From 424801df90aa7141ea05b069cada386987e14e73 Mon Sep 17 00:00:00 2001
From: Mike Small <smallm@sdf.org>
Date: Mon, 15 Jun 2020 19:31:19 -0400
Subject: [PATCH] Initial commit. Francais facile script.

---
 Makefile |  7 +++++++
 README   |  5 +++++
 frfac    | 63 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 75 insertions(+)
 create mode 100644 Makefile
 create mode 100644 README
 create mode 100755 frfac

diff --git a/Makefile b/Makefile
new file mode 100644
index 0000000..e141e82
--- /dev/null
+++ b/Makefile
@@ -0,0 +1,7 @@
+check:
+	perl -wc frfac
+	perlcritic frfac
+
+install:
+	sudo install -o root -g wheel -m 0755 -d /usr/local/bin
+	sudo install -o root -g wheel -m 0555 frfac /usr/local/bin
diff --git a/README b/README
new file mode 100644
index 0000000..b8d8d71
--- /dev/null
+++ b/README
@@ -0,0 +1,5 @@
+Scripts to extract media URLs from websites
+
+1. frfac - url of most recent Francais Facile episode from RFI.
+Usage: ftp $(frfac)
+(or substitute another downloader for ftp.)
diff --git a/frfac b/frfac
new file mode 100755
index 0000000..f070826
--- /dev/null
+++ b/frfac
@@ -0,0 +1,63 @@
+#!/usr/pkg/bin/perl
+# Print the URI of the most recent rfi francais facile episode.
+# Feed this to a downloader. (This script doesn't itself download.)
+
+use File::Temp;
+use HTML::LinkExtractor;
+use List::Util 'uniq';
+use Readonly;
+use URI;
+use URI::Fetch;
+use URI::Find;
+use strict;
+use warnings;
+use feature 'say';
+
+my $BASE_URI =
+	'http://www.rfi.fr/fr/podcasts/journal-fran%C3%A7ais-facile/';
+
+# find the list of available episodes
+
+my $res = URI::Fetch->fetch($BASE_URI) or die URI::Fetch->errstr;
+die "URI::Fetch reports failure, stopping" unless $res;
+my $epis_list_html = $res->content;
+
+my @episode_links;
+my $LX = HTML::LinkExtractor->new(undef, $BASE_URI);
+$LX->parse(\$epis_list_html);
+for my $link (@{$LX->links}) {
+    next unless $link->{href};
+    if ($link->{href} =~ /facile-\d+-\d+h\d+-gmt\s*$/) {
+	push @episode_links, $link->{href};
+    }
+}
+die "Français facile's episode list html has changed, stopping"
+    unless @episode_links > 0;
+@episode_links = sort @episode_links;
+#say for @episode_links;
+my $latest_episode = pop @episode_links;
+#say "Latest: ", $latest_episode;
+
+# fetch the mp3 from the most recent episode page
+
+my @mp3s;
+sub collect_mp3s {
+    my ($uri, $uri_text) = @_;
+    push @mp3s, $uri_text if $uri_text =~ /facile_\d+\.mp3$/;
+}
+
+$res = URI::Fetch->fetch($latest_episode) or die URI::Fetch->errstr;
+die "URI::Fetch reports failure, stopping" unless $res;
+my $episode_html = $res->content;
+
+# The page has an inline json script with forward slash escaped.
+$episode_html =~ s{\\/}{/}g;
+
+my $uri_finder = URI::Find->new(\&collect_mp3s);
+$uri_finder->find(\$episode_html);
+
+@mp3s = uniq @mp3s;
+die sprintf("Found %d show mp3s instead of one, stopping", scalar(@mp3s))
+    unless @mp3s == 1;
+say for @mp3s;
+