model/include/silent_tools/silentoopsdropcorruptmodels

#!/usr/bin/env python

import distutils.spawn
import os
import sys
sys.path.append(os.path.dirname(distutils.spawn.find_executable("silent_tools.py")))
import silent_tools
from silent_tools import eprint
import re

# Don't throw an error when someone uses head
from signal import signal, SIGPIPE, SIG_DFL
signal(SIGPIPE, SIG_DFL) 

if (len(sys.argv) == 1):
    eprint("")
    eprint('silentdropcorruptmodels by bcov - drop models with wrong number of residues')
    eprint("Usage:")
    eprint("        silentdropcorruptmodels myfile.silent > fixed.silent")
    eprint("")
    eprint("Note: Like the other oops series tools, this script performs a one-pass operation")
    eprint("        on your file without building an index. Currently only BINARY supported.")
    sys.exit(1)

silent_file = sys.argv[1]


scoreline, f = silent_tools.assert_is_silent_and_get_scoreline(silent_file, return_f=True, accept_garbage=True)

sys.stdout.write( silent_tools.silent_header_fix_corrupt_slim( "A", scoreline, "BINARY" ) )
sys.stdout.flush()


first_line = None
try:
    first_line = next(f)
except:
    pass

while (not first_line is None):
    # try:
    structure, first_line = silent_tools.rip_structure_by_lines_arbitrary_start(f, first_line)
    # except:
    #     break

    tag = structure[0].split()[-1]

    try:
        sequence_chunks = silent_tools.get_sequence_chunks( structure, tag )
    except:
        eprint("silentoopsdropcorruptmodels: Error reading sequence: %s"%(tag))
        continue
    if ( sequence_chunks is None ):
        continue
    sequence = "".join(sequence_chunks)
    seqlen = len(sequence)

    is_binary = True
    is_protein = False

    num_res_lines = 0
    for line in structure:
        if ( is_binary ):
            if ( len(line) == 0 ):
                continue
            if ( line[0] in "HEL" ):
                num_res_lines += 1
        if ( is_protein ):
            if ( len(line) < 6 ):
                continue
            if ( line[5] in "HEL" ):
                num_res_lines += 1


    if ( seqlen != num_res_lines ):
        eprint("silentoopsdropcorruptmodels: Found %5i res expected %5i res: %s"%
            (num_res_lines, seqlen, tag))
    else:
        sys.stdout.write("".join(structure))
        sys.stdout.flush()