Add TTSB feeds, close #8

feedgen.sh

@@ -78,6 +78,28 @@ curl -s 'https://en.havarikommissionen.dk/railway-archive/' \
 	| json2xml -s -ns 'http://search.yahoo.com/mrss/' media > $DIR/feeds/aibd/rail.xml.new \
 	&& mv $DIR/feeds/aibd/rail.xml.new $DIR/feeds/aibd/rail.xml
 
+log Building TTSB Aviation English feed to $DIR/feeds/ttsb/en/aviation.xml
+mkdir -p $DIR/feeds/ttsb/en
+curl -s 'https://www.ttsb.gov.tw/english/16051/16052/16053/16058/Lpsimplelist?PageSize=1000' \
+	| pup '#LP-ContentPage .TableList table tbody tr json{}' \
+	| jq -L $DIR/jq -f $DIR/jq/ttsb.jq \
+		--arg title 'TTSB Aviation' \
+		--arg lang 'en-TW' \
+		--arg link 'https://www.ttsb.gov.tw/english/16051/16052/16053/16058/Lpsimplelist?PageSize=1000' \
+	| json2xml > $DIR/feeds/ttsb/en/aviation.xml.new \
+	&& mv $DIR/feeds/ttsb/en/aviation.xml.new $DIR/feeds/ttsb/en/aviation.xml
+
+log Building TTSB Aviation Chinese feed to $DIR/feeds/ttsb/zh/aviation.xml
+mkdir -p $DIR/feeds/ttsb/zh
+curl -s 'https://www.ttsb.gov.tw/1133/1154/1155/1159/Lpsimplelist?PageSize=1000' \
+	| pup '#LP-ContentPage .TableList table tbody tr json{}' \
+	| jq -L $DIR/jq -f $DIR/jq/ttsb.jq \
+		--arg title '國家運輸安全調查委員會' \
+		--arg lang 'zh-Hant-TW' \
+		--arg link 'https://www.ttsb.gov.tw/1133/1154/1155/1159/Lpsimplelist?PageSize=1000' \
+	| json2xml > $DIR/feeds/ttsb/zh/aviation.xml.new \
+	&& mv $DIR/feeds/ttsb/zh/aviation.xml.new $DIR/feeds/ttsb/zh/aviation.xml
+
 log Building BFU English feed to $DIR/feeds/bfu/en.xml
 mkdir -p $DIR/feeds/bfu
 curl -s 'https://www.bfu-web.de/EN/Publications/Investigation%20Report/reports_node.html?cms_gts=238148_list%253DdateOfIssue_dt%252Bdesc' \

index.html

@@ -658,6 +658,42 @@
           </tr>
         </tbody>
       </table>
+
+      <h3>Taiwan Transportation Safety Board</h3>
+      <table>
+        <tbody>
+          <tr>
+            <td><strong>Country</strong></td>
+            <td>Taiwan</td>
+          </tr>
+          <tr>
+            <td><strong>Language</strong></td>
+            <td>Traditional Chinese, English</td>
+          </tr>
+          <tr>
+            <td><strong>Types</strong></td>
+            <td>Aviation, rail, marine, highway</td>
+          </tr>
+          <tr>
+            <td><strong>Frequency</strong></td>
+            <td>10 reports/year</td>
+          </tr>
+          <tr>
+            <td><strong>Feeds</strong></td>
+            <td>
+              <table>
+                <tr>
+                  <td>Aviation</td>
+                  <td>
+                    <a href="feeds/ttsb/zh/aviation.xml" target="_blank"><img src="img/rss-chinese.gif" alt="Chinese RSS" /></a>
+                    <a href="feeds/ttsb/en/aviation.xml" target="_blank"><img src="img/rss-english.gif" alt="English RSS" /></a>
+                  </td>
+                </tr>
+              </table>
+            </td>
+          </tr>
+        </tbody>
+      </table>
     </div>
   </body>
 </html>

jq/ttsb.jq

@@ -0,0 +1,49 @@
+# Taiwan Transportation Safety Board Aviation feed generator
+# Required arguments:
+#   $title: Feed title
+#   $lang: Feed language
+#   $link: Feed link
+# 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.
+import "./helpers" as helpers;
+
+{
+    "rss": {
+        "@version": "2.0",
+        "channel": {
+            "title": $title,
+            "description": $title,
+            "link": $link,
+            "language": $lang,
+            "pubDate": (now | strftime("%a, %d %b %Y %T %z")),
+            "docs": "https://cyber.harvard.edu/rss/rss.html",
+            "ttl": 86400,
+            "generator": "ITSB",
+            "item": [.[] | {
+                "title": .children[2].children[0].title,
+                "description": [
+                    .children[3:][]
+                    | (
+                        .["data-th"] + ": " + (
+                            .children[0].text
+                            // .children[0].children[0].text
+                            // "—"
+                        )
+                    )
+                ] | join("\n"),
+                "link": (.children[2].children[0].children[0].href | helpers::urlresolve($link)),
+                "pubDate": (
+                    .children[1].children[0].text
+                    | strptime("%Y-%m-%d")
+                    # strptime returns a [year, month, day, …] array
+                    # To handle the Chinese calendar, which is sometimes used by the TTSB feeds,
+                    # we add 1911 years to the parsed date when it is below 1998,
+                    # 1999 being the year of their earliest investigation.
+                    | .[0] |= if . <= 1998 then (. + 1911) else . end
+                    | mktime
+                    | strftime("%a, %d %b %Y %T %z")
+                )
+            }]
+        }
+    }
+}