From 4c8d9dbb638af51410702a3124caeda490b1e16a Mon Sep 17 00:00:00 2001
From: Tobie Morgan Hitchcock <tobie@abcum.com>
Date: Wed, 27 Apr 2022 16:21:51 +0100
Subject: [PATCH] implement COLLATE and NUMERIC ordering in ORDER BY clauses
---
Cargo.lock | 16 +++++++++++
lib/Cargo.toml | 1 +
lib/src/dbs/iterator.rs | 4 +--
lib/src/sql/value/compare.rs | 51 ++++++++++++++++++++++--------------
lib/src/sql/value/value.rs | 31 ++++++++++++++++++++++
5 files changed, 81 insertions(+), 22 deletions(-)
diff --git a/Cargo.lock b/Cargo.lock
index 391e2f36..ddba264c 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -32,6 +32,12 @@ dependencies = [
"alloc-no-stdlib",
]
+[[package]]
+name = "any_ascii"
+version = "0.1.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "70033777eb8b5124a81a1889416543dddef2de240019b674c81285a2635a7e1e"
+
[[package]]
name = "anyhow"
version = "1.0.56"
@@ -1123,6 +1129,15 @@ version = "1.3.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "830d08ce1d1d941e6b30645f1a0eb5643013d835ce3779a5fc208261dbe10f55"
+[[package]]
+name = "lexical-sort"
+version = "0.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c09e4591611e231daf4d4c685a66cb0410cc1e502027a20ae55f2bb9e997207a"
+dependencies = [
+ "any_ascii",
+]
+
[[package]]
name = "libc"
version = "0.2.121"
@@ -2242,6 +2257,7 @@ dependencies = [
"fuzzy-matcher",
"geo",
"indxdb",
+ "lexical-sort",
"log",
"md-5",
"nanoid",
diff --git a/lib/Cargo.toml b/lib/Cargo.toml
index 3eb09d88..4c864146 100644
--- a/lib/Cargo.toml
+++ b/lib/Cargo.toml
@@ -26,6 +26,7 @@ futures = "0.3.21"
fuzzy-matcher = "0.3.7"
geo = { version = "0.19.0", features = ["use-serde"] }
indxdb = { version = "0.2.0", optional = true }
+lexical-sort = "0.3.1"
log = "0.4.16"
md-5 = "0.10.1"
nanoid = "0.4.0"
diff --git a/lib/src/dbs/iterator.rs b/lib/src/dbs/iterator.rs
index 5f6916ec..d11f06c1 100644
--- a/lib/src/dbs/iterator.rs
+++ b/lib/src/dbs/iterator.rs
@@ -296,8 +296,8 @@ impl Iterator {
a.partial_cmp(&b)
}
false => match order.direction {
- true => a.compare(b, &order.order),
- false => b.compare(a, &order.order),
+ true => a.compare(b, &order.order, order.collate, order.numeric),
+ false => b.compare(a, &order.order, order.collate, order.numeric),
},
};
//
diff --git a/lib/src/sql/value/compare.rs b/lib/src/sql/value/compare.rs
index 4d9f8d8b..0dec7568 100644
--- a/lib/src/sql/value/compare.rs
+++ b/lib/src/sql/value/compare.rs
@@ -4,14 +4,20 @@ use crate::sql::value::Value;
use std::cmp::Ordering;
impl Value {
- pub fn compare(&self, other: &Self, path: &[Part]) -> Option<Ordering> {
+ pub fn compare(
+ &self,
+ other: &Self,
+ path: &[Part],
+ collate: bool,
+ numeric: bool,
+ ) -> Option<Ordering> {
match path.first() {
// Get the current path part
Some(p) => match (self, other) {
// Current path part is an object
(Value::Object(a), Value::Object(b)) => match p {
Part::Field(f) => match (a.value.get(&f.name), b.value.get(&f.name)) {
- (Some(a), Some(b)) => a.compare(b, path.next()),
+ (Some(a), Some(b)) => a.compare(b, path.next(), collate, numeric),
(Some(_), None) => Some(Ordering::Greater),
(None, Some(_)) => Some(Ordering::Less),
(_, _) => Some(Ordering::Equal),
@@ -22,7 +28,7 @@ impl Value {
(Value::Array(a), Value::Array(b)) => match p {
Part::All => {
for (a, b) in a.value.iter().zip(b.value.iter()) {
- match a.compare(b, path.next()) {
+ match a.compare(b, path.next(), collate, numeric) {
Some(Ordering::Equal) => continue,
None => continue,
o => return o,
@@ -35,20 +41,20 @@ impl Value {
}
}
Part::First => match (a.value.first(), b.value.first()) {
- (Some(a), Some(b)) => a.compare(b, path.next()),
+ (Some(a), Some(b)) => a.compare(b, path.next(), collate, numeric),
(Some(_), None) => Some(Ordering::Greater),
(None, Some(_)) => Some(Ordering::Less),
(_, _) => Some(Ordering::Equal),
},
Part::Last => match (a.value.first(), b.value.first()) {
- (Some(a), Some(b)) => a.compare(b, path.next()),
+ (Some(a), Some(b)) => a.compare(b, path.next(), collate, numeric),
(Some(_), None) => Some(Ordering::Greater),
(None, Some(_)) => Some(Ordering::Less),
(_, _) => Some(Ordering::Equal),
},
Part::Index(i) => {
match (a.value.get(i.to_usize()), b.value.get(i.to_usize())) {
- (Some(a), Some(b)) => a.compare(b, path.next()),
+ (Some(a), Some(b)) => a.compare(b, path.next(), collate, numeric),
(Some(_), None) => Some(Ordering::Greater),
(None, Some(_)) => Some(Ordering::Less),
(_, _) => Some(Ordering::Equal),
@@ -56,7 +62,7 @@ impl Value {
}
_ => {
for (a, b) in a.value.iter().zip(b.value.iter()) {
- match a.compare(b, path) {
+ match a.compare(b, path, collate, numeric) {
Some(Ordering::Equal) => continue,
None => continue,
o => return o,
@@ -70,10 +76,15 @@ impl Value {
}
},
// Ignore everything else
- (a, b) => a.compare(b, path.next()),
+ (a, b) => a.compare(b, path.next(), collate, numeric),
},
// No more parts so get the value
- None => self.partial_cmp(other),
+ None => match (collate, numeric) {
+ (true, true) => self.natural_lexical_cmp(other),
+ (true, false) => self.lexical_cmp(other),
+ (false, true) => self.natural_cmp(other),
+ _ => self.partial_cmp(other),
+ },
}
}
}
@@ -90,7 +101,7 @@ mod tests {
let idi = Idiom::default();
let one = Value::parse("{ test: { other: null, something: 456 } }");
let two = Value::parse("{ test: { other: null, something: 123 } }");
- let res = one.compare(&two, &idi);
+ let res = one.compare(&two, &idi, false, false);
assert_eq!(res, Some(Ordering::Greater));
}
@@ -99,7 +110,7 @@ mod tests {
let idi = Idiom::parse("test.something");
let one = Value::parse("{ test: { other: null, something: 456 } }");
let two = Value::parse("{ test: { other: null, something: 123 } }");
- let res = one.compare(&two, &idi);
+ let res = one.compare(&two, &idi, false, false);
assert_eq!(res, Some(Ordering::Greater));
}
@@ -108,7 +119,7 @@ mod tests {
let idi = Idiom::parse("test.something");
let one = Value::parse("{ test: { other: null } }");
let two = Value::parse("{ test: { other: null, something: 123 } }");
- let res = one.compare(&two, &idi);
+ let res = one.compare(&two, &idi, false, false);
assert_eq!(res, Some(Ordering::Less));
}
@@ -117,7 +128,7 @@ mod tests {
let idi = Idiom::parse("test.something");
let one = Value::parse("{ test: { other: null, something: 456 } }");
let two = Value::parse("{ test: { other: null } }");
- let res = one.compare(&two, &idi);
+ let res = one.compare(&two, &idi, false, false);
assert_eq!(res, Some(Ordering::Greater));
}
@@ -126,7 +137,7 @@ mod tests {
let idi = Idiom::parse("test.something.*");
let one = Value::parse("{ test: { other: null, something: [4, 5, 6] } }");
let two = Value::parse("{ test: { other: null, something: [1, 2, 3] } }");
- let res = one.compare(&two, &idi);
+ let res = one.compare(&two, &idi, false, false);
assert_eq!(res, Some(Ordering::Greater));
}
@@ -135,7 +146,7 @@ mod tests {
let idi = Idiom::parse("test.something.*");
let one = Value::parse("{ test: { other: null, something: [1, 2, 3, 4, 5, 6] } }");
let two = Value::parse("{ test: { other: null, something: [1, 2, 3] } }");
- let res = one.compare(&two, &idi);
+ let res = one.compare(&two, &idi, false, false);
assert_eq!(res, Some(Ordering::Greater));
}
@@ -144,7 +155,7 @@ mod tests {
let idi = Idiom::parse("test.something.*");
let one = Value::parse("{ test: { other: null, something: [1, 2, 3] } }");
let two = Value::parse("{ test: { other: null, something: [1, 2, 3, 4, 5, 6] } }");
- let res = one.compare(&two, &idi);
+ let res = one.compare(&two, &idi, false, false);
assert_eq!(res, Some(Ordering::Less));
}
@@ -153,7 +164,7 @@ mod tests {
let idi = Idiom::parse("test.something.*");
let one = Value::parse("{ test: { other: null, something: null } }");
let two = Value::parse("{ test: { other: null, something: [1, 2, 3] } }");
- let res = one.compare(&two, &idi);
+ let res = one.compare(&two, &idi, false, false);
assert_eq!(res, Some(Ordering::Less));
}
@@ -162,7 +173,7 @@ mod tests {
let idi = Idiom::parse("test.something.*");
let one = Value::parse("{ test: { other: null, something: [4, 5, 6] } }");
let two = Value::parse("{ test: { other: null, something: null } }");
- let res = one.compare(&two, &idi);
+ let res = one.compare(&two, &idi, false, false);
assert_eq!(res, Some(Ordering::Greater));
}
@@ -171,7 +182,7 @@ mod tests {
let idi = Idiom::parse("test.something.*");
let one = Value::parse("{ test: { other: null, something: [1, null, 3] } }");
let two = Value::parse("{ test: { other: null, something: [1, 2, 3] } }");
- let res = one.compare(&two, &idi);
+ let res = one.compare(&two, &idi, false, false);
assert_eq!(res, Some(Ordering::Less));
}
@@ -180,7 +191,7 @@ mod tests {
let idi = Idiom::parse("test.something.*");
let one = Value::parse("{ test: { other: null, something: [1, 2, 3] } }");
let two = Value::parse("{ test: { other: null, something: [1, null, 3] } }");
- let res = one.compare(&two, &idi);
+ let res = one.compare(&two, &idi, false, false);
assert_eq!(res, Some(Ordering::Greater));
}
}
diff --git a/lib/src/sql/value/value.rs b/lib/src/sql/value/value.rs
index ff86aca6..eeaec141 100644
--- a/lib/src/sql/value/value.rs
+++ b/lib/src/sql/value/value.rs
@@ -913,6 +913,37 @@ impl Value {
_ => false,
}
}
+
+ // -----------------------------------
+ // Sorting operations
+ // -----------------------------------
+
+ pub fn lexical_cmp(&self, other: &Value) -> Option<Ordering> {
+ match (self, other) {
+ (Value::Strand(a), Value::Strand(b)) => {
+ Some(lexical_sort::lexical_cmp(&a.value, &b.value))
+ }
+ _ => self.partial_cmp(other),
+ }
+ }
+
+ pub fn natural_cmp(&self, other: &Value) -> Option<Ordering> {
+ match (self, other) {
+ (Value::Strand(a), Value::Strand(b)) => {
+ Some(lexical_sort::natural_cmp(&a.value, &b.value))
+ }
+ _ => self.partial_cmp(other),
+ }
+ }
+
+ pub fn natural_lexical_cmp(&self, other: &Value) -> Option<Ordering> {
+ match (self, other) {
+ (Value::Strand(a), Value::Strand(b)) => {
+ Some(lexical_sort::natural_lexical_cmp(&a.value, &b.value))
+ }
+ _ => self.partial_cmp(other),
+ }
+ }
}
impl fmt::Display for Value {