Support inverse occupancy, post-dir in type-first street name

argyle-engineering · Apr 23, 2024 · 16ba3b4 · 16ba3b4
1 parent caee8ba
commit 16ba3b4
Show file tree

Hide file tree

Showing 2 changed files with 11 additions and 3 deletions.
diff --git a/pyap/source_US/data.py b/pyap/source_US/data.py
@@ -769,7 +769,7 @@ def str_list_to_upper_lower_regex(str_list: List[str]) -> str:
 street_type_leading_list = ["Camino", "El\ Camino"]
 
 
-def street_type_list_to_regex(street_type_list):
+def street_type_list_to_regex(street_type_list: list[str]) -> str:
     """Converts a list of street types into a regex"""
     street_types = str_list_to_upper_lower_regex(street_type_list)
 
@@ -803,14 +803,18 @@ def street_type_list_to_regex(street_type_list):
             (?:      
                 (?:{street_name_a}{space_div}{street_type_a})
                 |
-                (?:{street_type_b}{space_div}{street_name_b})
+                (?:
+                    (?:{post_direction_re}{space_div})?
+                    {street_type_b}{space_div}{street_name_b}
+                )
             )
 """.format(
     space_div=space_div,
     street_name_a=rf"(?P<street_name_a>{street_name_multi_word_re})",
     street_type_a=street_type_extended,
     street_type_b=rf"(?P<street_type_b>{street_types_leading_re})",
     street_name_b=rf"(?P<street_name_b>{street_name_one_word_re})",
+    post_direction_re=post_direction_re,
 )
 
 floor_indic = r"""
@@ -893,6 +897,8 @@ def street_type_list_to_regex(street_type_list):
                         (?:
                             [A-Za-z\#\&\-\d]{1,7}(?:\s?[SWNE])?
                         )?
+                        |
+                        \d{2,4}\ [Ss][Tt][Ee](?:\ \*)?
                     )
                     |
                     (?:
@@ -952,7 +958,7 @@ def street_type_list_to_regex(street_type_list):
                     |
                     (?:
                         {post_direction_re}\ 
-                        \d{{,3}}[A-Z][A-Za-z\-]{{,31}}
+                        \d{{,3}}[A-Za-z\-]{{1,31}}
                     )
                 )
                 (?:{space_div}{post_direction})?

diff --git a/tests/test_parser_us.py b/tests/test_parser_us.py
@@ -387,6 +387,7 @@ def test_po_box_positive(input, expected):
     "input,expected",
     [
         # positive assertions
+        ("800 W EL CAMINO REAL\n350 STE *", True),
         ("899 HEATHROW PARK LN", True),
         ("1806 Dominion Way Ste B", True),
         ("696 BEAL PKWY", True),
@@ -479,6 +480,7 @@ def test_full_street_positive(input, expected):
         ("696 BEAL PKWY NW\nFT WALTON BCH FL 32547", True),
         ("2633 Camino Ramon Ste. 400 San Ramon, CA 94583-2176", True),
         ("2951 El Camino Real Palo Alto, CA 94306", True),
+        ("800 W EL CAMINO REAL\n350 STE *\nMOUNTAIN VIEW, CA 94040", True),
         ("3821 ED DR, RALEIGH, NC 27612", True),
         ("213 WEST 35TH STREET SUITE, 400, NEW YORK, NY", True),
         ("326 33RD AVE., EAST, SEATTLE, WA 98112", True),