#!/usr/bin/env python

import argparse, sys, re
from argparse import RawTextHelpFormatter
from subprocess import Popen, PIPE, STDOUT

__author__ = "Colby Chiang (cc2qe@virginia.edu)"
__version__ = "$Revision: 0.0.2 $"
__date__ = "$Date: 2015-02-28 15:32 $"

# --------------------------------------
# define functions

def get_args():
    parser = argparse.ArgumentParser(formatter_class=RawTextHelpFormatter,
                                     add_help=True, description="\
vawk\n\
author: " + __author__ + "\n\
version: " + __version__ + "\n\
description: An awk-like VCF parser")
    parser.add_argument('-v', '--var', action='append', default=[], required=False, type=str, help='declare an external variable (e.g.: SIZE=10000)')
    parser.add_argument('-c', '--col', dest='info_col', required=False, default=8, type=int, help='column of the INFO field [8]')
    parser.add_argument('--header', required=False, action='store_true', help='print VCF header')
    parser.add_argument('cmd', nargs=1, help='''vawk command syntax is exactly the same as awk syntax with
    a few additional features. The INFO field can be split using
    the I$ prefix and the SAMPLE field can be split using
    the S$ prefix. For example, I$AF prints the allele frequency of
    each variant and S$NA12878 prints the entire SAMPLE field for the
    NA12878 individual for each variant. S$* returns all samples.

    The SAMPLE field can be further split based on the keys in the
    FORMAT field of the VCF (column 9). For example, S$NA12877$GT
    returns the genotype of the NA12878 individual.

    ex: '{ if (I$AF>0.5) print $1,$2,$3,I$AN,S$NA12878,S$NA12877$GT }'
    ''')
    parser.add_argument('vcf', nargs='?', type=str, default=None, help='VCF file (default: stdin)')
    parser.add_argument('--debug', action='store_true', help='debugging level verbosity')

    # parse the arguments
    args = parser.parse_args()

    # if no input, check if part of pipe and if so, read stdin.
    if args.vcf == None:
        if sys.stdin.isatty():
            parser.print_help()
            exit(1)
        else:
            args.vcf = '-'

    # send back the user input
    return args

# parse the vawk string into BEGIN PER-LINE and END portions
def parse(raw):
    begin = ''
    perline = ''
    end = ''
    op_brace = 0
    b = 0
    a = 0
    begin_idx = [0,-1]
    end_idx = [0,-1]

    # get BEGIN script
    try:
        begin_idx[0] = re.search(r'BEGIN\s*{', raw).start()
        a = begin_idx[0] + 5
        b = a
        while begin == '':
            if raw[b] == '{':
                op_brace += 1
            elif raw[b] == '}':
                op_brace -= 1
                if op_brace == 0:
                    begin = raw[a:b+1].strip()[1:-1]
                    begin_idx[1] = b
                    break
            b += 1
    except AttributeError:
        pass
    raw = raw[:begin_idx[0]] + raw[begin_idx[1]+1:]

    # get END script
    try:
        end_idx[0] = re.search(r'END\s*{', raw).start()
        a = re.search(r'END\s*{', raw).start() + 3
        b = a
        while end == '':
            if raw[b] == '{':
                op_brace += 1
            elif raw[b] == '}':
                op_brace -= 1
                if op_brace == 0:
                    end = raw[a:b+1].strip()[1:-1]
                    end_idx[1] = b
                    break
            b += 1
    except AttributeError:
        pass
    raw = raw[:end_idx[0]] + raw[end_idx[1]+1:]

    # get PER-LINE script (remainder of string)
    perline = raw.strip()
    if perline.startswith('{') and perline.endswith('}'):
        perline = perline[1:-1]
    else:
        perline = 'if (' + perline + ') print ; '

    return (begin, perline, end)

# a class for the tokenized sample string 
class Sample(object):
    def __init__(self, s):
        self.string_raw = s

        # tokenize the string
        s_tok = self.string_raw.split('$')
        self.name_raw = s_tok[0]
        self.name_clean = None
        self.name_escaped = None

        # set the field
        if len(s_tok) == 1:
            self.field = "ALL"
        elif len(s_tok) == 2:
            self.field = s_tok[1]
        else:
            print('Error: format should be S$[ID]$[FMT]')
            exit(1)
        return

    # clean weird punctuation from name
    def clean(self):
        s = self.name_raw
        s_esc = self.name_raw

        gremlin_list = ("-", ".")
        for gremlin in gremlin_list:
            s = s.replace(gremlin, "_")
            s_esc = s_esc.replace(gremlin, "\\" + gremlin)

        self.name_clean = s
        self.name_escaped = s_esc

        return

# primary function
def vawk(debug, header, var, info_col, vawk_string, vcf_file):
    # parse the external variables
    format_col = info_col + 1
    samples_start_col = info_col + 2
    
    var_list = []
    for v in var:
        var_list = var_list + ['-v', v]

    # parse the vawk string in to BEGIN, END, and PERLINE
    (begin, perline, end) = parse(vawk_string[0])

    if debug:
        print('begin:', begin)
        print('perline:', perline)
        print('end:', end)

    # get the requested info keys from the command line (format: I$FIELD)
    # [\w]: any word character  [a-zA-Z0-9_]
    # [\W]: any non-word character  [^a-zA-Z0-9_]
    info_keys = set(re.findall(r'[\W]I\$([\w]*)', perline))
    
    if re.search('I\$([^\W]*)',perline):
        # split and parse the relevant INFO fields
        # THE SPLIT INFO MATCHING CAN BE SPED UP BY DOING MULTI GREP AND MOVING THE OR STATEMENT INSIDE
        perline = 'split($' + str(info_col) + ',x,";"); for (i=1;i<=length(x);++i) { ' \
                  + ' '.join(['if (x[i]~"^' + mykey + '=" || x[i]=="' + mykey + '") { split(x[i],info,"="); if (length(info)==1) { INFO_' + mykey + '=1} else {INFO_' + mykey + '=info[2]} }' for mykey in info_keys]) \
                  + '} ' \
                  + perline
        # replace the toxic $ in the awk command
        for mykey in info_keys:
            perline = re.sub('I\$' + mykey, 'INFO_' + mykey, perline)

    # --------------------------------------
    # sample info
    samples_raw = set(re.findall(r'[\W]S\$([a-zA-Z0-9_\-\.*^$]*)', perline))
    sample_queries = []
    for s in samples_raw:
        # make new Sample object and clean it
        s_obj = Sample(s)
        s_obj.clean()
        # add it to the sample queries
        sample_queries.append(s_obj)

        if debug:
            print("name_raw: " + s_obj.name_raw)
            print("name_clean: " + s_obj.name_clean)
            print("name_escaped: " + s_obj.name_escaped)

    # parse the per-sample data fields
    for s_obj in sample_queries:
        # return all data for each sample
        if s_obj.field == 'ALL':
            # concatenate all sample columns with S$*
            if s_obj.name_clean == '*':
                perline =  'ALLSAMPLES_' + s_obj.field + '=""; for (COL=' + str(samples_start_col) + ';COL<=NF-1;++COL) { ALLSAMPLES_' + s_obj.field + '=ALLSAMPLES_' + s_obj.field + '""$COL"\\t" }; ALLSAMPLES_' + s_obj.field + '=ALLSAMPLES_' + s_obj.field + '""$NF; ' + perline
            # user-specified sample name
            else:
                perline = 'SAMPLE_' + '_'.join([s_obj.name_clean, s_obj.field]) + '=$SAMPLE["' + s_obj.name_raw + '"]; ' + perline
        # return user-specified sample fields
        else:
            # special case for wildcard (all samples)
            if s_obj.name_clean == '*':
                perline = 'ALLSAMPLES_' + s_obj.field + '=""; for(i=1;i<=length(fmt);++i) { if (fmt[i]=="' + s_obj.field + '") { fmt_key=i; break } }; for (COL=' + str(samples_start_col) + ';COL<=NF-1;++COL) { split($COL,samp,":"); ALLSAMPLES_' + s_obj.field + '=ALLSAMPLES_' + s_obj.field + '""samp[fmt_key]"\\t" }; split($NF,samp,":"); ALLSAMPLES_' + s_obj.field + '=ALLSAMPLES_' + s_obj.field + '""samp[fmt_key]; ' + perline
            else:
                perline = 'split($SAMPLE["' + s_obj.name_raw + '"],samp,":"); for(i=1;i<=length(fmt);++i) { if (fmt[i]=="' + s_obj.field + '") { SAMPLE_' + '_'.join([s_obj.name_clean, s_obj.field]) + '=samp[i]; break } } ' + perline

    # rename query (ex: S$NA12878$GT to SAMPLE_NA12878_GT)
    look_ahead = '(?![$_\-\.a-zA-Z0-9])'
    if re.search('S\$',perline):
        for s_obj in sample_queries:
            if s_obj.field == 'ALL':
                if s_obj.name_clean == '*':
                    perline = re.sub(r'S\$\*' + look_ahead, 'ALLSAMPLES_' + s_obj.field, perline)
                    perline = perline
                else:
                    perline = re.sub('S\$' + s_obj.name_escaped + look_ahead, 'SAMPLE_' + '_'.join([s_obj.name_clean, s_obj.field]), perline)
            else:
                if s_obj.name_clean == '*':
                    perline = re.sub('S\$\*\$' + s_obj.field + look_ahead, 'ALLSAMPLES_' + s_obj.field, perline)
                else:
                    perline = re.sub('S\$' + '\$'.join([s_obj.name_escaped, s_obj.field]) + look_ahead, 'SAMPLE_' + '_'.join([s_obj.name_clean, s_obj.field]), perline)

    # clear data from previous line
    for mykey in info_keys:
        perline = 'INFO_' + mykey + '=""; ' + perline
    for s_obj in sample_queries:
        if s_obj.name_clean != '*':
            perline = 'SAMPLE_' + '_'.join([s_obj.name_clean, s_obj.field]) + '=""; ' + perline

    # split up the FORMAT field and tack it to the beginning of the per-line statement
    perline = 'split($' + str(format_col) + ',fmt,":"); ' + perline

    # print the header if requested
    if header:
        perline = 'if ($0~"^#") {if ($0!~"^##") { for (i=' + str(samples_start_col) + ';i<=NF;++i) SAMPLE[$i]=i; }; print} else {' + perline + '}'
    else:
        perline = 'if ($0~"^#") {if ($0!~"^##") { for (i=' + str(samples_start_col) + ';i<=NF;++i) SAMPLE[$i]=i; }; next} else {' + perline + '}'

    vawk_string = 'BEGIN {FS="\\t"; OFS="\\t"; ' + begin + '}' + ' {' + perline + '} ' + 'END {' + end + '}'

    cmd = ['gawk'] + var_list + [vawk_string, vcf_file]
    if debug:
        print(' '.join(cmd))
    
    p = Popen(cmd, stdout=PIPE, stderr=STDOUT)

    for line in iter(p.stdout.readline, b''):
        print(line.rstrip().decode("utf-8"))

    return

# --------------------------------------
# main function

def main():
    # parse the command line args
    args = get_args()

    # call primary function
    vawk(args.debug, args.header, args.var, args.info_col, args.cmd, args.vcf)

# initialize the script
if __name__ == '__main__':
    try:
        sys.exit(main())
    except IOError as e:
        if e.errno != 32:  # ignore SIGPIPE
            raise
