From 2c90b359dd3800aeff1ac542aef054d1b2ebb4f1 Mon Sep 17 00:00:00 2001
From: Aki Kareha <aki@kareha.org>
Date: Wed, 26 Feb 2025 01:05:56 +0900
Subject: [PATCH] extract meanings of holidays and merge them into CSV

---
 extract | 50 ++++++++++++++++++++++++++++++++++++++++++++++++++
 format  |  4 +++-
 merge   | 41 +++++++++++++++++++++++++++++++++++++++++
 prepare |  3 +++
 update  |  3 ++-
 watch   |  4 +++-
 6 files changed, 102 insertions(+), 3 deletions(-)
 create mode 100755 extract
 create mode 100755 merge
 create mode 100755 prepare

diff --git a/extract b/extract
new file mode 100755
index 0000000..b767d78
--- /dev/null
+++ b/extract
@@ -0,0 +1,50 @@
+#!/usr/bin/perl
+use strict;
+use warnings;
+
+my $input_file = 'summary-jp.html';
+open my $fh_in, '<', $input_file
+  or die "Cannot open $input_file for reading: $!";
+my $output_file = 'meanings-jp.csv';
+open my $fh_out, '>', $output_file
+  or die "Cannot open $output_file for writing: $!";
+
+while ( my $line = <$fh_in> ) {
+    chomp $line;
+    if ( $line =~ m|<h2.*国民の祝日に関する法律.*</h2>| ) {
+        last;
+    }
+}
+
+my $state = "head";
+while ( my $line = <$fh_in> ) {
+    chomp $line;
+
+    if ( $line =~ m|</table>| ) {
+        last;
+    }
+
+    if ( $state eq "head" ) {
+        if ( $line =~ m|<th>(.*)</th>| ) {
+            print $fh_out "$1,";
+            $state = "rule";
+            next;
+        }
+    }
+    elsif ( $state eq "rule" ) {
+        if ( $line =~ m|<td>(.*)</td>| ) {
+            print $fh_out "$1,";
+            $state = "meaning";
+            next;
+        }
+    }
+    elsif ( $state eq "meaning" ) {
+        if ( $line =~ m|<td>(.*)</td>| ) {
+            print $fh_out "$1\n";
+            $state = "head";
+            next;
+        }
+    }
+}
+close $fh_in;
+close $fh_out;
diff --git a/format b/format
index edab75d..b384371 100755
--- a/format
+++ b/format
@@ -1,2 +1,4 @@
 #!/bin/sh
-shfmt -i 2 -ci -w update
+shfmt -i 2 -ci -w watch update
+perltidy -b extract merge
+rm *.bak
diff --git a/merge b/merge
new file mode 100755
index 0000000..dc26aca
--- /dev/null
+++ b/merge
@@ -0,0 +1,41 @@
+#!/usr/bin/perl
+use strict;
+use warnings;
+
+my $meanings_file = 'meanings-jp.csv';
+open my $fh_meanings, '<', $meanings_file
+  or die "Cannot open $meanings_file for reading: $!";
+my $holidays_file = 'holidays-jp.csv';
+open my $fh_holidays, '<', $holidays_file
+  or die "Cannot open $holidays_file for reading: $!";
+my $output_file = 'merged-jp.csv';
+open my $fh_out, '>', $output_file
+  or die "Cannot open $output_file for writing: $!";
+
+my @meanings;
+while ( my $line = <$fh_meanings> ) {
+    chomp $line;
+    my @fields = split /,/, $line;
+
+    push @meanings, \@fields;
+}
+
+while ( my $line = <$fh_holidays> ) {
+    chomp $line;
+    my @fields = split /,/, $line;
+
+    my $date_raw = $fields[0];
+    $date_raw =~ m|(\d+)/(\d+)/(\d+)|;
+    my $date    = sprintf( "%04d-%02d-%02d", $1, $2, $3 );
+    my $name    = $fields[1];
+    my $rule    = "振替休日";
+    my $meaning = "祝日法による休日。";
+    for my $row (@meanings) {
+        if ( $$row[0] eq $name ) {
+            $rule    = $$row[1];
+            $meaning = $$row[2];
+            last;
+        }
+    }
+    print $fh_out "$date,$name,$rule,$meaning\n";
+}
diff --git a/prepare b/prepare
new file mode 100755
index 0000000..3022c31
--- /dev/null
+++ b/prepare
@@ -0,0 +1,3 @@
+#!/bin/sh
+./watch
+./update
diff --git a/update b/update
index 1647d08..bb80672 100755
--- a/update
+++ b/update
@@ -72,9 +72,10 @@ if ! diff -q $RAW_CSV_FILE $TMP_CSV_FILE >/dev/null 2>&1; then
   mv $TMP_CSV_FILE $RAW_CSV_FILE
   nkf -w $RAW_CSV_FILE >$ALL_CSV_FILE
   current_year=$(date +'%Y')
-  tail -n +2 $ALL_CSV_FILE | awk -v cy="$current_year" -F'/' '{ if ($1 >= cy) print }' >$CSV_FILE
+  tail -n +2 $ALL_CSV_FILE | awk -v cy="$current_year" -F'/' '{ if ($1 >= cy) print }' | tr -d '\r' >$CSV_FILE
   log "Changed"
   send_mail
+  ./merge
 else
   log "No Change"
   rm $TMP_CSV_FILE
diff --git a/watch b/watch
index 22376c7..7b0bc65 100755
--- a/watch
+++ b/watch
@@ -63,11 +63,13 @@ curl -sS -L -o $TMP_HTML_FILE $SITE_URL
 if ! diff -q $HTML_FILE $TMP_HTML_FILE >/dev/null 2>&1; then
   CACHE_DIR="cache"
   mkdir -p "$CACHE_DIR"
-  CURRENT_DATETIME=`date "+%Y-%m-%d_%H:%M:%S%z"`
+  CURRENT_DATETIME=$(date "+%Y-%m-%d_%H:%M:%S%z")
   cp $TMP_HTML_FILE "$CACHE_DIR/summary-jp-$CURRENT_DATETIME.html"
   mv $TMP_HTML_FILE $HTML_FILE
   log "Changed"
   send_mail
+  ./extract
+  ./merge 2>/dev/null
 else
   log "No Change"
   rm $TMP_HTML_FILE