From 33d08970967802b375f6fe3ed88aee7534e8b9d3 Mon Sep 17 00:00:00 2001
From: Savanni D'Gerinel <savanni@luminescent-dreams.com>
Date: Sat, 16 Sep 2023 22:01:54 -0400
Subject: [PATCH] Add date parsing

---
 sgf/src/date.rs   |   2 +
 sgf/src/parser.rs | 131 ++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 133 insertions(+)

diff --git a/sgf/src/date.rs b/sgf/src/date.rs
index 3058174..09818d1 100644
--- a/sgf/src/date.rs
+++ b/sgf/src/date.rs
@@ -107,6 +107,7 @@ pub fn parse_date_field(s: &str) -> Result<Vec<Date>, Error> {
     Ok(dates)
 }
 
+/*
 #[cfg(test)]
 mod test {
     use super::*;
@@ -158,3 +159,4 @@ mod test {
         );
     }
 }
+*/
diff --git a/sgf/src/parser.rs b/sgf/src/parser.rs
index 3fc59f9..73fdb54 100644
--- a/sgf/src/parser.rs
+++ b/sgf/src/parser.rs
@@ -1,4 +1,5 @@
 use crate::{Color, Date, Error, GameResult};
+use chrono::{Datelike, NaiveDate};
 use nom::{
     branch::alt,
     bytes::complete::{escaped_transform, tag, take_until1},
@@ -815,6 +816,81 @@ fn parse_text<'a, E: ParseError<&'a str>>() -> impl Parser<&'a str, String, E> {
     }
 }
 
+enum DateSegment {
+    One(i32),
+    Two(i32, i32),
+    Three(i32, i32, i32),
+}
+
+fn parse_date_field<'a, E: ParseError<&'a str>>() -> impl Parser<&'a str, Vec<Date>, E> {
+    |input| {
+        let (input, first_date) = parse_date().parse(input)?;
+        let (input, mut more_dates) = many0(parse_date_segment())(input)?;
+
+        let mut date_segments = vec![first_date];
+        date_segments.append(&mut more_dates);
+
+        let mut dates = vec![];
+        let mut most_recent = None;
+
+        for date_segment in date_segments {
+            let new_date = match date_segment {
+                DateSegment::One(v) => match most_recent {
+                    Some(Date::Year(_)) => Date::Year(v),
+                    Some(Date::YearMonth(y, _)) => Date::YearMonth(y, v as u32),
+                    Some(Date::Date(d)) => Date::Date(d.clone().with_day(v as u32).unwrap()),
+                    None => Date::Year(v),
+                },
+                DateSegment::Two(y, m) => Date::YearMonth(y, m as u32),
+                DateSegment::Three(y, m, d) => {
+                    Date::Date(chrono::NaiveDate::from_ymd_opt(y, m as u32, d as u32).unwrap())
+                }
+            };
+            dates.push(new_date.clone());
+            most_recent = Some(new_date);
+        }
+
+        Ok((input, dates))
+    }
+}
+
+fn parse_date_segment<'a, E: ParseError<&'a str>>() -> impl Parser<&'a str, DateSegment, E> {
+    |input| {
+        let (input, _) = tag(",")(input)?;
+        let (input, element) = parse_date().parse(input)?;
+
+        Ok((input, element))
+    }
+}
+
+fn parse_date<'a, E: ParseError<&'a str>>() -> impl Parser<&'a str, DateSegment, E> {
+    alt((parse_three(), parse_two(), parse_one()))
+}
+
+fn parse_one<'a, E: ParseError<&'a str>>() -> impl Parser<&'a str, DateSegment, E> {
+    parse_number().map(|year| DateSegment::One(year))
+}
+
+fn parse_two<'a, E: ParseError<&'a str>>() -> impl Parser<&'a str, DateSegment, E> {
+    |input| {
+        let (input, year) = parse_number().parse(input)?;
+        let (input, _) = tag("-")(input)?;
+        let (input, month) = parse_number().parse(input)?;
+        Ok((input, DateSegment::Two(year, month)))
+    }
+}
+
+fn parse_three<'a, E: ParseError<&'a str>>() -> impl Parser<&'a str, DateSegment, E> {
+    |input| {
+        let (input, year) = parse_number().parse(input)?;
+        let (input, _) = tag("-")(input)?;
+        let (input, month) = parse_number().parse(input)?;
+        let (input, _) = tag("-")(input)?;
+        let (input, day) = parse_number().parse(input)?;
+        Ok((input, DateSegment::Three(year, month, day)))
+    }
+}
+
 #[cfg(test)]
 mod test {
     use super::*;
@@ -1059,3 +1135,58 @@ k<. Hard line breaks are all other linebreaks.",
         parse_tree::<nom::error::VerboseError<&str>>(&data).unwrap();
     }
 }
+
+#[cfg(test)]
+mod date_test {
+    use super::*;
+    use cool_asserts::assert_matches;
+
+    #[test]
+    fn it_parses_a_year() {
+        assert_matches!(parse_date_field::<nom::error::VerboseError<&str>>().parse("1996"), Ok((_, date)) => {
+            assert_eq!(date, vec![Date::Year(1996)]);
+        });
+    }
+
+    #[test]
+    fn it_parses_a_month() {
+        assert_matches!(
+            parse_date_field::<nom::error::VerboseError<&str>>().parse("1996-12"),
+            Ok((_, date)) => assert_eq!(date, vec![Date::YearMonth(1996, 12)])
+        );
+    }
+
+    #[test]
+    fn it_parses_a_date() {
+        assert_matches!(
+            parse_date_field::<nom::error::VerboseError<&str>>().parse("1996-12-27"),
+            Ok((_, date)) => assert_eq!(date, vec![Date::Date(
+                NaiveDate::from_ymd_opt(1996, 12, 27).unwrap()
+            )])
+        );
+    }
+
+    #[test]
+    fn it_parses_date_continuation() {
+        assert_matches!(
+            parse_date_field::<nom::error::VerboseError<&str>>().parse("1996-12-27,28"),
+            Ok((_, date)) => assert_eq!(date, vec![
+                Date::Date(NaiveDate::from_ymd_opt(1996, 12, 27).unwrap()),
+                Date::Date(NaiveDate::from_ymd_opt(1996, 12, 28).unwrap())
+            ])
+        );
+    }
+
+    #[test]
+    fn it_parses_date_crossing_year_boundary() {
+        assert_matches!(
+            parse_date_field::<nom::error::VerboseError<&str>>().parse("1996-12-27,28,1997-01-03,04"),
+            Ok((_, date)) => assert_eq!(date, vec![
+                Date::Date(NaiveDate::from_ymd_opt(1996, 12, 27).unwrap()),
+                Date::Date(NaiveDate::from_ymd_opt(1996, 12, 28).unwrap()),
+                Date::Date(NaiveDate::from_ymd_opt(1997, 1, 3).unwrap()),
+                Date::Date(NaiveDate::from_ymd_opt(1997, 1, 4).unwrap()),
+            ])
+        );
+    }
+}