are interpreted with the street being "W", the street type as "River"
(which abbreviates to "riv"), and the city as "RD \nCOLUMBIA STATION".
The example addresses are all in Ohio because that's my current data set,
but it's not an Ohio-specific phenomenon: for example, in Illinois,
there's a River Road that follows the Des Plaines River.
The erroneous parse appears to be from the "100 South Street"
special case in the street regexp. This commit adds a second special
case with higher precedence, matching [prefix, non-numeric street,
street type] sequences. The street match excludes numerics to preserve
the existing parse behavior for the "6641 N 2200 W Apt D304 Park City,
UT 84098" case.
Some addresses like
are interpreted with the street being "W", the street type as "River" (which abbreviates to "riv"), and the city as "RD \nCOLUMBIA STATION". The example addresses are all in Ohio because that's my current data set, but it's not an Ohio-specific phenomenon: for example, in Illinois, there's a River Road that follows the Des Plaines River.
The erroneous parse appears to be from the "100 South Street" special case in the street regexp. This commit adds a second special case with higher precedence, matching [prefix, non-numeric street, street type] sequences. The street match excludes numerics to preserve the existing parse behavior for the "6641 N 2200 W Apt D304 Park City, UT 84098" case.