X-Git-Url: http://gitweb.michael.orlitzky.com/?p=dead%2Fcensus-tools.git;a=blobdiff_plain;f=makefile;h=2631b6b11a8ed60dde2a492ca0f2db83bb2879b4;hp=829bbb0c25cb693a59c4c429c90297dcb2c1841f;hb=HEAD;hpb=e40c44272f87ff1025b0466b4cb17050aede45e0 diff --git a/makefile b/makefile index 829bbb0..2631b6b 100644 --- a/makefile +++ b/makefile @@ -1,84 +1,185 @@ -DB_NAME='census2000' -DB_USER='postgres' -TIGER_DATA_URL='http://www2.census.gov/cgi-bin/shapefiles/multi-file-download?files=24_MARYLAND%2Ftl_2008_24_tabblock00.zip' -TIGER_SRID='4269' +DB_NAME=census +DB_USER=postgres +TIGER_SRID=4269 + + +# 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. -# For now, we're only dealing with the Census 2000 Maryland Block -# data, so the filenames are hard-coded. Easy enough to change. -# data: - mkdir -p data/census-2000-block/maryland/ - if [ ! -f data/census-2000-block/maryland/tl_2008_24_tabblock00.shp ]; then \ - wget -O tiger.zip $(TIGER_DATA_URL); \ - unzip tiger.zip; \ - rm tiger.zip; \ - unzip srv/ftp/geo/tiger/TIGER2008/24_MARYLAND/tl_2008_24_tabblock00.zip \ - -d ./data/census-2000-block/maryland/; \ - rm -rf srv; \ - fi; - - -# This task does a couple of things. First, it drops and re-creates -# the DB_NAME database (or schema, whatever). Then, it adds PL/pgSQL -# support to the database. + bin/download_data + + +# 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. # -# At that point, we import the two PostGIS files, lwpostgis.sql and -# spatial_ref_sys.sql. The lwpostgis.sql file contains the geometry -# functions, while spatial_ref_sys.sql contains a table of SRIDs, and -# their associated properties. PostGIS requires both. +# 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 \ + $(SHP2PGSQL) \ + -a \ + -s $(TIGER_SRID) \ + -D \ + $$state/blocks/*.shp \ + tiger_blocks \ + | psql -U $(DB_USER) -d $(DB_NAME); \ + done; + +# Summary File 1 +# +# Run all of the geo (uf1) files through the import script. This has +# to happen after the blocks import since we impose a foreign key +# restriction. + for state in data/census2000/*; do \ + bin/sf1blocks2sql $$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 # -# Then, we import the Tiger data using shp2pgsql. The shapefiles +# Since the table and index already exist, we can utilize -a, +# and leave -I out. + 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; + + + +# This imports the Tiger data using shp2pgsql. The shapefiles # should exist, since this task depends on the "data" task, which # downloads said shapefiles. # -# Finally, we create the table for the demographic data (obtained from -# the geographic header records), and populate that table with the output -# of the sf1blocks2sql script. +# After the TIGER import is done, we use the sf1blocks2sql script to +# parse and import the geographic header record information. # -db: data - dropdb -U $(DB_USER) $(DB_NAME) - createdb -U $(DB_USER) $(DB_NAME) - createlang -U $(DB_USER) plpgsql $(DB_NAME) - - psql -d $(DB_NAME) \ - -U $(DB_USER) \ - -f /usr/share/postgresql/contrib/lwpostgis.sql +db: newdb blocks_data lines_data + # Do nothing except fulfill our prerequisites. - psql -d $(DB_NAME) \ - -U $(DB_USER) \ - -f /usr/share/postgresql/contrib/spatial_ref_sys.sql - shp2pgsql -I \ - -s $(TIGER_SRID) \ - data/census-2000-block/maryland/tl_2008_24_tabblock00.shp \ - tiger \ - | psql -U $(DB_USER) -d $(DB_NAME) - psql -d $(DB_NAME) \ - -U $(DB_USER) \ - -f sql/create-sf1_blocks-table.sql - - bin/sf1blocks2sql src/Tests/Fixtures/SummaryFile1/mdgeo.uf1 sf1_blocks \ - | psql -U postgres -d $(DB_NAME) +# First, we drop and re-create the DB_NAME database (or schema, +# whatever). Then, we add PL/pgSQL support to the database. +# +# At that point, we import the two PostGIS files, postgis.sql and +# spatial_ref_sys.sql. The postgis.sql file contains the geometry +# functions, while spatial_ref_sys.sql contains a table of SRIDs, and +# their associated properties. PostGIS requires both. +# +newdb: +# Ignore the result of dropdb when it fails. + dropdb -U $(DB_USER) $(DB_NAME) || true + createdb -U $(DB_USER) $(DB_NAME) + createlang -U $(DB_USER) plpgsql $(DB_NAME) + psql -d $(DB_NAME) \ + -U $(DB_USER) \ + -f $(POSTGIS_SQL) \ + > /dev/null + + psql -d $(DB_NAME) \ + -U $(DB_USER) \ + -f $(SPATIAL_REF_SYS_SQL) \ + > /dev/null + + +# This just runs the SQL script to create the sf1_blocks table. +sf1_blocks_table: + psql -d $(DB_NAME) \ + -U $(DB_USER) \ + -f sql/create-sf1_blocks-table.sql \ + > /dev/null + + +# Create the tiger_blocks table, and create its GiST index. Having the +# table already exist makes importing via shp2pgsql much easier. +# Any blocks file will work as an argument. +tiger_blocks_table: + $(SHP2PGSQL) \ + -p \ + -I \ + -s $(TIGER_SRID) \ + data/census2000/maryland/blocks/tl_2009_24_tabblock00.shp \ + tiger_blocks \ + | 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: + $(SHP2PGSQL) \ + -p \ + -I \ + -s $(TIGER_SRID) \ + data/census2000/maryland/lines/tl_2009_24510_edges.shp \ + tiger_lines \ + | 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