ics-to-html.cron

#!/usr/bin/env bash
#
# Cron job to update schedule.html and the pages for the tracks with
# events from the W3C Calendar, using the ics-to-html.awk script.
#
# (The events in the W3C Calendar are typically put there by the TPAC
# organizers. In the case of breakouts, they use a process that first
# collects breakouts as GithUb issues and then converts them to
# calendar events. See https://github.com/w3c/tpac-breakouts )
#
# To add "fake" events that are not in the W3C Calendar, edit the file
# extra-events.ics.
#
# To add extra categories (class names) to events beyond the
# categories that are in the W3C Calendar, edit the Google Docs
# spreadsheet:
#
# https://docs.google.com/spreadsheets/d/1629ZDKKClOWJkk3XOYCtQ8_TsllEl586IMzoMzyEh_4/edit?usp=sharing
#
# Command line options:
#
# -n
#     Run the commands, but do not actually update the pages. This is
#     useful to see if the commands succeed without any errors.
#     Implies -c.
# -c
#     Do not check the updated pages into CVS.
# -f
#     Force updates. Do not stop processing when a sanity check fails.
#     Warning: This may result in incorrect files, unless combined
#     with -n.
# -d directory
#     The working directory, i.e., the directory in which to check out
#     the relevant files from CVS. Default: $HOME/WWW/2025/11/TPAC
# -i url
#     The URL of the ICS file with the relevant entries from the W3C
#     group calendar. Default:
#     https://www.w3.org/calendar/tpac2025/export/"
#
# Created: 26 February 2023
# based on https://www.w3.org/2022/09/TPAC/ics-to-html.cron
# Author: Bert Bos <bert@w3.org>

set -u -e

# The default URL for the relevant subset of the W3C calendar, in
# icalendar format.
CALENDAR="https://www.w3.org/calendar/tpac2025/export/"

# Labels (‘tracks’) to add to certain events. To get this URL for the
# Google spreadsheet, open the spreadsheet in a browser and go to
# "File" -> "Share" -> "Publish to web". That opens a dialog box.
# Make sure that the checkbox next to "Restrict access to the
# following" is *not* checked. Then select "Tab-separated values
# (.tsv)" from the second dropdown menu and copy the long URL that is
# shown in the text box under that menu. It looks like
# "https://docs.google.com/spreadsheets/d/.../pub?output=tsv".
#TRACKS="https://docs.google.com/spreadsheets/d/e/2PACX-1vTm2XjfBK0AeekT1xnT-x-acsV73JOZ9aVTvjcENrihNuQK10ghmUhAM_0k1NgqX_atuaN9B3kcMjal/pub?output=tsv"
TRACKS=

# The default directory with checked-out copies of all relevant files from CVS.
# DIR=$HOME/WWW/2025/11/TPAC
DIR=.

# Extra events that aren't in $CALENDAR:
EXTRAEVENTS=extra-events.ics

# The files to update.
SCHEDULES=(schedule.html coming-up.html)

# The files in SCHEDULES are allowed to grow by at most 50% or shrink
# by at most 30% (without the -f option). If the limits for some files
# should be different, set them in the MAXSHRINK and MAXGROW
# associative arrays.
declare -A MAXSHRINK=(coming-up.html 90)
declare -A MAXGROW=(breakouts.html 200  coming-up.html 650)

# The Awk script that puts the calendars in the pages + its helpers.
AWKSCRIPT=ics-to-html.awk
AWKLIBS=(markdown.awk htmlmathml.awk)

# A file that assigns group names to industry tracks. Uncomment this
# to label events with tracks whenever the event involves a group that
# belongs to an "ecosystem" named like the track. (Ecosystems are
# pages on the W3C web site.) See documentation inside the file for
# how the file is made.
#ECOSYSTEMS=ecosystems.txt
ECOSYSTEMS=

# An optional text to include as a comment in the generated HTML.
COMMENT="Generated by $0 running as user ${LOGNAME:-$USER} on ${HOSTNAME:-$(hostname)}"

# An optional HTML fragment to include if a calendar is empty. Each
# page can have its own text.
declare -A EMPTYTEXT=(
  schedule.html "<span>(under construction)</span>"
  coming-up.html "<span>(no events)</span>")

# The HTML pages are assumed to be in US English and encoded in UTF-8.
# When the Awk program outputs dates and times, they must be in US
# English, too.
export LC_CTYPE=en_US.UTF-8
export LC_TIME=en_US.UTF-8

trap 'cat $LOG; rm -f $LOG $ICS $HTML $EXTRACATEGORIES' 0
LOG=$(mktemp /tmp/ics-to-html-log-XXXXXX) || exit 1
ICS=$(mktemp /tmp/ics-to-html-ics-XXXXXX) || exit 1
HTML=$(mktemp /tmp/ics-to-html-html-XXXXXX) || exit 1
EXTRACATEGORIES=$(mktemp /tmp/ics-to-html-html-XXXXXX) || exit 1

# Command line options.
do_cvs=true
dry_run=false
force=false
while getopts "cd:fi:n" opt; do
  case $opt in
    c) do_cvs=false;;
    d) DIR=$OPTARG;;
    f) force=true;;
    i) CALENDAR=$OPTARG;;
    n) dry_run=true; do_cvs=false;;
    ?) exit 1;;
  esac
done

# Capture error messages
exec 2>$LOG

# Define functions for progress reports and errors:
# trace -- print a message, but only if we are running in a terminal
# die -- print message and exit
# error -- print a warning (if $force=true) or print an error and exit
if tty -s; then function trace { printf "%s\n" "$@"; }
else function trace { :; }; fi
function die { printf "%s\n" "$*" >&2; exit 1; }
if $force; then function error { printf "Warning: %s\n" "$*" >&2; }
else function error { printf "Error: %s\n" "$*" >&2; exit 1; }; fi

# On MacOS, the awk isn't GNU awk. But MacPorts has it, under the name gawk.
if type -t gawk >/dev/null; then awk=gawk; else awk=awk; fi

trace "- Go to $DIR"
cd $DIR

trace "- Try to load SSH keys, if they exist."
[ -d ~/.keychain ] && . ~/.keychain/$(uname -n)-sh

# trace "- Update required files from CVS."
# cvs -Q up $EXTRAEVENTS $AWKSCRIPT "${AWKLIBS[@]}" $ECOSYSTEMS \
#     "${SCHEDULES[@]}" || die "CVS update failed. Is there a CVS conflict?"

trace "- Get $CALENDAR"
curl -f -L -s "$CALENDAR" >$ICS || die "curl failed"

if [[ -z "$TRACKS" ]]; then
  touch $EXTRACATEGORIES
else
  trace "- Get tracks from Google Doc"
  curl -f -L -s "$TRACKS" >$EXTRACATEGORIES || die "curl failed"
fi

# Loop over the schedule files.
for schedule in "${SCHEDULES[@]}"; do
  trace "- Update $schedule"

  trace "  - Update from the calendar and the extras."
  h=${EMPTYTEXT[$schedule]:-}
  $awk ${COMMENT:+-v comment="$COMMENT"} ${h:+-v emptytext="$h"} -f $AWKSCRIPT \
       $ICS $EXTRAEVENTS $EXTRACATEGORIES $ECOSYSTEMS $schedule >$HTML

  trace "  - Check that the result is plausible."
  set -- $(wc -c <$schedule); oldsize=$1
  set -- $(wc -c <$HTML); newsize=$1
  shrink=${MAXSHRINK[$schedule]:-30}
  grow=${MAXGROW[$schedule]:-50}
  ((minsize = (100 - shrink) * oldsize / 100))
  ((maxsize = (100 + grow) * oldsize / 100))
  if ((newsize > maxsize)); then
    error "$schedule has grown more than $shrink%."
  fi
  if ((newsize < minsize)); then
    error "$schedule has shrunk more than $grow%."
  fi
  if ! cmp <(head $schedule) <(head $HTML); then
    error "$schedule has unexpected text at start of file:"$'\n'"$(head $HTML)"
  fi

  if ! $dry_run; then
    cat $HTML >$schedule
  fi

done

# if $do_cvs; then
#   trace "- Commit files into CVS."
#   cvs -Q com -m "Updated by $0" "${SCHEDULES[@]}"
# fi

trace "- Done"