From a40e26f8cef4103e94be13edcc0044290ac141d7 Mon Sep 17 00:00:00 2001
From: Mx Kookie <kookie@spacekookie.de>
Date: Thu, 10 Dec 2020 14:30:38 +0000
Subject: cassiopeia: adding new format lexer built on logos

---
 apps/cassiopeia/src/format/lexer.rs | 116 ++++++++++++++++++++++++++++++++++++
 apps/cassiopeia/src/format/mod.rs   |   3 +
 2 files changed, 119 insertions(+)
 create mode 100644 apps/cassiopeia/src/format/lexer.rs
 create mode 100644 apps/cassiopeia/src/format/mod.rs

(limited to 'apps/cassiopeia/src/format')

diff --git a/apps/cassiopeia/src/format/lexer.rs b/apps/cassiopeia/src/format/lexer.rs
new file mode 100644
index 000000000000..f062ca4238c1
--- /dev/null
+++ b/apps/cassiopeia/src/format/lexer.rs
@@ -0,0 +1,116 @@
+//! Cassiopeia file lexer
+
+use logos::Logos;
+
+/// A basic line lexer type
+///
+/// This lexer distinguishes between comments, and keyword lines.  It
+/// does not attempt to parse the line specifics.  This is what the
+/// content lexer is for.
+#[derive(Logos, Debug, PartialEq)]
+enum Line {
+
+    #[token("HEADER")]
+    Header,
+
+    #[token("START")]
+    Start,
+
+    #[token("STOP")]
+    Stop,
+
+    #[token("INVOICE")]
+    Invoice,
+    
+    #[regex(r"\w+=[^,$]+[,$]")]
+    HeaderData,
+
+    // FIXME: this will have a leading whitespace that we could remove
+    // with ^\w, but logos does not support this at the moment
+    #[regex(r"[0-9-:+ ]+")]
+    Date,
+
+    #[token(" ", logos::skip)]
+    Space,
+    
+    #[error]
+    Error,
+}
+
+
+// pub fn test_this() {
+//     // let mut lex = Line::lexer("HEADER version=0.0.0,location=Berlin,");
+//     let mut lex = Line::lexer("START 2020-11-11 13:00:00+01:00");
+
+//     while let Some(t) = lex.next() {
+//         println!("{:?}: {}", t, lex.slice());
+//     }
+// }
+
+
+#[test]
+fn basic_header() {
+    let mut lex = Line::lexer("HEADER version=0.0.0,location=Berlin Lichtenberg,");
+
+    assert_eq!(lex.next(), Some(Line::Header));
+    assert_eq!(lex.span(), 0..6);
+    assert_eq!(lex.slice(), "HEADER");
+
+    assert_eq!(lex.next(), Some(Line::HeaderData));
+    assert_eq!(lex.span(), 7..21);
+    assert_eq!(lex.slice(), "version=0.0.0,");
+
+    assert_eq!(lex.next(), Some(Line::HeaderData));
+    assert_eq!(lex.span(), 21..49);
+    assert_eq!(lex.slice(), "location=Berlin Lichtenberg,");
+
+    assert_eq!(lex.next(), None);
+}
+
+
+#[test]
+fn basic_start() {
+    let mut lex = Line::lexer("START 2020-11-11 13:00:00+01:00");
+
+    assert_eq!(lex.next(), Some(Line::Start));
+    assert_eq!(lex.span(), 0..5);
+    assert_eq!(lex.slice(), "START");
+
+    assert_eq!(lex.next(), Some(Line::Date));
+    assert_eq!(lex.span(), 5..31);
+    assert_eq!(lex.slice(), " 2020-11-11 13:00:00+01:00");
+
+    assert_eq!(lex.next(), None);
+}
+
+
+#[test]
+fn basic_stop() {
+    let mut lex = Line::lexer("STOP 2020-11-11 13:00:00+01:00");
+
+    assert_eq!(lex.next(), Some(Line::Stop));
+    assert_eq!(lex.span(), 0..4);
+    assert_eq!(lex.slice(), "STOP");
+
+    assert_eq!(lex.next(), Some(Line::Date));
+    assert_eq!(lex.span(), 4..30);
+    assert_eq!(lex.slice(), " 2020-11-11 13:00:00+01:00");
+
+    assert_eq!(lex.next(), None);
+}
+
+
+#[test]
+fn basic_invoice() {
+    let mut lex = Line::lexer("INVOICE 2020-11-11 13:00:00+01:00");
+
+    assert_eq!(lex.next(), Some(Line::Invoice));
+    assert_eq!(lex.span(), 0..7);
+    assert_eq!(lex.slice(), "INVOICE");
+
+    assert_eq!(lex.next(), Some(Line::Date));
+    assert_eq!(lex.span(), 7..33);
+    assert_eq!(lex.slice(), " 2020-11-11 13:00:00+01:00");
+
+    assert_eq!(lex.next(), None);
+}
diff --git a/apps/cassiopeia/src/format/mod.rs b/apps/cassiopeia/src/format/mod.rs
new file mode 100644
index 000000000000..766bb22be13e
--- /dev/null
+++ b/apps/cassiopeia/src/format/mod.rs
@@ -0,0 +1,3 @@
+//! cassiopeia file format handling
+
+pub(crate) mod lexer;
-- 
cgit v1.2.3