Overhauled way that non-sequential case numbers are attempted to be handled on each run.

1 files changed, 38 insertions, 12 deletions

diff --git a/daily_run.sh b/daily_run.sh
index 2a6295f..6dfeb3f 100755
--- a/daily_run.sh
+++ b/daily_run.sh
@@ -7,25 +7,51 @@ DATE=$(date -d 'yesterday' +'%A, %B %d, %Y')
 # Get the highest current court_case_no from the database
 YEAR=$(date +'%y')
 PASSWD=$(cat ../mysql_passwd)
-QUERY="SELECT court_case_no FROM cases WHERE court_case_no LIKE 'f-$YEAR-%' ORDER BY id DESC LIMIT 1"
-CASE_NO=$(mysql -u root -p$PASSWD -ss -e "$QUERY" mdcc)
-CASE_ID=${CASE_NO:5}
+QUERY="SELECT court_case_no FROM cases WHERE court_case_no LIKE 'f-$YEAR-%' ORDER BY court_case_no DESC LIMIT 250"
+CASE_NOS=$(mysql -u root -p$PASSWD -ss -e "$QUERY" mdcc)
 
-# The case number in the database has leading zeros. Strip them off.
-START=$(echo $CASE_ID | sed 's/^0*//')
+# Turn list into an array
+cases=($CASE_NOS)
+
+MAX_CASE_NO="${cases[0]}"
+
+# Here we go back over the latest cases to fill in gaps, since the case numbers
+# are not always entered sequentially on any given day.
+idx=1
+shift
+for case in ${cases[@]} 
+do
+	SEQ_PAD=${case:5}
+	SEQ=$(echo $SEQ_PAD | sed 's/^0*//')
+	if [[ ! -z ${cases[$idx]} ]]
+	then
+		NEXT_SEQ_PAD=${cases[$idx]:5}
+		NEXT_SEQ=$(echo $NEXT_SEQ_PAD | sed 's/^0*//')
+		if [ $NEXT_SEQ != $((SEQ-1)) ]
+		then
+			let "gap = SEQ - NEXT_SEQ"
+			let "GAP_START = NEXT_SEQ + 1"
+			let "GAP_STOP = SEQ - 1"
+			./miami_dade_clerk_courts.py $GAP_START $GAP_STOP $YEAR
+		fi
+		let "idx = idx + 1"
+	fi
+done
 
-# At the very leading edge of cases they often enter case number out of
-# sequence. This is a minor attempt to go backward a bit to try to fill in the
-# blanks.
-((START = START - 25))
+CASE_SEQ_PAD=${MAX_CASE_NO:5}
+
+# The case number in the database has leading zeros. Strip them off.
+START=$(echo $CASE_SEQ_PAD | sed 's/^0*//')
 
-./miami_dade_clerk_courts.py $START $YEAR
+# Fetch all the latest entries
+./miami_dade_clerk_courts.py $START 0 $YEAR
 
-# Now get the new highest court_case_no after fetching all the latest
+# Get the new highest court_case_no after fetching all the latest
+QUERY="SELECT court_case_no FROM cases WHERE court_case_no LIKE 'f-$YEAR-%' ORDER BY court_case_no DESC LIMIT 1"
 NEW_CASE_NO=$(mysql -u root -p$PASSWD -ss -e "$QUERY" mdcc)
 
 # Send a link with the results to some people
 RCPT_TO="udighe@gmail.com nath@nkinka.de"
 SUBJECT="MDCC daily run for $DATE"
-BODY="http://li554-21.members.linode.com/search.php?between=$CASE_NO:$NEW_CASE_NO"
+BODY="http://dighelaw.com/mdcc/search.php?between=$MAX_CASE_NO:$NEW_CASE_NO"
 echo $BODY | mail -s "$SUBJECT" $RCPT_TO