obitools/obitools3
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 66 Wiki Activity

Cython API: when importing a file in a DMS, its length is computed

Browse Source 
beforehand for the progress bar
This commit is contained in:
Celine Mercier
2019-03-13 18:35:32 +01:00
parent 50e7cd61a6
commit d88390c6d8
5 changed files with 97 additions and 48 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
python/obitools3/uri/decode.pyx
View File

@ -20,7 +20,7 @@ from obitools3.format.fastq import FastqFormat
from obitools3.dms.obiseq import Nuc_Seq
from obitools3.apps.config import getConfiguration,logger
from obitools3.apps.temp import get_temp_dms
from obitools3.utils cimport tobytes # TODO because can't read options as bytes
from obitools3.utils cimport tobytes, count_entries # TODO tobytes because can't read options as bytes
from obitools3.dms.capi.obierrno cimport obi_errno, \
OBIVIEW_ALREADY_EXISTS_ERROR
@ -159,6 +159,7 @@ Reads an URI and returns a tuple containing:
(2) The opened view or iterator on the opened file or writer
(3) The class of object returned or handled by (2)
(4) The original URI in bytes
(5) The number of entries (if input URI) or -1 if unavailable
'''
def open_uri(uri,
bint input=True,
@ -209,7 +210,8 @@ def open_uri(uri,
return (dms[0],
dms[1],
type(dms[1]),
urlunparse(urip))
urlunparse(urip),
len(dms[0]))
try:
resource=open_dms_element(dms[0],
dms[1],
@ -230,7 +232,8 @@ def open_uri(uri,
return (resource[0],
resource[1],
type(resource[1]),
urlunparse(urip))
urlunparse(urip),
len(resource[1]))
except Exception as e:
global obi_errno
if obi_errno == OBIVIEW_ALREADY_EXISTS_ERROR:
@ -503,19 +506,19 @@ def open_uri(uri,
raise NotImplementedError('Output sequence file format not implemented')
else:
if input:
iseq, objclass = entryIteratorFactory(file,
skip, only,
seqtype,
offset,
noquality,
skiperror,
header,
sep,
dec,
nastring,
stripwhite,
blanklineskip,
commentchar)
iseq, objclass, format = entryIteratorFactory(file,
skip, only,
seqtype,
offset,
noquality,
skiperror,
header,
sep,
dec,
nastring,
stripwhite,
blanklineskip,
commentchar)
else: # default export is in fasta? or tab? TODO
objclass = Nuc_Seq # Nuc_Seq_Stored? TODO
iseq = FastaNucWriter(FastaFormat(printNAKeys=printna, NAString=nastring),
@ -524,7 +527,9 @@ def open_uri(uri,
only=only)
#tmpdms = get_temp_dms()
return (file, iseq, objclass, urib)
entry_count = -1
if input:
entry_count = count_entries(file, format)
return (file, iseq, objclass, urib, entry_count)