#!/usr/bin/env bash
#
# Copyright (c) 2016 10x Genomics, Inc. All rights reserved.
#
# Generate Martian info file (mri.tgz) for mrp pipestances.
#

folder="${1%/}"
ecode="$2"
cmdline="$3"
email="support@10xgenomics.com"
tarfile="$(basename "$folder").mri.tgz"
files="samplesheet.csv input_samplesheet.csv summary_cs.csv summary.json summary.txt summary_cs.txt metrics_summary_csv.csv metrics_summary_json.json web_summary.html Log.out _finalstate"
limit_filesize_patterns=(.json)

# Save the non-truncated version of these files if < 30M
do_not_truncate="$folder/CNV_CALLER_SINGLECELL_CS*DETECT_BREAKPOINTS*join*_stdout"

if [ -d "$folder" ]; then
    # Capture commandline into pipestance.
    echo "$cmdline" > "$folder/_cmdline"

    # Capture sitecheck into pipestance.
    sitecheck > "$folder/_sitecheck"

    # Capture data file paths into pipestance.
    find "$folder" -path "*files*" -ls > "$folder/_filelist"

    # Format summary file arguments.
    for f in $files; do
	# Limit filesize for a set of patterns
	more_options=""
	for pattern in "${limit_filesize_patterns[@]}"; do
	    if [[ "$f" =~ "$pattern" ]]; then
		more_options="-a -size -1024k"
	    fi
	done
	files_options="$files_options -o -name $f ${more_options}"
    done

    # Find large metadata files and make truncated versions of them.
    truncated_files=""
    for f in $(find "$folder" -type f \( -size +1024k -not -name "$tarfile" \
        -not -path "*files*" -not -path "*journal*" \
        -not -name "_finalstate" \
        -not -path "$do_not_truncate" -name '_*' \)); do
        _truncate "$f" > "$f._truncated_"
        truncated_files="$truncated_files \"$f._truncated_\""
    done

    # Generate log tar file.
    echo -e "Saving pipestance info to \"$folder/$tarfile\""

    # Use transform option if it exists to rewrite truncated file names to
    # their originals. CentOS 5's tar version lacks the --transform option,
    # however, so in that case we just leave the file names alone.
    tar --help | grep -e --transform > /dev/null 2>&1
    if [ $? -eq 0 ]; then
        find "$folder" -type f -not -name "$tarfile" -not -path "*files*" \
        -not -path "*journal*" \( \( -size -1024k  $files_options \) -o  \
        \( -path "$do_not_truncate" -size -30M \) \) -print0 | \
        tar zcf "$tarfile" --null --transform='s/\._truncated_$//' -T -
    else
        find "$folder" -type f -not -name "$tarfile" -not -path "*files*" \
        -not -path "*journal*" \( \( -size -1024k  $files_options \) -o  \
        \( -path "$do_not_truncate" -size -30M \) \) -print0 | \
        tar zcf "$tarfile" --null -T -
    fi

    # Clean up truncated versions of files, if any were created.
    if [ -n "$truncated_files" ]; then
        for f in $truncated_files; do
            rm -f "$f"
        done
    fi
    mv -t "$folder" "$tarfile"

    if [ -n "$TENX_AUTOSUPPORT_SENDER" ]; then
        echo -e "Automatically sending diagnostics to 10x Genomics."
        "$TENX_PRODUCT" upload "$TENX_AUTOSUPPORT_SENDER" "$folder/$tarfile"
    else
        # Help instructions if exit code is nonzero.
        if [ "$ecode" != "0" ]; then
            echo -e "For assistance, upload this file to 10x Genomics by running:\n\n$TENX_PRODUCT upload <your_email> \"$folder/$tarfile\"\n"
        fi
    fi
fi