fix and improve parser logic to handle errors better

bip/src/lib.rs

@@ -13,6 +13,9 @@ pub enum ParseError {
     ExpectedParens,
     ExpectedSym,
     ExpectedTerm,
+    BadInt,
+    BadStr,
+    BadSym(char),
 }
 
 impl fmt::Display for ParseError {
@@ -22,6 +25,9 @@ impl fmt::Display for ParseError {
             Self::ExpectedParens => write!(f, "expected parenthesis"),
             Self::ExpectedSym => write!(f, "expected symbol"),
             Self::ExpectedTerm => write!(f, "expected semicolon"),
+            Self::BadInt => write!(f, "invalid integer"),
+            Self::BadStr => write!(f, "unterminated string"),
+            Self::BadSym(c) => write!(f, "invalid symbol {c}"),
         }
     }
 }
@@ -75,22 +81,22 @@ impl Symbol {
     fn take<T: Iterator<Item = char>>(
         it: &mut Peekable<T>,
         predicate: fn(c: char) -> bool,
-    ) -> Vec<char> {
+    ) -> ParseResult<Vec<char>> {
         let mut v = Vec::new();
 
         while let Some(c) = it.peek() {
             if !predicate(*c) {
-                break;
+                return Ok(v);
             }
 
             v.push(*c);
             it.next();
         }
 
-        v
+        Err(ParseError::Eof)
     }
 
-    pub fn parse(s: &str) -> Result<Vec<Self>, String> {
+    pub fn parse(s: &str) -> ParseResult<Vec<Self>> {
         let mut syms = Vec::new();
         let mut it = s.chars().peekable();
 
@@ -109,7 +115,8 @@ impl Symbol {
                         it.next();
                     }
 
-                    let v = Symbol::take(&mut it, |c| c.is_numeric());
+                    let v = Symbol::take(&mut it, |c| c.is_numeric())
+                        .map_err(|_| ParseError::BadInt)?;
 
                     if let Ok(n) = v.iter().collect::<String>().parse::<u64>() {
                         syms.push(if sign > 0 {
@@ -118,7 +125,7 @@ impl Symbol {
                             Self::SInt((n as i64 * sign) as u64)
                         });
                     } else {
-                        return Err("invalid integer".to_string());
+                        return Err(ParseError::BadInt);
                     }
 
                     continue;
@@ -129,17 +136,18 @@ impl Symbol {
                 '"' => {
                     it.next();
 
-                    let v = Symbol::take(&mut it, |c| c != '"');
+                    let v = Symbol::take(&mut it, |c| c != '"').map_err(|_| ParseError::BadStr)?;
                     syms.push(Self::Str(v.iter().collect()));
                 }
                 ' ' | '\t' | '\r' => {}
                 _ => {
+                    let err = ParseError::BadSym(c.to_owned());
+
                     if !c.is_alphanumeric() {
-                        return Err(format!("invalid symbol {}", c));
+                        return Err(err);
                     }
 
-                    let v = Symbol::take(&mut it, |c| c.is_alphanumeric());
-
+                    let v = Symbol::take(&mut it, |c| c.is_alphanumeric()).map_err(|_| err)?;
                     syms.push(Self::Id(Some(v.iter().collect())));
                     continue;
                 }
@@ -195,11 +203,13 @@ impl<'a> Parser<'a> {
 
                 self.next();
                 loop {
-                    if *self.peek()? == close {
-                        break;
-                    }
+                    let sym = self.peek().map_err(|_| ParseError::ExpectedParens)?;
 
-                    node.entries.push(self.parse_expr()?);
+                    if *sym == close {
+                        break;
+                    } else {
+                        node.entries.push(self.parse_expr()?);
+                    }
                 }
                 self.expect(close, ParseError::ExpectedParens)?;
                 self.next();

bipc/src/lib.rs

@@ -8,7 +8,6 @@ pub type Result<T> = std::result::Result<T, Error>;
 pub enum Error {
     ReadError,
     BadFile,
-    BadRead(String),
     BadWrite(std::io::Error),
     BadPath(std::io::Error),
     BpError(bp::Error),
@@ -20,7 +19,6 @@ impl fmt::Display for Error {
         match self {
             Self::ReadError => write!(f, "errors found"),
             Self::BadFile => write!(f, "invalid text"),
-            Self::BadRead(e) => write!(f, "{e}"),
             Self::BadWrite(e) => write!(f, "{e}"),
             Self::BadPath(e) => write!(f, "{e}"),
             Self::BpError(e) => write!(f, "{e}"),
@@ -35,7 +33,7 @@ fn read_bip<R: BufRead>(buf: R) -> Result<Vec<bip::Node>> {
         .lines()
         .map(|l| {
             l.map_or(Err(Error::BadFile), |l| {
-                bip::Symbol::parse(&l).map_err(Error::BadRead)
+                bip::Symbol::parse(&l).map_err(Error::ParseError)
             })
         })
         .filter_map(|r| r.map_err(|e| errors.push(e)).ok())
@@ -48,7 +46,7 @@ fn read_bip<R: BufRead>(buf: R) -> Result<Vec<bip::Node>> {
     });
 
     if err_count > 0 {
-        eprintln!("{err_count} errors");
+        eprintln!("{err_count} error{}", if err_count == 1 { "" } else { "s" });
         Err(Error::ReadError)
     } else {
         bip::Parser::new(&syms).parse().map_err(Error::ParseError)