X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/2c7b8f8bee023c438c47aff2f35848bbf6c1ae5f..d25db7c02aa07e9d4812a029753c2b8606cf35b1:/crunch_scripts/split-fastq.py

diff --git a/crunch_scripts/split-fastq.py b/crunch_scripts/split-fastq.py
index a4444d0dd2..1c7a36871d 100755
--- a/crunch_scripts/split-fastq.py
+++ b/crunch_scripts/split-fastq.py
@@ -3,76 +3,65 @@
 import arvados
 import re
 import hashlib
+import string
 
 api = arvados.api('v1')
 
 piece = 0
 manifest_text = ""
 
-def put_manifest(manifest_text, sources=[]):
-    crm = arvados.CollectionReader(manifest_text)
-
-    combined = crm.manifest_text(strip=True)
-
-    m = hashlib.new('md5')
-    m.update(combined)
-
-    uuid = "{}+{}".format(m.hexdigest(), len(combined))
-
-    collection = arvados.api().collections().create(
-        body={
-            'uuid': uuid,
-            'manifest_text': crm.manifest_text(),
-        }).execute()
-
-    for s in sources:
-        l = arvados.api().links().create(body={
-            "link": {
-                "tail_uuid": s,
-                "head_uuid": uuid,
-                "link_class": "provenance",
-                "name": "provided"
-            }}).execute()
-
-    return uuid
-
 # Look for paired reads
 
-inp = arvados.CollectionReader(arvados.getjobparam('input'))
+inp = arvados.CollectionReader(arvados.getjobparam('reads'))
 
-with open("/home/peter/manifest") as f:
-    inp = arvados.CollectionReader(f.read())
+manifest_list = []
 
-prog = re.compile("(.*?)_1.fastq(.gz)?$")
+def nextline(reader, start):
+    n = -1
+    while True:
+        r = reader.readfrom(start, 128)
+        if r == '':
+            break
+        n = string.find(r, "\n")
+        if n > -1:
+            break
+        else:
+            start += 128
+    return n
 
-manifest_text = ""
+prog = re.compile(r'(.*?)(_[12])?\.fastq(\.gz)?$')
 
+# Look for fastq files
 for s in inp.all_streams():
-    if s.name() == ".":
-        for f in s.all_files():
-            result = prog.match(f.name())
-            if result != None:
-                p = [{}, {}]
-                p[0]["reader"] = s.files()[result.group(0)]
-                p[1]["reader"] = s.files()[result.group(1) + "_2.fastq" + result.group(2)]
-                m0 = p[0]["reader"].as_manifest()[1:]
-                m1 = p[1]["reader"].as_manifest()[1:]
-                manifest_text += "_" + str(piece) + m0
-                manifest_text += "_" + str(piece) + m1
+    for f in s.all_files():
+        name_pieces = prog.match(f.name())
+        if name_pieces is not None:
+            if s.name() != ".":
+                # The downstream tool (run-command) only iterates over the top
+                # level of directories so if there are fastq files in
+                # directories in the input, the choice is either to forget
+                # there are directories (which might lead to name conflicts) or
+                # just fail.
+                print >>sys.stderr, "fastq must be at the root of the collection"
+                sys.exit(1)
+
+            p = None
+            if name_pieces.group(2) is not None:
+                if name_pieces.group(2) == "_1":
+                    p = [{}, {}]
+                    p[0]["reader"] = s.files()[name_pieces.group(0)]
+                    p[1]["reader"] = s.files()[name_pieces.group(1) + "_2.fastq" + (name_pieces.group(3) if name_pieces.group(3) else '')]
+            else:
+                p = [{}]
+                p[0]["reader"] = s.files()[name_pieces.group(0)]
+
+            if p is not None:
+                for i in xrange(0, len(p)):
+                    m = p[i]["reader"].as_manifest().split()
+                    m[0] = "./_" + str(piece)
+                    manifest_list.append(m)
                 piece += 1
 
-# No pairs found so just put each fastq file into a separate directory
-if manifest_text == "":
-    for s in inp.all_streams():
-        prog = re.compile("(.*?).fastq(.gz)?$")
-        if s.name() == ".":
-            for f in s.all_files():
-                result = prog.match(f.name())
-                if result != None:
-                    p = [{}]
-                    p[0]["reader"] = s.files()[result.group(0)]
-                    m0 = p[0]["reader"].as_manifest()[1:]
-                    manifest_text += "_" + str(piece) + m0
-                    piece += 1
+manifest_text = "\n".join(" ".join(m) for m in manifest_list) + "\n"
 
-arvados.current_task().set_output(put_manifest(manifest_text, [arvados.get_job_param('input')]))
+arvados.current_task().set_output(manifest_text)