#!/bin/bash -u
# This script extracts the current weather information from the Web.
#------------------------------------------------------------
# Syntax:
#    $0 [ city ]
#------------------------------------------------------------
# Purpose:
#    Accept a city name on the command line.  Prompt and read the
#    city if it is missing.  If the city is empty, default to "Ottawa".
#    If the city is 3 UPPER-CASE letters, use it directly; otherwise,
#    convert the city to the 3-letter airport code using a built-in table.
#    Contact the Environment Canada Web site for the given airport code.
#    Download the current weather HTML page for that airport code.
#    Extract the current weather information and display it nicely.
#-------------------------------------------------------------
# Student Name:                  Ian! D. Allen
# Algonquin EMail Address:       alleni
# Student Number:                000-000-000
# Course Number:                 CST 8129
# Lab Section Number:            010
# Professor Name:                Ian! D. Allen
# Assignment Name/Number/Date:   Assignment #3 due December 6
# Comment:                       This is a sample assignment label.
# see http://idallen.com/teaching/assignment_standards.html
#-------------------------------------------------------------

# Set standard search path and umask.
#
PATH=/bin:/usr/bin ; export PATH
umask 022

# Function definitions go here ------------------------------------

# The CityToCode function turns its argument (a Canadian city) into
# the 3-character upper-case code for the city.  The argument city
# may be upper, lower, or mixed case.  Return 1 if not found (no output).
#
CityToCode () {
    # Translate the city name to lower-case for looking up in our table.
    #
    city=$( echo "$1" | tr 'A-Z' 'a-z' )

    # Translate the argument city to the 3-letter city code.
    #
    case "$city" in
    ottawa)   code=YOW ;;
    montreal) code=YUL ;;
    toronto)  code=YYZ ;;
    *) return 1 ;;
    esac
    echo "$code"  # function exits with return status of last command
}

# The GetField function locates the first occurrence of a field in the
# web page and returns the text following the field at the end of the line.
#   First argument: user-supplied title of this field, for error messages
#   Second argument: regular expression to match at start of line
#   Third argument: name of file in which to look for the regexp
# The function echoes all the text found after the regexp.
# Return code is 1 if nothing was found; nothing is echoed to stdout.
#
GetField () {
    title="$1"
    regexp="$2"
    file="$3"

    # Scan the whole input file trying the sed substitution on every line.
    # Turn off default sed output; only print if the substitute works.
    # The "head -1" makes sure we only get the first occurrence.
    # We pick off everything that follows the regexp and echo it.
    # (Another way, perhaps less clear: sed -n -e "s/$regexp//p")
    #
    field=$( sed -n -e "s/$regexp"'\(.*\)/\1/p' "$file" | head -1 )
    if [ -z "$field" ] ; then
        echo 1>&2 "$0: Error fetching '$title' using '$regexp' from '$file'."
        return 1  # failed
    fi
    echo "$field"
    return 0  # success
}
# End of function definitions ---------------------------------------

# Main program starts here.

# Pick up a missing city by prompting on stderr and reading
# it from standard input.  Accept only zero or one city name.
# If more than 1, echo the count and incorrect input on stderr and exit.
#
case "$#" in
0)  echo 1>&2 "Enter a Canadian City or UPPER-CASE 3-letter City Code:"
    read city
    ;;
1)  city="$1" ;;
*)  echo 1>&2 "$0: Expecting one City argument, found $# ($*)"
    exit 1
    ;;
esac

# The default, if no city is given, is to use this city.
# Echo the city that is going to be used.
#
if [ -z "$city" ] ; then
    city='Ottawa'
    echo "Using default city of '$city'."
fi
echo "Now retrieving weather for '$city'."

# If the argument is exactly 3 upper-case letters, assume it is
# already a City Code.  If not 3 upper-case letters, assume it is a city
# name and look up the code.  (Some cities are only 3 letters, e.g. Oka.)
#
case "$city" in
[A-Z][A-Z][A-Z])
    code="$city"
    echo "Using UPPER-CASE City Code '$code'."
    ;;
*)
    if ! code=$( CityToCode "$city" ) ; then
        echo 1>&2 "$0: Cannot find the City Code for '$city'."
        exit 1
    fi
    echo "City Name '$city' has City Code '$code'."
esac

# Establish a few temp file names for downloading and editing web pages.
#
tmp=/tmp/weather1$$
tmp2=/tmp/weather2$$
rm -f "$tmp" "$tmp2"

# Download the weather for the given City Code.
#
weatherpage='http://weatheroffice.ec.gc.ca/scripts/citygen.pl?client=ECCDN_e'
wget --quiet -O "$tmp" "$weatherpage&city=$code"

# Check that the file is non-empty and that wget worked.
#
if [ ! -s "$tmp" -o "$?" -ne 0 ] ; then
    echo 1>&2 "$0: wget of web page '$weatherpage' into '$tmp' failed"
    echo 1>&2 "...try again later?"
    exit 1
fi

# Pre-process the downloaded web page:
#   Remove carriage returns from the pages.
#   Change <br> and <p> (any case) to spaces (improves output).
#   Remove all the other HTML tags.
#   Substitute out a few common web-escaped special characters.
# (Note that the input and output files MUST NOT be the same!)
#
tr -d '\r' <"$tmp" \
        | sed -e 's/<[Bb][Rr]>/ /g' -e 's/<[Pp]>/ /g' \
              -e 's/<[^<>]*>//g' \
              -e 's/&deg;//g' \
              -e 's/&ccedil;/c/g' \
              -e 's/&nbsp;/ /g' >$tmp2
mv $tmp2 $tmp  # rename the secondary output file back to $tmp

# Look for errors in the downloaded web page in $tmp.
# This puts the error messages on the screen, which you may not like.
# You could redirect stdout to /dev/null to suppress the output.
#
if grep -i -w error "$tmp" ; then
    echo 1>&2 "$0: Error fetching weather for City Code '$code'."
    exit 1
fi

# Extract a fiew fields from the downloaded web page.
#
title=$( GetField 'City Title' '^Weather for ' "$tmp" )
temp=$( GetField 'Temperature' '^Temp\.:' "$tmp" )
pressure=$( GetField 'Pressure' '^Pressure:' "$tmp" )
visibility=$( GetField 'Visibility' '^Visibility:' "$tmp" )
humidity=$( GetField 'Humidity' '^Humidity:' "$tmp" )
wind=$( GetField 'Wind' '^Wind:' "$tmp" )

# Some pages have Dew Point, some have Wind Chill.
# I don't think they are ever both there at once.
# Try both.  Only print an error if neither is found.
# (Note how stderr is redirected on both these lines.)
#
dewpoint=$( GetField 2>/dev/null 'Dew Point' '^DewPoint:' "$tmp" )
windchill=$( GetField 2>/dev/null 'Wind Chill' '^WindChill:' "$tmp" )
if [ -z "$dewpoint" -a -z "$windchill" ] ; then
    echo 1>&2 "$0: Cannot find Dew Point or Wind Chill in $tmp"
fi

# Remove the temporary file when done.
#
rm "$tmp"

# Output the current weather.
# Only output the Dew Point or Wind Chill if they are found.
#
echo " "
echo "The Current Weather for $title ($code)"
echo "    Temperature: $temp"
echo "    Pressure:    $pressure"
echo "    Visibility:  $visibility"
echo "    Humidity:    $humidity"
if [ ! -z "$dewpoint" ] ; then
    echo "    Dew Point:   $dewpoint"
fi
if [ ! -z "$windchill" ] ; then
    echo "    Wind Chill:  $windchill"
fi
echo "    Wind:        $wind"