| 1 | //! Inline rule chain |
| 2 | use once_cell::sync::OnceCell; |
| 3 | use std::collections::HashMap; |
| 4 | |
| 5 | mod state; |
| 6 | pub use state::*; |
| 7 | |
| 8 | mod rule; |
| 9 | pub use rule::*; |
| 10 | |
| 11 | #[doc(hidden)] |
| 12 | pub mod builtin; |
| 13 | |
| 14 | pub use builtin::inline_parser::InlineRoot; |
| 15 | pub use builtin::skip_text::{Text, TextSpecial}; |
| 16 | use builtin::skip_text::TextScannerImpl; |
| 17 | |
| 18 | use crate::{MarkdownIt, Node}; |
| 19 | use crate::common::TypeKey; |
| 20 | use crate::common::ruler::Ruler; |
| 21 | use crate::parser::extset::{InlineRootExtSet, RootExtSet}; |
| 22 | |
| 23 | use super::node::NodeEmpty; |
| 24 | |
| 25 | type RuleFns = ( |
| 26 | fn (&mut InlineState) -> Option<usize>, |
| 27 | fn (&mut InlineState) -> Option<(Node, usize)>, |
| 28 | ); |
| 29 | |
| 30 | #[derive(Debug, Default)] |
| 31 | /// Inline-level tokenizer. |
| 32 | pub struct InlineParser { |
| 33 | ruler: Ruler<TypeKey, RuleFns>, |
| 34 | text_charmap: HashMap<char, Vec<TypeKey>>, |
| 35 | text_impl: OnceCell<TextScannerImpl>, |
| 36 | } |
| 37 | |
| 38 | impl InlineParser { |
| 39 | pub fn new() -> Self { |
| 40 | Self::default() |
| 41 | } |
| 42 | |
| 43 | /// Skip single token by running all rules in validation mode; |
| 44 | /// returns `true` if any rule reported success |
| 45 | /// |
| 46 | pub fn skip_token(&self, state: &mut InlineState) { |
| 47 | stacker::maybe_grow(64*1024, 1024*1024, || { |
| 48 | let mut ok = None; |
| 49 | |
| 50 | if state.level < state.md.max_nesting { |
| 51 | for rule in self.ruler.iter() { |
| 52 | ok = rule.0(state); |
| 53 | if ok.is_some() { |
| 54 | break; |
| 55 | } |
| 56 | } |
| 57 | } else { |
| 58 | // Too much nesting, just skip until the end of the paragraph. |
| 59 | // |
| 60 | // NOTE: this will cause links to behave incorrectly in the following case, |
| 61 | // when an amount of `[` is exactly equal to `maxNesting + 1`: |
| 62 | // |
| 63 | // [[[[[[[[[[[[[[[[[[[[[foo]() |
| 64 | // |
| 65 | // TODO: remove this workaround when CM standard will allow nested links |
| 66 | // (we can replace it by preventing links from being parsed in |
| 67 | // validation mode) |
| 68 | // |
| 69 | state.pos = state.pos_max; |
| 70 | } |
| 71 | |
| 72 | if let Some(len) = ok { |
| 73 | state.pos += len; |
| 74 | } else { |
| 75 | let ch = state.src[state.pos..state.pos_max].chars().next().unwrap(); |
| 76 | state.pos += ch.len_utf8(); |
| 77 | } |
| 78 | }); |
| 79 | } |
| 80 | |
| 81 | /// Generate tokens for input range |
| 82 | /// |
| 83 | pub fn tokenize(&self, state: &mut InlineState) { |
| 84 | stacker::maybe_grow(64*1024, 1024*1024, || { |
| 85 | let end = state.pos_max; |
| 86 | |
| 87 | while state.pos < end { |
| 88 | // Try all possible rules. |
| 89 | // On success, rule should: |
| 90 | // |
| 91 | // - update `state.pos` |
| 92 | // - update `state.tokens` |
| 93 | // - return true |
| 94 | let mut ok = None; |
| 95 | |
| 96 | if state.level < state.md.max_nesting { |
| 97 | for rule in self.ruler.iter() { |
| 98 | ok = rule.1(state); |
| 99 | if ok.is_some() { |
| 100 | break; |
| 101 | } |
| 102 | } |
| 103 | } |
| 104 | |
| 105 | if let Some((mut node, len)) = ok { |
| 106 | state.pos += len; |
| 107 | if !node.is::<NodeEmpty>() { |
| 108 | node.srcmap = state.get_map(state.pos - len, state.pos); |
| 109 | state.node.children.push(node); |
| 110 | if state.pos >= end { break; } |
| 111 | } |
| 112 | continue; |
| 113 | } |
| 114 | |
| 115 | let ch = state.src[state.pos..state.pos_max].chars().next().unwrap(); |
| 116 | let len = ch.len_utf8(); |
| 117 | state.trailing_text_push(state.pos, state.pos + len); |
| 118 | state.pos += len; |
| 119 | } |
| 120 | }); |
| 121 | } |
| 122 | |
| 123 | /// Process input string and push inline tokens into `out_tokens` |
| 124 | /// |
| 125 | pub fn parse( |
| 126 | &self, |
| 127 | src: String, |
| 128 | srcmap: Vec<(usize, usize)>, |
| 129 | node: Node, |
| 130 | md: &MarkdownIt, |
| 131 | root_ext: &mut RootExtSet, |
| 132 | inline_ext: &mut InlineRootExtSet, |
| 133 | ) -> Node { |
| 134 | let mut state = InlineState::new(src, srcmap, md, root_ext, inline_ext, node); |
| 135 | self.tokenize(&mut state); |
| 136 | state.node |
| 137 | } |
| 138 | |
| 139 | pub fn add_rule<T: InlineRule>(&mut self) -> RuleBuilder<RuleFns> { |
| 140 | if T::MARKER != '\0' { |
| 141 | let charvec = self.text_charmap.entry(T::MARKER).or_default(); |
| 142 | charvec.push(TypeKey::of::<T>()); |
| 143 | } |
| 144 | |
| 145 | let item = self.ruler.add(TypeKey::of::<T>(), (T::check, T::run)); |
| 146 | RuleBuilder::new(item) |
| 147 | } |
| 148 | |
| 149 | pub fn has_rule<T: InlineRule>(&mut self) -> bool { |
| 150 | self.ruler.contains(TypeKey::of::<T>()) |
| 151 | } |
| 152 | |
| 153 | pub fn remove_rule<T: InlineRule>(&mut self) { |
| 154 | if T::MARKER != '\0' { |
| 155 | let mut charvec = self.text_charmap.remove(&T::MARKER).unwrap_or_default(); |
| 156 | charvec.retain(|x| *x != TypeKey::of::<T>()); |
| 157 | self.text_charmap.insert(T::MARKER, charvec); |
| 158 | } |
| 159 | |
| 160 | self.ruler.remove(TypeKey::of::<T>()); |
| 161 | } |
| 162 | } |