From d575d031bdd92db1437a28e2a6a30360ca1f66a3 Mon Sep 17 00:00:00 2001
From: jspeis <jspeis@gmail.com>
Date: Wed, 7 Sep 2016 16:12:57 -0400
Subject: [PATCH] adding scripts to build api joins table

---
 scripts/bidirectional_geo_crosswalk.py | 24 +++++++
 scripts/crosswalk_geo_containment.py   | 96 ++++++++++++++++++++++++++
 2 files changed, 120 insertions(+)
 create mode 100644 scripts/bidirectional_geo_crosswalk.py
 create mode 100644 scripts/crosswalk_geo_containment.py

diff --git a/scripts/bidirectional_geo_crosswalk.py b/scripts/bidirectional_geo_crosswalk.py
new file mode 100644
index 0000000..3345a34
--- /dev/null
+++ b/scripts/bidirectional_geo_crosswalk.py
@@ -0,0 +1,24 @@
+import pandas as pd
+
+df = pd.read_csv("crosswalk_geo_containment.csv")
+
+my_connections = {}
+
+headers = [str(x) for x in df.columns]
+for row in list(df.itertuples()):
+    my_set = frozenset([row.parent_geoid, row.child_geoid])
+    if not my_set in my_connections:
+        my_connections[my_set] = True
+
+import csv
+with open('eggs.csv', 'wb') as csvfile:
+    spamwriter = csv.writer(csvfile, delimiter=',',
+                            quotechar='|', quoting=csv.QUOTE_MINIMAL)
+    spamwriter.writerow(["geo_a", "geo_b"])
+    for row in my_connections.keys():
+        my_row = list(row)
+        if len(my_row) == 1:
+            my_row = 2*my_row
+        spamwriter.writerow(my_row)
+
+    
diff --git a/scripts/crosswalk_geo_containment.py b/scripts/crosswalk_geo_containment.py
new file mode 100644
index 0000000..c26b12c
--- /dev/null
+++ b/scripts/crosswalk_geo_containment.py
@@ -0,0 +1,96 @@
+table_name = 'attrs.crosswalk_geo_containment'
+print '''
+CREATE TABLE {} (
+    child_geoid varchar(40),
+    parent_geoid varchar(40),
+    percent_covered real,
+    area_covered real
+);\n'''.format(table_name)
+
+
+# list of child/parent
+containments = [
+    ("160", "040"),
+    ("160", "050"),
+    ("160", "310"),
+    ("160", "795"),
+
+    ("140", "160"),
+
+    ("050", "040"),
+    ("050", "160"),
+    ("050", "310"),
+    ("050", "795"),
+
+    ("310", "040"),
+    ("310", "050"),
+    ("310", "795"),
+    ("310", "160"),
+
+    ("795", "040"),
+    ("795", "050"),
+    ("795", "160"),
+    ("795", "310"),
+
+]
+
+# --  (160) in Counties (050)
+
+name_map = {
+    "160": "place",
+    "050": "county",
+    "310": "cbsa",
+    "795": "puma",
+    "140": "tract",
+    "040": "state",
+}
+
+def gen_sql(child_level, parent_level):
+    child_gid_name = "geoid" if child_level != "795" else "geoid10"
+    parent_gid_name = "geoid" if parent_level != "795" else "geoid10"
+    child_name = name_map[child_level]
+    parent_name = name_map[parent_level]
+    sql = '''INSERT INTO {4} (
+        SELECT
+            '{5}00US' || {0}.{2} AS child_geoid,
+            '{6}00US' || {1}.{3} AS parent_geoid,
+            ST_Area(ST_Intersection({0}.geom,{1}.geom))/ST_Area({0}.geom)*100 as percent_covered,
+            ST_Area(ST_Intersection({0}.geom,{1}.geom)) as area_covered
+        FROM tiger2013.{0}
+        JOIN tiger2013.{1} ON ST_Intersects({0}.geom, {1}.geom)
+        WHERE ST_Area(ST_Intersection({0}.geom,{1}.geom))/ST_Area({0}.geom) > 0
+    );'''.format(child_name, parent_name, child_gid_name, parent_gid_name, table_name, child_level, parent_level)
+    return sql
+
+
+
+# identity rows for geo crosswalk joins
+def states_sql(table_name):
+    sql = '''INSERT INTO {0} (
+        select distinct parent_geoid as child_geoid, '01000US' parent_geoid,
+                        100 as percent_covered, NULL::real as area_covered
+        FROM {0}
+        WHERE parent_geoid LIKE '040%'
+    );
+    '''
+    return sql.format(table_name)
+
+def ident_sql(table_name):
+    sql = '''INSERT INTO {0} (
+        select distinct child_geoid, child_geoid as parent_geoid,
+                        NULL::real as percent_covered, NULL::real as area_covered
+        FROM {0}
+    );
+    INSERT INTO {0} (child_geoid, parent_geoid, percent_covered, area_covered) VALUES ('01000US', '01000US', NULL, NULL);
+    '''
+    return sql.format(table_name)
+
+for parent, child in containments:
+    print gen_sql(parent, child)
+
+print states_sql(table_name)
+print ident_sql(table_name)
+
+table_name_no_dot = table_name.replace(".", "_")
+print "CREATE INDEX {}_idx_child_geoid ON {} (child_geoid);".format(table_name_no_dot, table_name)
+print "CREATE INDEX {}_idx_parent_geoid ON {} (parent_geoid);".format(table_name_no_dot, table_name)