html_outliner/
heading.rs

1use kuchiki::{NodeData, NodeRef};
2use once_cell::sync::Lazy;
3use regex::Regex;
4
5static RE_NEW_LINE: Lazy<Regex> = Lazy::new(|| Regex::new("\n[\n\t ]*").unwrap());
6
7#[derive(Debug, Clone)]
8pub enum Heading {
9    Header { level: u8, text: String },
10    Group(Vec<Heading>),
11}
12
13impl Heading {
14    #[inline]
15    pub fn get_end_level(&self) -> u8 {
16        match self {
17            Heading::Header {
18                level, ..
19            } => *level,
20            Heading::Group(headings) => headings[headings.len() - 1].get_end_level(),
21        }
22    }
23
24    #[inline]
25    pub fn get_start_level(&self) -> u8 {
26        match self {
27            Heading::Header {
28                level, ..
29            } => *level,
30            Heading::Group(headings) => headings[0].get_end_level(),
31        }
32    }
33}
34
35pub(crate) fn create_heading(node: NodeRef, depth: usize, max_depth: usize) -> Option<Heading> {
36    if depth > max_depth {
37        return None;
38    }
39
40    let mut heading = if let NodeData::Element(element_data) = node.data() {
41        let local_name: &str = &element_data.name.local;
42
43        let local_name_length = local_name.len();
44
45        match local_name_length {
46            2 => {
47                if let Some(stripped_local_name) = local_name.strip_prefix('h') {
48                    match stripped_local_name.parse::<u8>() {
49                        Ok(level) if (1..=6).contains(&level) => Heading::Header {
50                            level,
51                            text: String::new(),
52                        },
53                        _ => return None,
54                    }
55                } else {
56                    return None;
57                }
58            },
59            6 => {
60                if local_name.eq("hgroup") {
61                    Heading::Group(Vec::with_capacity(2))
62                } else {
63                    return None;
64                }
65            },
66            _ => return None,
67        }
68    } else {
69        return None;
70    };
71
72    match &mut heading {
73        Heading::Header {
74            text, ..
75        } => {
76            for child in node.children() {
77                create_text(text, child, depth + 1, max_depth);
78            }
79        },
80        Heading::Group(headings) => {
81            for child in node.children() {
82                if let Some(heading) = create_heading(child, depth + 1, max_depth) {
83                    headings.push(heading);
84                }
85            }
86
87            if headings.is_empty() {
88                return None;
89            }
90        },
91    }
92
93    Some(heading)
94}
95
96impl From<Heading> for String {
97    #[inline]
98    fn from(heading: Heading) -> String {
99        match heading {
100            Heading::Header {
101                text, ..
102            } => text,
103            Heading::Group(headings) => {
104                let mut iter = headings.into_iter();
105
106                let mut text: String = iter.next().unwrap().into();
107
108                for heading in iter {
109                    text.push_str(" — ");
110
111                    let t: String = heading.into();
112                    text.push_str(&t);
113                }
114
115                text
116            },
117        }
118    }
119}
120
121#[inline]
122pub(crate) fn create_text(text: &mut String, node: NodeRef, depth: usize, max_depth: usize) {
123    if depth > max_depth {
124        return;
125    }
126
127    if let NodeData::Text(t) = node.data() {
128        let t = t.borrow();
129
130        text.push_str(RE_NEW_LINE.replace(t.as_str(), " ").trim());
131    } else {
132        for child in node.children() {
133            create_text(text, child, depth + 1, max_depth);
134        }
135    }
136}
html_outliner/heading.rs

html_outliner/
heading.rs
