Add dates

src/address.pest

@@ -3,13 +3,15 @@
 WSP = _{ " " | "\t" } // todo: such alias slow down the parser
 CRLF = _{ "\r\n" }
 
-// 3.2.1.   Quoted characters
+// 3.2.  Lexical Tokens
+
+// 3.2.1.    Quoted characters
 
 VCHAR = _{ '\x21'..'\x7E' }
 vchar_seq = { VCHAR+ }
 quoted_pair = { "\\" ~ (VCHAR | WSP) }
 
-// 3.2.2.   Folding White Space and Comments
+// 3.2.2.    Folding White Space and Comments
 
 FWS = _{ (WSP* ~ CRLF)? ~ WSP+ }
 CTEXT = _{ '!'..'\'' | '*'..'[' | ']'..'~' }
@@ -18,7 +20,7 @@ ccontent = _{ ctext_seq | quoted_pair | comment }
 comment = { "(" ~ (FWS? ~ ccontent)* ~ FWS? ~ ")" }
 CFWS = _{ ((FWS? ~ comment)+ ~ FWS?) | FWS }
 
-// 3.2.3.   Atom
+// 3.2.3.    Atom
 
 ATEXT = _{ ASCII_ALPHANUMERIC | "!" | "#" | "$" | "%" | "&" | "'" | "*" | "+" | "-" | "/" | "=" | "?" | "^" | "_" | "`" | "{" | "|" | "}" | "~" }
 atext_seq = { ATEXT+ }
@@ -27,14 +29,30 @@ dot_atom_text = { atext_seq ~ ("." ~ atext_seq)* }
 dot_atom = _{ CFWS? ~ dot_atom_text ~ CFWS? }
 specials = _{ "(" | ")" | "<" | ">" | "[" | "]" | ":" | ";" | "@" | "\\" | "," | "." | "\"" }
 
-// 3.2.4.   Quoted Strings
+// 3.2.4.    Quoted Strings
 
 QTEXT = _{ "!" | '#'..'[' | ']'..'~' }
 qtext_seq = { QTEXT+ }
 quoted_string = { CFWS? ~ "\"" ~ (FWS? ~ (qtext_seq | quoted_pair))* ~ FWS? ~ "\"" ~ CFWS? }
 
-// 3.2.5.   Miscellaneous Tokens
+// 3.2.5.    Miscellaneous Tokens
 
 word = { atom | quoted_string }
 phrase = { word+ }
 unstructured = { (FWS? ~ vchar_seq)* ~ WSP*}
+
+// 3.3.  Date and Time Specification
+
+date_time = { (day_of_week ~ ",")? ~ date ~ time ~ CFWS? }
+day_of_week = _{ FWS? ~ day_name }
+day_name = { "Mon" | "Tue" | "Wed" | "Thu" | "Fri" | "Sat" | "Sun" }
+date = _{ (FWS? ~ day ~ FWS) ~ month ~ (FWS ~ year ~ FWS) } 
+day = { ASCII_DIGIT{1,2} } // In the spec FWS are included in day, year and zone but I moved them one level up
+month = { "Jan" | "Feb" | "Mar" | "Apr" | "May" | "Jun" | "Jul" | "Aug" | "Sep" | "Oct" | "Nov" | "Dec" }
+year = { ASCII_DIGIT{4} }
+time = _{ time_of_day ~ (FWS ~ zone) }
+time_of_day = _{ hour ~ ":" ~ minute ~ (":" ~ second)? }
+hour = { ASCII_DIGIT{2} }
+minute = { ASCII_DIGIT{2} }
+second = { ASCII_DIGIT{2} }
+zone = { ("+" | "-") ~ ASCII_DIGIT{4} }

src/lib.rs

@@ -103,3 +103,24 @@ fn test_quoted_string() {
     let qtext_seqs = quoted_string.children().map(|c| c.as_str()).collect::<Vec<_>>();
     assert_eq!(qtext_seqs, vec!["quoted", "\\ ", "test"]);
 }
+
+#[test]
+fn test_date() {
+    let input = "Mon, 5 May 2003 18:58:55 +0100 trail";
+    let output = Parser::parse_date_time(input).map_err(|e| e.print(input)).unwrap();
+    let date_time = output.into_iter().next().unwrap();
+    let children = date_time.children().map(|c| c.as_str()).collect::<Vec<_>>();
+    assert_eq!(children, vec!["Mon", "5", "May", "2003", "18", "58", "55", "+0100"]);
+
+    let input = "3 Jan 2009 18:15:05 +0000";
+    let output = Parser::parse_date_time(input).map_err(|e| e.print(input)).unwrap();
+    let date_time = output.into_iter().next().unwrap();
+    let children = date_time.children().map(|c| c.as_str()).collect::<Vec<_>>();
+    assert_eq!(children, vec!["3", "Jan", "2009", "18", "15", "05", "+0000"]);
+
+    let input = "14 Jul 1789   14:00:00 +0100";
+    let output = Parser::parse_date_time(input).map_err(|e| e.print(input)).unwrap();
+    let date_time = output.into_iter().next().unwrap();
+    let children = date_time.children().map(|c| c.as_str()).collect::<Vec<_>>();
+    assert_eq!(children, vec!["14", "Jul", "1789", "14", "00", "00", "+0100"]);
+}