vortex_array/
context.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use std::fmt::Display;
5use std::sync::Arc;
6
7use itertools::Itertools;
8use parking_lot::RwLock;
9use vortex_error::{VortexExpect, VortexResult, vortex_err};
10use vortex_utils::aliases::hash_map::HashMap;
11
12use crate::EncodingRef;
13use crate::arrays::{
14    BoolEncoding, ChunkedEncoding, ConstantEncoding, DecimalEncoding, ExtensionEncoding,
15    FixedSizeListEncoding, ListEncoding, NullEncoding, PrimitiveEncoding, StructEncoding,
16    VarBinEncoding, VarBinViewEncoding,
17};
18
19/// A collection of array encodings.
20// TODO(ngates): it feels weird that this has interior mutability. I think maybe it shouldn't.
21pub type ArrayContext = VTableContext<EncodingRef>;
22pub type ArrayRegistry = VTableRegistry<EncodingRef>;
23
24impl ArrayRegistry {
25    pub fn canonical_only() -> Self {
26        let mut this = Self::empty();
27
28        // Register the canonical encodings
29        this.register_many([
30            EncodingRef::new_ref(NullEncoding.as_ref()) as EncodingRef,
31            EncodingRef::new_ref(BoolEncoding.as_ref()),
32            EncodingRef::new_ref(PrimitiveEncoding.as_ref()),
33            EncodingRef::new_ref(DecimalEncoding.as_ref()),
34            EncodingRef::new_ref(StructEncoding.as_ref()),
35            EncodingRef::new_ref(ListEncoding.as_ref()),
36            EncodingRef::new_ref(FixedSizeListEncoding.as_ref()),
37            EncodingRef::new_ref(VarBinEncoding.as_ref()),
38            EncodingRef::new_ref(VarBinViewEncoding.as_ref()),
39            EncodingRef::new_ref(ExtensionEncoding.as_ref()),
40        ]);
41
42        // Register the utility encodings
43        this.register_many([
44            EncodingRef::new_ref(ConstantEncoding.as_ref()) as EncodingRef,
45            EncodingRef::new_ref(ChunkedEncoding.as_ref()),
46        ]);
47
48        this
49    }
50}
51
52/// A collection of encodings that can be addressed by a u16 positional index.
53/// This is used to map array encodings and layout encodings when reading from a file.
54#[derive(Debug, Clone)]
55pub struct VTableContext<T>(Arc<RwLock<Vec<T>>>);
56
57impl<T: Clone + Eq> VTableContext<T> {
58    pub fn empty() -> Self {
59        Self(Arc::new(RwLock::new(Vec::new())))
60    }
61
62    pub fn with(self, encoding: T) -> Self {
63        {
64            let mut write = self.0.write();
65            if write.iter().all(|e| e != &encoding) {
66                write.push(encoding);
67            }
68        }
69        self
70    }
71
72    pub fn with_many<E: IntoIterator<Item = T>>(self, items: E) -> Self {
73        items.into_iter().fold(self, |ctx, e| ctx.with(e))
74    }
75
76    pub fn encodings(&self) -> Vec<T> {
77        self.0.read().clone()
78    }
79
80    /// Returns the index of the encoding in the context, or adds it if it doesn't exist.
81    pub fn encoding_idx(&self, encoding: &T) -> u16 {
82        let mut write = self.0.write();
83        if let Some(idx) = write.iter().position(|e| e == encoding) {
84            return u16::try_from(idx).vortex_expect("Cannot have more than u16::MAX encodings");
85        }
86        assert!(
87            write.len() < u16::MAX as usize,
88            "Cannot have more than u16::MAX encodings"
89        );
90        write.push(encoding.clone());
91        u16::try_from(write.len() - 1).vortex_expect("checked already")
92    }
93
94    /// Find an encoding by its position.
95    pub fn lookup_encoding(&self, idx: u16) -> Option<T> {
96        self.0.read().get(idx as usize).cloned()
97    }
98}
99
100/// A registry of encodings that can be used to construct a context for serde.
101///
102/// In the future, we will support loading encodings from shared libraries or even from within
103/// the Vortex file itself. This registry will be used to manage the available encodings.
104#[derive(Clone, Debug)]
105pub struct VTableRegistry<T>(HashMap<String, T>);
106
107// TODO(ngates): define a trait for `T` that requires an `id` method returning a `Arc<str>` and
108//  auto-implement `Display` and `Eq` for it.
109impl<T: Clone + Display + Eq> VTableRegistry<T> {
110    pub fn empty() -> Self {
111        Self(Default::default())
112    }
113
114    /// Create a new [`VTableContext`] with the provided encodings.
115    pub fn new_context<'a>(
116        &self,
117        encoding_ids: impl Iterator<Item = &'a str>,
118    ) -> VortexResult<VTableContext<T>> {
119        let mut ctx = VTableContext::<T>::empty();
120        for id in encoding_ids {
121            let encoding = self.0.get(id).ok_or_else(|| {
122                vortex_err!(
123                    "Array encoding {} not found in registry {}",
124                    id,
125                    self.0.values().join(", ")
126                )
127            })?;
128            ctx = ctx.with(encoding.clone());
129        }
130        Ok(ctx)
131    }
132
133    /// List the vtables in the registry.
134    pub fn vtables(&self) -> impl Iterator<Item = &T> + '_ {
135        self.0.values()
136    }
137
138    /// Find the encoding with the given ID.
139    pub fn get(&self, id: &str) -> Option<&T> {
140        self.0.get(id)
141    }
142
143    /// Register a new encoding, replacing any existing encoding with the same ID.
144    pub fn register(&mut self, encoding: T) {
145        self.0.insert(encoding.to_string(), encoding);
146    }
147
148    /// Register a new encoding, replacing any existing encoding with the same ID.
149    pub fn register_many<I: IntoIterator<Item = T>>(&mut self, encodings: I) {
150        self.0
151            .extend(encodings.into_iter().map(|e| (e.to_string(), e)));
152    }
153}
vortex_array/context.rs

vortex_array/
context.rs
