#! /bin/bash
# Last edited on 2008-06-14 11:17:37 by stolfi

# Usage: "extract-all {LISTNAME} {RAWDIR} {OUTDIR}"

# Converts Reese human genome files from the Genbank format to the 
# standard ".bas", ".lab", and ".ama" files.

# The input files are "{RAWDIR}/{ITEM}" where 
# {ITEM} is each of the names listed in the "{LISTNAME}.cands" file.

# The output files are "{OUTDIR}/{ITEM}.bas", 
# "{OUTDIR}/{ITEM}.lab", "{OUTDIR}/{ITEM}.aba", 
# "{OUTDIR}/{ITEM}.doc"

# Diagnostic messages are written to "{LISTNAME}.log".
# A list of the {ITEM}s that were not converted,
# and the reasons, is written to "{LISTNAME}.bugs".

listName="$1"; shift
rawDir="$1"; shift
outDir="$1"; shift

progDir="${STOLFIHOME}/programs/c/DNA/dnabayes"
itemList="${listName}.cands"
logFile="${listName}.log"
bugFile="${listName}.bugs"

echo "cleaning ${outDir} ..."
/bin/rm -f ${outDir}/*.bas
/bin/rm -f ${outDir}/*.lab
/bin/rm -f ${outDir}/*.ama
/bin/rm -f ${outDir}/*.doc
/bin/rm -f ${outDir}/*.trn
ls -l ${outDir}

/bin/rm -f ${logFile}

for item in `cat ${itemList}` ; do
  echo "=== ${item} ==="
  ${progDir}/dbd_digest_Reese \
      -v outName="${outDir}/${item}" \
      ${rawDir}/${item} \
    >> ${logFile} \
    2>&1
done

cat ${logFile} \
  | gawk \
      ' /^[*][*]/ { \
          gsub(/[:][0-9]+[:]/ , " | ", $0); \
          gsub(/^.*[\/]/, "", $0); \
          print; \
        } \
      ' \
  > ${bugFile}