1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
//! Tier-B speech-synthesis voice persona library (G099).
//!
//! Real browsers expose a list of speech-synthesis voices via
//! `speechSynthesis.getVoices()`. The exact set and names are browser- and
//! platform-specific, making them a subtle fingerprint. This loader lets an
//! caller maintain platform-typical voice lists from TOML drop-ins without
//! recompiling, so the voice persona stays coherent with the claimed OS.
//!
//! A malformed file fails closed (Law 10): the first bad entry rejects the whole
//! load, no voice is ever silently skipped, because a skipped entry would
//! create a coverage hole in the voice fingerprint.
/// One speech-synthesis voice entry.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct VoicePersona {
/// Display name (e.g. `"Samantha"`).
pub name: &'static str,
/// BCP 47 language tag (e.g. `"en-US"`).
pub lang: &'static str,
/// Whether this voice is the platform default.
pub default: bool,
}
#[cfg(feature = "tier-b-toml")]
mod loader_impl {
use super::VoicePersona;
use std::path::Path;
/// Upper bound on a Tier-B voice TOML (64 KiB).
const MAX_VOICE_TOML_BYTES: u64 = 64 * 1024;
/// Error loading a Tier-B voice persona library. Every variant is a loud,
/// fail-closed outcome.
#[derive(Debug)]
pub enum VoiceLoadError {
/// The file could not be read.
Read(String),
/// The file exceeds [`MAX_VOICE_TOML_BYTES`].
TooLarge {
/// The offending path.
path: String,
/// Actual size in bytes.
bytes: u64,
},
/// The TOML did not parse.
Parse(String),
/// A voice entry was malformed (carries index + reason).
Invalid {
/// One-based entry index in the file.
index: usize,
/// Why it was rejected.
reason: String,
},
}
impl std::fmt::Display for VoiceLoadError {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
match self {
Self::Read(e) => write!(f, "tier-b voices: read failed: {e}"),
Self::TooLarge { path, bytes } => write!(
f,
"tier-b voices: {path} is {bytes} bytes, over the {MAX_VOICE_TOML_BYTES}-byte cap"
),
Self::Parse(e) => write!(f, "tier-b voices: TOML parse failed: {e}"),
Self::Invalid { index, reason } => {
write!(f, "tier-b voices: entry #{index} invalid: {reason}")
}
}
}
}
impl std::error::Error for VoiceLoadError {}
#[derive(serde::Deserialize)]
struct VoiceDoc {
name: String,
lang: String,
#[serde(default)]
default: bool,
}
#[derive(serde::Deserialize)]
struct VoicesDoc {
/// `[[voice]]` array-of-tables; empty/absent is a successful load of zero
/// voices, not an error.
#[serde(default)]
voice: Vec<VoiceDoc>,
}
/// Load + validate Tier-B speech-synthesis voice personas from a TOML file.
///
/// Every entry is validated: non-empty name and a non-empty, plausibly
/// formatted BCP 47 lang tag. The first malformed entry fails the whole load
/// (fail-closed).
///
/// # Errors
/// [`VoiceLoadError`] on read failure, oversize, parse failure, or a
/// malformed voice entry.
pub fn load_voices_from_toml(path: &Path) -> Result<Vec<VoicePersona>, VoiceLoadError> {
let meta = std::fs::metadata(path)
.map_err(|e| VoiceLoadError::Read(format!("{}: {e}", path.display())))?;
if meta.len() > MAX_VOICE_TOML_BYTES {
return Err(VoiceLoadError::TooLarge {
path: path.display().to_string(),
bytes: meta.len(),
});
}
let raw = std::fs::read_to_string(path)
.map_err(|e| VoiceLoadError::Read(format!("{}: {e}", path.display())))?;
let doc: VoicesDoc =
toml::from_str(&raw).map_err(|e| VoiceLoadError::Parse(e.to_string()))?;
let mut out = Vec::with_capacity(doc.voice.len());
for (idx, d) in doc.voice.into_iter().enumerate() {
let index = idx + 1;
let invalid = |reason: &str| VoiceLoadError::Invalid {
index,
reason: reason.to_string(),
};
let name = d.name.trim();
let lang = d.lang.trim();
if name.is_empty() {
return Err(invalid("empty name"));
}
if lang.is_empty() {
return Err(invalid("empty lang"));
}
// Plausibility: a BCP 47 tag has at least a 2-letter primary subtag
// and no spaces.
if lang.contains(' ') || lang.len() < 2 {
return Err(invalid(&format!(
"lang `{lang}` is not a plausible BCP 47 tag"
)));
}
out.push(VoicePersona {
name: Box::leak(name.to_string().into_boxed_str()),
lang: Box::leak(lang.to_string().into_boxed_str()),
default: d.default,
});
}
Ok(out)
}
/// Load every `*.toml` file in a Tier-B voice directory, merging them into
/// one pooled library. Files are processed in lexicographic order; the first
/// malformed file fails the whole load.
pub fn load_voice_directory(path: &Path) -> Result<Vec<VoicePersona>, VoiceLoadError> {
let mut entries: Vec<_> = std::fs::read_dir(path)
.map_err(|e| VoiceLoadError::Read(format!("{}: {e}", path.display())))?
.filter_map(|e| e.ok())
.filter(|e| {
e.path()
.extension()
.is_some_and(|ext| ext.eq_ignore_ascii_case("toml"))
})
.map(|e| e.path())
.collect();
entries.sort();
let mut out = Vec::new();
for path in entries {
out.extend(load_voices_from_toml(&path)?);
}
Ok(out)
}
}
#[cfg(feature = "tier-b-toml")]
pub use loader_impl::{load_voice_directory, load_voices_from_toml, VoiceLoadError};
#[cfg(all(test, feature = "tier-b-toml"))]
#[path = "voice_tier_b/tests.rs"]
mod tests;