extend parser

9 files changed, 215 insertions, 51 deletions

diff --git a/code/Cargo.lock b/code/Cargo.lock
index 0bf0ca0..a41b2d4 100644
--- a/code/Cargo.lock
+++ b/code/Cargo.lock
@@ -48,7 +48,6 @@ dependencies = [
  "clap",
  "iso8601",
  "laby",
- "markdown",
  "synoptic",
 ]
 
@@ -171,29 +170,12 @@ dependencies = [
 ]
 
 [[package]]
-name = "lazy_static"
-version = "1.4.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e2abad23fbc42b3700f2f279844dc832adb2b2eb069b2df918f455c4e18cc646"
-
-[[package]]
 name = "libc"
 version = "0.2.132"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "8371e4e5341c3a96db127eb2465ac681ced4c433e01dd0e938adbef26ba93ba5"
 
 [[package]]
-name = "markdown"
-version = "0.3.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ef3aab6a1d529b112695f72beec5ee80e729cb45af58663ec902c8fac764ecdd"
-dependencies = [
- "lazy_static",
- "pipeline",
- "regex",
-]
-
-[[package]]
 name = "memchr"
 version = "2.5.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
@@ -228,12 +210,6 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9ff7415e9ae3fff1225851df9e0d9e4e5479f947619774677a63572e55e80eff"
 
 [[package]]
-name = "pipeline"
-version = "0.5.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d15b6607fa632996eb8a17c9041cb6071cb75ac057abd45dece578723ea8c7c0"
-
-[[package]]
 name = "proc-macro-error"
 version = "1.0.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
diff --git a/code/Cargo.toml b/code/Cargo.toml
index c528649..c2081a5 100644
--- a/code/Cargo.toml
+++ b/code/Cargo.toml
@@ -7,6 +7,5 @@ edition = "2021"
 laby = "0.2.4"                                       # html generation macros
 clap = { version = "3.2.17", features = ["derive"] } # bloated argument parser
 anyhow = "1.0.62"                                    # error stuff
-markdown = "0.3.0"                                   # markdown parser
 iso8601 = "0.5.0"                                    # date parsing
 synoptic = "1.2.0"                                   # syntax highlighting
diff --git a/code/src/atom.rs b/code/src/atom.rs
index 7912b97..c231c94 100644
--- a/code/src/atom.rs
+++ b/code/src/atom.rs
@@ -1,4 +1,4 @@
-use crate::{get_articles, markdown::escape, Args, ArticleMeta, BLOG_BASE};
+use crate::{get_articles, markdown::render::escape, Args, ArticleMeta, BLOG_BASE};
 use std::process::{Command, Stdio};
 
 pub fn generate_atom(args: &Args) -> String {
diff --git a/code/src/html.rs b/code/src/html.rs
index 5bcc245..d310ad3 100644
--- a/code/src/html.rs
+++ b/code/src/html.rs
@@ -1,4 +1,6 @@
-use crate::{article_metadata, file_history, get_articles, markdown::blocks_to_html, ArticleMeta};
+use crate::{
+    article_metadata, file_history, get_articles, markdown::{render::blocks_to_html, self}, ArticleMeta,
+};
 use laby::{frag, html, iter, li, ul, Render};
 use std::fs::read_to_string;
 
@@ -26,7 +28,7 @@ pub fn article(path: String) -> impl Render {
     scaffold(
         article_metadata(path.clone().into()).title,
         frag!(
-            laby::raw!(blocks_to_html(markdown::tokenize(
+            laby::raw!(blocks_to_html(markdown::parser::parse(
                 &read_to_string(&path).unwrap()
             ))),
             hr!(),
diff --git a/code/src/markdown/parser.rs b/code/src/markdown/parser.rs
index e69de29..cb025a6 100644
--- a/code/src/markdown/parser.rs
+++ b/code/src/markdown/parser.rs
@@ -0,0 +1,188 @@
+#[derive(Debug, Clone)]
+pub enum Block {
+    Header(usize, Vec<Span>),
+    Paragraph(Vec<Span>),
+    Blockquote(Vec<Block>),
+    CodeBlock(Option<String>, String),
+    LatexBlock(String),
+    OrderedList(Vec<Vec<Block>>),
+    UnorderedList(Vec<Vec<Block>>),
+    Raw(String),
+    Hr,
+}
+#[derive(Debug, Clone)]
+pub enum Span {
+    Break,
+    Text(String),
+    Code(String),
+    Link(String, String),
+    Image(String, String),
+    Emphasis(Vec<Span>),
+    Strong(Vec<Span>),
+    Latex(String),
+}
+
+pub fn parse(mut s: &str) -> Vec<Block> {
+    let mut blocks = Vec::new();
+    while s.len() != 0 {
+        if s.starts_with("\n") {
+            s = &s[1..];
+            continue;
+        }
+        // TODO bad code here
+        if let Some((block, rest)) = try_header(s) {
+            s = rest;
+            blocks.push(block);
+            continue;
+        }
+        if let Some((block, rest)) = try_latex_block(s) {
+            s = rest;
+            blocks.push(block);
+            continue;
+        }
+        if let Some((block, rest)) = try_code_block(s) {
+            s = rest;
+            blocks.push(block);
+            continue;
+        }
+        if let Some((block, rest)) = try_list(s) {
+            s = rest;
+            blocks.push(block);
+            continue;
+        }
+        let lf = [s.find("\n\n"), s.find("\n-")]
+            .iter()
+            .filter_map(|e| *e)
+            .min()
+            .unwrap_or(s.len());
+        let span = Span::parse(&s[..lf]);
+        blocks.push(Block::Paragraph(span));
+        if lf >= s.len() {
+            break;
+        }
+        s = &s[lf + 1..];
+    }
+    blocks
+}
+
+fn try_code_block(mut s: &str) -> Option<(Block, &str)> {
+    if !s.starts_with("```") {
+        return None;
+    }
+    s = &s[3..];
+    let lf = s.find('\n')?;
+    let syntax = if lf != 0 {
+        Some(String::from(&s[0..lf]))
+    } else {
+        None
+    };
+    s = &s[lf..];
+    let end = s.find("\n```\n")?;
+    Some((
+        Block::CodeBlock(syntax, String::from(&s[..end])),
+        &s[end + 4..],
+    ))
+}
+fn try_latex_block(mut s: &str) -> Option<(Block, &str)> {
+    if !s.starts_with("$$") {
+        return None;
+    }
+    s = &s[2..];
+    let end = s.find("$$")?;
+    Some((Block::LatexBlock(String::from(&s[..end])), &s[end + 2..]))
+}
+fn try_list(mut s: &str) -> Option<(Block, &str)> {
+    if !s.starts_with("-") {
+        return None;
+    };
+    let mut blocks = vec![];
+    loop {
+        if !s.starts_with("-") || s.len() == 0 {
+            break Some((Block::UnorderedList(blocks), s));
+        }
+        s = &s[1..];
+        let mut lf = s.find("\n").unwrap();
+        while s[lf + 1..].starts_with("  -") {
+            lf += 2 + s[lf + 2..].find("\n").unwrap();
+        }
+        eprintln!("{:?}", &s[..lf]);
+        let mut k = s[..lf]
+            .split("\n")
+            .map(|l| if l.starts_with("  ") { &l[2..] } else { &l })
+            .collect::<Vec<_>>()
+            .join("\n");
+        k.push('\n');
+        blocks.push(parse(&k));
+        s = &s[lf + 1..];
+    }
+}
+fn try_header(s: &str) -> Option<(Block, &str)> {
+    if s.starts_with("#") {
+        let mut u = 0;
+        while s.chars().nth(u)? == '#' {
+            u += 1;
+        }
+        let lf = s.find('\n')?;
+        Some((Block::Header(u, Span::parse(&s[u..lf])), &s[lf + 1..]))
+    } else {
+        None
+    }
+}
+
+impl Span {
+    pub fn parse(mut s: &str) -> Vec<Span> {
+        let mut spans = Vec::new();
+        while s.len() != 0 {
+            let nt = s.find(&['*', '_', '`', '[', '$']);
+
+            if let Some(nt) = nt {
+                spans.push(Span::Text(String::from(&s[..nt])));
+                s = &s[nt..];
+                if s.starts_with("**") {
+                    s = &s[2..];
+                    let end = s.find("**").expect("** not ended");
+                    spans.push(Span::Strong(Span::parse(&s[..end])));
+                    s = &s[end + 2..];
+                    continue;
+                }
+                if s.starts_with("_") {
+                    s = &s[1..];
+                    let end = s.find("_").expect("_ not ended");
+                    spans.push(Span::Emphasis(Span::parse(&s[..end])));
+                    s = &s[end + 1..];
+                    continue;
+                }
+                if s.starts_with("`") {
+                    s = &s[1..];
+                    let end = s.find("`").expect("` not ended");
+                    spans.push(Span::Code(String::from(&s[..end])));
+                    s = &s[end + 1..];
+                    continue;
+                }
+                if s.starts_with("$") {
+                    s = &s[1..];
+                    let end = s.find("$").expect("$ not ended");
+                    spans.push(Span::Latex(String::from(&s[..end])));
+                    s = &s[end + 1..];
+                    continue;
+                }
+                if s.starts_with("[") {
+                    s = &s[1..];
+                    let del = s.find("](").expect("]( expected");
+                    let end = del + s[del..].find(")").expect(") expected");
+                    spans.push(Span::Link(
+                        String::from(&s[..del]),
+                        String::from(&s[del + 2..end]),
+                    ));
+                    s = &s[end + 1..];
+                    continue;
+                }
+                panic!("{s:?}")
+            } else {
+                spans.push(Span::Text(String::from(s)));
+                break;
+            }
+        }
+        spans
+    }
+}
diff --git a/code/src/markdown/render.rs b/code/src/markdown/render.rs
index 22934c5..39f204f 100644
--- a/code/src/markdown/render.rs
+++ b/code/src/markdown/render.rs
@@ -1,7 +1,7 @@
-use markdown::{Block, ListItem, Span};
-
 use crate::syntax_highlight::syntax_highlight;
 
+use super::parser::{Block, Span};
+
 pub fn span_to_html(ss: Vec<Span>) -> String {
     let mut out = String::new();
     for s in ss {
@@ -9,12 +9,13 @@ pub fn span_to_html(ss: Vec<Span>) -> String {
             Span::Break => format!("<br/>"),
             Span::Text(t) => escape(&t),
             Span::Code(c) => format!("<code>{}</code>", escape(&c)),
-            Span::Link(text, url, _) => {
+            Span::Link(text, url) => {
                 format!("<a href=\"{}\">{}</a>", escape(&url), escape(&text))
             }
-            Span::Image(_, _, _) => todo!(),
+            Span::Image(_, _) => todo!(),
             Span::Emphasis(c) => format!("<i>{}</i>", span_to_html(c)),
             Span::Strong(c) => format!("<b>{}</b>", span_to_html(c)),
+            Span::Latex(_) => format!("TODO: Inline Latex"),
         }
         .as_str()
     }
@@ -24,7 +25,7 @@ pub fn blocks_to_html(blocks: Vec<Block>) -> String {
     let mut out = String::new();
     for e in blocks {
         out += match e {
-            Block::Header(text, level) => {
+            Block::Header(level, text) => {
                 format!("<h{level}>{}</h{level}>", span_to_html(text))
             }
             Block::Paragraph(p) => format!("<p>{}</p>", span_to_html(p)),
@@ -36,16 +37,10 @@ pub fn blocks_to_html(blocks: Vec<Block>) -> String {
                     format!("<pre>{}</pre>", escape(&content))
                 }
             }
-            Block::OrderedList(els, _) => format!(
+            Block::OrderedList(els) => format!(
                 "<ol>{}</ol>",
                 els.into_iter()
-                    .map(|e| format!(
-                        "<li>{}</li>",
-                        match e {
-                            ListItem::Simple(s) => span_to_html(s),
-                            ListItem::Paragraph(b) => blocks_to_html(b),
-                        }
-                    ))
+                    .map(|e| format!("<li>{}</li>", blocks_to_html(e)))
                     .collect::<Vec<_>>()
                     .join("")
             ),
@@ -53,19 +48,14 @@ pub fn blocks_to_html(blocks: Vec<Block>) -> String {
                 format!(
                     "<ul>{}</ul>",
                     els.into_iter()
-                        .map(|e| format!(
-                            "<li>{}</li>",
-                            match e {
-                                ListItem::Simple(s) => span_to_html(s),
-                                ListItem::Paragraph(b) => blocks_to_html(b),
-                            }
-                        ))
+                        .map(|e| format!("<li>{}</li>", blocks_to_html(e)))
                         .collect::<Vec<_>>()
                         .join("")
                 )
             }
             Block::Raw(r) => r,
             Block::Hr => format!("<hr/>"),
+            Block::LatexBlock(_) => format!("TODO: Latex block"),
         }
         .as_str();
     }
diff --git a/code/src/syntax_highlight/grammar.rs b/code/src/syntax_highlight/grammar.rs
index 18ba9e2..0878b33 100644
--- a/code/src/syntax_highlight/grammar.rs
+++ b/code/src/syntax_highlight/grammar.rs
@@ -5,7 +5,7 @@ pub fn grammar_for(syntax: &str) -> &'static [(&'static str, &'static [&'static
                 "keyword",
                 &[
                     "fn", "pub", "async", "return", "if", "else", "let", "for", "in", "while",
-                    "loop", "impl", "for", "trait", "struct", "enum",
+                    "loop", "impl", "for", "trait", "struct", "enum", "dyn",
                 ],
             ),
             (
diff --git a/code/src/syntax_highlight/mod.rs b/code/src/syntax_highlight/mod.rs
index 46bf2a0..62a2536 100644
--- a/code/src/syntax_highlight/mod.rs
+++ b/code/src/syntax_highlight/mod.rs
@@ -1,7 +1,7 @@
 pub mod grammar;
 pub mod theme;
 
-use crate::{markdown::escape, syntax_highlight::theme::theme};
+use crate::{markdown::render::escape, syntax_highlight::theme::theme};
 use grammar::grammar_for;
 use synoptic::{Highlighter, Token};
 
diff --git a/readme.md b/readme.md
index c96fee1..8417c90 100644
--- a/readme.md
+++ b/readme.md
@@ -1,6 +1,15 @@
 # metamuffin's blog
 
-TODO
+My personal blog's source code.
+
+## Ideas
+
+## Todo (blog-tool)
+
+- proper markdown parsing
+- markdown images
+- parser error handling
+- syntax highlighting via AST from grammar
 
 ## License