From eb32a3f06e45cfeeb1be552664723ba46944dc4b Mon Sep 17 00:00:00 2001
From: Hector Castro <43639+hectcastro@users.noreply.github.com>
Date: Tue, 2 Feb 2021 16:09:39 -0500
Subject: [PATCH] Add Docker configuration to process OTP data files (#1281)

* Add Docker configuration to process OTP data files

* Add curl
---
 deployment/dataproc/Dockerfile           |  8 ++++++++
 deployment/dataproc/combine-and-clip.sh  | 20 ++++++++++++++++++++
 deployment/dataproc/docker-compose.yml   |  8 ++++++++
 deployment/dataproc/download-osm-data.sh |  8 ++++++++
 4 files changed, 44 insertions(+)
 create mode 100644 deployment/dataproc/Dockerfile
 create mode 100755 deployment/dataproc/combine-and-clip.sh
 create mode 100644 deployment/dataproc/docker-compose.yml
 create mode 100755 deployment/dataproc/download-osm-data.sh

diff --git a/deployment/dataproc/Dockerfile b/deployment/dataproc/Dockerfile
new file mode 100644
index 000000000..20fc8bade
--- /dev/null
+++ b/deployment/dataproc/Dockerfile
@@ -0,0 +1,8 @@
+FROM ubuntu:20.04
+
+RUN mkdir -p /usr/local/src
+WORKDIR /usr/local/src
+
+RUN apt update && apt install -y curl osmctools
+
+COPY . /usr/local/src
diff --git a/deployment/dataproc/combine-and-clip.sh b/deployment/dataproc/combine-and-clip.sh
new file mode 100755
index 000000000..8d088b69a
--- /dev/null
+++ b/deployment/dataproc/combine-and-clip.sh
@@ -0,0 +1,20 @@
+#!/bin/bash
+
+BOUNDS=-76.209582,38.441753,-74.243725,40.725449
+
+declare -a STATES=("pennsylvania" "new-jersey" "delaware")
+
+for STATE in "${STATES[@]}"; do
+    echo "extracting OSM data for the region served by CAC for ${STATE}..."
+    osmconvert ${STATE}.pbf -b=${BOUNDS} \
+        --complete-ways --complex-ways --complete-boundaries --complete-multipolygons \
+        -o=${STATE}.o5m
+done
+
+echo "combining state region extracts..."
+
+INPUTS=$(printf "%s.o5m " "${STATES[@]}")
+
+osmconvert ${INPUTS} -o=cac.pbf
+
+echo "all done!"
diff --git a/deployment/dataproc/docker-compose.yml b/deployment/dataproc/docker-compose.yml
new file mode 100644
index 000000000..ea7acca97
--- /dev/null
+++ b/deployment/dataproc/docker-compose.yml
@@ -0,0 +1,8 @@
+version: "3"
+services:
+  osm:
+    build:
+      context: .
+      dockerfile: Dockerfile
+    volumes:
+      - ./:/usr/local/src
diff --git a/deployment/dataproc/download-osm-data.sh b/deployment/dataproc/download-osm-data.sh
new file mode 100755
index 000000000..f7be1d8b5
--- /dev/null
+++ b/deployment/dataproc/download-osm-data.sh
@@ -0,0 +1,8 @@
+#!/bin/bash
+
+declare -a STATES=("pennsylvania" "new-jersey" "delaware")
+
+for STATE in "${STATES[@]}"; do
+    echo "downloading OSM data for ${STATE}..."
+    curl https://download.geofabrik.de/north-america/us/${STATE}-latest.osm.pbf -o ${STATE}.pbf
+done