From: Pierre Choffet <peuc@wanadoo.fr>
Date: Fri, 24 Jan 2020 20:38:29 +0000 (-0500)
Subject: Add required scripts to generate almanac JSONs
X-Git-Url: https://git.wikimedia.ca/?a=commitdiff_plain;h=0673c779d48ec15e66ab666ffc573daf56e8e714;p=eccc_to_commons.git

Add required scripts to generate almanac JSONs
---

diff --git a/commons_rules.xsd b/commons_rules.xsd
index 169dd3e..49a50d3 100644
--- a/commons_rules.xsd
+++ b/commons_rules.xsd
@@ -3,7 +3,12 @@
      historical data prior of its import into Wikidata.
 
      All tests done in ECCC's bulkschema.xsd aren't replicated here as we're
-     focusing into pure Wikimedia Commons data quality -->
+     focusing into pure Wikimedia Commons data quality.
+
+     Schema is restrictive on purpose: up to date extreme values are hard coded
+     just like current year. It WILL break starting 2021 so we can make sure
+     someone has checked potential upstream structure change before running
+     Commons data update. -->
 <xsd:schema xmlns:xsd="http://www.w3.org/2001/XMLSchema">
 	<!-- Define elements -->
 	<xsd:element name="climatedata">
diff --git a/eccc_to_commons.sh b/eccc_to_commons.sh
index c580fbb..5d7b010 100755
--- a/eccc_to_commons.sh
+++ b/eccc_to_commons.sh
@@ -92,8 +92,7 @@ do
 	elif [ ${STATIONDATA_COUNT} -eq 0 ]&& \
 	     [ ${AL_MONTH_COUNT} -gt 0 ]
 	then
-		echo "${REPLY}: Almanac data not compatible yet. Ignoring." >&2
-		continue
+		STYLESHEET_PATH='almanac_to_commons.xslt'
 		DESTINATION_PATH="${DESTINATION}/weather.gc.ca/Almanac/${STATION_ID}.tab"
 	else
 		echo "${REPLY}: Cannot detect file type. Exiting."
@@ -102,5 +101,5 @@ do
 
 	echo "Processing ${REPLY} to ${DESTINATION_PATH}…" >&2
 	mkdir -p "$(dirname "${DESTINATION_PATH}")"
-	xmlstarlet tr "${STYLESHEET_PATH}" "${REPLY}" > "${DESTINATION_PATH}"
+	xmlstarlet tr "${STYLESHEET_PATH}" "${REPLY}" | jq > "${DESTINATION_PATH}"
 done 9< <( find "${SOURCE}" -type f -name '*.xml' -print0 )