filter out whitespace at the beginning and end of lines

harper-latex/src/lib.rs

@@ -1,10 +1,11 @@
 use harper_core::parsers::{self, Parser, PlainEnglish};
+use harper_core::Span;
 use harper_core::Token;
 use harper_tree_sitter::TreeSitterMasker;
 use tree_sitter::Node;
 
 pub struct LatexParser {
-    inner: parsers::Mask<TreeSitterMasker, PlainEnglish>
+    inner: parsers::Mask<TreeSitterMasker, PlainEnglish>,
 }
 
 impl LatexParser {
@@ -35,14 +36,73 @@ impl Default for LatexParser {
         Self {
             inner: parsers::Mask::new(
                 TreeSitterMasker::new(tree_sitter_latex::language(), Self::node_condition),
-                PlainEnglish
-            )
+                PlainEnglish,
+            ),
         }
     }
 }
 
 impl Parser for LatexParser {
     fn parse(&mut self, source: &[char]) -> Vec<Token> {
-        self.inner.parse(source)
+        let mut tokens = Vec::new();
+
+        let mut chars_traversed = 0;
+
+        for line in source.split(|c| *c == '\n') {
+            let mut new_tokens = parse_line(line, &mut self.inner);
+
+            new_tokens.push(Token::new(
+                Span::new_with_len(line.len(), 1),
+                harper_core::TokenKind::Newline(1),
+            ));
+
+            new_tokens
+                .iter_mut()
+                .for_each(|t| t.span.push_by(chars_traversed));
+
+            chars_traversed += line.len() + 1;
+            tokens.append(&mut new_tokens);
+        }
+
+        tokens
     }
 }
+
+fn parse_line(
+    source: &[char],
+    parser: &mut parsers::Mask<TreeSitterMasker, PlainEnglish>,
+) -> Vec<Token> {
+    let actual = without_leading(source);
+
+    if actual.is_empty() {
+        return Vec::new();
+    }
+
+    let source = actual.get_content(source);
+
+    let mut new_tokens = parser.parse(source);
+
+    new_tokens
+        .iter_mut()
+        .for_each(|t| t.span.push_by(actual.start));
+
+    new_tokens
+}
+
+fn without_leading(source: &[char]) -> Span {
+    // Skip over the comment start characters
+    let actual_start = source
+        .iter()
+        .position(|c| !c.is_whitespace())
+        .unwrap_or(source.len());
+
+    // Chop off the end
+    let actual_end = source.len()
+        - source
+            .iter()
+            .rev()
+            .position(|c| !c.is_whitespace())
+            .unwrap_or(0);
+
+    Span::new(actual_start, actual_end)
+}