From 845719346eb226e33effc1861347c88f9c342e4c Mon Sep 17 00:00:00 2001
From: Andy Chosak <andy.chosak@cfpb.gov>
Date: Thu, 2 Nov 2023 11:14:52 -0400
Subject: [PATCH] Remove or rename existing "crawsqueal" references

With the renaming of this repository from "crawsqueal" to
"website-indexer", there are a few "crawsqueal" references that need
to be removed or renamed.
---
 fabfile.py                                    | 24 ++++++++++---------
 package.json                                  |  8 +++----
 .../static_src/{crawsqueal.css => main.css}   |  0
 viewer/static_src/{crawsqueal.js => main.js}  |  0
 viewer/templates/viewer/base.html             |  4 ++--
 viewer/templates/viewer/page_list.html        |  2 +-
 6 files changed, 20 insertions(+), 18 deletions(-)
 rename viewer/static_src/{crawsqueal.css => main.css} (100%)
 rename viewer/static_src/{crawsqueal.js => main.js} (100%)

diff --git a/fabfile.py b/fabfile.py
index 0d97059..2a40444 100644
--- a/fabfile.py
+++ b/fabfile.py
@@ -27,26 +27,28 @@
 PYTHON_INSTALL_ROOT = f"{DEPLOY_ROOT}/{PYTHON_BASENAME}"
 
 SOURCE_PARENT = f"{DEPLOY_ROOT}/cfpb"
-SOURCE_REPO = "https://github.com/cfpb/crawsqueal.git"
-SOURCE_DIRNAME = "crawsqueal"
+SOURCE_REPO = "https://github.com/cfpb/website-indexer.git"
+SOURCE_DIRNAME = "website-indexer"
 SOURCE_ROOT = f"{SOURCE_PARENT}/{SOURCE_DIRNAME}"
 
-CRAWL_DATABASE = "/var/tmp/crawl.sqlite3"
+CRAWL_DIR = "/var/tmp"
+CRAWL_DATABASE = f"{CRAWL_DIR}/crawl.sqlite3"
+CRAWL_DATABASE_TMP = f"{CRAWL_DIR}/crawl-new.sqlite3"
 
 LOGROTATE_DIR = "/etc/logrotate.d"
-LOGROTATE_NAME = "crawsqueal"
+LOGROTATE_NAME = "website-indexer"
 LOGROTATE_PATH = f"{LOGROTATE_DIR}/{LOGROTATE_NAME}"
 
 SYSTEMD_DIR = "/etc/systemd/system"
-SYSTEMD_SERVICE = "crawsqueal"
+SYSTEMD_SERVICE = "website-indexer"
 SYSTEMD_NAME = f"{SYSTEMD_SERVICE}.service"
 SYSTEMD_PATH = f"{SYSTEMD_DIR}/{SYSTEMD_NAME}"
 
-CRONTAB_NAME = "crawsqueal"
+CRONTAB_NAME = "website-indexer"
 CRONTAB_DIR = "/etc/cron.d"
 CRONTAB_PATH = f"{CRONTAB_DIR}/{CRONTAB_NAME}"
 
-LOG_DIR = "/var/log/crawsqueal"
+LOG_DIR = "/var/log/website-indexer"
 
 
 @task
@@ -144,10 +146,10 @@ def deploy(conn):
         "SHELL=/bin/bash\n"
         f"0 0 * * * {conn.user} "
         f"cd {SOURCE_ROOT} && "
-        f"./wget_crawl.sh https://www.consumerfinance.gov/ && "
-        f"PYTHONPATH=. DJANGO_SETTINGS_MODULE=settings ./venv/bin/django-admin "
-        "warc_to_db --recreate ./crawl.warc.gz ./crawl.sqlite3 && "
-        f"mv crawl.{{cdx,sqlite3,warc.gz}} wget.log /var/tmp/\n"
+        f"./venv/bin/python manage.py crawl --recreate "
+        f"https://www.consumerfinance.gov {CRAWL_DATABASE_TMP} "
+        f"> {CRAWL_DIR}/crawl.log 2>&1 && "
+        f"mv {CRAWL_DATABASE_TMP} {CRAWL_DATABASE}\n"
         "EOF'"
     )
 
diff --git a/package.json b/package.json
index 951d34c..755f4eb 100644
--- a/package.json
+++ b/package.json
@@ -1,6 +1,6 @@
 {
-  "name": "crawsqueal",
-  "version": "1.0.0",
+  "name": "website-indexer",
+  "version": "1.1.0",
   "license": "CC0-1.0",
   "type": "module",
   "engines": {
@@ -10,8 +10,8 @@
     "prettier": "prettier --check 'viewer/static_src/*.{css,js}' 'viewer/**/*.html'",
     "fix": "npm run prettier -- --write",
     "fonts": "cp -r viewer/static_src/fonts viewer/static/fonts",
-    "styles": "curl -o viewer/static/main.css https://www.consumerfinance.gov/static/css/main.css && curl https://www.consumerfinance.gov/static/apps/regulations3k/css/main.css >> viewer/static/main.css && cp viewer/static_src/crawsqueal.css viewer/static/crawsqueal.css",
-    "scripts": "esbuild viewer/static_src/crawsqueal.js --bundle --outfile=viewer/static/crawsqueal.js",
+    "styles": "curl -o viewer/static/cfgov.css https://www.consumerfinance.gov/static/css/main.css && curl https://www.consumerfinance.gov/static/apps/regulations3k/css/main.css >> viewer/static/cfgov.css && cp viewer/static_src/main.css viewer/static/main.css",
+    "scripts": "esbuild viewer/static_src/main.js --bundle --outfile=viewer/static/main.js",
     "build": "yarn fonts && yarn styles && yarn scripts"
   },
   "dependencies": {
diff --git a/viewer/static_src/crawsqueal.css b/viewer/static_src/main.css
similarity index 100%
rename from viewer/static_src/crawsqueal.css
rename to viewer/static_src/main.css
diff --git a/viewer/static_src/crawsqueal.js b/viewer/static_src/main.js
similarity index 100%
rename from viewer/static_src/crawsqueal.js
rename to viewer/static_src/main.js
diff --git a/viewer/templates/viewer/base.html b/viewer/templates/viewer/base.html
index 55b7c73..3f33cea 100644
--- a/viewer/templates/viewer/base.html
+++ b/viewer/templates/viewer/base.html
@@ -12,8 +12,8 @@
     <title>{% block title %}Consumerfinance.gov web page index{% endblock %}</title>
     <!-- prettier-ignore -->
     <link rel="icon" href="data:image/svg+xml,<svg xmlns=%22http://www.w3.org/2000/svg%22 viewBox=%220 0 100 100%22><text y=%22.9em%22 font-size=%2290%22>🪱</text></svg>" />
+    <link rel="stylesheet" href="{% static 'cfgov.css' %}" />
     <link rel="stylesheet" href="{% static 'main.css' %}" />
-    <link rel="stylesheet" href="{% static 'crawsqueal.css' %}" />
   </head>
   <body>
     <div class="skip-nav">
@@ -39,6 +39,6 @@
         </div>
       </div>
     </footer>
-    <script src="{% static 'crawsqueal.js' %}" type="module"></script>
+    <script src="{% static 'main.js' %}" type="module"></script>
   </body>
 </html>
diff --git a/viewer/templates/viewer/page_list.html b/viewer/templates/viewer/page_list.html
index c869b31..7e03ad5 100644
--- a/viewer/templates/viewer/page_list.html
+++ b/viewer/templates/viewer/page_list.html
@@ -184,7 +184,7 @@ <h4>
       ({{ crawl_stats.database_size | filesizeformat }}) to
       <a
         class="m-list_link a-link a-link__icon"
-        href="https://github.com/cfpb/crawsqueal#searching-the-crawl-database">
+        href="https://github.com/cfpb/website-indexer#searching-the-crawl-database">
         <span class="a-link_text">query the data locally</span>
         <svg
           class="cf-icon-svg"