commit 42b8056e27057500a36d6d114b41b4d90ea3f9e7
Author: Remy Moll <me@moll.re>
Date:   Mon Nov 24 20:59:33 2025 +0100

    initial working but bare setup

diff --git a/README.md b/README.md
new file mode 100644
index 0000000..d1bf00f
--- /dev/null
+++ b/README.md
@@ -0,0 +1,18 @@
+# OSM on postgres
+
+## Key components
+- a stable, completely standardized database that can easily be scaled backed by postgres + postgis
+    - this is backed a by a persistent, node-independent storage claim
+- a specialized container (running as cronjob/on demand job) that ingests the data from OSM dumps.
+    - this is backed by its own storage to prevent excessive re-downloads
+
+
+## Prerequisites
+- A storage class provisioner
+- A CNPG operator that will provision this.
+
+
+## Results
+A well defined database with a table and entries defined [base/importer.configmap.yaml](base/importer.configmap.yaml) that can be reached from within the cluster.
+- Acccess data can be read from the secret `<postgres-cluster-name>-app`.
+
diff --git a/base/importer.configmap.yaml b/base/importer.configmap.yaml
new file mode 100644
index 0000000..01e549f
--- /dev/null
+++ b/base/importer.configmap.yaml
@@ -0,0 +1,205 @@
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: importer-config
+data:
+  simple.lua: |
+    -- This config example file is released into the Public Domain.
+
+    -- This is a very simple Lua config for the Flex output not intended for
+    -- real-world use. Use it do understand the basic principles of the
+    -- configuration. After reading and understanding this, have a look at
+    -- "geometries.lua".
+
+    -- The global variable "osm2pgsql" is used to talk to the main osm2pgsql code.
+    -- You can, for instance, get the version of osm2pgsql:
+    print('osm2pgsql version: ' .. osm2pgsql.version)
+
+    -- A place to store the SQL tables we will define shortly.
+    local tables = {}
+
+    -- Create a new table called "pois" with the given columns. When running in
+    -- "create" mode, this will do the `CREATE TABLE`, when running in "append"
+    -- mode, this will only declare the table for use.
+    --
+    -- This is a "node table", it can only contain data derived from nodes and will
+    -- contain a "node_id" column (SQL type INT8) as first column. When running in
+    -- "append" mode, osm2pgsql will automatically update this table using the node
+    -- ids.
+    tables.pois = osm2pgsql.define_node_table('pois', {
+        { column = 'tags', type = 'jsonb' },
+        -- In most cases we'll need a column for the geometry. The default
+        -- projection is Web Mercator (3857), so this will result in an SQL
+        -- type `geometry(Point, 3857)`.
+        { column = 'geom', type = 'point', not_null = true },
+    })
+
+    -- A special table for restaurants to demonstrate that we can have any tables
+    -- with any columns we want.
+    tables.restaurants = osm2pgsql.define_node_table('restaurants', {
+        { column = 'name', type = 'text' },
+        { column = 'cuisine', type = 'text' },
+        -- We declare all geometry columns as "NOT NULL". If osm2pgsql encounters
+        -- an invalid geometry (for whatever reason) it will generate a null
+        -- geometry which will not be written to the database if "not_null" is
+        -- set. The result is that broken geometries will just be silently
+        -- ignored.
+        { column = 'geom', type = 'point', not_null = true },
+    })
+
+    -- This is a "way table", it can only contain data derived from ways and will
+    -- contain a "way_id" column. When running in "append" mode, osm2pgsql will
+    -- automatically update this table using the way ids.
+    tables.ways = osm2pgsql.define_way_table('ways', {
+        { column = 'tags', type = 'jsonb' },
+        { column = 'geom', type = 'linestring', not_null = true },
+    })
+
+    -- This is an "area table", it can contain data derived from ways or relations
+    -- and will contain an "area_id" column. Way ids will be stored "as is" in the
+    -- "area_id" column, for relations the negative id will be stored. When
+    -- running in "append" mode, osm2pgsql will automatically update this table
+    -- using the way/relation ids.
+    tables.polygons = osm2pgsql.define_area_table('polygons', {
+        { column = 'type', type = 'text' },
+        { column = 'tags', type = 'jsonb' },
+        -- The type of the `geom` column is `geometry`, because we need to store
+        -- polygons AND multipolygons
+        { column = 'geom', type = 'geometry', not_null = true },
+    })
+
+    -- Debug output: Show definition of tables
+    for name, dtable in pairs(tables) do
+        print("\ntable '" .. name .. "':")
+        print("  name='" .. dtable:name() .. "'")
+    end
+
+    -- Called for every node in the input. The `object` argument contains all the
+    -- attributes of the node like `id`, `version`, etc. as well as all tags as a
+    -- Lua table (`object.tags`).
+    function osm2pgsql.process_node(object)
+        if object.tags.amenity == 'restaurant' then
+            -- Add a row to the SQL table. The keys in the parameter table
+            -- correspond to the table columns, if one is missing the column will
+            -- be NULL. The id column will be filled automatically.
+            tables.restaurants:insert({
+                name = object.tags.name,
+                cuisine = object.tags.cuisine,
+                geom = object:as_point()
+            })
+        else
+            tables.pois:insert({
+                -- We know `tags` is of type `jsonb` so this will do the
+                -- right thing.
+                tags = object.tags,
+                geom = object:as_point()
+            })
+        end
+    end
+
+    -- Called for every way in the input. The `object` argument contains the same
+    -- information as with nodes and additionally a boolean `is_closed` flag and
+    -- the list of node IDs referenced by the way (`object.nodes`).
+    function osm2pgsql.process_way(object)
+        -- Very simple check to decide whether a way is a polygon or not, in a
+        -- real stylesheet we'd have to also look at the tags...
+        if object.is_closed then
+            tables.polygons:insert({
+                type = object.type,
+                tags = object.tags,
+                geom = object:as_polygon()
+            })
+        else
+            tables.ways:insert({
+                tags = object.tags,
+                geom = object:as_linestring()
+            })
+        end
+    end
+
+    -- Called for every relation in the input. The `object` argument contains the
+    -- same information as with nodes and additionally an array of members
+    -- (`object.members`).
+    function osm2pgsql.process_relation(object)
+        -- Store multipolygons and boundaries as polygons
+        if object.tags.type == 'multipolygon' or
+          object.tags.type == 'boundary' then
+            tables.polygons:insert({
+                type = object.type,
+                tags = object.tags,
+                geom = object:as_multipolygon()
+            })
+        end
+    end
+
+  anyway.lua: |
+    -- This config file mirrors our previous overpass selectors for overpass. They follow an arbitrary selection from https://taginfo.openstreetmap.org
+
+    local tables = {}
+
+    tables.landmark = osm2pgsql.define_node_table('landmark', {
+        { column = 'type', type = 'text' },
+        { column = 'name', type = 'text' },
+        { column = 'name_en', type = 'text' },
+        { column = 'description', type = 'text' },
+        { column = 'tags', type = 'jsonb' },
+        { column = 'geom', type = 'point', not_null = true },
+    })
+
+
+
+    function name_in_list(tag, list)
+        if tag == nil then
+            return false
+        end
+        for _, v in ipairs(list) do
+            if tag == v then
+                return true
+            end
+        end
+        return false
+    end
+
+    function osm2pgsql.process_node(object)
+        -- we assign landmarks a type based on their tags. so we specify a list of tags to check and assign a type accordingly
+
+
+        -- nature landmark
+        if name_in_list(object.tags.natural, {'geyser', 'hot_spring', 'arch', 'volcano', 'stone'}) or name_in_list(object.tags.tourism, {'alpine_hut', 'viewpoint', 'zoo', 'resort', 'picnic_site'}) or name_in_list(object.tags.water, {'pond', 'lake', 'river', 'basin', 'stream', 'lagoon', 'rapids'}) or name_in_list(object.tags.waterway, {'waterfall', 'river', 'canal', 'dam', 'dock', 'boatyard'}) then
+            tables.landmark:insert({
+                type = 'nature',
+                name = object.tags.name,
+                name_en = object.tags['name:en'],
+                description = object.tags.description,
+                tags = object.tags,
+                geom = object:as_point()
+            })
+            return
+        end
+
+        -- shopping landmark
+        if name_in_list(object.tags.shop, {'mall', 'department_store', 'clothes', 'shoes', 'jewelry', 'furniture', 'electronics', 'books'}) then
+            tables.landmark:insert({
+                type = 'shopping',
+                name = object.tags.name,
+                name_en = object.tags['name:en'],
+                description = object.tags.description,
+                tags = object.tags,
+                geom = object:as_point()
+            })
+            return
+        end
+
+        -- sightseeing landmark
+        if name_in_list(object.tags.tourism, {'museum', 'attraction', 'artwork', 'gallery', 'aquarium', 'castle', 'monument', 'memorial'}) or object.tags.historic ~= nil or name_in_list(object.tags.amenity, {'theatre', 'planetarium', 'library', 'place_of_worship', 'fountain', 'townhall'}) or name_in_list(object.tags.bridge, {'aqueduct', 'viaduct', 'suspension', 'boardwalk', 'cantilever', 'abandoned'}) or name_in_list(object.tags.building, {'cathedral', 'church', 'mosque', 'temple', 'synagogue', 'chapel'}) then
+            tables.landmark:insert({
+                type = 'sightseeing',
+                name = object.tags.name,
+                name_en = object.tags['name:en'],
+                description = object.tags.description,
+                tags = object.tags,
+                geom = object:as_point()
+            })
+            return
+        end
+    end
diff --git a/base/importer.cronjob.yaml b/base/importer.cronjob.yaml
new file mode 100644
index 0000000..29adac5
--- /dev/null
+++ b/base/importer.cronjob.yaml
@@ -0,0 +1,67 @@
+apiVersion: batch/v1
+kind: CronJob
+metadata:
+  name: importer
+spec:
+  schedule: '0 0 1 1 1'
+  concurrencyPolicy: Forbid
+  jobTemplate:
+    spec:
+      template:
+        spec:
+          containers:
+            - name: download
+              image: download
+              command: ["/bin/sh", "-c"]
+              args:
+                - '[ ! -f /data/latest.osm.pbf ] && curl -o /data/latest.osm.pbf http://download.geofabrik.de/europe/switzerland-251123.osm.pbf || echo "File already exists, skipping download"'
+              volumeMounts:
+                - name: data
+                  mountPath: /data
+
+
+
+            - name: importer
+              image: importer
+              args:
+                - --create
+                - --input-reader=pbf
+                - --output=flex
+                - --style=/rules/anyway.lua
+                - /data/latest.osm.pbf
+
+                # connection details are implictly loaded from env vars, not set as args
+
+              # Environment Variables
+              env:
+                - name: PGHOST
+                  value: "osm-postgresql-r"
+                - name: PGPORT
+                  value: "5432"
+                - name: PGDATABASE
+                  value: "osm"
+                - name: PGUSER
+                  valueFrom:
+                    secretKeyRef:
+                      name: osm-postgresql-app
+                      key: username
+                - name: PGPASSWORD
+                  valueFrom:
+                    secretKeyRef:
+                      name: osm-postgresql-app
+                      key: password
+              volumeMounts:
+                - name: data
+                  mountPath: /data
+                - name: rules
+                  mountPath: /rules
+
+          restartPolicy: Never
+
+          volumes:
+            - name: data
+              persistentVolumeClaim:
+                claimName: worldfile
+            - name: rules
+              configMap:
+                name: importer-config
diff --git a/base/kustomization.yaml b/base/kustomization.yaml
new file mode 100644
index 0000000..44c5a16
--- /dev/null
+++ b/base/kustomization.yaml
@@ -0,0 +1,22 @@
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+
+
+
+resources:
+  - namespace.yaml
+
+  - postgres.yaml
+
+  - worldfile.pvc.yaml
+  - importer.cronjob.yaml
+  - importer.configmap.yaml
+
+
+images:
+  - name: importer
+    newName: iboates/osm2pgsql
+    newTag: 2.1.0-nightly
+  - name: download
+    newName: curlimages/curl
+    newTag: 8.4.0
diff --git a/base/namespace.yaml b/base/namespace.yaml
new file mode 100644
index 0000000..0a074bd
--- /dev/null
+++ b/base/namespace.yaml
@@ -0,0 +1,4 @@
+apiVersion: v1
+kind: Namespace
+metadata:
+  name: placeholder
diff --git a/base/postgres.yaml b/base/postgres.yaml
new file mode 100644
index 0000000..2d26793
--- /dev/null
+++ b/base/postgres.yaml
@@ -0,0 +1,28 @@
+apiVersion: postgresql.cnpg.io/v1
+kind: Cluster
+metadata:
+  name: osm-postgresql
+spec:
+  instances: 1
+  imageName: ghcr.io/cloudnative-pg/postgis:18-3.6-system-trixie
+
+  bootstrap:
+    initdb:
+      owner: osm
+      database: osm
+
+      postInitTemplateSQL:
+        - CREATE EXTENSION postgis;
+        # - CREATE EXTENSION postgis_topology;
+        # - CREATE EXTENSION fuzzystrmatch;
+        # - CREATE EXTENSION postgis_tiger_geocoder;
+        # - CREATE EXTENSION hstore;
+
+
+  storage:
+    size: 10Gi
+    pvcTemplate:
+      accessModes:
+        - ReadWriteOnce
+      storageClassName: nfs-client
+      volumeMode: Filesystem
diff --git a/base/worldfile.pvc.yaml b/base/worldfile.pvc.yaml
new file mode 100644
index 0000000..361e095
--- /dev/null
+++ b/base/worldfile.pvc.yaml
@@ -0,0 +1,11 @@
+kind: PersistentVolumeClaim
+apiVersion: v1
+metadata:
+  name: worldfile
+spec:
+  storageClassName: "nfs-client"
+  accessModes:
+    - ReadWriteOnce
+  resources:
+    requests:
+      storage: 10Gi
diff --git a/overlays/stg/kustomization.yaml b/overlays/stg/kustomization.yaml
new file mode 100644
index 0000000..2a66422
--- /dev/null
+++ b/overlays/stg/kustomization.yaml
@@ -0,0 +1,8 @@
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+
+
+namespace: anydev-osm-stg
+
+resources:
+  - ../../base/