minify-html/build.rs

use std::collections::HashMap;
use std::env;
use std::fs::File;
use std::io::Write;
use std::path::Path;

use serde::{Deserialize, Serialize};

fn create_byte_string_literal(bytes: &[u8]) -> String {
    format!("b\"{}\"", bytes
        .iter()
        .map(|&b| if b >= b' ' && b <= b'~' && b != b'\\' && b != b'"' {
            (b as char).to_string()
        } else {
            format!("\\x{:02x}", b)
        })
        .collect::<String>())
}

fn read_json<T>(name: &str) -> T
    where for<'de> T: Deserialize<'de> {
    let patterns_path = Path::new("gen").join(format!("{}.json", name));
    let patterns_file = File::open(patterns_path).unwrap();
    serde_json::from_reader(patterns_file).unwrap()
}

fn write_rs(name: &str, code: String) -> () {
    let out_dir = env::var("OUT_DIR").unwrap();
    let dest_path = Path::new(&out_dir).join(format!("gen_{}.rs", name));
    let mut dest_file = File::create(&dest_path).unwrap();
    dest_file.write_all(code.as_bytes()).unwrap();
}

struct AutoIncrement {
    next_val: usize,
}

impl AutoIncrement {
    fn new() -> AutoIncrement {
        AutoIncrement { next_val: 0 }
    }

    fn next(&mut self) -> usize {
        let v = self.next_val;
        self.next_val += 1;
        v
    }
}

struct TrieBuilderNode {
    children: HashMap<char, TrieBuilderNode>,
    value_as_code: Option<String>,
}

impl TrieBuilderNode {
    fn new() -> TrieBuilderNode {
        TrieBuilderNode {
            children: HashMap::new(),
            value_as_code: None,
        }
    }

    fn add(&mut self, pat: &str, val: String) -> () {
        let mut current = self;
        for c in pat.chars() {
            if !current.children.contains_key(&c) {
                current.children.insert(c, TrieBuilderNode::new());
            };
            current = current.children.get_mut(&c).unwrap();
        };
        assert!(current.value_as_code.is_none());
        current.value_as_code = Some(val);
    }

    fn build(&self, ai: &mut AutoIncrement, value_type: &'static str, out: &mut String) -> usize {
        let child_ids: Vec<(char, usize)> = self.children
            .iter()
            .map(|(&c, n)| (c, n.build(ai, value_type, out)))
            .collect();
        let id = ai.next();

        out.push_str(format!("static N{}: &TrieNode<{}> = &TrieNode::<{}> {{\n", id, value_type, value_type).as_str());
        out.push_str(format!("children: phf::phf_map! {{\n").as_str());
        for (c, n) in child_ids {
            debug_assert!(c as u32 <= 0x7f);
            out.push_str(format!("{}u8 => N{},\n", c as u8, n).as_str());
        }
        out.push_str("},\n");
        out.push_str("value: ");
        match &self.value_as_code {
            Some(v) => {
                out.push_str(format!("Some({})", v).as_str());
            }
            None => out.push_str("None"),
        };
        out.push_str(",\n};\n");

        id
    }
}

fn build_pattern(pattern: String) -> String {
    assert!(pattern.is_ascii());
    let seq = pattern.as_bytes();
    let mut max_prefix_len = 0usize;
    let mut table = vec![0usize; seq.len()];

    let mut i = 1;
    while i < seq.len() {
        if seq[i] == seq[max_prefix_len] {
            max_prefix_len += 1;
            table[i] = max_prefix_len;
            i += 1;
        } else {
            if max_prefix_len != 0 {
                max_prefix_len = table[max_prefix_len - 1];
            } else {
                table[i] = 0;
                i += 1;
            };
        };
    };

    format!("SinglePattern {{ seq: {}, table: &[{}] }}",
        create_byte_string_literal(pattern.as_bytes()),
        table.iter().map(|v| v.to_string()).collect::<Vec<String>>().join(", "))
}

#[derive(Serialize, Deserialize, Debug)]
struct Entity {
    codepoints: Vec<u32>,
    characters: String,
}

fn generate_entities() {
    // Read named entities map from JSON file.
    let entities: HashMap<String, Entity> = read_json("entities");

    // Add entities to trie builder.
    let mut trie_builder = TrieBuilderNode::new();
    for (rep, entity) in entities {
        trie_builder.add(&rep[1..], create_byte_string_literal(entity.characters.as_bytes()));
    };
    // Generate trie code from builder.
    let mut trie_code = String::new();
    let trie_root_id = trie_builder.build(&mut AutoIncrement::new(), "&'static [u8]", &mut trie_code);

    // Write trie code to output Rust file.
    // Make trie root public and use proper variable name.
    write_rs("entities", trie_code.replace(
        format!("static N{}:", trie_root_id).as_str(),
        "pub static ENTITY_REFERENCES:",
    ));
}

fn generate_patterns() {
    let patterns: HashMap<String, String> = read_json("patterns");

    for (name, pattern) in patterns {
        let mut code = String::new();
        code.push_str(format!("static {}: &SinglePattern = &{};", name, build_pattern(pattern)).as_str());
        write_rs(format!("pattern_{}", name).as_str(), code);
    };
}

fn generate_tries() {
    let tries: HashMap<String, HashMap<String, String>> = read_json("tries");

    for (name, values) in tries {
        let mut trie_builder = TrieBuilderNode::new();
        for (seq, value_code) in values {
            trie_builder.add(seq.as_str(), value_code);
        }
        let mut trie_code = String::new();
        let trie_root_id = trie_builder.build(&mut AutoIncrement::new(), "ContentType", &mut trie_code);

        write_rs(format!("trie_{}", name).as_str(), trie_code.replace(
            format!("static N{}:", trie_root_id).as_str(),
            format!("static {}:", name).as_str(),
        ));
    }
}

fn main() {
    generate_entities();
    generate_patterns();
    generate_tries();
}