#!/usr/bin/env python3

"""
Runs the C version of NCBI xtract. Input file is an XML Document Summary returned from 
the Eutils Python library. 
Output is a TAB-separated value file with Accession numbers, Titles and Sequence Lengths.

Brian Fristensky, Plant Science, University of Manitoba
brian.fristensky@umanitoba.ca
"""

import subprocess
RFN="FristenskyInput.xml"
OFN="FristenskyOutput.tsv"
ELEMENTS="Caption,Title,Slen"
h_OUT = open(OFN,"w")

# -----------------METHOD 1 -------------------
# Method 1 gives the following error in the output file:
# ERROR: No data supplied to xtract from stdin or file, mode is '-rw-------'
#h_RES = open(RFN,"r")
#p = subprocess.Popen(["xtract", "-pattern", "DocumentSummary", "-element", ELEMENTS ], stdin=h_RES, stdout=h_OUT, stderr=h_OUT)

# -----------------METHOD 2 -------------------
# Correctly generates a report as a TSV file.
p = subprocess.Popen(["xtract", "-input", RFN, "-pattern", "DocumentSummary", "-element", ELEMENTS ], stdout=h_OUT, stderr=h_OUT)

p.wait()
h_OUT.close()