added some tests for geom

fixed some bugs in geom

.gitignore

@@ -8,3 +8,4 @@ build/
 dist/
 pdftabextract.egg-info/
 ipynotes.py
+.cache/

pdftabextract/geom.py

@@ -34,7 +34,7 @@ def vecrotate(v, theta, about=np.array((0,0))):
     """rotate a vector v by angle theta (in radians) about point <about>"""
     cth = math.cos(theta)
     sth = math.sin(theta)
-        
+
     return pt(
         cth * v[0] - sth * v[1] + about[0] - cth * about[0] + sth * about[1],
         sth * v[0] + cth * v[1] + about[1] - sth * about[0] - cth * about[1]
@@ -54,12 +54,12 @@ def overlap(a1, a2, b1, b2):
            a_min <= b_max <= a_max or b_min <= a_max <= b_max
 
 
-def pointintersect(p1, p2, p3, p4, check_in_segm=True):
+def lineintersect(p1, p2, p3, p4, check_in_segm=True):
     """
     Check if two lines made from (p1, p2) and (p3, p4) respectively, intersect.
     
     If check_in_segm is True, will check that the line segments actually intersect,
-    will calculate intersection of inifite lines.
+    otherwise will calculate intersection of inifite lines.
     
     If no intersection is found, returns None
     For parallel lines, returns pt(np.nan, np.nan) if they are coincident.
@@ -99,7 +99,7 @@ def pointintersect(p1, p2, p3, p4, check_in_segm=True):
         return P
 
     if parallel:  # check if parallel segments are coincident
-        if overlap(p1[0], p2[0], p3[0], p4[0]) and overlap(p1[1], p2[1], p3[1], p4[1]):
+        if not check_in_segm or (overlap(p1[0], p2[0], p3[0], p4[0]) and overlap(p1[1], p2[1], p3[1], p4[1])):
             return P      # lines coincident -> return pt(np.nan, np.nan)
         else:
             return None   # no intersection in segment, only parallel
@@ -152,7 +152,7 @@ def rectcenter_dist(r1, r2):
 
 
 def rectarea(r):
-    """Return the area of rectangle <r>""""
+    """Return the area of rectangle <r>"""
     return (r[1][0] - r[0][0]) * (r[1][1] - r[0][1])
 
 

tests/test_geom.py

@@ -0,0 +1,96 @@
+# -*- coding: utf-8 -*-
+"""
+Created on Mon Feb 13 09:50:51 2017
+
+@author: mkonrad
+"""
+
+import math
+
+import numpy as np
+
+from pdftabextract.geom import pt, ptdist, vecangle, vecrotate, overlap, lineintersect
+
+
+def test_pt():
+    x = 0
+    y = 1
+    pt0 = pt(x, y)
+    assert pt0.dtype == np.float
+    assert pt0[0] == x
+    assert pt0[1] == y
+
+    pt1 = pt(x, y, np.int)
+    assert pt1.dtype == np.int
+    assert pt1[0] == x
+    assert pt1[1] == y
+
+
+def test_ptdist():
+    p1 = pt(0, 0)
+    p2 = pt(1, 0)
+    p3 = pt(1, 1)
+
+    assert ptdist(p1, p1) == 0
+    assert ptdist(p1, p2) == 1
+    assert ptdist(p2, p1) == ptdist(p1, p2)
+
+    assert ptdist(p1, p3) == math.sqrt(2)
+
+
+def test_vecangle():
+    v1 = pt(1, 0)
+    v2 = pt(2, 0)
+    v3 = pt(1, 1)
+    v4 = pt(0, 1)
+    v5 = pt(0, -1)
+
+    assert np.isnan(vecangle(pt(0, 0), v1))   # pt(0, 0) is vec of no length
+    assert vecangle(v1, v2) == 0
+    assert round(vecangle(v1, v3), 4) == round(math.radians(45), 4)
+    assert vecangle(v2, v4) == vecangle(v1, v4) == math.radians(90)
+    assert vecangle(v2, v5) == math.radians(90)   # always the smaller angle
+
+
+def test_vecrotate():
+    assert np.array_equal(vecrotate(pt(0, 0), 0.123), pt(0, 0))
+    assert np.allclose(vecrotate(pt(1, 0), math.radians(90)), pt(0, 1))
+    assert np.allclose(vecrotate(pt(1, 0), math.radians(90), about=pt(1, 1)), pt(2, 1))
+
+
+def test_overlap():
+    assert overlap(0, 1, 0, 1) is True
+    assert overlap(0, 0, 1, 1) is False
+    assert overlap(0, 10, 5, 15) is True
+    assert overlap(-10, 10, -20, -10) is True
+    assert overlap(-9, 10, -20, -10) is False
+
+
+def test_lineintersect():
+    # first with check_in_segm = True
+    X = lineintersect(pt(0, 0), pt(0, 0), pt(0, 0), pt(0, 0))   # coincident I
+    assert sum(np.isnan(X)) == len(X)
+
+    X = lineintersect(pt(0, 0), pt(0, 1), pt(0, 0), pt(0, 1))   # coincident II
+    assert sum(np.isnan(X)) == len(X)
+
+    assert lineintersect(pt(0, 0), pt(0, 1), pt(1, 0), pt(1, 1)) is None  # parallel, non coincident
+    assert lineintersect(pt(0, 0), pt(0, 1), pt(1, 1), pt(2, 2)) is None  # non-parellel, no intersection
+    assert lineintersect(pt(0, 0), pt(2, 2), pt(0, 5), pt(5, 0)) is None  # non-parellel, no intersection II
+    assert np.array_equal(lineintersect(pt(0, 0), pt(0, 1), pt(0, 1), pt(2, 2)), pt(0, 1))  # intersection - touch
+    assert np.array_equal(lineintersect(pt(0, 0), pt(2, 2), pt(0, 2), pt(2, 0)), pt(1, 1))  # intersection
+
+    # now with check_in_segm = False
+    X = lineintersect(pt(0, 0), pt(0, 0), pt(0, 0), pt(0, 0), False)   # coincident I
+    assert sum(np.isnan(X)) == len(X)
+
+    X = lineintersect(pt(0, 0), pt(0, 1), pt(0, 0), pt(0, 1), False)   # coincident II
+    assert sum(np.isnan(X)) == len(X)
+
+    X = lineintersect(pt(0, 0), pt(1, 1), pt(2, 2), pt(3, 3), False)   # coincident III
+    assert sum(np.isnan(X)) == len(X)
+
+    assert np.array_equal(lineintersect(pt(0, 0), pt(0, 1), pt(1, 1), pt(2, 2), False), pt(0, 0))  # intersection (out of segments)
+    assert np.array_equal(lineintersect(pt(0, 0), pt(0, 1), pt(0, 1), pt(2, 2), False), pt(0, 1))  # intersection - touch
+    assert np.array_equal(lineintersect(pt(0, 0), pt(2, 2), pt(0, 2), pt(2, 0), False), pt(1, 1))  # intersection
+