### Makefile --- Oak Park IL address grid
## Author: Dan Jacobson https://www.jidanni.org/
## Copyright: https://www.gnu.org/licenses/gpl.html
## Permalink: https://www.jidanni.org/geo/house_numbering/grids/us/il/cook/oak_park/Makefile
## Created: 2024-05-29T21:45:05+0000
## Last-Updated: 2024-06-12T09:57:06+0000
##     Update #: 505

# Welcome to the address grid project. I will be your guide today.
# My name is Nerdsome.

# Today we will see how, by "simply" entering a few parameters into
# this Makefile, we'll have your grid on the screen in "no time".

# Grid? See
# https://www.jidanni.org/geo/house_numbering/grids/define.html ,
# where we "define" the "problem".

# Actually for places like this we would usually use a PLSS based
# setup. But this is a demo for a future user interface where the user clicks
# on the map, and the interface fills out this Makefile, and the runs it.

# OK, let's get started.

# What project name file stem should we use?
fs=oak_park_il
# What shall we use for these KML/KMZ tags? They'll be embedded in the
# file you'll show your grandkids.

DO = -dsco NAME="Oak Park IL USA address grid (north sector), by \
Dan Jacobson" -dsco DESCRIPTION="Quirk: 100S\
is across the street from 100N. I.e., two tiers of city blocks\
swallowed up by a black hole. We will just map the part of town\
north of the hole today.\
\
One would think we could just make a translator algorithm but then\
there also is this long 1100 S block...\
\
One might think Oh, we'll just use Chicago's system and translate from\
that. But hey, Chicago also has a long 0-1200S mile that should have\
been 0-800S.\
"

# That's correct. Just for "spite" today we chose a town where they
# folded out those two "00 blocks". Gone into a black hole. So we must
# deal with it. I bet those "village mothers and fathers" thought they
# were "real smart" back a hundred years ago.

# We will try to squeeze it out, with fancy math. No we will absolutely
# not. We will instead make two sectors. I.e., two projects here
# instead of one. And in fact do just one sector today.

# What are the four edges of the rectangle of grids you want?
# And remember, e.g., 500S and 500W are written -500.
N=1400
E=100
S=0
W=-1300

# (By the way, don't "freak out" that on boundary streets, Oak Park uses
# the system on the other side, instead of their own.)

# And how many numbers per grid (block, cell)?
NPG=100

# And where, may I ask, are your Ground Control Points?
GCPs=\
-gcp -1200  1300 -87.80576 41.90887\
-gcp  -800   100 -87.79448 41.88709\
-gcp     0  1300 -87.77559 41.90925

# Here's the southern corners of the Village
#-gcp -1200 -1200 -87.80422 41.86502
#-gcp     0 -1200 -87.77416 41.86549
#usable on day after we figure out how we want to deal with quirks.

# that's X and Y of an address point, followed by Longitude, Latitude.
# Need at least three. Maybe just two next year with new GDAL.

# The final ones fave fake negative Y values...

# Oh yeah. Forgot to tell you to be sure you have installed the
# gdal-bin (and viking) packages, then do "make all.kmz". Then do
# "viking all.kmz", or use your own favorite KML/KMZ viewer, like
# Google Earth (Pro).

# The following is just internal stuff. Ho hum. No need to read it.

# Never mind this special Dan J. junk:
dd=$(HOME)/jidanni.org/geo/house_numbering/grids/us/il/cook/oak_park
H=ssh $(remote) viking $(dd)/$<
remote=n3
target=all.vik
f: clean
	rm all.kmz||:
	if pidof -q viking; then killall --wait --signal SIGHUP viking; fi
	case `uname --nodename` in dan1) false;; esac
	ssh dan1 make -C $$PWD $(target)
%.vik:%.kmz
	cd /tmp && nohup ssh $(remote) viking --map 13 -e $(dd)/$? & sleep 11
# End of special Dan J. junk.

# Here are the order of operations, backwards, naturally:

publish:all.kmz; mv $< $(fs).kmz #yes, when you're satisfied, run "make publish"
all.kmz: $(addsuffix .kml, $(addsuffix .lines,0 1) $(addsuffix .labels,0 1))
	ogrmerge -f LIBKML -overwrite_ds -o $@ $^ $(DO) #Yes, get warnings as of 2024.
%.labels.kml:%.labels.csv
	ogr2ogr -order 1 -f LIBKML $@ $< \
	-oo X_POSSIBLE_NAMES=field_1 \
        -oo Y_POSSIBLE_NAMES=field_2 \
        -sql 'SELECT Name FROM "$(basename $<)"'
%.labels.csv:%.lines.kml
# We use the nodes that the contour operation gave us as convenient places for labels:
	ogr2ogr -explodecollections -dialect SQLITE -lco GEOMETRY=AS_WKT -sql \
	'SELECT Name AS Name, ST_DissolvePoints(GEOMETRY) AS WKT FROM "$*"' $@ $<
%.lines.kml:%.csv #Apply the Ground Control Points!:
	ogr2ogr -f LIBKML $@ $< $(GCPs) -sql 'SELECT Name FROM "$*"'
%.csv:%.raw.csv #Here we hack the hell out of the lines the contour operation made.
# Indeed our Makefile is one big hack. Real sorry about that.
	sed \$$d $<|perl -F'\(|\)' -anwle \
	'if($$#F){@M=split /,/, $$F[1]; $$N=join "," ,$(\
	)@M[1..$$#M-1]; print "$$F[0]($$N)$$F[2]";}else{print}'|\
	perl -F, -apwle '$(nwse)' > $@
# In Oak Park there is no E / W differentiation:
	perl -pwle 's/W$$//' $@ > $@.tmp && mv $@.tmp $@
%.raw.csv:%.f2.xyz #drape contours over the wedge
	gdal_contour -i $(NPG) -q -a Name $< $@ -lco GEOMETRY=AS_WKT
%.f2.xyz:f2.xy #Here's where we fork, making one "wedge" for each axis
	perl -anwle '$$F[2]=$$F[$*]; print "@F"' $< > $@
f2.xy: #It's our very own version of "gdal_create -outsize 2 2"!:
	echo $W $N,$E $N,$W $S,$E $S | tr , \\n > $@
clean:; for i in xy xyz csv kml; do find -maxdepth 1 -name \*.$$i -delete; done
# A function to make e.g., -500 into 500W or 500S:
nwse=\
next if /WKT/;\
$$F[-1] =\
  abs( $$F[-1] ) . ( $* ? ( ( $$F[-1] < 0 ) ? "S" : "N" ) : ( ( $$F[-1] < 0 ) ? "W" : "E" ) );\
$$_ = join ",", @F;
.SECONDARY:

# Discussion: So what have we done? We have discovered that we can
# model addressing grids _one axis at a time_, as the slope on a
# wedge, and drape contour lines on it!

# Yes, we will still have to determine why it's not perfect here in
# our implementation for Oak Park.

# First of course we note the streets themselves aren't too regularly
# spaced. Can't blame us for that one.

# Then what about geographic datums, etc.? Seems like we haven't
# figured them in yet.

# Then of course note we have PLSS-based solutions for this are that
# we will also try soon.