From 9372b46a1636eb6b18b3df627eb62b324dfce3f6 Mon Sep 17 00:00:00 2001
From: Kévin Le Gouguec <kevin.legouguec@gmail.com>
Date: Tue, 22 Feb 2022 22:24:31 +0100
Subject: [wip] Add publication dates

---
 build-feed.py | 57 +++++++++++++++++++++++++++++++++++++++++++++++----------
 1 file changed, 47 insertions(+), 10 deletions(-)

(limited to 'build-feed.py')

diff --git a/build-feed.py b/build-feed.py
index fb85cab..2359344 100755
--- a/build-feed.py
+++ b/build-feed.py
@@ -1,6 +1,8 @@
 #!/usr/bin/env python3
 
 from datetime import datetime
+import json
+import re
 from sys import argv
 from urllib.parse import urljoin
 
@@ -16,6 +18,20 @@ from helpers import (
 )
 
 
+# TODO: handle timezones correctly.
+# Places to disambiguate:
+#
+# - concerts.in:
+#   either add the zone explicitly, or deduce it from the place,
+#   assuming all times in concerts.in are local times.
+#
+# - concerts-pubdates.json:
+#   just add the zone explicitly.
+#
+# Until then, assume all these "naive times" describe the same timezone
+# (CET/CEST).
+
+
 TIMEZONE = datetime.now().astimezone().tzinfo
 NOW = datetime.now(tz=TIMEZONE)
 DATE_FORMAT = '%-d %b %Y %H:%M %z'
@@ -59,6 +75,9 @@ def join(sequence, joiner_factory):
         result.append(joiner_factory())
 
 
+CDATA_INDENT = 8*' '
+
+
 def cdata_concert(concert, lang):
     formatters = DATE_FORMATTERS[lang]
 
@@ -77,12 +96,15 @@ def cdata_concert(concert, lang):
         *(E.p(line) for line in concert.instructions.splitlines()),
     ))
 
-    return CDATA('\n'.join(
-        tostring(b, encoding='utf-8').decode() for b in blocks
-    ))
+    html_blocks = (tostring(b, encoding='utf-8').decode() for b in blocks)
+
+    cdata = '\n' + '\n'.join(html_blocks) + '\n'
+    cdata = re.sub('^', CDATA_INDENT, cdata, flags=re.MULTILINE)
+
+    return CDATA(cdata)
 
 
-def generate_concert(concert, concerts_url, lang):
+def generate_concert(concert, concerts_url, pubdates, lang):
     formatters = LOCALIZED_FORMATS[lang]
 
     with tmplocale(lang):
@@ -90,21 +112,32 @@ def generate_concert(concert, concerts_url, lang):
 
     anchor = f'concert-{concert.time.strftime("%F")}'
 
-    return E.item(
+    item = E.item(
         E.title(title),
         E.link(f'{concerts_url}#{anchor}'),
         E.description(cdata_concert(concert, lang)),
     )
 
+    pubdate_str = pubdates[concert.time.isoformat(timespec='minutes')]
+
+    if pubdate_str is not None:
+        pubdate = datetime.fromisoformat(pubdate_str).replace(tzinfo=TIMEZONE)
+        item.append(E.pubDate(pubdate.strftime(DATE_FORMAT)))
+
+    return item
+
+
+def generate_concerts(concerts_src, concerts_url, concerts_pubdates, lang):
+    with open(concerts_pubdates) as pubdates_file:
+        pubdates = json.load(pubdates_file)
 
-def generate_concerts(concerts_src, concerts_url, lang):
     return tuple(
-        generate_concert(c, concerts_url, lang)
+        generate_concert(c, concerts_url, pubdates, lang)
         for c in read_concerts(concerts_src)
     )
 
 
-def main(concerts_src, feed_dst, domain):
+def main(concerts_src, feed_dst, concerts_pubdates, domain):
     lang = guess_language(concerts_src)
     text = LOCALIZED_TEXT[lang]
 
@@ -114,6 +147,10 @@ def main(concerts_src, feed_dst, domain):
 
     now_formatted = NOW.strftime(DATE_FORMAT)
 
+    concerts = generate_concerts(
+        concerts_src, concerts_url, concerts_pubdates, lang
+    )
+
     rss = E.rss(
         E.channel(
             E.title(text['title']),
@@ -126,7 +163,7 @@ def main(concerts_src, feed_dst, domain):
             E.lastBuildDate(now_formatted),
             E.pubDate(now_formatted),
             E.language(lang),
-            *generate_concerts(concerts_src, concerts_url, lang),
+            *concerts,
         ),
         version='2.0',
     )
@@ -138,4 +175,4 @@ def main(concerts_src, feed_dst, domain):
 
 
 if __name__ == '__main__':
-    main(argv[1], argv[2], argv[3])
+    main(*argv[1:])
-- 
cgit v1.2.3