Makefile



infiles := $(wildcard data/*.ttl data/*/*.ttl)
turtlefiles := $(infiles:data/%.ttl=build/%/index.ttl)
textfiles = $(turtlefiles:%.ttl=%.txt)
rdfxmlfiles = $(turtlefiles:%.ttl=%.rdf)
# TODO: resolve event:Event with prog:Programme being foaf:primaryTopic of <>
htmlprogrammefiles = \
 build/2011/cambridge/index.html \
 build/2011/banjaluka/index.html \
 build/2011/asia/index.html \
 build/2014/europe/index.html \
 build/2014/bruxelles-05/index.html \
 build/2014/bruxelles-05-2/index.html \
 build/2014/bruxelles-07/index.html \
 build/2014/europe-08/index.html \
 build/2015/brussels-03/index.html \
 build/2015/peru/index.html \
 build/2015/heidelberg/index.html
htmlfiles = $(htmlprogrammefiles)

getbase = $(shell perl -ne '/^\@base\s+<(http.+)>/ and print $$1 and exit;' "$1")

all: $(textfiles) $(rdfxmlfiles) $(htmlfiles)

clean:
    rm -rf build

$(turtlefiles): build/%/index.ttl : data/%.ttl
    mkdir -p $(dir $@)
    cp $< $@
#   $(eval base = $(call getbase, $<))
#   rapper $(base:%=-I "%") -i turtle -o turtle "$<" > "$@"
#   perl -i -pe 's/ {4}/\t/g' "$@"

$(textfiles): %.txt : %.ttl
    ln -fs $(notdir $<) $@

$(rdfxmlfiles): %.rdf : %.ttl
    $(eval base = $(call getbase, $<))
    rapper $(base:%=-I "%") -i turtle -o rdfxml-abbrev "$<" > "$@"

# Fetch and tidy programme rendering
#  * add doctype and title to validate
#  * add charset to not rely on server hinting
#  * refer to plaintext serialization (not self) of dataset
$(htmlprogrammefiles): %.html : %.ttl
    $(eval base = $(call getbase, $<))
    wget -O"$@" "http://programme.ecs.soton.ac.uk/tool.php?src=$(base)"
    perl -pi -0 \
        -e 's,^(.*?<h1[^>]*>)(.*?)(?=</h1>),<!DOCTYPE html>\n<meta charset="utf-8">\n<title>$$2</title>$$1$$2,ms;' \
        -e 's,$(base)\K,/index.txt,;' \
        -e "s,class='[^'] *\Kprogramme_current',',;" \
        -e "s,<div class='programme_now_message'>now</div>,,g;" \
        "$@"