X-Git-Url: http://gitweb.michael.orlitzky.com/?p=dead%2Fcensus-tools.git;a=blobdiff_plain;f=makefile;h=2631b6b11a8ed60dde2a492ca0f2db83bb2879b4;hp=30e876aba03cc57a1f7fa69e426425cd8dc87620;hb=HEAD;hpb=4da15c8385b990b4f51842debe2ebc4595d22a00 diff --git a/makefile b/makefile index 30e876a..2631b6b 100644 --- a/makefile +++ b/makefile @@ -2,144 +2,114 @@ DB_NAME=census DB_USER=postgres TIGER_SRID=4269 -# Root folder for the shapefiles. -TIGER_ROOT=http://www2.census.gov/geo/tiger/TIGER2009 - -# State-specific folders. -DC_ROOT=$(TIGER_ROOT)/11_DISTRICT_OF_COLUMBIA -MD_ROOT=$(TIGER_ROOT)/24_MARYLAND -VA_ROOT=$(TIGER_ROOT)/51_VIRGINIA -PA_ROOT=$(TIGER_ROOT)/42_PENNSYLVANIA -NY_ROOT=$(TIGER_ROOT)/36_NEW_YORK - -# URLs for the TIGER/Line block-level shapefiles. -DC_BLOCKS_URL=$(DC_ROOT)/tl_2009_11_tabblock00.zip -MD_BLOCKS_URL=$(MD_ROOT)/tl_2009_24_tabblock00.zip -VA_BLOCKS_URL=$(VA_ROOT)/tl_2009_51_tabblock00.zip -PA_BLOCKS_URL=$(PA_ROOT)/tl_2009_42_tabblock00.zip -NY_BLOCKS_URL=$(NY_ROOT)/tl_2009_36_tabblock00.zip - -# Starting with PostGIS 1.4.0, these paths are calculated at install -# time using the pg_config utility. Rather than try to guess where -# PostGIS will wind up installed, we can just check the output of -# pg_config ourselves. -PG_BINDIR=`pg_config --bindir` -PG_SHAREDIR=`pg_config --sharedir` + +# Dark magic. We set these makefile variables to be the result of the +# 'shell' function. The shell function, in turn, executes a Python +# script which determines the locations of these files. +SHP2PGSQL := $(shell bin/find_file_paths --root /usr --single shp2pgsql) +POSTGIS_SQL := $(shell bin/find_file_paths --root /usr lwpostgis.sql postgis.sql) +SPATIAL_REF_SYS_SQL := $(shell bin/find_file_paths --root /usr spatial_ref_sys.sql) # Necessary to run test/data without prerequisites. # -.PHONY : test data +.PHONY : test data lib # The default task, since it comes first in the list. # -all: clean test +all: clean lib test test: ./bin/run_tests +# Download or check out any third-party libraries. +lib: + make -C lib/ + + # Remove byte-compiled python code. # clean: find ./ -name '*.pyc' -print0 | xargs -0 rm -f -# Download the shapefiles from Tiger if they don't already exist. -data: tiger_blocks tiger_lines - -tiger_blocks: dc_blocks md_blocks va_blocks pa_blocks ny_blocks - -dc_blocks: - mkdir -p data/census2000/dc/block - if [ ! -f data/census2000/dc/block/tl_2009_11_tabblock00.shp ]; \ - then \ - wget -O dcblocks.zip $(DC_BLOCKS_URL); \ - unzip dcblocks.zip -d ./data/census2000/dc/block; \ - rm dcblocks.zip; \ - fi; - -md_blocks: - mkdir -p data/census2000/maryland/block - if [ ! -f data/census2000/maryland/block/tl_2009_24_tabblock00.shp ]; \ - then \ - wget -O mdblocks.zip $(MD_BLOCKS_URL); \ - unzip mdblocks.zip -d ./data/census2000/maryland/block; \ - rm mdblocks.zip; \ - fi; - -va_blocks: - mkdir -p data/census2000/virginia/block - if [ ! -f data/census2000/virginia/block/tl_2009_51_tabblock00.shp ]; \ - then \ - wget -O vablocks.zip $(VA_BLOCKS_URL); \ - unzip vablocks.zip -d ./data/census2000/virginia/block; \ - rm vablocks.zip; \ - fi; - -pa_blocks: - mkdir -p data/census2000/pennsylvania/block - if [ ! -f data/census2000/pennsylvania/block/tl_2009_42_tabblock00.shp ]; \ - then \ - wget -O pablocks.zip $(PA_BLOCKS_URL); \ - unzip pablocks.zip -d ./data/census2000/pennsylvania/block; \ - rm pablocks.zip; \ - fi; - -ny_blocks: - mkdir -p data/census2000/new_york/block - if [ ! -f data/census2000/new_york/block/tl_2009_36_tabblock00.shp ]; \ - then \ - wget -O nyblocks.zip $(NY_BLOCKS_URL); \ - unzip nyblocks.zip -d ./data/census2000/new_york/block; \ - rm nyblocks.zip; \ - fi; - - -tiger_lines: +data: bin/download_data -# This imports the Tiger data using shp2pgsql. The shapefiles -# should exist, since this task depends on the "data" task, which -# downloads said shapefiles. +# There is a small issue here with the blocks_db and lines_db +# targets. Each of these requires that the database exists, and might +# therefore depend on the newdb target. However, if /each/ of them +# depends on newdb, the database will be dropped twice and the data +# from one of {blocks, lines} would be lost. # -# After the TIGER import is done, we use the sf1blocks2sql script to -# parse and import the geographic header record information. -# -db: data newdb tiger_blocks_table tiger_lines_table sf1_blocks_table +# We therefore assume that the database already exists when blocks_db +# or lines_db are initiated. +blocks_db: data blocks_table # All Blocks # # The table already exists, so we can append to it, and we don't have # to create the GiST index. for state in data/census2000/*; do \ - $(PG_BINDIR)/shp2pgsql \ + $(SHP2PGSQL) \ -a \ -s $(TIGER_SRID) \ -D \ - $$state/block/*.shp \ + $$state/blocks/*.shp \ tiger_blocks \ | psql -U $(DB_USER) -d $(DB_NAME); \ done; -# MD Lines +# Summary File 1 +# +# Run all of the geo (uf1) files through the import script. This has +# to happen after the blocks import since we impose a foreign key +# restriction. + for state in data/census2000/*; do \ + bin/sf1blocks2sql $$state/sf1/*.uf1 sf1_blocks \ + | psql -U $(DB_USER) -d $(DB_NAME) \ + > /dev/null; \ + done; + +# Run the query to combine the two blocks tables, and drop the +# constituents. + psql -U $(DB_USER) \ + -d $(DB_NAME) \ + -f sql/combine-block-tables.sql + + +lines_db: data tiger_lines_table +# All Lines # # Since the table and index already exist, we can utilize -a, # and leave -I out. - for x in data/census2000/maryland/lines/*.shp; do \ - $(PG_BINDIR)/shp2pgsql \ - -a \ - -s $(TIGER_SRID) \ - -D \ - $$x \ - tiger_lines \ - | psql -U $(DB_USER) -d $(DB_NAME); \ + for state in data/census2000/*; do \ + for shapefile in $$state/lines/*.shp; do \ + echo "Importing $$shapefile."; \ + $(SHP2PGSQL) \ + -a \ + -s $(TIGER_SRID) \ + $$shapefile \ + tiger_lines \ + | bin/filter-transactions \ + | psql -U $(DB_USER) -d $(DB_NAME) \ + > /dev/null; \ + done; \ done; - bin/sf1blocks2sql src/Tests/Fixtures/SummaryFile1/mdgeo.uf1 sf1_blocks \ - | psql -U postgres -d $(DB_NAME) \ - > /dev/null + + +# This imports the Tiger data using shp2pgsql. The shapefiles +# should exist, since this task depends on the "data" task, which +# downloads said shapefiles. +# +# After the TIGER import is done, we use the sf1blocks2sql script to +# parse and import the geographic header record information. +# +db: newdb blocks_data lines_data + # Do nothing except fulfill our prerequisites. @@ -157,14 +127,14 @@ newdb: createdb -U $(DB_USER) $(DB_NAME) createlang -U $(DB_USER) plpgsql $(DB_NAME) - psql -d $(DB_NAME) \ - -U $(DB_USER) \ - -f $(PG_SHAREDIR)/contrib/postgis.sql \ + psql -d $(DB_NAME) \ + -U $(DB_USER) \ + -f $(POSTGIS_SQL) \ > /dev/null - psql -d $(DB_NAME) \ - -U $(DB_USER) \ - -f $(PG_SHAREDIR)/contrib/spatial_ref_sys.sql \ + psql -d $(DB_NAME) \ + -U $(DB_USER) \ + -f $(SPATIAL_REF_SYS_SQL) \ > /dev/null @@ -180,24 +150,36 @@ sf1_blocks_table: # table already exist makes importing via shp2pgsql much easier. # Any blocks file will work as an argument. tiger_blocks_table: - $(PG_BINDIR)/shp2pgsql \ + $(SHP2PGSQL) \ -p \ -I \ -s $(TIGER_SRID) \ - data/census2000/maryland/block/tl_2009_24_tabblock00.shp \ + data/census2000/maryland/blocks/tl_2009_24_tabblock00.shp \ tiger_blocks \ - | psql -U postgres -d $(DB_NAME) \ + | psql -U $(DB_USER) -d $(DB_NAME) \ > /dev/null +# Create the "blocks" table, which is the result of joining +# the tiger_blocks and sf1_blocks tables. +blocks_table: tiger_blocks_table sf1_blocks_table + psql -U $(DB_USER) \ + -d $(DB_NAME) \ + -f sql/create-blocks-table.sql + # Prepare the tiger_lines table, and create the GiST index on its # geometry column. Any lines shapefile will do here. tiger_lines_table: - $(PG_BINDIR)/shp2pgsql \ + $(SHP2PGSQL) \ -p \ -I \ -s $(TIGER_SRID) \ data/census2000/maryland/lines/tl_2009_24510_edges.shp \ tiger_lines \ - | psql -U postgres -d $(DB_NAME) \ + | psql -U $(DB_USER) -d $(DB_NAME) \ > /dev/null + +# Add a unique index on the "tlid" column. + psql -U $(DB_USER) \ + -d $(DB_NAME) \ + -f sql/create_tlid_unique_index.sql