Add basic support for block quotes

4 files changed, 101 insertions, 13 deletions

diff --git a/src/construct/block_quote.rs b/src/construct/block_quote.rs
new file mode 100644
index 0000000..cd5b872
--- /dev/null
+++ b/src/construct/block_quote.rs
@@ -0,0 +1,58 @@
+//! To do.
+
+use crate::constant::TAB_SIZE;
+use crate::construct::partial_space_or_tab::space_or_tab_min_max;
+use crate::tokenizer::{Code, State, StateFnResult, TokenType, Tokenizer};
+
+pub fn start(tokenizer: &mut Tokenizer, code: Code) -> StateFnResult {
+    // To do: allow arbitrary when code (indented) is turned off.
+    tokenizer.go(space_or_tab_min_max(0, TAB_SIZE - 1), before)(tokenizer, code)
+}
+
+fn before(tokenizer: &mut Tokenizer, code: Code) -> StateFnResult {
+    match code {
+        Code::Char('>') => {
+            tokenizer.enter(TokenType::BlockQuote);
+            cont_before(tokenizer, code)
+        }
+        _ => cont_before(tokenizer, code),
+    }
+}
+
+pub fn cont(tokenizer: &mut Tokenizer, code: Code) -> StateFnResult {
+    // To do: allow arbitrary when code (indented) is turned off.
+    tokenizer.go(space_or_tab_min_max(0, TAB_SIZE - 1), cont_before)(tokenizer, code)
+}
+
+fn cont_before(tokenizer: &mut Tokenizer, code: Code) -> StateFnResult {
+    match code {
+        Code::Char('>') => {
+            tokenizer.enter(TokenType::BlockQuotePrefix);
+            tokenizer.enter(TokenType::BlockQuoteMarker);
+            tokenizer.consume(code);
+            tokenizer.exit(TokenType::BlockQuoteMarker);
+            (State::Fn(Box::new(cont_after)), None)
+        }
+        _ => (State::Nok, None),
+    }
+}
+
+fn cont_after(tokenizer: &mut Tokenizer, code: Code) -> StateFnResult {
+    match code {
+        Code::VirtualSpace | Code::Char('\t' | ' ') => {
+            tokenizer.enter(TokenType::BlockQuotePrefixWhitespace);
+            tokenizer.consume(code);
+            tokenizer.exit(TokenType::BlockQuotePrefixWhitespace);
+            tokenizer.exit(TokenType::BlockQuotePrefix);
+            (State::Ok, None)
+        }
+        _ => {
+            tokenizer.exit(TokenType::BlockQuotePrefix);
+            (State::Ok, Some(vec![code]))
+        }
+    }
+}
+
+pub fn end() -> Vec<TokenType> {
+    vec![TokenType::BlockQuote]
+}
diff --git a/src/construct/heading_setext.rs b/src/construct/heading_setext.rs
index 211434f..440baa8 100644
--- a/src/construct/heading_setext.rs
+++ b/src/construct/heading_setext.rs
@@ -60,7 +60,7 @@
 use crate::constant::TAB_SIZE;
 use crate::construct::partial_space_or_tab::{space_or_tab, space_or_tab_min_max};
 use crate::tokenizer::{Code, Event, EventType, State, StateFnResult, TokenType, Tokenizer};
-use crate::util::edit_map::EditMap;
+use crate::util::{edit_map::EditMap, skip::opt_back as skip_opt_back};
 
 /// Kind of underline.
 #[derive(Debug, Clone, PartialEq)]
@@ -116,11 +116,26 @@ impl Kind {
 /// ```
 pub fn start(tokenizer: &mut Tokenizer, code: Code) -> StateFnResult {
     let index = tokenizer.events.len();
-    let paragraph_before = index > 3
-        && tokenizer.events[index - 1].token_type == TokenType::LineEnding
-        && tokenizer.events[index - 3].token_type == TokenType::Paragraph;
-
-    if paragraph_before {
+    let previous = if index > 1 {
+        skip_opt_back(
+            &tokenizer.events,
+            index - 1,
+            &[TokenType::SpaceOrTab, TokenType::BlockQuotePrefix],
+        )
+    } else {
+        0
+    };
+    let previous = skip_opt_back(&tokenizer.events, previous, &[TokenType::LineEnding]);
+    let paragraph_before =
+        previous > 1 && tokenizer.events[previous].token_type == TokenType::Paragraph;
+
+    println!(
+        "setext-start: {:?} {:?} {:?}",
+        tokenizer.interrupt, tokenizer.lazy, paragraph_before
+    );
+
+    // Require a paragraph before and do not allow on a lazy line.
+    if paragraph_before && !tokenizer.lazy {
         // To do: allow arbitrary when code (indented) is turned off.
         tokenizer.go(space_or_tab_min_max(0, TAB_SIZE - 1), before)(tokenizer, code)
     } else {
diff --git a/src/construct/mod.rs b/src/construct/mod.rs
index 66b2a3c..936ecf6 100644
--- a/src/construct/mod.rs
+++ b/src/construct/mod.rs
@@ -64,6 +64,7 @@
 pub mod attention;
 pub mod autolink;
 pub mod blank_line;
+pub mod block_quote;
 pub mod character_escape;
 pub mod character_reference;
 pub mod code_fenced;
diff --git a/src/construct/paragraph.rs b/src/construct/paragraph.rs
index 4f5e662..ace174f 100644
--- a/src/construct/paragraph.rs
+++ b/src/construct/paragraph.rs
@@ -35,7 +35,7 @@
 use crate::tokenizer::{
     Code, ContentType, Event, EventType, State, StateFnResult, TokenType, Tokenizer,
 };
-use crate::util::edit_map::EditMap;
+use crate::util::{edit_map::EditMap, skip::opt as skip_opt};
 
 /// Before a paragraph.
 ///
@@ -90,19 +90,27 @@ pub fn resolve(tokenizer: &mut Tokenizer) -> Vec<Event> {
         if event.event_type == EventType::Enter && event.token_type == TokenType::Paragraph {
             // Exit:Paragraph
             let mut exit_index = index + 3;
+            let mut enter_next_index =
+                skip_opt(&tokenizer.events, exit_index + 1, &[TokenType::LineEnding]);
             // Enter:Paragraph
-            let mut enter_next_index = exit_index + 3;
+            enter_next_index = skip_opt(
+                &tokenizer.events,
+                enter_next_index,
+                &[TokenType::SpaceOrTab, TokenType::BlockQuotePrefix],
+            );
 
             // Find future `Paragraphs`.
-            // There will be `LineEnding` between.
-            while enter_next_index < len
+            while enter_next_index < tokenizer.events.len()
                 && tokenizer.events[enter_next_index].token_type == TokenType::Paragraph
             {
                 // Remove Exit:Paragraph, Enter:LineEnding, Exit:LineEnding, Enter:Paragraph.
-                edit_map.add(exit_index, 4, vec![]);
+                edit_map.add(exit_index, 3, vec![]);
+
+                // Remove Enter:Paragraph.
+                edit_map.add(enter_next_index, 1, vec![]);
 
                 // Add Exit:LineEnding position info to Exit:Data.
-                let line_ending_exit = &tokenizer.events[enter_next_index - 1];
+                let line_ending_exit = &tokenizer.events[exit_index + 2];
                 let line_ending_point = line_ending_exit.point.clone();
                 let line_ending_index = line_ending_exit.index;
                 let data_exit = &mut tokenizer.events[exit_index - 1];
@@ -117,7 +125,13 @@ pub fn resolve(tokenizer: &mut Tokenizer) -> Vec<Event> {
 
                 // Potential next start.
                 exit_index = enter_next_index + 3;
-                enter_next_index = exit_index + 3;
+                enter_next_index =
+                    skip_opt(&tokenizer.events, exit_index + 1, &[TokenType::LineEnding]);
+                enter_next_index = skip_opt(
+                    &tokenizer.events,
+                    enter_next_index,
+                    &[TokenType::SpaceOrTab, TokenType::BlockQuotePrefix],
+                );
             }
 
             // Move to `Exit:Paragraph`.