diff options
author | Titus Wormer <tituswormer@gmail.com> | 2022-08-09 17:59:23 +0200 |
---|---|---|
committer | Titus Wormer <tituswormer@gmail.com> | 2022-08-09 18:38:17 +0200 |
commit | 5e28dd2143022115e347cf8944adf39ff7ff2136 (patch) | |
tree | b20ed6210d196e752c7a7f5eef08388312d76e27 | |
parent | 9bab2951cd3e9d492f3aa9f127531643323a1de1 (diff) | |
download | markdown-rs-5e28dd2143022115e347cf8944adf39ff7ff2136.tar.gz markdown-rs-5e28dd2143022115e347cf8944adf39ff7ff2136.tar.bz2 markdown-rs-5e28dd2143022115e347cf8944adf39ff7ff2136.zip |
Refactor to share some code
-rw-r--r-- | src/content/document.rs | 105 | ||||
-rw-r--r-- | src/subtokenize.rs | 142 |
2 files changed, 99 insertions, 148 deletions
diff --git a/src/content/document.rs b/src/content/document.rs index cc83415..2c340f2 100644 --- a/src/content/document.rs +++ b/src/content/document.rs @@ -9,7 +9,7 @@ //! * [List][crate::construct::list] use crate::parser::ParseState; -use crate::subtokenize::subtokenize; +use crate::subtokenize::{divide_events, subtokenize}; use crate::token::Token; use crate::tokenizer::{ Container, ContainerState, ContentType, Event, EventType, Link, Point, State, StateName, @@ -550,101 +550,22 @@ fn exit_containers(tokenizer: &mut Tokenizer, phase: &Phase) { // Inject the container events. fn resolve(tokenizer: &mut Tokenizer) { let mut child = tokenizer.tokenize_state.child_tokenizer.take().unwrap(); - child.map.consume(&mut child.events); // To do: see if we can do this less. tokenizer.map.consume(&mut tokenizer.events); - let mut link_index = skip::to(&tokenizer.events, 0, &[Token::Data]); - // To do: share this code with `subtokenize`. - // Now, loop through all subevents to figure out which parts - // belong where and fix deep links. - let mut subindex = 0; - let mut slices = vec![]; - let mut slice_start = 0; - let mut old_prev: Option<usize> = None; - - while subindex < child.events.len() { - // Find the first event that starts after the end we’re looking - // for. - if child.events[subindex].event_type == EventType::Enter - && child.events[subindex].point.index >= tokenizer.events[link_index + 1].point.index - { - slices.push((link_index, slice_start)); - slice_start = subindex; - link_index = tokenizer.events[link_index] - .link - .as_ref() - .unwrap() - .next - .unwrap(); - } - - // Fix sublinks. - if let Some(sublink_curr) = &child.events[subindex].link { - if sublink_curr.previous.is_some() { - let old_prev = old_prev.unwrap(); - let prev_event = &mut child.events[old_prev]; - // The `index` in `events` where the current link is, - // minus one to get the previous link, - // minus 2 events (the enter and exit) for each removed - // link. - let new_link = if slices.is_empty() { - old_prev + link_index + 2 - } else { - old_prev + link_index - (slices.len() - 1) * 2 - }; - prev_event.link.as_mut().unwrap().next = Some(new_link); - } - } - - // If there is a `next` link in the subevents, we have to change - // its `previous` index to account for the shifted events. - // If it points to a next event, we also change the next event’s - // reference back to *this* event. - if let Some(sublink_curr) = &child.events[subindex].link { - if let Some(next) = sublink_curr.next { - let sublink_next = child.events[next].link.as_mut().unwrap(); - - old_prev = sublink_next.previous; - - sublink_next.previous = sublink_next - .previous - // The `index` in `events` where the current link is, - // minus 2 events (the enter and exit) for each removed - // link. - .map(|previous| previous + link_index - (slices.len() * 2)); - } - } + divide_events( + &mut tokenizer.map, + &tokenizer.events, + skip::to(&tokenizer.events, 0, &[Token::Data]), + &mut child.events, + ); - subindex += 1; - } - - if !child.events.is_empty() { - slices.push((link_index, slice_start)); - } - - // Finally, inject the subevents. - let mut index = slices.len(); - - while index > 0 { - index -= 1; - let start = slices[index].0; - tokenizer.map.add( - start, - if start == tokenizer.events.len() { - 0 - } else { - 2 - }, - child.events.split_off(slices[index].1), - ); - } - // To do: share the above code with `subtokenize`. - - let mut resolvers = child.resolvers.split_off(0); - let mut resolver_ids = child.resolver_ids.split_off(0); - tokenizer.resolvers.append(&mut resolvers); - tokenizer.resolver_ids.append(&mut resolver_ids); + tokenizer + .resolvers + .append(&mut child.resolvers.split_off(0)); + tokenizer + .resolver_ids + .append(&mut child.resolver_ids.split_off(0)); // To do: see if we can do this less. tokenizer.map.consume(&mut tokenizer.events); diff --git a/src/subtokenize.rs b/src/subtokenize.rs index b080b46..10f34d0 100644 --- a/src/subtokenize.rs +++ b/src/subtokenize.rs @@ -108,72 +108,102 @@ pub fn subtokenize(events: &mut Vec<Event>, parse_state: &ParseState) -> bool { tokenizer.flush(state, true); - // Now, loop through all subevents to figure out which parts - // belong where and fix deep links. - let mut subindex = 0; - let mut link_index = index; - let mut slices = vec![]; - let mut slice_start = 0; - - while subindex < tokenizer.events.len() { - let subevent = &mut tokenizer.events[subindex]; - - // Find the first event that starts after the end we’re looking - // for. - if subevent.event_type == EventType::Enter - && subevent.point.index >= events[link_index + 1].point.index - { - slices.push((link_index, slice_start)); - slice_start = subindex; - link_index = events[link_index].link.as_ref().unwrap().next.unwrap(); - } + divide_events(&mut map, events, index, &mut tokenizer.events); - if subevent.link.is_some() { - // Need to call `subtokenize` again. - done = false; - } + // To do: check `tokenizer.events` if there is a deep content type? + done = false; + } + } - // If there is a `next` link in the subevents, we have to change - // its index to account for the shifted events. - // If it points to a next event, we also change the next event’s - // reference back to *this* event. - if let Some(sublink_curr) = &mut subevent.link { - if let Some(next) = sublink_curr.next { - // The `index` in `events` where the current link is, - // minus 2 events (the enter and exit) for each removed - // link. - let shift = link_index - (slices.len() * 2); - sublink_curr.next = sublink_curr.next.map(|next| next + shift); - let next_ev = &mut tokenizer.events[next]; - let sublink_next = next_ev.link.as_mut().unwrap(); - sublink_next.previous = - sublink_next.previous.map(|previous| previous + shift); - } - } + index += 1; + } - subindex += 1; - } + map.consume(events); + + done +} - slices.push((link_index, slice_start)); +/// Parse linked events. +/// +/// Supposed to be called repeatedly, returns `1: true` when done. +pub fn divide_events( + map: &mut EditMap, + events: &[Event], + mut link_index: usize, + child_events: &mut Vec<Event>, +) { + // Now, loop through all subevents to figure out which parts + // belong where and fix deep links. + let mut subindex = 0; + let mut slices = vec![]; + let mut slice_start = 0; + let mut old_prev: Option<usize> = None; + + while subindex < child_events.len() { + // Find the first event that starts after the end we’re looking + // for. + if child_events[subindex].event_type == EventType::Enter + && child_events[subindex].point.index >= events[link_index + 1].point.index + { + slices.push((link_index, slice_start)); + slice_start = subindex; + link_index = events[link_index].link.as_ref().unwrap().next.unwrap(); + } - // Finally, inject the subevents. - let mut index = slices.len(); + // Fix sublinks. + if let Some(sublink_curr) = &child_events[subindex].link { + if sublink_curr.previous.is_some() { + let old_prev = old_prev.unwrap(); + let prev_event = &mut child_events[old_prev]; + // The `index` in `events` where the current link is, + // minus one to get the previous link, + // minus 2 events (the enter and exit) for each removed + // link. + let new_link = if slices.is_empty() { + old_prev + link_index + 2 + } else { + old_prev + link_index - (slices.len() - 1) * 2 + }; + prev_event.link.as_mut().unwrap().next = Some(new_link); + } + } - while index > 0 { - index -= 1; - map.add( - slices[index].0, - 2, - tokenizer.events.split_off(slices[index].1), - ); - } + // If there is a `next` link in the subevents, we have to change + // its `previous` index to account for the shifted events. + // If it points to a next event, we also change the next event’s + // reference back to *this* event. + if let Some(sublink_curr) = &child_events[subindex].link { + if let Some(next) = sublink_curr.next { + let sublink_next = child_events[next].link.as_mut().unwrap(); + + old_prev = sublink_next.previous; + + sublink_next.previous = sublink_next + .previous + // The `index` in `events` where the current link is, + // minus 2 events (the enter and exit) for each removed + // link. + .map(|previous| previous + link_index - (slices.len() * 2)); } } - index += 1; + subindex += 1; } - map.consume(events); + if !child_events.is_empty() { + slices.push((link_index, slice_start)); + } - done + // Finally, inject the subevents. + let mut index = slices.len(); + + while index > 0 { + index -= 1; + let start = slices[index].0; + map.add( + start, + if start == events.len() { 0 } else { 2 }, + child_events.split_off(slices[index].1), + ); + } } |