pinakes/crates/pinakes-core/src/search.rs
NotAShelf 3d9f8933d2
pinakes-core: update remaining modules and tests
Signed-off-by: NotAShelf <raf@notashelf.dev>
Change-Id: I9e0ff5ea33a5cf697473423e88f167ce6a6a6964
2026-03-08 00:43:30 +03:00

547 lines
14 KiB
Rust

use serde::{Deserialize, Serialize};
use winnow::{
ModalResult,
Parser,
combinator::{alt, delimited, preceded, repeat},
token::{take_till, take_while},
};
/// Represents a parsed search query.
#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
pub enum SearchQuery {
FullText(String),
FieldMatch {
field: String,
value: String,
},
And(Vec<Self>),
Or(Vec<Self>),
Not(Box<Self>),
Prefix(String),
Fuzzy(String),
TypeFilter(String),
TagFilter(String),
/// Range query: field:start..end (inclusive)
RangeQuery {
field: String,
start: Option<i64>,
end: Option<i64>,
},
/// Comparison query: field:>value, field:<value, field:>=value, field:<=value
CompareQuery {
field: String,
op: CompareOp,
value: i64,
},
/// Date query: created:today, modified:last-week, etc.
DateQuery {
field: String,
value: DateValue,
},
}
/// Comparison operators for range queries.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
pub enum CompareOp {
GreaterThan,
GreaterOrEqual,
LessThan,
LessOrEqual,
}
/// Date values for date-based queries.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub enum DateValue {
Today,
Yesterday,
ThisWeek,
LastWeek,
ThisMonth,
LastMonth,
ThisYear,
LastYear,
/// Days ago: last-7d, last-30d
DaysAgo(u32),
}
/// Request for executing a search.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct SearchRequest {
pub query: SearchQuery,
pub sort: SortOrder,
pub pagination: crate::model::Pagination,
}
/// Results of a search operation.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct SearchResults {
pub items: Vec<crate::model::MediaItem>,
pub total_count: u64,
}
/// Sorting options for search results.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
#[serde(rename_all = "snake_case")]
#[derive(Default)]
pub enum SortOrder {
#[default]
Relevance,
DateAsc,
DateDesc,
NameAsc,
NameDesc,
SizeAsc,
SizeDesc,
}
fn ws<'i>(input: &mut &'i str) -> ModalResult<&'i str> {
take_while(0.., ' ').parse_next(input)
}
fn quoted_string(input: &mut &str) -> ModalResult<String> {
delimited('"', take_till(0.., '"'), '"')
.map(|s: &str| s.to_string())
.parse_next(input)
}
fn bare_word(input: &mut &str) -> ModalResult<String> {
take_while(1.., |c: char| !c.is_whitespace() && c != ')' && c != '(')
.map(|s: &str| s.to_string())
.parse_next(input)
}
fn word_or_quoted(input: &mut &str) -> ModalResult<String> {
alt((quoted_string, bare_word)).parse_next(input)
}
fn not_expr(input: &mut &str) -> ModalResult<SearchQuery> {
preceded(('-', ws), atom)
.map(|q| SearchQuery::Not(Box::new(q)))
.parse_next(input)
}
/// Parse a date value like "today", "yesterday", "last-week", "last-30d"
fn parse_date_value(s: &str) -> Option<DateValue> {
match s.to_lowercase().as_str() {
"today" => Some(DateValue::Today),
"yesterday" => Some(DateValue::Yesterday),
"this-week" | "thisweek" => Some(DateValue::ThisWeek),
"last-week" | "lastweek" => Some(DateValue::LastWeek),
"this-month" | "thismonth" => Some(DateValue::ThisMonth),
"last-month" | "lastmonth" => Some(DateValue::LastMonth),
"this-year" | "thisyear" => Some(DateValue::ThisYear),
"last-year" | "lastyear" => Some(DateValue::LastYear),
other => {
// Try to parse "last-Nd" format (e.g., "last-7d", "last-30d")
if let Some(rest) = other.strip_prefix("last-")
&& let Some(days_str) = rest.strip_suffix('d')
&& let Ok(days) = days_str.parse::<u32>()
{
return Some(DateValue::DaysAgo(days));
}
None
},
}
}
/// Parse size strings like "10MB", "1GB", "500KB" to bytes
///
/// Returns `None` if the input is invalid or if the value would overflow.
fn parse_size_value(s: &str) -> Option<i64> {
let s = s.to_uppercase();
let (num_str, multiplier): (&str, i64) = s
.strip_suffix("GB")
.map(|n| (n, 1024 * 1024 * 1024_i64))
.or_else(|| s.strip_suffix("MB").map(|n| (n, 1024 * 1024)))
.or_else(|| s.strip_suffix("KB").map(|n| (n, 1024)))
.or_else(|| s.strip_suffix('B').map(|n| (n, 1)))
.unwrap_or((s.as_str(), 1));
let num: i64 = num_str.parse().ok()?;
num.checked_mul(multiplier)
}
fn field_match(input: &mut &str) -> ModalResult<SearchQuery> {
let field_name = take_while(1.., |c: char| c.is_alphanumeric() || c == '_')
.map(|s: &str| s.to_string());
(field_name, ':', word_or_quoted)
.map(|(field, _, value)| {
// Handle special field types
match field.as_str() {
"type" => return SearchQuery::TypeFilter(value),
"tag" => return SearchQuery::TagFilter(value),
_ => {},
}
// Check for range queries: field:start..end
if value.contains("..") {
let parts: Vec<&str> = value.split("..").collect();
if parts.len() == 2 {
let start = if parts[0].is_empty() {
None
} else if field == "size" {
parse_size_value(parts[0])
} else {
parts[0].parse().ok()
};
let end = if parts[1].is_empty() {
None
} else if field == "size" {
parse_size_value(parts[1])
} else {
parts[1].parse().ok()
};
return SearchQuery::RangeQuery { field, start, end };
}
}
// Check for comparison queries: >=, <=, >, <
if let Some(rest) = value.strip_prefix(">=") {
let val = if field == "size" {
parse_size_value(rest).unwrap_or(0)
} else {
rest.parse().unwrap_or(0)
};
return SearchQuery::CompareQuery {
field,
op: CompareOp::GreaterOrEqual,
value: val,
};
}
if let Some(rest) = value.strip_prefix("<=") {
let val = if field == "size" {
parse_size_value(rest).unwrap_or(0)
} else {
rest.parse().unwrap_or(0)
};
return SearchQuery::CompareQuery {
field,
op: CompareOp::LessOrEqual,
value: val,
};
}
if let Some(rest) = value.strip_prefix('>') {
let val = if field == "size" {
parse_size_value(rest).unwrap_or(0)
} else {
rest.parse().unwrap_or(0)
};
return SearchQuery::CompareQuery {
field,
op: CompareOp::GreaterThan,
value: val,
};
}
if let Some(rest) = value.strip_prefix('<') {
let val = if field == "size" {
parse_size_value(rest).unwrap_or(0)
} else {
rest.parse().unwrap_or(0)
};
return SearchQuery::CompareQuery {
field,
op: CompareOp::LessThan,
value: val,
};
}
// Check for date queries on created/modified fields
if (field == "created" || field == "modified")
&& let Some(date_val) = parse_date_value(&value)
{
return SearchQuery::DateQuery {
field,
value: date_val,
};
}
// Default: simple field match
SearchQuery::FieldMatch { field, value }
})
.parse_next(input)
}
fn prefix_expr(input: &mut &str) -> ModalResult<SearchQuery> {
let word = take_while(1.., |c: char| {
!c.is_whitespace() && c != ')' && c != '(' && c != '*'
})
.map(|s: &str| s.to_string());
(word, '*')
.map(|(w, _)| SearchQuery::Prefix(w))
.parse_next(input)
}
fn fuzzy_expr(input: &mut &str) -> ModalResult<SearchQuery> {
let word = take_while(1.., |c: char| {
!c.is_whitespace() && c != ')' && c != '(' && c != '~'
})
.map(|s: &str| s.to_string());
(word, '~')
.map(|(w, _)| SearchQuery::Fuzzy(w))
.parse_next(input)
}
fn paren_expr(input: &mut &str) -> ModalResult<SearchQuery> {
delimited(('(', ws), or_expr, (ws, ')')).parse_next(input)
}
fn not_or_keyword(input: &mut &str) -> ModalResult<()> {
if let Some(rest) = input.strip_prefix("OR")
&& (rest.is_empty() || rest.starts_with(' ') || rest.starts_with(')'))
{
return Err(winnow::error::ErrMode::Backtrack(
winnow::error::ContextError::new(),
));
}
Ok(())
}
fn full_text(input: &mut &str) -> ModalResult<SearchQuery> {
not_or_keyword.parse_next(input)?;
word_or_quoted.map(SearchQuery::FullText).parse_next(input)
}
fn atom(input: &mut &str) -> ModalResult<SearchQuery> {
alt((
paren_expr,
not_expr,
field_match,
prefix_expr,
fuzzy_expr,
full_text,
))
.parse_next(input)
}
fn and_expr(input: &mut &str) -> ModalResult<SearchQuery> {
let first = atom.parse_next(input)?;
let rest: Vec<SearchQuery> =
repeat(0.., preceded(ws, atom)).parse_next(input)?;
if rest.is_empty() {
Ok(first)
} else {
let mut terms = vec![first];
terms.extend(rest);
Ok(SearchQuery::And(terms))
}
}
fn or_expr(input: &mut &str) -> ModalResult<SearchQuery> {
let first = and_expr.parse_next(input)?;
let rest: Vec<SearchQuery> =
repeat(0.., preceded((ws, "OR", ws), and_expr)).parse_next(input)?;
if rest.is_empty() {
Ok(first)
} else {
let mut terms = vec![first];
terms.extend(rest);
Ok(SearchQuery::Or(terms))
}
}
/// Parses a search query string into a structured query.
///
/// Supports full-text search, field matches, operators (AND/OR/NOT),
/// prefixes, fuzzy matching, and type/tag filters.
///
/// # Arguments
///
/// * `input` - Raw query string
///
/// # Returns
///
/// Parsed query tree
///
/// # Errors
///
/// Returns `SearchParse` error for invalid syntax
pub fn parse_search_query(input: &str) -> crate::error::Result<SearchQuery> {
let trimmed = input.trim();
if trimmed.is_empty() {
return Ok(SearchQuery::FullText(String::new()));
}
let mut input = trimmed;
or_expr
.parse_next(&mut input)
.map_err(|e| crate::error::PinakesError::SearchParse(format!("{e}")))
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_simple_text() {
let q = parse_search_query("hello").unwrap();
assert_eq!(q, SearchQuery::FullText("hello".into()));
}
#[test]
fn test_field_match() {
let q = parse_search_query("artist:Beatles").unwrap();
assert_eq!(q, SearchQuery::FieldMatch {
field: "artist".into(),
value: "Beatles".into(),
});
}
#[test]
fn test_type_filter() {
let q = parse_search_query("type:pdf").unwrap();
assert_eq!(q, SearchQuery::TypeFilter("pdf".into()));
}
#[test]
fn test_tag_filter() {
let q = parse_search_query("tag:music").unwrap();
assert_eq!(q, SearchQuery::TagFilter("music".into()));
}
#[test]
fn test_and_implicit() {
let q = parse_search_query("hello world").unwrap();
assert_eq!(
q,
SearchQuery::And(vec![
SearchQuery::FullText("hello".into()),
SearchQuery::FullText("world".into()),
])
);
}
#[test]
fn test_or() {
let q = parse_search_query("hello OR world").unwrap();
assert_eq!(
q,
SearchQuery::Or(vec![
SearchQuery::FullText("hello".into()),
SearchQuery::FullText("world".into()),
])
);
}
#[test]
fn test_not() {
let q = parse_search_query("-excluded").unwrap();
assert_eq!(
q,
SearchQuery::Not(Box::new(SearchQuery::FullText("excluded".into())))
);
}
#[test]
fn test_prefix() {
let q = parse_search_query("hel*").unwrap();
assert_eq!(q, SearchQuery::Prefix("hel".into()));
}
#[test]
fn test_fuzzy() {
let q = parse_search_query("hello~").unwrap();
assert_eq!(q, SearchQuery::Fuzzy("hello".into()));
}
#[test]
fn test_quoted() {
let q = parse_search_query("\"hello world\"").unwrap();
assert_eq!(q, SearchQuery::FullText("hello world".into()));
}
#[test]
fn test_range_query_year() {
let q = parse_search_query("year:2020..2023").unwrap();
assert_eq!(q, SearchQuery::RangeQuery {
field: "year".into(),
start: Some(2020),
end: Some(2023),
});
}
#[test]
fn test_range_query_open_start() {
let q = parse_search_query("year:..2023").unwrap();
assert_eq!(q, SearchQuery::RangeQuery {
field: "year".into(),
start: None,
end: Some(2023),
});
}
#[test]
fn test_range_query_open_end() {
let q = parse_search_query("year:2020..").unwrap();
assert_eq!(q, SearchQuery::RangeQuery {
field: "year".into(),
start: Some(2020),
end: None,
});
}
#[test]
fn test_compare_greater_than() {
let q = parse_search_query("year:>2020").unwrap();
assert_eq!(q, SearchQuery::CompareQuery {
field: "year".into(),
op: CompareOp::GreaterThan,
value: 2020,
});
}
#[test]
fn test_compare_less_or_equal() {
let q = parse_search_query("year:<=2023").unwrap();
assert_eq!(q, SearchQuery::CompareQuery {
field: "year".into(),
op: CompareOp::LessOrEqual,
value: 2023,
});
}
#[test]
fn test_size_compare_mb() {
let q = parse_search_query("size:>10MB").unwrap();
assert_eq!(q, SearchQuery::CompareQuery {
field: "size".into(),
op: CompareOp::GreaterThan,
value: 10 * 1024 * 1024,
});
}
#[test]
fn test_size_range_gb() {
let q = parse_search_query("size:1GB..2GB").unwrap();
assert_eq!(q, SearchQuery::RangeQuery {
field: "size".into(),
start: Some(1024 * 1024 * 1024),
end: Some(2 * 1024 * 1024 * 1024),
});
}
#[test]
fn test_date_query_today() {
let q = parse_search_query("created:today").unwrap();
assert_eq!(q, SearchQuery::DateQuery {
field: "created".into(),
value: DateValue::Today,
});
}
#[test]
fn test_date_query_last_week() {
let q = parse_search_query("modified:last-week").unwrap();
assert_eq!(q, SearchQuery::DateQuery {
field: "modified".into(),
value: DateValue::LastWeek,
});
}
#[test]
fn test_date_query_days_ago() {
let q = parse_search_query("created:last-30d").unwrap();
assert_eq!(q, SearchQuery::DateQuery {
field: "created".into(),
value: DateValue::DaysAgo(30),
});
}
}