ENH Add xpipe, csvview and tabview

README.rst

@@ -9,6 +9,9 @@ EMBL utilities
 
 ``docker_wrap.sh`` - a convenience script to start a container that shares the user's home and creates a user to ensure consistent permissions.
 
+``tabview`` and ``csvview`` - quick preview of tab and csv delimited files. Aligns columns based on content. Works fine with streams and large files. Uses xpipe.
+
+``xpipe`` - like xargs but for splitting large or infinite streams into chunks and execute commands on them.
 
 Examples
 ========

bin/csvview

+#!/usr/bin/env sh
+##
+# Preview comma-delimited files regardless of their size
+#
+# Copyright 2017 Renato Alves
+# Licensed under MIT (https://en.wikipedia.org/wiki/MIT_License)
+##
+
+xpipe 'column -t -s","' -f "$1" | less -S

bin/tabview

+#!/usr/bin/env sh
+##
+# Preview tab-delimited files regardless of their size
+#
+# Copyright 2017 Renato Alves
+# Licensed under MIT (https://en.wikipedia.org/wiki/MIT_License)
+##
+
+xpipe 'column -t -s"	"' -f "$1" | less -S

bin/xpipe

+#!/usr/bin/env sh
+##
+# Split a stream/file and pipe it to a command in chunks.
+#
+# Copyright 2017 Renato Alves
+# Licensed under MIT (https://en.wikipedia.org/wiki/MIT_License)
+##
+
+ERROR="ERROR:"
+CHUNK=10000
+FILE="/dev/stdin"
+
+usage () {
+    echo >&2 ""
+    echo >&2 "Split a stream/file and pipe it to a command in chunks of X lines."
+    echo >&2 "Think xargs with piped streams - hence xpipe"
+    echo >&2 ""
+    echo >&2 "Usage:"
+    echo >&2 "  xpipe [options] 'command ...'"
+    echo >&2 ""
+    echo >&2 "  Options:"
+    echo >&2 "    -f --file    = file/handle to read. Defaults to STDIN"
+    echo >&2 "    -l --lines   = number of lines to chunk. Defaults to ${CHUNK}"
+    echo >&2 ""
+    echo >&2 "NOTE: Specified command must accept input via STDIN"
+    echo >&2 ""
+}
+
+generic_error () {
+    usage
+    echo >&2 "${ERROR} $1"
+    echo >&2 ""
+    exit 1
+}
+
+required_arg () {
+    usage
+    echo >&2 "${ERROR} '$1' is a required argument."
+    echo >&2 ""
+    exit 1
+}
+
+command -v getopt   >/dev/null 2>&1 || generic_error "getopt is required but was not found."
+command -v parallel >/dev/null 2>&1 || generic_error "GNU Parallel is required but was not found."
+
+ARG_PARSE="getopt -o f:l:h -l file:lines:,help -n $0 --"
+
+# We process arguments twice to handle any argument parsing error:
+ARG_ERROR=$($ARG_PARSE "$@" 2>&1 1>/dev/null)
+
+if [ $? -ne 0 ]; then
+    generic_error "$ARG_ERROR"
+fi
+
+# Abort on any errors from this point onwards
+set -e
+
+# Parse args using getopt (instead of getopts) to allow arguments before options
+ARGS=$($ARG_PARSE "$@")
+
+# reorganize arguments as returned by getopt
+eval set -- "$ARGS"
+
+while true; do
+    case "$1" in
+        # Shift before to throw away option
+        # Shift after if option has a required positional argument
+        -l|--lines)
+            shift
+            CHUNK="$1"
+            shift
+            ;;
+        -f|--file)
+            shift
+            FILE="$1"
+            shift
+            ;;
+        -h|--help)
+            shift
+            usage
+            exit 1
+            ;;
+
+        --)
+            shift
+            break
+            ;;
+    esac
+done
+
+COMMAND="$1"
+
+[ -n "$COMMAND" ] || required_arg "command"
+
+# TODO For now parallel does all the heavy lifting. Ideally we would only depend on the shell or unix commands.
+parallel --pipe -N "$CHUNK" -k "$1" < "$FILE"