ferrous_opencc_compiler/
lib.rs1use std::{
2 collections::BTreeMap,
3 fs::File,
4 io::{
5 BufRead,
6 BufReader,
7 Write,
8 },
9 path::Path,
10};
11
12use anyhow::{
13 Context,
14 Result,
15};
16use fst::MapBuilder;
17use rkyv::{
18 Archive,
19 Deserialize,
20 Serialize,
21};
22
23#[derive(Archive, Serialize, Deserialize, Debug)]
24pub enum Delta {
25 CharDiffs(Vec<(u16, char)>),
26 FullReplacement(String),
27}
28
29#[derive(Archive, Serialize, Deserialize, Debug)]
30pub struct SerializableFstDict {
31 pub values: Vec<Vec<Delta>>,
32 pub max_key_length: u32,
33}
34
35fn compute_delta(key: &str, value: &str) -> Delta {
36 if key.chars().count() != value.chars().count() {
37 return Delta::FullReplacement(value.to_string());
38 }
39
40 let diffs: Vec<(u16, char)> = key
41 .chars()
42 .zip(value.chars())
43 .enumerate()
44 .filter_map(|(i, (k, v))| if k != v { Some((i as u16, v)) } else { None })
45 .collect();
46
47 if diffs.len() * 6 > value.len() {
48 return Delta::FullReplacement(value.to_string());
49 }
50
51 Delta::CharDiffs(diffs)
52}
53
54pub fn compile_dictionary(input_path: &Path) -> Result<Vec<u8>> {
55 let file = File::open(input_path)
56 .with_context(|| format!("Failed to open input dictionary: {}", input_path.display()))?;
57 let reader = BufReader::new(file);
58
59 let mut entries = BTreeMap::new();
60 let mut max_key_length = 0u32;
61
62 for line in reader.lines() {
63 let line = line.with_context(|| "Failed to read line from dictionary")?;
64
65 let trimmed_line = line.trim();
66 if trimmed_line.is_empty() || trimmed_line.starts_with('#') {
67 continue;
68 }
69 let parts: Vec<&str> = line.split('\t').collect();
70
71 if parts.len() == 2 {
72 let key = parts[0];
73 let values: Vec<&str> = parts[1].split(' ').collect();
74
75 if !key.is_empty() && !values.is_empty() && !values.iter().any(|s| s.is_empty()) {
76 max_key_length = max_key_length.max(key.chars().count() as u32);
77 let delta_values = values.into_iter().map(|v| compute_delta(key, v)).collect();
78 entries.insert(key.to_string(), delta_values);
79 }
80 }
81 }
82
83 let mut values_vec: Vec<Vec<Delta>> = Vec::with_capacity(entries.len());
84 let mut builder = MapBuilder::memory();
85
86 for (key, values) in entries {
87 let index = values_vec.len() as u64;
88 values_vec.push(values);
89 builder
90 .insert(key, index)
91 .with_context(|| "Failed to insert key-value pair into FST")?;
92 }
93
94 let fst_map_bytes = builder
95 .into_inner()
96 .with_context(|| "Failed to finalize FST construction")?;
97
98 let metadata = SerializableFstDict {
99 values: values_vec,
100 max_key_length,
101 };
102
103 let metadata_bytes = rkyv::to_bytes::<rkyv::rancor::Error>(&metadata)
104 .map_err(|e| anyhow::anyhow!("Rkyv serialization failed: {e}"))?
105 .into_vec();
106
107 let mut final_bytes = Vec::new();
108
109 final_bytes.write_all(&(metadata_bytes.len() as u64).to_le_bytes())?;
110 final_bytes.write_all(&metadata_bytes)?;
111 final_bytes.write_all(&fst_map_bytes)?;
112
113 Ok(final_bytes)
114}