use crate::{
analysis::{
linting::{IssueCategory, IssueSeverity, LintIssue, LintRule},
ScriptAnalysis,
},
parser::Section,
};
use alloc::{format, string::ToString, vec::Vec};
pub struct EncodingRule;
impl LintRule for EncodingRule {
fn id(&self) -> &'static str {
"encoding"
}
fn name(&self) -> &'static str {
"Encoding"
}
fn description(&self) -> &'static str {
"Detects potential encoding or character issues"
}
fn default_severity(&self) -> IssueSeverity {
IssueSeverity::Warning
}
fn category(&self) -> IssueCategory {
IssueCategory::Encoding
}
fn check_script(&self, analysis: &ScriptAnalysis) -> Vec<LintIssue> {
let mut issues = Vec::new();
if let Some(Section::Events(events)) = analysis
.script()
.sections()
.iter()
.find(|s| matches!(s, Section::Events(_)))
{
for event in events {
self.check_event_encoding(&mut issues, event);
}
}
self.check_script_info_encoding(&mut issues, analysis.script());
issues
}
}
impl EncodingRule {
fn check_event_encoding(&self, issues: &mut Vec<LintIssue>, event: &crate::parser::Event) {
if event
.text
.chars()
.any(|c| c.is_control() && c != '\n' && c != '\r' && c != '\t')
{
let issue = LintIssue::new(
self.default_severity(),
IssueCategory::Encoding,
self.id(),
"Event contains non-printable control characters".to_string(),
)
.with_description(
"Control characters may cause display issues in subtitle renderers".to_string(),
)
.with_suggested_fix(
"Remove or replace control characters with appropriate text".to_string(),
);
issues.push(issue);
}
if event.text.contains('\u{FFFD}') {
let issue = LintIssue::new(
self.default_severity(),
IssueCategory::Encoding,
self.id(),
"Event contains Unicode replacement character (�)".to_string(),
)
.with_description(
"Replacement characters indicate corrupted or invalid encoding".to_string(),
)
.with_suggested_fix("Check source file encoding and re-import".to_string());
issues.push(issue);
}
let char_count = event.text.chars().count();
let byte_count = event.text.len();
if char_count > 0 && byte_count > char_count * 3 {
let issue = LintIssue::new(
IssueSeverity::Hint,
IssueCategory::Encoding,
self.id(),
"Event contains many multi-byte characters".to_string(),
)
.with_description(
"Heavy use of multi-byte characters may impact performance".to_string(),
);
issues.push(issue);
}
}
fn check_script_info_encoding(
&self,
issues: &mut Vec<LintIssue>,
script: &crate::parser::Script,
) {
if let Some(Section::ScriptInfo(info)) = script
.sections()
.iter()
.find(|s| matches!(s, Section::ScriptInfo(_)))
{
for (key, value) in &info.fields {
if value
.chars()
.any(|c| c.is_control() && c != '\n' && c != '\r')
{
let issue = LintIssue::new(
self.default_severity(),
IssueCategory::Encoding,
self.id(),
format!("Script info field '{key}' contains control characters"),
);
issues.push(issue);
}
}
}
}
}