threading is coo

main.py

@@ -2,12 +2,27 @@ import debug
 import cluster
 import config
 import pdfplumber
+from multiprocessing.pool import ThreadPool
+import queue
 
 def main():
+   de_column_ify()
+
+def log(*args):
+   print(*args, flush=True)
+
+def de_column_ify():
+   q = queue.Queue(maxsize=4)
    with pdfplumber.open(config.INPUT) as pdf:
-      for page in pdf.pages:
-         got = cluster.Chars(page.chars, page).divide_into_columns()
-         debug.draw_boxes(page, [
+      with ThreadPool(4) as pool:
+         n = 0
+         for i in range(len(pdf.pages)):
+            pool.apply_async(de_columnify_page, (q, n, pdf.pages[i], ))
+            n += 1
+         for i in range(n):
+            log("getting", i, "of", n)
+            got = q.get()
+            debug.draw_boxes(got[1] , [
                {
                   "x0": i.chars[0]["x0"],
                   "x1": i.chars[0]["x1"],
@@ -15,8 +30,13 @@ def main():
                   "y1": i.chars[0]["y1"],
                   "debug_label": i.n,
                }
-            for i in got if i.chars
+               for i in got[2] if i.chars
             ])
 
+def de_columnify_page(q, idx, page):
+   result = cluster.Chars(page.chars, page).divide_into_columns()
+   log("putting", idx)
+   q.put((idx, page, result))
+
 if __name__ == "__main__":
    main()