From f05a8a79bcf42999cdd9dbd7caa4d0799d990b2d Mon Sep 17 00:00:00 2001
From: JustAnotherArchivist <JustAnotherArchivist@users.noreply.github.com>
Date: Mon, 29 Mar 2021 23:16:40 +0000
Subject: [PATCH] Clean up wpull DB commands

---
 wpull2-extract-remaining | 19 ++++++++++++++++---
 wpull2-url-origin        | 31 ++++++++++++++++++++++++++++++-
 2 files changed, 46 insertions(+), 4 deletions(-)

diff --git a/wpull2-extract-remaining b/wpull2-extract-remaining
index 0aeba51..13a43f9 100755
--- a/wpull2-extract-remaining
+++ b/wpull2-extract-remaining
@@ -1,8 +1,21 @@
 #!/bin/bash
-# Usage: wpull2-extract-remaining FILENAME
-#  FILENAME points to a wpull 2.x SQLite DB
+# Usage: wpull2-extract-remaining [FILENAME]
+#  FILENAME points to a wpull 2.x SQLite DB; if not specified, defaults to wpull.db
 # Prints all remaining URLs from the DB on stdout
+
+if [[ $# -eq 1 ]]
+then
+	filename="$1"
+else
+	filename=wpull.db
+fi
+if [[ ! -f "${filename}" ]]
+then
+	echo "Error: ${filename} does not exist or is not a regular file" >&2
+	exit 1
+fi
+
 for status in in_progress todo error
 do
-	sqlite3 "$1" 'SELECT url_strings.url FROM queued_urls JOIN url_strings ON url_string_id = url_strings.id WHERE status = "'$status'"'
+	sqlite3 "${filename}" 'SELECT url_strings.url FROM queued_urls JOIN url_strings ON url_string_id = url_strings.id WHERE status = "'$status'"'
 done
diff --git a/wpull2-url-origin b/wpull2-url-origin
index 57d7672..9758f8a 100755
--- a/wpull2-url-origin
+++ b/wpull2-url-origin
@@ -1,3 +1,32 @@
 #!/bin/bash
+# Usage: wpull2-url-origin [FILENAME] URL
+#  FILENAME defaults to wpull.db
 # Trace back where a URL was discovered, all the way back to the root
-url="$1"; curId=$(sqlite3 wpull.db 'SELECT id FROM url_strings WHERE url = "'"${url}"'"'); while :; do sqlite3 wpull.db 'SELECT queued_urls.*, url_strings.* FROM queued_urls JOIN url_strings ON queued_urls.url_string_id = url_strings.id WHERE url_strings.id = '$curId; IFS='|' read -r curId level < <(sqlite3 wpull.db 'SELECT parent_url_string_id, level FROM queued_urls WHERE url_string_id = '$curId); if [[ ${level} -eq 0 ]]; then break; fi done
+if [[ $# -eq 2 ]]
+then
+	filename="$1"
+	shift
+else
+	filename=wpull.db
+fi
+if [[ ! -f "${filename}" ]]
+then
+	echo "Error: ${filename} does not exist or is not a regular file" >&2
+	exit 1
+fi
+url="$1"
+curId=$(sqlite3 "${filename}" 'SELECT id FROM url_strings WHERE url = "'"${url}"'"')
+if [[ -z "${curId}" ]]
+then
+	echo "Error: ${url} not found" >&2
+	exit 1
+fi
+while :
+do
+	sqlite3 "${filename}" 'SELECT queued_urls.*, url_strings.* FROM queued_urls JOIN url_strings ON queued_urls.url_string_id = url_strings.id WHERE url_strings.id = '${curId}
+	IFS='|' read -r curId level < <(sqlite3 "${filename}" 'SELECT parent_url_string_id, level FROM queued_urls WHERE url_string_id = '${curId})
+	if [[ ${level} -eq 0 ]]
+	then
+		break
+	fi
+done