From 6825fc13ab77d47834c34400526db716636076eb Mon Sep 17 00:00:00 2001
From: Celine Mercier <celine.mercier@metabarcoding.org>
Date: Wed, 21 Mar 2018 16:41:25 +0100
Subject: [PATCH] Cython API: added ngsfilter file parser

---
 python/obitools3/parsers/ngsfilter.pxd |  8 +++
 python/obitools3/parsers/ngsfilter.pyx | 84 ++++++++++++++++++++++++++
 2 files changed, 92 insertions(+)
 create mode 100644 python/obitools3/parsers/ngsfilter.pxd
 create mode 100644 python/obitools3/parsers/ngsfilter.pyx

diff --git a/python/obitools3/parsers/ngsfilter.pxd b/python/obitools3/parsers/ngsfilter.pxd
new file mode 100644
index 0000000..4422655
--- /dev/null
+++ b/python/obitools3/parsers/ngsfilter.pxd
@@ -0,0 +1,8 @@
+#cython: language_level=3
+
+from ..utils cimport str2bytes
+from ..files.universalopener cimport uopen
+from ..files.linebuffer cimport LineBuffer
+
+
+    
\ No newline at end of file
diff --git a/python/obitools3/parsers/ngsfilter.pyx b/python/obitools3/parsers/ngsfilter.pyx
new file mode 100644
index 0000000..bdaea16
--- /dev/null
+++ b/python/obitools3/parsers/ngsfilter.pyx
@@ -0,0 +1,84 @@
+#cython: language_level=3
+
+'''
+Created on march 8th 2018
+
+@author: cmercier
+'''
+
+from .tab import tabIterator
+from obitools3.utils cimport bytes2str
+import types
+
+
+def ngsfilterIterator(lineiterator, 
+                      bytes sep = None,
+                      bytes dec = b".",
+                      bint stripwhite=True,
+                      bint blanklineskip=True,
+                      bytes commentchar=b"#",
+                      int skip=0,
+                      only=None,
+                      firstline=None,
+                      int buffersize=100000000
+                      ):
+    
+    cdef list  all_lines
+    cdef str   header 
+    cdef str   sep_str
+    cdef bytes out_sep
+    cdef str   out_sep_str
+        
+    out_sep = b"\t"
+    out_sep_str = "\t"
+    
+    if sep is not None:
+        sep_str = bytes2str(sep)
+    else:
+        sep_str = None
+
+    if isinstance(lineiterator, (str, bytes)):
+        lineiterator=uopen(lineiterator)        
+    if isinstance(lineiterator, LineBuffer):
+        iterator = iter(lineiterator)
+    else:
+        if hasattr(lineiterator, "readlines"):
+            iterator = iter(LineBuffer(lineiterator, buffersize))
+        elif hasattr(lineiterator, '__next__'):
+            iterator = lineiterator
+        else:
+            raise Exception("Invalid line iterator")
+    
+    all_lines = [line for line in iterator]
+    new_lines = []
+        
+    if firstline is not None:
+        all_lines.insert(0, firstline)
+        
+    # Insert header for column names
+    column_names = ["experiment", "sample", "forward_tag", "reverse_tag", "forward_primer", "reverse_primer"]
+    header = out_sep_str.join(column_names)
+    
+    new_lines.append(header)
+        
+    for line in all_lines:
+        split_line = line.split(sep_str)
+        tags = split_line.pop(2)
+        tags = tags.split(":")
+        if len(tags) == 1:          # Forward and reverse tags are the same
+            tags.append(tags[0])
+        split_line.insert(2, tags[0])
+        split_line.insert(3, tags[1])
+        new_lines.append(out_sep_str.join(split_line[0:6]))
+    
+    return tabIterator(iter(new_lines),
+                       header = True,
+                       sep = out_sep,
+                       dec = dec,
+                       stripwhite = stripwhite,
+                       blanklineskip = blanklineskip,
+                       commentchar = commentchar,
+                       skip = skip,
+                       only = only,
+                       firstline = None)
+