Update to 50 (#25)

pyranges · Nov 11, 2019 · 4cc843b · 4cc843b
1 parent 2e97975
commit 4cc843b
Show file tree

Hide file tree

Showing 6 changed files with 1,899 additions and 2,149 deletions.
diff --git a/CHANGELOG b/CHANGELOG
@@ -1,4 +1,7 @@
-# 0.0.48 (08.11.19)
+# 0.0.50 (11.11.19)
+- fix last_overlaps_both
+
+# 0.0.49 (08.11.19)
 - add last_overlaps_both
 
 # 0).0.47 (30.10.19)

diff --git a/ncls/src/ncls.c b/ncls/src/ncls.c
diff --git a/ncls/src/ncls.pyx b/ncls/src/ncls.pyx
@@ -95,6 +95,8 @@ cdef class NCLS64:
         if not self.im: # if empty
             return [], []
 
+        # from time import time
+        # start = time()
         it_alloc = cn.interval_iterator_alloc()
         it = it_alloc
         for loop_counter in range(length):
@@ -138,6 +140,9 @@ cdef class NCLS64:
             it = it_alloc
 
         cn.free_interval_iterator(it_alloc)
+        # end = time()
+
+        # print("ncls time:", end - start)
 
         return output_arr[:nfound], output_arr_other[:nfound]
 
@@ -152,6 +157,7 @@ cdef class NCLS64:
         cdef int length = len(starts)
         cdef int loop_counter = 0
         cdef int nfound = 0
+        cdef int max_end = -1
 
         output_arr = np.zeros(length, dtype=long)
         output_arr_other = np.zeros(length, dtype=long)
@@ -172,35 +178,25 @@ cdef class NCLS64:
         it = it_alloc
         for loop_counter in range(length):
 
+            # print("loop_counter", loop_counter)
             # remember first pointer for dealloc
             while it:
                 i = 0
                 cn.find_intervals(it, starts[loop_counter], ends[loop_counter], self.im, self.ntop,
                                 self.subheader, self.nlists, im_buf, 1024,
                                 &(nhit), &(it)) # GET NEXT BUFFER CHUNK
-
-                # print("nhit", nhit)
-                # print("length", length)
-                # print("nfound", nfound)
-                # print(nfound + nhit >= length)
-                if nfound + nhit >= length:
-
-                    length = (length + nhit) * 2
-                    output_arr = np.resize(output_arr, length)
-                    output_arr_other = np.resize(output_arr_other, length)
-                    output = output_arr
-                    output_other = output_arr_other
-
                 if nhit:
+                    max_end = -1
+                    while i < nhit:
+                        if im_buf[i].end >= max_end:
+                            # print("max_end", im_buf[i].end)
+                            output[nfound] = indexes[loop_counter]
+                            output_other[nfound] = im_buf[i].target_id
+                            max_end = im_buf[i].end
 
-                    # print("length", length)
-                    # print("nfound", nfound)
-                    # print("loop_counter", loop_counter)
-                    output[nfound] = indexes[loop_counter]
-                    output_other[nfound] = im_buf[nhit - 1].target_id
+                        i += 1
 
                     nfound += 1
-                    i += 1
 
             cn.reset_interval_iterator(it_alloc)
             it = it_alloc