refactor md parser, TODO: parse_str

.gitea/workflows/cargo-test.yml

@@ -0,0 +1,23 @@
+name: Test
+
+on:
+  push:
+    branches: [ "master" ]
+  pull_request:
+    branches: [ "master" ]
+
+    env:
+      CARGO_TERM_COLOR: always
+
+    jobs:
+      build:
+        name: Cargo test
+        runs-on: ubuntu-latest
+
+        steps:
+          - name: Build
+            uses: actions/checkout@v4
+            run: cargo build --verbose
+          
+          - name: Run tests
+            run: cargo test --verbose

cracked_md/src/ast.rs

@@ -171,8 +171,11 @@ mod convert_md_to_html_test {
     #[test]
     fn single_header() {
         let md = "# Header 1";
-
-        let html = parse(md).to_html();
+        let ast = match parse(md) {
+            Ok(a) => a,
+            Err(e) => panic!("{}", e),
+        };
+        let html = ast.to_html();
 
         assert_eq!(
             html,
@@ -180,11 +183,22 @@ mod convert_md_to_html_test {
         );
     }
 
+    #[test]
+    fn single_header_wrong_format() {
+        let md = "#Whoops";
+        let ast = parse(md);
+
+        assert!(ast.is_err());
+    }
+
     #[test]
     fn nested_bold_headers_and_nested_code_paragraph() {
         let md = "# *Bold* header 1\n## Header 2\nrun `sudo rm -rf /` on your computer";
-
-        let html = parse(md).to_html();
+        let ast = match parse(md) {
+            Ok(a) => a,
+            Err(e) => panic!("{}", e),
+        };
+        let html = ast.to_html();
 
         assert_eq!(
             html,
@@ -192,3 +206,21 @@ mod convert_md_to_html_test {
         );
     }
 }
+
+#[cfg(test)]
+mod parse_real_md {
+    use std::fs;
+
+    use crate::parser::parse;
+
+    #[test]
+    fn go() {
+        let file = "./test.md";
+        let md = fs::read_to_string(file).expect("reading ./test.md failed");
+
+        let _ast = match parse(&md).map_err(|e| e.set_file(file.into())) {
+            Ok(a) => a,
+            Err(e) => panic!("{}", e),
+        };
+    }
+}

cracked_md/src/lib.rs

@@ -1,4 +1,4 @@
-#![deny(dead_code, unused_imports)]
+#![deny(unused_imports)]
 
 use fstools::crawl_fs;
 use parser::parse;
@@ -11,9 +11,91 @@ use std::{
 use to_html::ToHtml;
 
 pub mod ast;
+mod parse_trait;
 pub mod parser;
 pub mod to_html;
 
+#[derive(Debug)]
+pub struct MdParseError {
+    file: Option<PathBuf>,
+    line: Option<usize>,
+    //col: Option<usize>,
+    expected: String,
+    got: String,
+}
+
+impl MdParseError {
+    pub fn new(expected: impl ToString, got: impl ToString) -> Self {
+        Self {
+            file: None,
+            line: None,
+            //col: None,
+            expected: expected.to_string(),
+            got: got.to_string(),
+        }
+    }
+
+    pub fn from_line(line: usize, expected: impl ToString, got: impl ToString) -> Self {
+        Self {
+            file: None,
+            line: Some(line),
+            //col: None,
+            expected: expected.to_string(),
+            got: got.to_string(),
+        }
+    }
+
+    /*
+    pub fn from_col(col: usize, expected: impl ToString, got: impl ToString) -> Self {
+        Self {
+            file: None,
+            line: None,
+            col: Some(col),
+            expected: expected.to_string(),
+            got: got.to_string(),
+        }
+    }
+    */
+
+    pub fn set_line(self, line: usize) -> Self {
+        Self {
+            file: self.file,
+            line: Some(line),
+            //col: self.col,
+            expected: self.expected,
+            got: self.got,
+        }
+    }
+
+    pub fn set_file(self, file: PathBuf) -> Self {
+        Self {
+            file: Some(file),
+            line: self.line,
+            //col: self.col,
+            expected: self.expected,
+            got: self.got,
+        }
+    }
+}
+
+impl Display for MdParseError {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        // no error message :/
+        let file = self.file.clone().unwrap_or("<unknown>".into());
+        write!(
+            f,
+            "Parse error in '{}' on line {}: expected '{}', got '{}'",
+            file.display(),
+            self.line.unwrap_or(0),
+            //self.col.unwrap_or(0),
+            self.expected,
+            self.got
+        )
+    }
+}
+
+impl std::error::Error for MdParseError {}
+
 #[derive(Debug)]
 pub enum Error {
     OutDirIsNotEmpty,
@@ -24,6 +106,7 @@ pub enum Error {
     FileWrite,
     FileCreate,
     DirCreate,
+    Parse(MdParseError),
 }
 
 impl Display for Error {
@@ -32,6 +115,12 @@ impl Display for Error {
     }
 }
 
+impl From<MdParseError> for Error {
+    fn from(value: MdParseError) -> Self {
+        Error::Parse(value)
+    }
+}
+
 impl std::error::Error for Error {}
 
 type Result<T> = std::result::Result<T, crate::Error>;
@@ -44,7 +133,7 @@ pub fn generate(indir: &PathBuf, outdir: &PathBuf, force: bool) -> Result<()> {
 
         // read and parse md file
         let content = fs::read_to_string(&fullpath).map_err(|_e| Error::FileRead)?;
-        let html = parse(&content).to_html();
+        let html = parse(&content)?.to_html();
 
         // write html data to file
         let mut newpath = outdir.to_owned();

cracked_md/src/parse_trait.rs

@@ -0,0 +1,111 @@
+/*
+
+use crate::MdParseError;
+
+pub type Pattern<T> = Vec<PatternToken<T>>;
+
+pub enum PatternToken<T> {
+    Once(T),
+    Optional(T),
+    AtLeastOnce(T),
+    NTimes(T),
+}
+
+/// panics: on invalid pattern
+pub fn char_pattern(s: &str) -> Pattern<char> {
+    let mut s_chars = s.chars().peekable();
+    let mut pat: Pattern<char> = Vec::new();
+    while let Some(token) = s_chars.next() {
+        pat.push(if let Some(&next) = s_chars.peek() {
+            match next {
+                '?' => {
+                    s_chars.next().unwrap();
+                    PatternToken::Optional(token)
+                }
+                '+' => {
+                    s_chars.next().unwrap();
+                    PatternToken::AtLeastOnce(token)
+                }
+                '*' => {
+                    s_chars.next().unwrap();
+                    PatternToken::NTimes(token)
+                }
+                _ => PatternToken::Once(token),
+            }
+        } else {
+            PatternToken::Once(token)
+        });
+    }
+    pat
+}
+
+pub trait ParsePattern: Iterator + Clone {
+    fn parse<T>(&mut self, expect: Pattern<T>) -> Result<Vec<Self::Item>, MdParseError>
+    where
+        T: PartialEq<<Self as Iterator>::Item>,
+    {
+        let mut consumed = Vec::new();
+        let mut cloned = self.clone();
+
+        for pat_token in expect {
+            match pat_token {
+                PatternToken::Once(c) => {
+                    if !cloned.next().map(|v| c == v).unwrap_or(false) {
+                        return None;
+                    }
+                }
+                PatternToken::Optional(c) => if cloned.peek().map(|v| c == *v).unwrap_or(false) {},
+            }
+        }
+
+        *self = cloned;
+
+        Some(consumed)
+    }
+}
+*/
+
+pub trait Parse: Iterator {
+    fn follows(&mut self, token: char) -> bool;
+
+    fn parse_token(&mut self, token: char) -> bool {
+        if self.follows(token) {
+            let _ = self.next();
+            true
+        } else {
+            false
+        }
+    }
+
+    fn parse_str(&mut self, _tokens: &str) -> bool {
+        todo!()
+    }
+}
+
+impl Parse for std::iter::Peekable<std::str::Chars<'_>> {
+    fn follows(&mut self, token: char) -> bool {
+        self.peek().map(|c| c == &token).unwrap_or(false)
+    }
+}
+
+impl Parse for std::iter::Peekable<std::iter::Enumerate<std::str::Chars<'_>>> {
+    fn follows(&mut self, token: char) -> bool {
+        self.peekable()
+            .peek()
+            .map(|&(_i, c)| c == token)
+            .unwrap_or(false)
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+
+    #[test]
+    fn chars_parse_follows_double() {
+        let mut c = "abc".chars().peekable();
+
+        assert!(c.follows('a'));
+        assert!(c.follows('a'));
+    }
+}

cracked_md/src/parser.rs

@@ -3,14 +3,15 @@ mod inline;
 
 use block::parse_blocks;
 
-use crate::ast::Document;
+use crate::{MdParseError, ast::Document};
 
-pub fn parse(s: &str) -> Document {
-    Document {
-        blocks: parse_blocks(s),
-    }
+pub fn parse(s: &str) -> Result<Document, MdParseError> {
+    Ok(Document {
+        blocks: parse_blocks(s)?,
+    })
 }
 
+/*
 #[cfg(test)]
 mod test {
     use crate::ast::*;
@@ -145,3 +146,4 @@ mod test {
         );
     }
 }
+// */

cracked_md/src/parser/block.rs

@@ -1,22 +1,106 @@
-use crate::ast::Block;
-
 use super::inline::parse_inlines;
+use crate::{MdParseError, ast::Block};
 
-pub fn parse_blocks(input: &str) -> Vec<Block> {
+use crate::parse_trait::Parse;
+
+pub fn parse_blocks(input: &str) -> Result<Vec<Block>, MdParseError> {
+    let mut blocks = Vec::new();
+    let mut lines = input.lines().enumerate().peekable();
+
+    while let Some((i, line)) = lines.next() {
+        let mut line_chars = line.chars().peekable();
+
+        // header
+        let mut heading_level = 0;
+        while line_chars.parse_token('#') {
+            if heading_level < 6 {
+                heading_level += 1;
+            }
+        }
+        if heading_level > 0 {
+            if !line_chars.parse_token(' ') {
+                Err(MdParseError::from_line(
+                    i + 1,
+                    "<space> after #",
+                    "no <space>",
+                ))?;
+            }
+            let line_content: String = line_chars.collect();
+            blocks.push(Block::Heading {
+                level: heading_level,
+                content: parse_inlines(&line_content)?,
+            });
+            continue;
+        }
+
+        // quote TODO
+        /*
+        if line_chars.parse_str("> ") {
+            let content: String = line_chars.collect();
+            let quote_blocks = parse_blocks(&content).map_err(|e| e.set_line(i + 1))?;
+            blocks.push(Block::Quote(quote_blocks));
+            continue;
+        }
+        */
+
+        // code
+        if line_chars.parse_str("```") {
+            let lang_line: String = line_chars.collect();
+            let lang = if lang_line.is_empty() {
+                None
+            } else {
+                Some(lang_line)
+            };
+            let mut code = String::new();
+
+            for (j, line) in lines.by_ref() {
+                let mut code_line_chars = line.chars().peekable();
+                // code block end
+                if code_line_chars.parse_str("```") {
+                    let remaining: String = code_line_chars.collect();
+                    if remaining.is_empty() {
+                        blocks.push(Block::Code {
+                            language: lang,
+                            content: code,
+                        });
+                        break;
+                    } else {
+                        Err(MdParseError::from_line(
+                            j + 1,
+                            "```",
+                            format!("```{}", remaining),
+                        ))?;
+                    }
+                } else {
+                    code.push_str(line);
+                    code.push('\n');
+                }
+            }
+            Err(MdParseError::from_line(i + 1, "a terminating '```'", ""))?;
+        }
+
+        // lists TODO
+    }
+
+    Ok(blocks)
+}
+
+/*
+pub fn parse_blocks(input: &str) -> Result<Vec<Block>, MdParseError> {
     let mut blocks = Vec::new();
 
-    let mut lines = input.lines().peekable();
+    let mut lines = input.lines().enumerate().peekable();
 
-    while let Some(line) = lines.next() {
+    while let Some((i, line)) = lines.next() {
         if line.starts_with("#") {
             let level = line.chars().take_while(|&c| c == '#').count() as u8;
             let text = line[level as usize..].trim();
             blocks.push(Block::Heading {
                 level,
-                content: parse_inlines(text),
+                content: parse_inlines(text).map_err(|e| e.set_line(i + 1))?,
             });
         } else if let Some(quote_body) = line.strip_prefix(">") {
-            let quote_blocks = parse_blocks(quote_body);
+            let quote_blocks = parse_blocks(quote_body).map_err(|e| e.set_line(i + 1))?;
             blocks.push(Block::Quote(quote_blocks));
         } else if line.starts_with("```") {
             let lang_line = line.strip_prefix("```").unwrap().to_string();
@@ -26,8 +110,16 @@ pub fn parse_blocks(input: &str) -> Vec<Block> {
                 Some(lang_line)
             };
             let mut code = String::new();
-            while lines.peek().is_some() && !lines.peek().unwrap().starts_with("```") {
-                code.push_str(&format!("{}\n", lines.next().unwrap()));
+            while lines.peek().is_some()
+                && !lines
+                    .peek()
+                    .ok_or(MdParseError::from_line(i + 1, "a line", ""))?
+                    .1
+                    .starts_with("```")
+            {
+                if let Some((_i, l)) = lines.next() {
+                    code.push_str(&format!("{}\n", l));
+                }
             }
             lines.next();
             blocks.push(Block::Code {
@@ -37,9 +129,12 @@ pub fn parse_blocks(input: &str) -> Vec<Block> {
         } else if line.trim().is_empty() {
             continue;
         } else {
-            blocks.push(Block::Paragraph(parse_inlines(line)));
+            blocks.push(Block::Paragraph(
+                parse_inlines(line).map_err(|e| e.set_line(i + 1))?,
+            ));
         }
     }
 
-    blocks
+    Ok(blocks)
 }
+*/

cracked_md/src/parser/inline.rs

@@ -1,61 +1,65 @@
-use crate::ast::Inline;
+use crate::{MdParseError, ast::Inline};
 
-pub fn parse_inlines(input: &str) -> Vec<Inline> {
+pub fn parse_inlines(input: &str) -> Result<Vec<Inline>, MdParseError> {
     let mut inlines = Vec::new();
     let mut chars = input.chars().peekable();
 
     while let Some(c) = chars.next() {
         match c {
             '*' => {
-                let inner = collect_until(&mut chars, '*');
-                inlines.push(Inline::Bold(parse_inlines(&inner)));
+                let inner = collect_until(&mut chars, '*')?;
+                inlines.push(Inline::Bold(parse_inlines(&inner)?));
             }
             '_' => {
-                let inner = collect_until(&mut chars, '_');
-                inlines.push(Inline::Italic(parse_inlines(&inner)));
+                let inner = collect_until(&mut chars, '_')?;
+                inlines.push(Inline::Italic(parse_inlines(&inner)?));
             }
             '`' => {
-                let code = collect_until(&mut chars, '`');
+                let code = collect_until(&mut chars, '`')?;
                 inlines.push(Inline::Code(code));
             }
             '[' => {
-                let text = collect_until(&mut chars, ']');
-                if chars.next() == Some('(') {
-                    let href = collect_until(&mut chars, ')');
+                let text = collect_until(&mut chars, ']')?;
+                if let Some('(') = chars.next() {
+                    let href = collect_until(&mut chars, ')')?;
                     inlines.push(Inline::Link {
-                        text: parse_inlines(&text),
+                        text: parse_inlines(&text)?,
                         href,
                     });
+                } else {
+                    Err(MdParseError::new(
+                        "(<href>)",
+                        chars.next().unwrap_or_default(),
+                    ))?;
                 }
             }
             _ => {
                 let mut text = String::new();
                 text.push(c);
-                while let Some(&nc) = chars.peek() {
+                while let Some(nc) = chars.next() {
                     if matches!(nc, '*' | '_' | '`' | '[') {
                         break;
                     }
-                    text.push(chars.next().unwrap());
+                    text.push(nc);
                 }
                 inlines.push(Inline::Text(text));
             }
         }
     }
 
-    inlines
+    Ok(inlines)
 }
 
 fn collect_until<I: Iterator<Item = char>>(
     chars: &mut std::iter::Peekable<I>,
     end: char,
-) -> String {
+) -> Result<String, MdParseError> {
     let mut s = String::new();
-    while let Some(&c) = chars.peek() {
+    while let Some(c) = chars.next() {
         if c == end {
-            chars.next();
-            break;
+            return Ok(s);
         }
-        s.push(chars.next().unwrap());
+        s.push(c);
     }
-    s
+    Err(MdParseError::new(end, ""))
 }

cracked_md/test.md

@@ -0,0 +1,7 @@
+# Header *1kkkkkkkkkkkkkkkkkkkkkk*
+
+this is some code: `abc
+
+```code
+
+oiajwefoijao089uaoisdjfoijasdfoijasdofij