From d0c663d90e8d47abfa26a3be4d70559da7c50373 Mon Sep 17 00:00:00 2001
From: Jochen Topf <jochen@topf.org>
Date: Thu, 3 Jul 2025 15:01:18 +0200
Subject: [PATCH] Add core/extract topic using the new locator feature of
 osm2pgsql

This allows filtering all data by bounding box or polygon.
---
 themes/core/README.md          | 33 +++++++++++++++++
 themes/core/topics/extract.lua | 66 ++++++++++++++++++++++++++++++++++
 2 files changed, 99 insertions(+)
 create mode 100644 themes/core/topics/extract.lua

diff --git a/themes/core/README.md b/themes/core/README.md
index 0f8811c..ec94c53 100644
--- a/themes/core/README.md
+++ b/themes/core/README.md
@@ -54,6 +54,39 @@ If a valid elevation tag is found, the attributes `core.ele_m` and
 `core.ele_ft` are added and contain the elevation in meters and feet,
 respectively, as a number, rounded to the nearest integer.
 
+## Topic 'extract'
+
+### Description
+
+Restrict imported data to a bounding box or polygon using the Locator feature
+of osm2pgsql. Use this topic *before* any other topics you want to restrict.
+Objects intersecting the locator area are kept, all others are discarded. For
+closed ways the area of the polygon is used for intersection, for unclosed ways
+the linestring. For relations that can be turned into multipolygons, the
+multipolygon is used, otherwise all node and way members of the polygon are
+checked as point and linestring features, respectively.
+
+Geometries are *not* clipped to the locator area.
+
+This only works in osm2pgsql version 2.2.0 and above.
+
+### Tables
+
+This topic doesn't add any output tables.
+
+### Configuration
+
+* `locator`: Set to an existing osm2pgsql.Locator (optional, if not set, a new
+  locator will be created).
+* `bbox`: A table with the bounding box to be added to the locator. The table
+  must contain four elements, in that order: min lon, min lat, max lon, max
+  lat.
+
+### Attributes
+
+No attributes are added.
+
+
 ## Topic 'layer'
 
 ### Description
diff --git a/themes/core/topics/extract.lua b/themes/core/topics/extract.lua
new file mode 100644
index 0000000..a3a3ec8
--- /dev/null
+++ b/themes/core/topics/extract.lua
@@ -0,0 +1,66 @@
+-- ---------------------------------------------------------------------------
+--
+-- Theme: core
+-- Topic: extract
+--
+-- ---------------------------------------------------------------------------
+
+local themepark, theme, cfg = ...
+
+-- ---------------------------------------------------------------------------
+
+local locator
+
+if cfg.locator == nil then
+    locator = osm2pgsql.define_locator({
+        name = 'themepark-core-extract-' .. math.random(100000000)
+    })
+else
+    locator = cfg.locator
+end
+
+if cfg.bbox ~= nil then
+    local b = cfg.bbox
+    locator:add_bbox('inside', b[1], b[2], b[3], b[4])
+end
+
+-- ---------------------------------------------------------------------------
+
+themepark:add_proc('node', function(object)
+    if not locator:first_intersecting(object:as_point()) then
+        return 'stop'
+    end
+end)
+
+themepark:add_proc('way', function(object)
+    local geom
+    if object.is_closed then
+        geom = object:as_polygon()
+    end
+
+    if geom == nil or geom:is_null() then
+        geom = object:as_linestring()
+    end
+
+    if not locator:first_intersecting(geom) then
+        return 'stop'
+    end
+end)
+
+themepark:add_proc('relation', function(object)
+    local geom = object:as_multipolygon()
+
+    if geom:is_null() then
+        geom = object:as_geometrycollection()
+    end
+
+    if geom:is_null() then
+        return 'stop'
+    end
+
+    if not locator:first_intersecting(geom) then
+        return 'stop'
+    end
+end)
+
+-- ---------------------------------------------------------------------------