mk.pdf-data

# Rules to extract bookmarks from the PDFs distributed by the
# Thesaurus Linguae Latinae.  Pass the directory containing the PDFs
# on the command-line as TLLDIR for TLL and as OLDFILE for OLD.
# Requires qpdf and pdftk.

.DEFAULT_GOAL := all

include mk.common

DATAFILES = \
	$(DATA)/tll-bookmarks.txt \
	$(DATA)/tll-bookmarks.idt \
	$(DATA)/tll-pdf-list.txt \
	$(DATA)/old-bookmarks.txt

$(BUILD)/index.json:
	curl -k -o $@ https://publikationen.badw.de/de/thesaurus/-/index.json

all: $(DATAFILES)

.SUFFIXES: .txt .idt

.txt.idt:
	./utils/make_index.pl < $< > $@

mkfile_path := $(abspath $(lastword $(MAKEFILE_LIST)))
mkfile_dir := $(dir $(mkfile_path))

# Former code that extracted info from bookmarks in the PDFs; doesn't work for newest files.
# $(DATA)/tll-bookmarks.txt: utils/make_pdf_bookmarks.pl
#	utils/make_pdf_bookmarks.pl -t "$(TLLDIR)" -l "$(mkfile_dir)/$(DATA)" > $(DATA)/tll-bookmarks.txt

# New code that uses the index.json file from the TLL
$(DATA)/tll-bookmarks.txt: utils/get_tll_bookmarks.pl $(BUILD)/index.json
	utils/get_tll_bookmarks.pl -t "$(TLLDIR)" -l "$(mkfile_dir)/$(DATA)" -i "$(BUILD)/index.json" > $(DATA)/tll-bookmarks.txt

$(DATA)/old-bookmarks.txt: utils/make_pdf_bookmarks.pl
	utils/make_pdf_bookmarks.pl -o "$(OLDFILE)" > $(DATA)/old-bookmarks.txt

$(DATA)/tll-bookmarks.idt: $(DATA)/tll-bookmarks.txt

$(DATA)/tll-pdf-list.txt: $(DATA)/tll-bookmarks.txt

clean:
	rm -f $(DATAFILES)