From c67ec83de98b26e3264ebd6b45bc620d63de311e Mon Sep 17 00:00:00 2001
From: Bel LaPointe <breel@qualtrics.com>
Date: Sun, 19 Feb 2023 08:39:18 -0700
Subject: [PATCH] whee

---
 main.py | 125 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 125 insertions(+)
 create mode 100644 main.py

diff --git a/main.py b/main.py
new file mode 100644
index 0000000..703ae20
--- /dev/null
+++ b/main.py
@@ -0,0 +1,125 @@
+import pdfplumber
+import os
+import time
+import subprocess
+
+DEBUG = os.environ.get("DEBUG", "")
+INPUT = os.environ.get("INPUT", "./testdata/input.pdf")
+
+def main():
+   print("main")
+   with pdfplumber.open(INPUT) as pdf:
+      for page in pdf.pages[4:5]:
+         for splitpage in v_split(page):
+            print(splitpage.extract_text())
+   print("/main")
+
+def crop(page, x0, y0, x1, y1):
+   if DEBUG:
+      im = debug_im(page)
+      im.draw_lines([
+         ((x0, y0), (x0, y1)),
+         ((x0, y1), (x1, y1)),
+         ((x1, y1), (x1, y0)),
+         ((x1, y0), (x0, y0)),
+      ], stroke_width=5)
+      debug_show(im)
+   return page.crop((x0, y0, x1, y1))
+
+def debug_im(page):
+   return page.to_image(height=800)
+
+def debug_show(im):
+   im.save("/tmp/out.jpg")
+   go("qlmanage -p /tmp/out.jpg &> /dev/null")
+
+def v_split(page):
+   clusters = cluster(page)
+   points = [i.x0 for i in clusters]
+   points += [i.x1 for i in clusters]
+   x_clusters = []
+   for point in points:
+      merged = False
+      for x_cluster in x_clusters:
+         if (x_cluster - point) ** 2 < 100:
+            merged = True
+            break
+      if not merged:
+         x_clusters.append(point)
+   x_clusters = sorted(x_clusters)
+   if len(x_clusters) != 4:
+      return [page]
+   x = (x_clusters[2] + x_clusters[1]) / 2
+   result = [page.within_bbox((0, 0, x, page.height)), page.within_bbox((x, 0, page.width, page.height))]
+   if DEBUG:
+      for page in result:
+         debug_show(debug_im(page))
+   return result
+
+def cluster(page):
+   points = [(i["x0"], i["y0"]) for i in page.chars]
+   points += [(i["x1"], i["y1"]) for i in page.chars]
+   clusters = []
+   class cluster:
+      def __init__(self, x, y):
+         self.x0 = x
+         self.y0 = y
+         self.x1 = x+1
+         self.y1 = y+1
+         self.len = 1
+      def merge(self, x, y):
+         if x < self.x0:
+            self.x0 = x
+         elif x > self.x1:
+            self.x1 = x
+         if y < self.y0:
+            self.y0 = y
+         elif y > self.y1:
+            self.y1 = y
+         self.len += 1
+      def dist(self, x, y):
+         x_delta = 0
+         y_delta = 0
+         if   x < self.x0:
+            x_delta = self.x0 - x
+         elif x > self.x1:
+            x_delta = x - self.x1
+         if   y < self.y0 :
+            y_delta = self.y0 - y
+         elif y > self.y1:
+            y_delta = y - self.y1
+         return x_delta**2 + y_delta**2
+      def __str__(self):
+         return f'({int(self.x0)}, {int(self.y0)}, {int(self.x1)}, {int(self.y1)})'
+   for point in points:
+      merged = False
+      for a_cluster in clusters:
+         if a_cluster.dist(point[0], point[1]) < (page.width/50)**2:
+            a_cluster.merge(point[0], point[1])
+            merged = True
+            break
+      if not merged:
+         clusters.append(cluster(point[0], point[1]))
+   clusters = [i for i in clusters if i.len > 100]
+   if DEBUG:
+      im = debug_im(page)
+      for i in clusters:
+         im.draw_lines([
+            ((i.x0, page.height-i.y0), (i.x0, page.height-i.y1)),
+            ((i.x0, page.height-i.y1), (i.x1, page.height-i.y1)),
+            ((i.x1, page.height-i.y1), (i.x1, page.height-i.y0)),
+            ((i.x1, page.height-i.y0), (i.x0, page.height-i.y0)),
+         ], stroke_width=5)
+      debug_show(im)
+   return clusters
+
+__subprocesses__ = []
+def go(cmd):
+   global __subprocesses__
+   __subprocesses__.append(subprocess.Popen(cmd, shell=True))
+
+if __name__ == "__main__":
+   main()
+   for p in __subprocesses__:
+      p.wait()
+      p.terminate()