Implement SHA1 checksums and optional download for documents

1 files changed, 25 insertions, 6 deletions

diff --git a/pcb/download.sh b/pcb/download.sh
index dfe1be9..4aa5f49 100755
--- a/pcb/download.sh
+++ b/pcb/download.sh
@@ -1,11 +1,15 @@
 #!/bin/sh
 
 
+# This script creates a cache directory and downloads all external documents
+# to this folder. In addition the SHA1 checksums are checked and the user is
+# warned if a checksum does not match.
+
+
 # constants
 DOCUMENTS_FILE='documents.tsv'
 CACHE='./cache'  # target folder for all downloads
 
-
 # strip first line of table file
 TABLE=$(sed '1d' "${DOCUMENTS_FILE}")
 
@@ -18,14 +22,29 @@ do
 
 	FILENAME=$(echo "$line" | awk -F '\t' '{print $1}')  # column 1 -> filename
 	URL=$(echo "$line" | awk -F '\t' '{print $2}')       # column 2 -> URL
+	SHA1=$(echo "$line" | awk -F '\t' '{print $3}')      # column 3 -> SHA1 hash
+
+	FILEPATH="${CACHE}/${FILENAME}"
+
+	# download file if it does not exist in cache
+	if [ ! -f "${FILEPATH}" ]
+	then
+		printf 'Downloading %s ...' "${FILEPATH}"
+		if wget -qO "${FILEPATH}" "${URL}" 2> /dev/null
+		then
+			printf " ok\n"        # if wget returned with exit code 0
+		else
+			printf " error!\n"    # in any other case
+		fi
+	fi
 
-	# actual download of one document
-	printf 'Downloading %s/%s ...' "${CACHE}" "${FILENAME}"
-	if wget -qO "${CACHE}/${FILENAME}" "${URL}" 2> /dev/null
+	# generate and compare SHA1 checksum of file
+	printf 'Checking SHA1 sum of %s ...' "${FILEPATH}"
+	if printf '%s %s' "${SHA1}" "${FILEPATH}" | sha1sum -c - > /dev/null 2>&1
 	then
-		printf " ok\n"        # if wget returned with exit code 0
+		echo " ok"
 	else
-		printf " error!\n"    # in any other case
+		echo " file changed!"
 	fi
 
 done