Skip to main content

muffy_validation_macro/
lib.rs

1//! Macros for document validation.
2
3extern crate alloc;
4
5mod attribute;
6mod compiler;
7mod content;
8mod definition;
9mod error;
10mod name;
11mod pattern;
12
13use self::{
14    attribute::AttributeSet,
15    compiler::Compiler,
16    content::{children, generate_content},
17    definition::load_definitions,
18    error::MacroError,
19    name::class_names,
20    pattern::Pattern,
21};
22use alloc::collections::BTreeMap;
23use itertools::Itertools;
24use muffy_rnc::{NameClass, Pattern as RncPattern};
25use proc_macro::TokenStream;
26use proc_macro2::Span;
27use quote::{format_ident, quote};
28
29/// Generates HTML validation functions.
30#[proc_macro]
31pub fn html(_input: TokenStream) -> TokenStream {
32    generate_validation("html", &["html.rnc"]).unwrap_or_else(|error| {
33        syn::Error::new(Span::call_site(), error)
34            .to_compile_error()
35            .into()
36    })
37}
38
39fn generate_validation(language: &str, files: &[&str]) -> Result<TokenStream, MacroError> {
40    let definitions = load_definitions(files)?;
41    let mut compiler = Compiler::new(&definitions);
42    let mut element_rules = BTreeMap::<String, Vec<(Vec<AttributeSet>, Pattern)>>::new();
43
44    for definition in definitions.values() {
45        for (name_class, pattern) in collect_elements(definition) {
46            let names = class_names(name_class, false);
47
48            if names.is_empty() {
49                continue;
50            }
51
52            for (attribute_sets, content_pattern) in compiler.compile(pattern)? {
53                for name in &names {
54                    let variants = element_rules.entry(name.clone()).or_default();
55                    let variant = (attribute_sets.clone(), content_pattern.clone());
56
57                    if !variants.contains(&variant) {
58                        variants.push(variant);
59                    }
60                }
61            }
62        }
63    }
64
65    let mut attribute_set_indexes = BTreeMap::<Vec<AttributeSet>, usize>::new();
66    let mut content_indexes = BTreeMap::<Pattern, usize>::new();
67    let mut element_matches = vec![];
68
69    for (name, variants) in &element_rules {
70        let attributes = variants
71            .iter()
72            .flat_map(|(sets, _)| sets)
73            .flat_map(|set| set.required.iter().chain(&set.optional))
74            .unique()
75            .sorted()
76            .map(|name| quote!(#name));
77        let children = variants
78            .iter()
79            .flat_map(|(_, content)| children(content))
80            .unique()
81            .sorted()
82            .map(|name| quote!(#name));
83
84        let variants = variants
85            .iter()
86            .map(|(sets, content)| {
87                let index = attribute_set_indexes.len();
88                let sets = format_ident!(
89                    "ATTRIBUTE_SETS_{}",
90                    *attribute_set_indexes.entry(sets.clone()).or_insert(index)
91                );
92
93                let index = content_indexes.len();
94                let content = format_ident!(
95                    "CONTENT_{}",
96                    *content_indexes.entry(content.clone()).or_insert(index)
97                );
98
99                quote!(Variant { attributes: #sets, content: &#content })
100            })
101            .collect::<Vec<_>>();
102
103        element_matches.push(quote! {
104            #name => {
105                const RULE: Rule = Rule {
106                    attributes: &[#(#attributes),*],
107                    children: &[#(#children),*],
108                    variants: &[#(#variants),*],
109                };
110
111                validate_rule(element, ignored_attributes, ignored_elements, &RULE)
112            }
113        });
114    }
115
116    let attribute_set_definitions = sort_by_index(attribute_set_indexes).map(|(sets, index)| {
117        let identifier = format_ident!("ATTRIBUTE_SETS_{index}");
118        let sets = sets.iter().map(|set| {
119            let required = set.required.iter().map(|name| quote!(#name));
120            let optional = set.optional.iter().map(|name| quote!(#name));
121
122            quote!(AttributeSet {
123                required: &[#(#required),*],
124                optional: &[#(#optional),*],
125            })
126        });
127
128        quote!(const #identifier: &[AttributeSet] = &[#(#sets),*];)
129    });
130    let content_definitions = sort_by_index(content_indexes)
131        .map(|(content, index)| {
132            let identifier = format_ident!("CONTENT_{index}");
133            let content = generate_content(&content)?;
134
135            Ok(quote!(const #identifier: Content = #content;))
136        })
137        .collect::<Result<Vec<_>, MacroError>>()?;
138
139    let function_name = format_ident!("validate_{language}_element");
140    let documentation = format!("Validates an {} element.", language.to_uppercase());
141
142    Ok(quote! {
143        #[doc = #documentation]
144        pub fn #function_name(
145            element: &Element,
146            ignored_attributes: &[::regex::Regex],
147            ignored_elements: &[::regex::Regex],
148        ) -> Result<(), MarkupError> {
149            #(#attribute_set_definitions)*
150            #(#content_definitions)*
151
152            match element.name() {
153                name if ignored_elements.iter().any(|pattern| pattern.is_match(name)) => Ok(()),
154                #(#element_matches)*
155                _ => Err(MarkupError::UnknownTag(element.name().to_string())),
156            }
157        }
158    }
159    .into())
160}
161
162fn sort_by_index<T>(indexes: BTreeMap<T, usize>) -> impl Iterator<Item = (T, usize)> {
163    let mut entries = indexes.into_iter().collect::<Vec<_>>();
164
165    entries.sort_by_key(|(_, index)| *index);
166
167    entries.into_iter()
168}
169
170// TODO Skip element definitions gated by not-allowed flag conjuncts.
171fn collect_elements(pattern: &RncPattern) -> Vec<(&NameClass, &RncPattern)> {
172    match pattern {
173        RncPattern::Element {
174            name_class,
175            pattern,
176        } => vec![(name_class, pattern)],
177        RncPattern::Choice(patterns)
178        | RncPattern::Group(patterns)
179        | RncPattern::Interleave(patterns) => patterns.iter().flat_map(collect_elements).collect(),
180        RncPattern::Many0(pattern) | RncPattern::Many1(pattern) | RncPattern::Optional(pattern) => {
181            collect_elements(pattern)
182        }
183        RncPattern::Attribute { .. }
184        | RncPattern::Data { .. }
185        | RncPattern::Empty
186        | RncPattern::External(_)
187        | RncPattern::Grammar(_)
188        | RncPattern::List(_)
189        | RncPattern::Name(_)
190        | RncPattern::NotAllowed
191        | RncPattern::Text
192        | RncPattern::Value { .. } => vec![],
193    }
194}