From 8385103796e0bdb6281c4c8303cc2862b218252e Mon Sep 17 00:00:00 2001
From: iou1name <iou1name@national.shitposting.agency>
Date: Mon, 20 Jan 2020 13:18:47 -0500
Subject: [PATCH] add email and id fields

---
 scrape_quest.py | 20 +++++++++++++-------
 voyage.sql      |  6 ++++--
 2 files changed, 17 insertions(+), 9 deletions(-)

diff --git a/scrape_quest.py b/scrape_quest.py
index 0aaada1..b9e5308 100644
--- a/scrape_quest.py
+++ b/scrape_quest.py
@@ -52,13 +52,19 @@ def scrape_posts(root_dir):
 			for post in posts:
 				# information gathering
 				post_id = int(post.get('id')[2:])
-				name = post.find(class_='name').text
-				tripcode = post.find(class_='postertrip')
-				if tripcode:
-					tripcode = tripcode.text
 				subject = post.find(class_='subject')
 				if subject:
 					subject = subject.text
+				name = post.find(class_='name').text
+				user_email = post.find(class_='useremail')
+				if user_email:
+					user_email = user_email.get('href')
+				tripcode = post.find(class_='postertrip')
+				if tripcode:
+					tripcode = tripcode.text
+				user_id = post.find(class_='hand')
+				if user_id:
+					user_id = user_id.text
 				post_time = int(post.find(class_='dateTime').get('data-utc'))
 				post_time = datetime.datetime.utcfromtimestamp(post_time)
 				post_time = post_time.replace(tzinfo=datetime.timezone.utc)
@@ -124,9 +130,9 @@ def scrape_posts(root_dir):
 
 				# database insert
 				cur.execute(
-					"INSERT INTO post VALUES (%s,%s,%s,%s,%s,%s,%s)",
-					(thread_id, post_id, name, tripcode, subject,
-					post_time, post_body)
+					"INSERT INTO post VALUES (%s,%s,%s,%s,%s,%s,%s,%s,%s)",
+					(thread_id, post_id, subject, name, user_email, tripcode,
+					user_id, post_time, post_body)
 				)
 				for link in links:
 					cur.execute("INSERT INTO link VALUES (%s,%s)",
diff --git a/voyage.sql b/voyage.sql
index e36ec8a..a08e960 100644
--- a/voyage.sql
+++ b/voyage.sql
@@ -7,9 +7,11 @@ CREATE TABLE IF NOT EXISTS thread (
 CREATE TABLE IF NOT EXISTS post (
 	thread_id INTEGER REFERENCES thread(id) ON DELETE CASCADE NOT NULL,
 	id INTEGER PRIMARY KEY,
-	name TEXT NOT NULL,
-	tripcode TEXT,
 	subject TEXT,
+	name TEXT NOT NULL,
+	email TEXT,
+	tripcode TEXT,
+	user_id TEXT,
 	time TIMESTAMP WITH TIME ZONE NOT NULL,
 	body TEXT NOT NULL
 );