From eb431dc9bfd0c5962b1b7ee6fbd2d2c04f20dcb5 Mon Sep 17 00:00:00 2001
From: Erwan Rouchet <rouchet@teklia.com>
Date: Tue, 28 Sep 2021 09:06:49 +0200
Subject: [PATCH] Rewrite DAAI feed, close #87

---
 itsb.xml   |  9 +++++++--
 jq/daai.jq | 24 +++++++++++++++---------
 2 files changed, 22 insertions(+), 11 deletions(-)

diff --git a/itsb.xml b/itsb.xml
index 35078e9..473abb5 100644
--- a/itsb.xml
+++ b/itsb.xml
@@ -820,9 +820,14 @@
       <frequency>3-5 reports/year</frequency>
       <feed format="rss" lang="English" id="daai-rss">
         <curl>
-          <url verify-ssl="false">https://mwt.gov.na/published-daai-report/-/document_library/PVB0Gp2amxBo/view/1344073?_com_liferay_document_library_web_portlet_DLPortlet_INSTANCE_PVB0Gp2amxBo_displayStyle=list</url>
+          <url>https://mwt.gov.na/published-daai-report</url>
         </curl>
-        <pup>tr[data-title]</pup>
+        <!--
+        We would need both :not(:first-child) and :not(:nth-child(2)) to
+        properly remove all the header rows since there are two of them,
+        but pup does not allow that so the jq script will ignore the first row it gets.
+        -->
+        <pup>#our-content tr:not(:first-child)</pup>
         <jq path="daai.jq" />
         <json2xml />
         <output>daai.xml</output>
diff --git a/jq/daai.jq b/jq/daai.jq
index 5466657..599d1c4 100644
--- a/jq/daai.jq
+++ b/jq/daai.jq
@@ -9,22 +9,28 @@ import "./helpers" as helpers;
         "channel": {
             "title": "DAAI",
             "description": "Namibian Directorate of Aircraft Accidents and Investigation reports",
-            "link": "https://mwt.gov.na/web/mwt/published-reports",
+	    "link": "https://mwt.gov.na/published-daai-report",
             "language": "en-na",
             "pubDate": (now | strftime("%a, %d %b %Y %T %z")),
             "docs": "https://www.rssboard.org/rss-specification",
             "ttl": 1440,
             "generator": "ITSB",
             "item": [
-                .[]
+                .[1:][].children
                 | {
-                    "title": .["data-title"],
-                    "link": (.children[0].children[0].href | sub("\\?.*$"; "")),
-                    "pubDate": (.children[3].children[0].onmouseover | match("&#39;(.*)&#39;").captures[0].string | strptime("%e/%m/%y %I:%M %p") | mktime | strftime("%a, %d %b %Y %T %z")),
-                    "guid": {
-                        "@isPermaLink": "true",
-                        "#text": (.children[0].children[0].href | sub("\\?.*$"; ""))
-                    }
+                    "title": "\(.[1].children[0].text) - \(.[2].children[0].text) \(.[3].children[0].children[0].text) - \(.[-1].children[-1].text)",
+		    "link": (.[3].children[0].children[0].href | helpers::urlresolve("https://mwt.gov.na/published-daai-report")),
+                    "category": [
+                      [.[1].children[0].text, .[-1].children[-1].text][]
+                      | {
+                        "@domain": "https://mwt.gov.na",
+                        "#text": .
+                      }
+                    ]
+                }
+                | .guid = {
+                  "@isPermaLink": "true",
+                  "#text": .link
                 }
             ]
         }