Play around with line-wise parsing

src/parse.rs

@@ -5,7 +5,9 @@ use crate::commands::Command;
 
 use self::line::parse_lines;
 
+mod error;
 mod line;
+mod parser;
 
 pub fn parse(file: &Path) -> anyhow::Result<Vec<Command>> {
     let content = fs::read_to_string(file)?;

src/parse/error.rs

@@ -0,0 +1,32 @@
+use std::error;
+
+#[derive(Debug, thiserror::Error)]
+#[error("line {line}: {reason}")]
+pub struct ParseError {
+    line: usize,
+    reason: Box<dyn error::Error>,
+}
+
+impl ParseError {
+    #[must_use]
+    pub fn new(line: usize, reason: impl error::Error + 'static) -> Self {
+        Self {
+            line,
+            reason: Box::new(reason),
+        }
+    }
+
+    #[must_use]
+    pub fn pack<T>(line: usize, reason: impl error::Error + 'static) -> Result<T, Self> {
+        Err(Self::new(line, reason))
+    }
+}
+
+pub trait ToParseError: error::Error + 'static + Sized {
+    #[must_use]
+    fn at(self, line: usize) -> ParseError {
+        ParseError::new(line, self)
+    }
+}
+
+impl<E: error::Error + 'static> ToParseError for E {}

src/parse/line.rs

@@ -4,6 +4,8 @@ use chrono::NaiveDate;
 
 use crate::commands::{BirthdaySpec, Done, Spec};
 
+use super::error::ParseError;
+
 #[derive(Debug)]
 pub enum Line {
     Empty,
@@ -21,14 +23,16 @@ pub enum Line {
 }
 
 #[derive(Debug, thiserror::Error)]
-pub enum Error {
+pub enum Reason {
-    #[error("line {line}: unknown command {name:?}")]
+    #[error("unknown format")]
-    UnknownCommand { line: usize, name: String },
+    UnknownFormat,
-    #[error("line {line}: unknown format")]
+    #[error("unknown command {0:?}")]
-    UnknownFormat { line: usize },
+    UnknownCommand(String),
+    #[error("empty command body")]
+    EmptyCommand,
 }
 
-type Result<T> = result::Result<T, Error>;
+type Result<T> = result::Result<T, ParseError>;
 
 pub fn parse_lines(content: &str) -> Result<Vec<Line>> {
     content
@@ -46,6 +50,10 @@ fn parse_line(line: usize, content: &str) -> Result<Line> {
     } else if content.starts_with('\t') || content.starts_with(' ') {
         Ok(Line::Indented(content.to_string()))
     } else if let Some((name, rest)) = parse_command(content) {
+        let rest = rest.trim();
+        if rest.is_empty() {
+            return ParseError::pack(line, Reason::EmptyCommand);
+        }
         match name {
             "TASK" => Ok(Line::Task(rest.to_string())),
             "NOTE" => Ok(Line::Note(rest.to_string())),
@@ -56,13 +64,10 @@ fn parse_line(line: usize, content: &str) -> Result<Line> {
             "UNTIL" => parse_datum(rest).map(Line::Until),
             "EXCEPT" => parse_datum(rest).map(Line::Except),
             "DONE" => parse_done(rest),
-            _ => Err(Error::UnknownCommand {
+            _ => ParseError::pack(line, Reason::UnknownCommand(name.to_string())),
-                line,
-                name: name.to_string(),
-            }),
         }
     } else {
-        Err(Error::UnknownFormat { line })
+        ParseError::pack(line, Reason::UnknownFormat)
     }
 }
 

src/parse/parser.rs

@@ -0,0 +1,88 @@
+pub struct Parser<'d> {
+    data: &'d str,
+    index: usize,
+}
+
+#[derive(Debug, thiserror::Error)]
+pub enum Reason {
+    #[error("expected character {expected:?} at {rest:?}")]
+    ExpectedChar { expected: char, rest: String },
+    #[error("expected string {expected:?} at {rest:?}")]
+    ExpectedStr { expected: String, rest: String },
+    #[error("expected whitespace at {rest:?}")]
+    ExpectedWhitespace { rest: String },
+}
+
+impl<'d> Parser<'d> {
+    pub fn new(data: &'d str) -> Self {
+        Self { data, index: 0 }
+    }
+
+    fn rest(&self) -> &'d str {
+        &self.data[self.index..]
+    }
+
+    pub fn peek(&self) -> Option<char> {
+        self.rest().chars().next()
+    }
+
+    pub fn take(&mut self) -> Option<char> {
+        if let Some(c) = self.peek() {
+            self.index += c.len_utf8();
+            Some(c)
+        } else {
+            None
+        }
+    }
+
+    pub fn take_exact(&mut self, c: char) -> Result<(), Reason> {
+        if self.peek() == Some(c) {
+            self.take();
+            Ok(())
+        } else {
+            Err(Reason::ExpectedChar {
+                expected: c,
+                rest: self.rest().to_string(),
+            })
+        }
+    }
+
+    pub fn take_any_whitespace(&mut self) {
+        while let Some(c) = self.peek() {
+            if c.is_whitespace() {
+                self.take();
+            } else {
+                break;
+            }
+        }
+    }
+
+    pub fn take_some_whitespace(&mut self) -> Result<(), Reason> {
+        match self.peek() {
+            Some(c) if c.is_whitespace() => {
+                self.take();
+                self.take_any_whitespace();
+                Ok(())
+            }
+            _ => Err(Reason::ExpectedWhitespace {
+                rest: self.rest().to_string(),
+            }),
+        }
+    }
+
+    pub fn starts_with(&self, pattern: &str) -> bool {
+        self.data.starts_with(pattern)
+    }
+
+    pub fn take_starting_with(&mut self, pattern: &str) -> Result<(), Reason> {
+        if self.starts_with(pattern) {
+            self.index += pattern.len();
+            Ok(())
+        } else {
+            Err(Reason::ExpectedStr {
+                expected: pattern.to_string(),
+                rest: self.rest().to_string(),
+            })
+        }
+    }
+}