Makefile

MKDIR = mkdir -p
RM = rm -rfv
UNZIP = unzip -o
WGET = wget --progress=bar:force --no-verbose
PIP = pip3

SAXON_HE_VERSION_MAJOR = 11
SAXON_HE_VERSION_MINOR = 2
SAXON_HE_ZIP = SaxonHE$(SAXON_HE_VERSION_MAJOR)-$(SAXON_HE_VERSION_MINOR)J.zip
SAXON_HE_URL = https://netcologne.dl.sourceforge.net/project/saxon/Saxon-HE/$(SAXON_HE_VERSION_MAJOR)/Java/$(SAXON_HE_ZIP)
SAXON_HE_JAR = saxon-he-$(SAXON_HE_VERSION_MAJOR).$(SAXON_HE_VERSION_MINOR).jar

PAGE_SCHEMA_REPO = page-schema
PAGE_SCHEMA_VERSIONS = 2009-03-16 2010-01-12 2010-03-19 2013-07-15 2016-07-15 2017-07-15 2018-07-15 2019-07-15
PAGE_SCHEMA_BASE_URL = https://raw.githubusercontent.com/PRImA-Research-Lab/PAGE-XML/master/PAGE-release/gts/pagecontent

ABBYY_SCHEMA_REPO = abbyy-schema
ABBYY_SCHEMA_BASE_URL = https://fr7.abbyy.com/FineReader_xml/FineReader
ABBYY_SCHEMA_VERSIONS = 6-schema-v1 8-schema-v2 9-schema-v1 10-schema-v1

ALTO2PAGE_VERSION_MAJOR_MINOR = 1.5
ALTO2PAGE_VERSION = $(ALTO2PAGE_VERSION_MAJOR_MINOR).06
ALTO2PAGE_ZIP = JPageConverter.zip
ALTO2PAGE_URL = https://github.com/UB-Mannheim/prima-page-converter/releases/download/$(ALTO2PAGE_VERSION)/JPageConverter_$(ALTO2PAGE_VERSION).zip
ALTO2PAGE_DIR = JPageConverter

# {{{
# SAXON_BROWSER_VERSION = 1.1
# SAXON_BROWSER_ZIP = Saxon-CE_$(SAXON_BROWSER_VERSION).zip
# SAXON_BROWSER_JS =  TODO
# SAXON_BROWSER_URL = http://www.saxonica.com/ce/download/$(SAXON_BROWSER_ZIP)

# $(SAXON_BROWSER_JS): $(SAXON_BROWSER_ZIP)

# $(SAXON_BROWSER_ZIP):
#     wget -O '$@' '$(SAXON_BROWSER_URL)'
#}}}

.PHONY: all check $(PAGE_SCHEMA_REPO) $(ABBYY_SCHEMA_REPO) gcv2hocr page-to-alto textract2page

all:\
	$(PAGE_SCHEMA_REPO)\
	$(ABBYY_SCHEMA_REPO)\
	gcv2hocr \
	saxon.jar \
	$(ALTO2PAGE_DIR) \
	page-to-alto \
	textract2page

clean:
	$(RM) $(SAXON_HE_JAR) saxon.jar
	$(RM) $(SAXON_HE_ZIP)
	$(RM) $(PAGE_SCHEMA_REPO)
	$(RM) $(ALTO2PAGE_DIR)
	$(RM) $(ALTO2PAGE_ZIP)

check:
	@which wget >/dev/null || (echo "Missing wget. Please install package wget." && exit 1)
	@which unzip >/dev/null || (echo "Missing unzip. Please install package unzip." && exit 1)

$(ABBYY_SCHEMA_REPO):
	@$(MKDIR) "$@" && cd "$@" && \
		for version in $(ABBYY_SCHEMA_VERSIONS);do \
			xsd=abbyy-$$version.xsd; if [ ! -e $$xsd ];then \
				$(WGET) -O $$xsd $(ABBYY_SCHEMA_BASE_URL)$$version.xml; \
			fi; \
		done;

$(PAGE_SCHEMA_REPO):
	@$(MKDIR) "$@" && cd "$@" && \
		for version in $(PAGE_SCHEMA_VERSIONS);do \
			xsd=page-$$version.xsd; if [ ! -e $$xsd ];then \
				$(WGET) -O $$xsd $(PAGE_SCHEMA_BASE_URL)/$$version/pagecontent.xsd; \
			fi; \
		done;

saxon.jar: $(SAXON_HE_JAR)
	ln -sf "$<" "$@"

$(SAXON_HE_JAR): $(SAXON_HE_ZIP)
	$(UNZIP) "$<"

$(SAXON_HE_ZIP):
	$(WGET) -O "$@" "$(SAXON_HE_URL)"

gcv2hocr:
	$(MAKE) -C $@

$(ALTO2PAGE_ZIP):
	$(WGET) -O "$@" "$(ALTO2PAGE_URL)"

$(ALTO2PAGE_DIR): $(ALTO2PAGE_ZIP)
	$(UNZIP) "$<"
	rm -rf "$@"
	mv "JPageConverter $(ALTO2PAGE_VERSION)" "$@"

page-to-alto:
	cd "$@"; $(PIP) install .

textract2page:
	cd "$@"; $(PIP) install .