From 4a8fba32741a8a9b866efce5b5f3233c5e1a3d32 Mon Sep 17 00:00:00 2001
From: Sijia Wang <wangsijia0628@gmail.com>
Date: Thu, 5 Aug 2021 17:22:28 -0400
Subject: [PATCH] fix bug in add_counts

---
 lasso/roadway.py | 29 ++++++++++++++++++++++++++++-
 1 file changed, 28 insertions(+), 1 deletion(-)

diff --git a/lasso/roadway.py b/lasso/roadway.py
index f8568ec..38514be 100644
--- a/lasso/roadway.py
+++ b/lasso/roadway.py
@@ -575,6 +575,10 @@ def add_variable_using_shst_reference(
         )
 
         var_shst_df = pd.read_csv(var_shst_csvdata)
+        # there are aadt = 0 in the counts, drop them
+        var_shst_df = var_shst_df[var_shst_df[shst_csv_variable] > 0].copy()
+        # count station to shared street match - there are many-to-one matches, keep just one match
+        var_shst_df.drop_duplicates(subset = ["shstReferenceId"], inplace = True)
 
         if "shstReferenceId" not in var_shst_df.columns:
             msg = "'shstReferenceId' required but not found in {}".format(var_shst_data)
@@ -598,7 +602,7 @@ def add_variable_using_shst_reference(
         join_gdf[shst_csv_variable].fillna(0, inplace=True)
 
         if network_variable in self.links_df.columns and not overwrite:
-            join_gdf.loc[join_gdf[network_variable] > 0, network_variable] = join_gdf[
+            join_gdf.loc[join_gdf[network_variable] == 0, network_variable] = join_gdf[
                 shst_csv_variable
             ].astype(network_var_type)
         else:
@@ -608,6 +612,28 @@ def add_variable_using_shst_reference(
 
         self.links_df[network_variable] = join_gdf[network_variable]
 
+        # MN and WI counts are vehicles using the segment in both directions, no directional counts
+        # we will make sure both direction has the same daily AADT
+        dir_link_count_df = self.links_df[
+            (self.links_df[network_variable] > 0) &
+            (self.links_df["drive_access"] == 1)
+        ][["A", "B", network_variable]].copy()
+        reverse_dir_link_count_df = dir_link_count_df.rename(columns = {"A":"B", "B":"A"}).copy()
+
+        link_count_df = pd.concat(
+            [dir_link_count_df, reverse_dir_link_count_df],
+            sort = False,
+            ignore_index = True
+        )
+        link_count_df.drop_duplicates(subset = ["A", "B"], inplace = True)
+
+        self.links_df = pd.merge(
+            self.links_df.drop(network_variable, axis = 1),
+            link_count_df[["A", "B", network_variable]],
+            how = "left",
+            on = ["A", "B"]
+        )
+        self.links_df[network_variable].fillna(0, inplace = True)
         WranglerLogger.info(
             "Added variable: {} using Shared Streets Reference".format(network_variable)
         )
@@ -722,6 +748,7 @@ def add_counts(
         self.links_df["count_NT"] = self.links_df[network_variable] / 4
 
         self.links_df["count_daily"] = self.links_df[network_variable]
+        # add COUNTYEAR
         self.links_df["count_year"] = 2017
 
         WranglerLogger.info(