[build] adding Automake file in src, including rule to download data dir tarball

configure.ac

@@ -48,7 +48,7 @@ AC_CONFIG_FILES([Makefile
                  src/Makefile
                  src/sparkey/Makefile])
 
-AM_CONDITIONAL([USE_DATE_STAT], [date -r . >/dev/null 2>&1])
+AM_CONDITIONAL([HAVE_DATE_STAT], [date -r . >/dev/null 2>&1])
-AM_CONDITIONAL([USE_STAT], [stat -f %Sm . >/dev/null 2>&1])
+AM_CONDITIONAL([HAVE_STAT], [stat -f %Sm . >/dev/null 2>&1])
 
 AC_OUTPUT

src/Makefile.am

@@ -0,0 +1,95 @@
+SUBDIRS = sparkey
+
+CFLAGS_BASE = -Wfloat-equal -Wpointer-arith -Werror
+CFLAGS_O0 = $(CFLAGS_BASE) -O0
+CFLAGS_O1 = $(CFLAGS_BASE) -O1
+CFLAGS_O2 = $(CFLAGS_BASE) -O2
+CFLAGS_O3 = $(CFLAGS_BASE) -O3
+
+CFLAGS = $(CFLAGS_BASE) -DLIBPOSTAL_DATA_DIR='"$(LIBPOSTAL_DATA_DIR)"'
+
+lib_LTLIBRARIES = libpostal.la
+libpostal_la_SOURCES = libpostal.c address_dictionary.c transliterate.c tokens.c trie.c trie_search.c string_utils.c file_utils.c numex.c utf8proc/utf8proc.c cmp/cmp.c geodb.c geo_disambiguation.c normalize.c bloom.c features.c geonames.c geohash/geohash.c unicode_scripts.c msgpack_utils.c
+libpostal_la_LIBADD = libscanner.la sparkey/libsparkey.la
+libpostal_la_CFLAGS = $(CFLAGS_O2)
+
+# Scanner can take a very long time to compile with higher optimization levels, so always use -O0, scanner is fast enough
+noinst_LTLIBRARIES = libscanner.la
+libscanner_la_SOURCES = scanner.c
+libscanner_la_CFLAGS =  $(CFLAGS_O0)
+
+pkginclude_HEADERS = libpostal.h
+pkgdata_DATA = libpostal_data.tar.gz
+
+LIBPOSTAL_S3_BUCKET_NAME = libpostal
+LIBPOSTAL_S3_BUCKET_URL = http://$(LIBPOSTAL_S3_BUCKET_NAME).s3.amazonaws.com
+LIBPOSTAL_DATA_FILE = libpostal_data.tar.gz
+
+EPOCH_DATE = "Jan  1 00:00:00 1970"
+
+LIBPOSTAL_DATA_DIR = $(datadir)/libpostal/
+LIBPOSTAL_DATA_LAST_UPDATED = $(LIBPOSTAL_DATA_DIR)libpostal_data_last_updated
+
+LAST_MODIFIED = $(EPOCH_DATE)
+
+if HAVE_DATE_STAT
+USE_DATE_STAT = 1
+
+define set_last_modified
+$(eval LAST_MODIFIED := $(shell date -d "@$(shell cat $1)"))
+endef
+
+else
+if HAVE_STAT
+
+USE_STAT = 1
+
+define next_ts
+$(shell date -r $(shell stat -f %m $1) -v+1S +%s)
+endef
+
+else
+    $(error Cannot get file modification date on this platform);
+endif
+
+endif
+
+
+libpostal_data_mkdir:
+	mkdir -p $(LIBPOSTAL_DATA_DIR)
+
+libpostal_data_last_updated: libpostal_data_mkdir
+	if [ -e $(LIBPOSTAL_DATA_LAST_UPDATED) ]; then						\
+		$(call set_last_modified,$(LIBPOSTAL_DATA_LAST_UPDATED))		\
+		echo "LAST_MODIFIED=$(LAST_MODIFIED)";							\
+	else																\
+		echo "LAST_MODIFIED=$(LAST_MODIFIED)";							\
+	fi;
+
+
+libpostal_data.tar.gz: libpostal_data_mkdir libpostal_data_last_updated
+	if [ `curl $(LIBPOSTAL_S3_BUCKET_URL)/$(LIBPOSTAL_DATA_FILE) -z "$(LAST_MODIFIED)" --silent --remote-time -o $(LIBPOSTAL_DATA_DIR)$(LIBPOSTAL_DATA_FILE) -w %{http_code}` = "200" ]; then		\
+		if [ "x$(USE_DATE_STAT)" != "x" ]; then																																						\
+			echo $$(date -d"$$(date -d "@$$(date -r $(LIBPOSTAL_DATA_DIR)$(LIBPOSTAL_DATA_FILE)) +%s") + 1 second" +%s) > $(LIBPOSTAL_DATA_LAST_UPDATED);											\
+		elif [ "x$(USE_STAT)" != "x" ]; then																																						\
+			echo $$(date -r $$(stat -f %m $(LIBPOSTAL_DATA_DIR)$(LIBPOSTAL_DATA_FILE)) -v+1S +%s) > $(LIBPOSTAL_DATA_LAST_UPDATED);																	\
+		fi;																																															\
+		tar -xvzf $(LIBPOSTAL_DATA_DIR)$(LIBPOSTAL_DATA_FILE) -C $(LIBPOSTAL_DATA_LAST_UPDATED); 																									\
+	fi;
+
+address_dictionary_builder: address_dictionary_builder.c
+	$(CC) $(CFLAGS_O2) $(LDFLAGS) address_dictionary_builder.c address_dictionary.c file_utils.c string_utils.c trie.c trie_search.c utf8proc/utf8proc.c -o build_address_dictionary
+
+geodb_builder: geodb_builder.c
+	$(CC) $(CFLAGS_O2) $(LDFLAGS) geodb_builder.c geodb.c geo_disambiguation.c normalize.c bloom.c features.c geonames.c geohash/geohash.c sparkey/*.c unicode_scripts.c transliterate.c trie.c trie_search.c string_utils.c msgpack_utils.c file_utils.c utf8proc/utf8proc.c cmp/cmp.c -o build_geodb $(LIBS)
+
+transliteration_table_builder: transliteration_table_builder.c
+	$(CC) $(CFLAGS_O2) $(LDFLAGS) transliteration_table_builder.c transliterate.c trie.c trie_search.c file_utils.c string_utils.c utf8proc/utf8proc.c -o build_trans_table
+
+numex_table_builder: numex_table_builder.c numex_data.c
+	$(CC) $(CFLAGS_O2) $(LDFLAGS) numex_table_builder.c numex.c file_utils.c string_utils.c tokens.c trie.c trie_search.c utf8proc/utf8proc.c -o build_numex_table
+
+lexer: scanner.re
+	re2c -F -s -b -8 -o scanner.c scanner.re
+
+.PHONY: address_dictionary_builder geodb_builder transliteration_table_builder numex_table_builder lexer