heuristics, whoop whoop

2 files changed, 121 insertions, 15 deletions

diff --git a/convert.zsh b/convert.zsh
index eb5bb25..c37a208 100755
--- a/convert.zsh
+++ b/convert.zsh
@@ -4,11 +4,7 @@ setopt err_exit no_unset
 
 mkdir -p svg pdf png
 
-# bbox:     left bottom right top
-
 # S. 6, 11ff
-ice1_bbox='78 45 830 198'
-#ice1_bbox='95 59 825 150' # doesn't work for 8029, 8031
 ice1_offset=11
 # Stelle 6 bis 9 der UIC-Wagennr.
 typeset -a ice1_types=(
@@ -253,7 +249,7 @@ x
 1101 # DBpdzfa # 110.A
 )
 
-function extract_wagons() {
+function extract_wagons {
 	start=$1
 	shift
 	bbox=$1
@@ -301,18 +297,38 @@ function extract_wagons() {
 	done
 }
 
+function extract_wagons_heuristic {
+	start=$1
+	shift
+
+	for i in {1..$#}; do
+		target=$@[$i]
+
+		if [[ $target == x ]]; then
+			continue
+		fi
+
+		echo "Page $(( start + i - 1 )): $target"
+		inkscape --export-filename=svg/${target}.svg tmp-$(( start + i - 1 )).pdf &> /dev/null
+		lib/export-carriage.py svg/${target}.svg png/${target}.png
+	done
+}
+
 pdfseparate Fahrzeuglexikon_2020.pdf tmp-%d.pdf
 
-extract_wagons $ice1_offset $ice1_bbox $ice1_types
-extract_wagons $ice2_offset $ice2_bbox $ice2_types
-extract_wagons $ice3_403_1_offset $ice3_403_1_bbox $ice3_403_1_types
-extract_wagons $ice3_403_2_offset $ice3_403_2_bbox $ice3_403_2_types
-extract_wagons $ice3_403_r_offset $ice3_403_r_bbox $ice3_403_r_types
-extract_wagons $ice3_406_offset   $ice3_406_bbox   $ice3_406_types
-extract_wagons $ice3_406_r_offset $ice3_406_r_bbox $ice3_406_r_types
-extract_wagons $ice3_407_offset   $ice3_407_bbox   $ice3_407_types
-extract_wagons $ice4_offset $ice4_bbox $ice4_types
-extract_wagons $ic1_offset $ic1_bbox $ic1_types
+extract_wagons_heuristic $ice1_offset        $ice1_types
+extract_wagons_heuristic $ice2_offset        $ice2_types
+extract_wagons_heuristic $ice3_403_1_offset  $ice3_403_1_types
+extract_wagons_heuristic $ice3_403_2_offset  $ice3_403_2_types
+extract_wagons_heuristic $ice3_403_r_offset  $ice3_403_r_types
+extract_wagons_heuristic $ice3_406_offset    $ice3_406_types
+extract_wagons_heuristic $ice3_406_r_offset  $ice3_406_r_types
+extract_wagons_heuristic $ice3_407_offset    $ice3_407_types
+extract_wagons_heuristic $ice4_offset        $ice4_types
+extract_wagons_heuristic $icet_411_s1_offset $icet_411_s1_types
+extract_wagons_heuristic $icet_411_s2_offset $icet_411_s2_types
+extract_wagons_heuristic $icet_415_offset    $icet_415_types
+extract_wagons_heuristic $ic1_offset         $ic1_types
 extract_wagons $ic2_bt_offset $ic2_bt_bbox $ic2_bt_types
 extract_wagons $ic2_sk_offset $ic2_sk_bbox $ic2_sk_types
 
diff --git a/lib/export-carriage.py b/lib/export-carriage.py
new file mode 100755
index 0000000..69e5133
--- /dev/null
+++ b/lib/export-carriage.py
@@ -0,0 +1,90 @@
+#!/usr/bin/env python3
+
+import numpy as np
+import subprocess
+import sys
+
+
+class SVGObject:
+    def __init__(self, attrlist):
+        self.id = attrlist[0]
+        self.x = float(attrlist[1])
+        self.y = float(attrlist[2])
+        self.w = float(attrlist[3])
+        self.h = float(attrlist[4])
+        self.x_ = self.x + self.w
+        self.y_ = self.y + self.h
+
+    def is_path(self):
+        return self.id.startswith("path")
+
+    def is_tspan(self):
+        return self.id.startswith("tspan")
+
+    def __repr__(self):
+        return "{}<x={}, y={}, w={}, h={}>".format(
+            self.id, self.x, self.y, self.w, self.h
+        )
+
+
+def main(infile, outfile):
+    ret = subprocess.run(
+        ["inkscape", "--query-all", infile], capture_output=True, check=True
+    )
+
+    objects = ret.stdout.decode("ascii").split("\n")
+
+    # remove last (empty) line
+    objects.pop()
+
+    # objects = [[ID, X, Y, W, H], ... ]
+    objects = list(map(lambda x: x.split(","), objects))
+    objects = list(map(lambda x: SVGObject(x), objects))
+
+    ys = list()
+    for o in objects:
+        if o.is_path() and o.h > 2 and o.h < 4 and o.w > 10:
+            ys.append(o.y)
+    bins = np.bincount(ys)
+
+    candidates = list(filter(lambda i: bins[i] > 2, range(len(bins))))
+
+    crop_y_min = np.min(candidates) - 1
+    crop_y_max = np.max(candidates) + 4
+
+    objects_to_delete = list()
+    x_positions = list()
+
+    for o in objects:
+        if (o.is_path or o.is_tspan) and (o.y < crop_y_min or o.y_ > crop_y_max):
+            objects_to_delete.append(o.id)
+        elif o.is_path and o.y >= crop_y_min and o.y_ <= crop_y_max:
+            x_positions.append(o.x)
+            x_positions.append(o.x_)
+
+    crop_x_min = int(np.min(x_positions))
+    crop_x_max = int(np.max(x_positions)) + 1
+
+    objects_to_delete = ",".join(objects_to_delete)
+
+    export_area = f"--export-area={crop_x_min}:{crop_y_min}:{crop_x_max}:{crop_y_max}"
+    select_objects = f"--select={objects_to_delete}"
+
+    subprocess.run(
+        [
+            "xvfb-run",
+            "inkscape",
+            export_area,
+            select_objects,
+            "--verb=EditDelete",
+            "--batch-process",
+            "--export-dpi=600",
+            "-o",
+            outfile,
+            infile,
+        ]
+    )
+
+
+if __name__ == "__main__":
+    main(*sys.argv[1:])