feat(sync): 完成 S-08 数据分类

This commit is contained in:
2026-09-09 12:07:45 +08:00
parent b08b7815e5
commit bafec01c41
14 changed files with 1036 additions and 26 deletions
@@ -101,6 +101,60 @@ struct Preferences {
markdown: Markdown,
presets: Vec<Preset>,
}
#[derive(Deserialize)]
#[serde(deny_unknown_fields)]
struct ConversationMessage {
message_id: String,
parent_message_id: Option<String>,
role: String,
content: String,
thinking: Option<String>,
attachments: Vec<String>,
created_at_ms: i64,
}
#[derive(Deserialize)]
#[serde(deny_unknown_fields)]
struct Conversation {
title: String,
active_leaf: Option<String>,
created_at_ms: i64,
updated_at_ms: i64,
messages: Vec<ConversationMessage>,
}
#[derive(Deserialize)]
#[serde(deny_unknown_fields)]
struct AgentHistory {
status: String,
input: String,
output: Option<String>,
model: String,
skill_id: Option<String>,
error_code: Option<String>,
error_message: Option<String>,
token_usage: u64,
created_at_ms: i64,
updated_at_ms: i64,
}
#[derive(Deserialize)]
#[serde(deny_unknown_fields)]
struct ProviderSettings {
version: u64,
provider_type: String,
name: String,
base_url: Option<String>,
default_model: Option<String>,
enabled: bool,
capabilities: Vec<String>,
}
#[derive(Deserialize)]
#[serde(deny_unknown_fields)]
struct ExtensionInstallation {
package_kind: String,
package_id: String,
source: String,
version: String,
sha256: String,
}
fn decode<T: serde::de::DeserializeOwned>(value: &Value) -> Result<T> {
serde_json::from_value(value.clone()).map_err(|_| HostError::new("RECORD_SCHEMA_INVALID"))
}
@@ -138,6 +192,44 @@ fn unique_bounded(values: &[String], max_items: usize, max_chars: usize) -> bool
.enumerate()
.all(|(index, value)| !values[..index].contains(value))
}
fn bounded_identifier(value: &str, max_chars: usize) -> bool {
!value.is_empty()
&& value.chars().count() <= max_chars
&& value
.bytes()
.all(|b| b.is_ascii_alphanumeric() || b"._:-".contains(&b))
}
fn portable_reference(value: &str, max_chars: usize) -> bool {
!value.is_empty()
&& value.chars().count() <= max_chars
&& !value.starts_with('/')
&& !value.contains(['\\', '\0'])
&& value
.split('/')
.all(|part| !part.is_empty() && part != "." && part != "..")
}
fn model_identifier(value: &str) -> bool {
!value.is_empty()
&& value.chars().count() <= 256
&& value
.bytes()
.all(|byte| byte.is_ascii_alphanumeric() || b"._:/+-".contains(&byte))
}
fn timestamp(value: i64) -> bool {
(0..=253402300799999).contains(&value)
}
fn safe_base_url(value: &str) -> bool {
let Some(rest) = value
.strip_prefix("https://")
.or_else(|| value.strip_prefix("http://"))
else {
return false;
};
!rest.is_empty()
&& !rest.bytes().any(|byte| byte.is_ascii_control())
&& !rest.contains(['@', '?', '#', '\\'])
&& rest.split('/').next().is_some_and(|host| !host.is_empty())
}
pub fn validate(kind: &str, value: &Value) -> Result<()> {
let valid = match kind {
"persona" => {
@@ -252,6 +344,138 @@ pub fn validate(kind: &str, value: &Value) -> Result<()> {
&& markdown(&p.preferences)
})
}
"conversation" => {
let value: Conversation = decode(value)?;
let message_id = |id: &str| bounded_identifier(id, 128);
!value.title.trim().is_empty()
&& value.title.chars().count() <= 120
&& timestamp(value.created_at_ms)
&& timestamp(value.updated_at_ms)
&& value.updated_at_ms >= value.created_at_ms
&& value.messages.len() <= 2_000
&& value.active_leaf.as_deref().is_none_or(|id| {
message_id(id) && value.messages.iter().any(|v| v.message_id == id)
})
&& value.messages.iter().enumerate().all(|(index, message)| {
message_id(&message.message_id)
&& !value.messages[..index]
.iter()
.any(|previous| previous.message_id == message.message_id)
&& message.parent_message_id.as_deref().is_none_or(|id| {
message_id(id)
&& id != message.message_id
&& value
.messages
.iter()
.any(|candidate| candidate.message_id == id)
})
&& matches!(message.role.as_str(), "user" | "assistant" | "system")
&& message.content.chars().count() <= 262_144
&& message
.thinking
.as_ref()
.is_none_or(|text| text.chars().count() <= 262_144)
&& message.attachments.len() <= 64
&& message
.attachments
.iter()
.all(|path| portable_reference(path, 512))
&& message
.attachments
.iter()
.enumerate()
.all(|(index, path)| !message.attachments[..index].contains(path))
&& timestamp(message.created_at_ms)
})
}
"agent_history" => {
let value: AgentHistory = decode(value)?;
matches!(value.status.as_str(), "completed" | "failed" | "cancelled")
&& value.input.chars().count() <= 262_144
&& value
.output
.as_ref()
.is_none_or(|text| text.chars().count() <= 524_288)
&& model_identifier(&value.model)
&& value
.skill_id
.as_deref()
.is_none_or(|id| bounded_identifier(id, 128))
&& value
.error_code
.as_deref()
.is_none_or(|code| bounded_identifier(code, 128))
&& value
.error_message
.as_ref()
.is_none_or(|message| message.chars().count() <= 4_096)
&& value.token_usage <= 9_007_199_254_740_991
&& timestamp(value.created_at_ms)
&& timestamp(value.updated_at_ms)
&& value.updated_at_ms >= value.created_at_ms
}
"provider_settings" => {
let value: ProviderSettings = decode(value)?;
let _ = value.enabled;
let known_capabilities = [
"chat",
"vision",
"tool_calling",
"reasoning",
"streaming",
"structured_output",
"embedding",
"transcription",
"speaker_matching",
];
value.version <= 9_007_199_254_740_991
&& matches!(
value.provider_type.as_str(),
"mock"
| "openai_responses"
| "openai_chat"
| "openai_compatible"
| "anthropic_messages"
| "ollama"
)
&& !value.name.trim().is_empty()
&& value.name.chars().count() <= 128
&& value
.base_url
.as_deref()
.is_none_or(|url| url.len() <= 2_048 && safe_base_url(url))
&& value.default_model.as_deref().is_none_or(model_identifier)
&& unique_bounded(&value.capabilities, 16, 64)
&& value
.capabilities
.iter()
.all(|capability| known_capabilities.contains(&capability.as_str()))
}
"extension_installation" => {
let value: ExtensionInstallation = decode(value)?;
matches!(value.package_kind.as_str(), "skill" | "plugin" | "theme")
&& bounded_identifier(&value.package_id, 128)
&& !value.source.is_empty()
&& value.source.len() <= 512
&& value
.source
.bytes()
.all(|byte| byte.is_ascii_alphanumeric() || b"._:/-".contains(&byte))
&& !value.source.contains("..")
&& !value.source.contains('@')
&& !value.source.contains(['?', '#', '\\'])
&& !value.version.is_empty()
&& value.version.len() <= 128
&& value
.version
.bytes()
.all(|byte| byte.is_ascii_alphanumeric() || b".+-_".contains(&byte))
&& value.sha256.len() == 64
&& value
.sha256
.bytes()
.all(|byte| byte.is_ascii_digit() || (b'a'..=b'f').contains(&byte))
}
_ => return Err(HostError::new("RECORD_SCHEMA_UNSUPPORTED")),
};
if valid {
@@ -347,4 +571,71 @@ mod tests {
"RECORD_DATA_INVALID"
);
}
#[test]
fn s08_optional_schemas_are_strict_and_installations_cannot_authorize() {
let fixtures = [
(
"conversation",
"conversation_00000000000000000000000000000001",
json!({"title":"Review","active_leaf":"message_1","created_at_ms":1,"updated_at_ms":2,"messages":[{"message_id":"message_1","parent_message_id":null,"role":"user","content":"Hello","thinking":null,"attachments":[],"created_at_ms":1}]}),
),
(
"agent_history",
"agent_run_00000000000000000000000000000001",
json!({"status":"completed","input":"Review","output":"Done","model":"gpt-5.6","skill_id":null,"error_code":null,"error_message":null,"token_usage":42,"created_at_ms":1,"updated_at_ms":2}),
),
(
"provider_settings",
"provider_00000000000000000000000000000001",
json!({"version":1,"provider_type":"openai_responses","name":"OpenAI","base_url":"https://api.openai.com/v1","default_model":"gpt-5.6","enabled":true,"capabilities":["chat","tool_calling"]}),
),
(
"extension_installation",
"extension_00000000000000000000000000000001",
json!({"package_kind":"plugin","package_id":"opennexus.review","source":"community/opennexus.review","version":"1.2.3","sha256":"a".repeat(64)}),
),
];
for (kind, id, data) in fixtures {
let path = crate::records::path_for(kind, id).unwrap();
let record = json!({"schema":1,"kind":kind,"id":id,"data":data});
crate::records::validate(&path, &serde_json::to_vec(&record).unwrap()).unwrap();
assert!(crate::records::allowed(&path));
assert!(!crate::sync_scope::OptionalScope::default().includes(&path));
let mut unknown = record.clone();
unknown["data"]["api_key"] = json!("must-never-sync");
assert_eq!(
crate::records::validate(&path, &serde_json::to_vec(&unknown).unwrap())
.err()
.unwrap()
.code,
"RECORD_SCHEMA_INVALID"
);
}
let installation = json!({"package_kind":"plugin","package_id":"opennexus.review","source":"community/opennexus.review","version":"1.2.3","sha256":"a".repeat(64)});
for forbidden in [
"permissions",
"granted_permissions",
"enabled",
"trusted",
"device_grants",
"package_path",
] {
let mut invalid = installation.clone();
invalid[forbidden] = json!(true);
assert_eq!(
validate("extension_installation", &invalid)
.unwrap_err()
.code,
"RECORD_SCHEMA_INVALID"
);
}
let mut provider = json!({"version":1,"provider_type":"openai_responses","name":"OpenAI","base_url":"https://api.openai.com/v1","default_model":"gpt-5.6","enabled":true,"capabilities":["chat"]});
provider["credential_id"] = json!("device-secret-reference");
assert_eq!(
validate("provider_settings", &provider).unwrap_err().code,
"RECORD_SCHEMA_INVALID"
);
}
}