AAC custom feed, closes #23

itsb.xml

@@ -25,7 +25,7 @@
   <section name="official feeds" id="official-feeds">
     <content><![CDATA[<p>those feeds are provided by the agencies themselves.]]></content>
 
-    <source id="ntsb">
+    <!--<source id="ntsb">
       <name>National Transport Safety Board</name>
       <region>United States</region>
       <type>Road, rail, aviation, marine, pipeline</type>
@@ -1172,6 +1172,22 @@
         <json2xml />
         <output>aaia/zh-hant.xml</output>
       </feed>
+    </source>-->
+
+    <source id="aac">
+      <name>Autoridad de Aviación Civil</name>
+      <region>El Salvador</region>
+      <type>Aviation</type>
+      <frequency>0-2 reports/year</frequency>
+      <feed type="aviation" lang="Spanish" format="rss" id="aac-rss">
+        <curl>
+          <url>https://www.aac.gob.sv/?page_id=1006</url>
+        </curl>
+	<pup>tr:not(:first-child)</pup>
+        <jq path="aac.jq" />
+        <json2xml />
+        <output>aac.xml</output>
+      </feed>
     </source>
   </section>
 </itsb>

jq/aac.jq

@@ -0,0 +1,34 @@
+# AAC feed generator
+# Expects pup JSON output holding <tr> tags, outputs xmltodict-compatible JSON
+# WARNING: Dates are locale-sensitive; the RSS feed might not generate correctly with another locale.
+
+{
+    "rss": {
+        "@version": "2.0",
+        "channel": {
+            "title": "AAC",
+            "description": "El Salvador's Autordid de Aviación Civil",
+            "link": "https://www.aac.gob.sv/?page_id=1006",
+            "language": "es-SV",
+            "pubDate": (now | strftime("%a, %d %b %Y %T %z")),
+            "docs": "https://www.rssboard.org/rss-specification",
+            "ttl": 1440,
+            "generator": "ITSB",
+            "item": [
+                .[].children | {
+                    "title": .[0].children[0].text,
+                    "link": (
+                        # The links are held in onclick attributes which set location.href to a string,
+                        # so we extract the string.
+                        .[3].children[0].onclick
+                        | match("&#39;(.*)&#39;").captures[0].string
+                        # & is escaped as we were within an HTML attribute, so this restores it
+                        | sub("&amp;"; "&")
+                    ),
+                    "description": .[1].children[0].text,
+                    "pubDate": (.[2].text | strptime("%d/%m/%Y") | mktime | strftime("%a, %d %b %Y %T %z"))
+                }
+            ]
+        }
+    }
+}