn't always play well with pip install.
 install:
   - "if [[ $TRAVIS_PYTHON_VERSION != 'pypy' ]]; then pip install -q numpy; fi"
-  - "pip install -q biopython nose"
+  - "pip install -q biopython nose pygtrie"
   - "pip install ."
 
 script:

requirements-rtd.txt

@@ -1,2 +1,3 @@
 biopython
 sphinx
+pygtrie

requirements.txt

@@ -1,4 +1,5 @@
 biopython>=1.70
+pygtrie
 
 # for development
 wheel

seqmagick/subcommands/quality_filter.py

@@ -11,16 +11,29 @@
 import time
 
 from Bio import SeqIO
-try:
-    from Bio import trie, triefind
-except ImportError:
-    trie = None
-    triefind = None
+import pygtrie as trie
 from Bio.SeqIO import QualityIO
 
 from seqmagick import fileformat, __version__
 from .common import typed_range, FileType
 
+
+def trie_match(string, trie):
+    def has_prefix(teststring, trie):
+        for key in trie.keys():
+            if key.startswith(teststring):
+                return True
+        return False
+    longest = None
+    for i in range(len(string)):
+        substr = string[:i + 1]
+        if not has_prefix(substr, trie):
+            break
+        if trie.has_key(substr):
+            longest = substr
+    return longest
+
+
 # Default minimummean quality score
 DEFAULT_MEAN_SCORE = 25.0
 
@@ -626,7 +639,7 @@ def __init__(self,
         self.trie = trie
 
     def filter_record(self, record):
-        m = triefind.match(str(record.seq), self.trie)
+        m = trie_match(str(record.seq), self.trie)
         if m:
             if self.listener:
                 self.listener(
@@ -646,7 +659,7 @@ def parse_barcode_file(fp, primer=None, header=False):
 
     Any additional columns are ignored
     """
-    tr = trie.trie()
+    tr = trie.StringTrie()
     reader = csv.reader(fp)
 
     if header:
@@ -680,10 +693,6 @@ def action(arguments):
         raise ValueError("--quality-window-mean-qual specified without "
                          "--quality-window")
 
-    if trie is None or triefind is None:
-        raise ValueError(
-            'Missing Bio.trie and/or Bio.triefind modules. Cannot continue')
-
     filters = []
     input_type = fileformat.from_handle(arguments.sequence_file)
     output_type = fileformat.from_handle(arguments.output_file)

seqmagick/test/test_subcommands_quality_filter.py

@@ -7,8 +7,6 @@
 
 from seqmagick.subcommands import quality_filter
 
-from Bio import triefind
-
 IS_PYPY = hasattr(sys, 'pypy_version_info')
 
 
@@ -194,7 +192,6 @@ def test_all_truncated(self):
                          [i.id for i in actual])
 
 
-@unittest.skipIf(IS_PYPY, "Bio.trie not available on pypy.")
 class PrimerBarcodeFilterTestCase(unittest.TestCase):
     def setUp(self):
         self.sequences = [
@@ -240,7 +237,6 @@ def e_handler(record, n=1):
         self.assertEqual(events, [1, 5])
 
 
-@unittest.skipIf(IS_PYPY, "Bio.trie not available on pypy.")
 class BarcodePrimerTrieTestCase(unittest.TestCase):
     def setUp(self):
         self.barcode_str = """p1d1bc205,TACTAGCG,CATTGCCTATG
@@ -258,9 +254,9 @@ def test_primer_provided(self):
         res = quality_filter.parse_barcode_file(self.fp, primer='CATTGCCTATG')
         self.assertEqual(9, len(list(res.keys())))
         self.assertEqual('p1d1bc210', res['TACAGTCGCATTGCCTATG'])
-        self.assertEqual(None, triefind.match('TACAGTCGCATTGCCTAT', res))
+        self.assertEqual(None, quality_filter.trie_match('TACAGTCGCATTGCCTAT', res))
         self.assertEqual('TACAGTCGCATTGCCTATG',
-                         triefind.match('TACAGTCGCATTGCCTATGCTACCTA', res))
+                         quality_filter.trie_match('TACAGTCGCATTGCCTATGCTACCTA', res))
 
     def test_primer_in_file(self):
         res = quality_filter.parse_barcode_file(self.fp, primer=None)