summary_to_wiki.py

#!/usr/bin/env python3

### Note: This script is unlikely to be useful outside of Edinburgh Genomics ###

import os, sys, re

"""A simple script.  Reads the 10000targets_all_lanes.txt summary text file
   as generated by the Snakefile and spits it out as wiki markup for Confluence.
   Note the file is actually created by running 'tail' on the individual results
   which is where the ==> Headings <== comes from.
   I could make the code output Wiki markup directly but I don't want to have
   that in the scripts, so this is done in the style of old school "perl -p"
   regex-powered munging.
"""

def munge(line):
    header_mo = re.search(r"==> [\w/]*?(\d+)targets_lane(\d+)([TB]?)\.txt <==", line)
    if header_mo:
        return "h3. %s targets per tile on lane %s%s" % header_mo.groups()

    sum_mo = re.match(r"LaneSummary:.*(Tiles:.*)", line)
    if sum_mo:
        return re.sub(r"\t", r"   ", sum_mo.group(1))

    lev1_mo = re.match("Level: 1\s", line)
    if lev1_mo:
        headings = [i.split(": ")[0] for i in line.split("\t")]
        return "||".join([""] + headings + [""]) + "\n" + fmtline(line)

    levn_mo = re.match("Level:", line)
    if levn_mo:
        return fmtline(line)

def fmtline(line):
    """Formats a Level:.* line
       On line 1, hilights the last box, as that's the figure we care about
    """
    levn_mo = re.match("Level: (\d+)", line)
    vals = [i.split(": ",1)[1] for i in line.split("\t")]

    if levn_mo.group(1) == '1':
        vals[-1] = "{color:red}%s{color}" % vals[-1]

    return "|".join([""] + vals + [""])

if __name__ == '__main__':
    #pipeline mode
    from signal import signal, SIGPIPE, SIG_DFL
    signal(SIGPIPE,SIG_DFL)

    for line in (x.rstrip("\n") for x in sys.stdin):
        munged = munge(line)
        if munged is not None: print(munged)