From 72d0b64b6ed8d581baff7f798cf8cfaa0b015852 Mon Sep 17 00:00:00 2001
From: showerst <showerst@gmail.com>
Date: Wed, 6 Sep 2023 12:22:43 -0400
Subject: [PATCH] Events: Misc Add locations, improved addresses (#4695)

* SC: Events: geocode locations

* NH: Events: Geocode common locations, and add committees
---
 scrapers/nh/events.py | 43 +++++++++++++++++++++++++++++++++++++------
 scrapers/sc/events.py | 22 ++++++++++++++++++++++
 2 files changed, 59 insertions(+), 6 deletions(-)

diff --git a/scrapers/nh/events.py b/scrapers/nh/events.py
index 186fa350ed..da7deef361 100644
--- a/scrapers/nh/events.py
+++ b/scrapers/nh/events.py
@@ -7,6 +7,7 @@
 import lxml
 import datetime
 from openstates.scrape import Scraper, Event
+from utils.events import match_coordinates
 import re
 
 bill_re = re.compile(
@@ -64,12 +65,6 @@ def scrape_chamber(self, chamber):
             end = dateutil.parser.parse(row["end"])
             end = self._tz.localize(end)
 
-            if (
-                "cancelled" in row["title"].lower()
-                or "canceled" in row["title"].lower()
-            ):
-                status = "cancelled"
-
             if start < self._tz.localize(datetime.datetime.now()):
                 status = "passed"
 
@@ -81,6 +76,14 @@ def scrape_chamber(self, chamber):
                 classification = "other"
 
             location = row["title"].split(":")[-1].strip()
+            location = location.replace(
+                "LOB",
+                "Legislative Office Building, 33 North State Street, Concord, NH 03301",
+            )
+            location = location.replace(
+                "SH",
+                "New Hampshire State House, 107 North Main Street, Concord, NH 03301",
+            )
 
             event_name = f"{event_url}#{location}#{start}"
             if event_name in event_objects:
@@ -88,6 +91,19 @@ def scrape_chamber(self, chamber):
                 continue
             event_objects.add(event_name)
 
+            title = row["title"].split(":")[0].strip()
+
+            title = re.sub(
+                r"==(revised|time change|room change)==", "", title, flags=re.IGNORECASE
+            )
+
+            if (
+                "cancelled" in row["title"].lower()
+                or "canceled" in row["title"].lower()
+            ):
+                status = "cancelled"
+                title = re.sub("==Cancell?ed==", "", title, flags=re.IGNORECASE)
+
             event = Event(
                 name=title,
                 start_date=start,
@@ -99,7 +115,22 @@ def scrape_chamber(self, chamber):
             event.dedupe_key = event_name
             event.add_source(event_url)
 
+            if "commission" not in title.lower():
+                prefix = chamber_names[chamber].title()
+                if title.isupper():
+                    prefix = prefix.upper()
+                event.add_committee(f"{prefix} {title}")
+
             self.scrape_event_details(event, event_url)
+
+            match_coordinates(
+                event,
+                {
+                    "Legislative Office Building": ("43.20662", "-71.53938"),
+                    "State House": ("43.20699", "-71.53811"),
+                },
+            )
+
             yield event
 
     def scrape_event_details(self, event, url):
diff --git a/scrapers/sc/events.py b/scrapers/sc/events.py
index 09ef3cdb17..c1194982dd 100644
--- a/scrapers/sc/events.py
+++ b/scrapers/sc/events.py
@@ -5,6 +5,7 @@
 
 from openstates.scrape import Scraper, Event
 from spatula import PdfPage, URL
+from utils.events import match_coordinates
 
 
 def normalize_time(time_string):
@@ -223,6 +224,18 @@ def scrape_single_chamber(self, chamber=None, session=None):
                 else:
                     self.event_keys.add(event_key)
 
+                location = location.replace(
+                    "Blatt", "Blatt Building, 1105 Pendleton St, Columbia, SC 29201"
+                )
+                location = location.replace(
+                    "Gressette",
+                    "Gressette Building, 1101 Pendleton St, Columbia, SC 29201",
+                )
+                location = location.replace(
+                    "State House",
+                    "South Carolina State House, 1100 Gervais St, Columbia, SC 29208",
+                )
+
                 event = Event(
                     name=description,  # Event Name
                     start_date=date_time,  # When the event will take place
@@ -287,4 +300,13 @@ def scrape_single_chamber(self, chamber=None, session=None):
                         media_type="text/html",
                     )
 
+                match_coordinates(
+                    event,
+                    {
+                        "Blatt Building": ("33.99860", "-81.03323"),
+                        "Gressette Building": ("33.99917", "-81.03306"),
+                        "State House": ("34.00028", "-81.032954"),
+                    },
+                )
+
                 yield event