Implement zoneinfo parsing and tzif structures

It's time to finally tackle the zoneinfo and tzif sourcing
issue. The work here lays the baseline for to handle sourcing.
There's still a lot more to add, and maybe some bugs to
uncover. But this is also major progress.

Author: nekevss

Cargo.lock

@@ -1,6 +1,6 @@
 [workspace]
 resolver = "2"
-members = ["diplomat-gen", "bakeddata", "provider", "temporal_capi"]
+members = ["diplomat-gen", "bakeddata", "provider", "temporal_capi", "zoneinfo"]
 
 [workspace.package]
 edition = "2021"

Cargo.toml

@@ -3,6 +3,7 @@
 //!
 
 mod tzdb;
+mod tzif;
 
 pub use tzdb::{IanaDataError, IanaIdentifierNormalizer};
 

provider/src/lib.rs

@@ -53,8 +53,8 @@ const ZONE_INFO_FILES: [&str; 9] = [
 ];
 
 pub struct TzdbDataProvider {
-    version: String,
-    data: Table,
+    pub(crate) version: String,
+    pub(crate) data: Table,
 }
 
 impl TzdbDataProvider {

provider/src/tzdb.rs

@@ -0,0 +1,87 @@
+//! A compact, zero copy TZif file.
+//!
+//! NOTE: This representation does not follow the TZif specification
+//! to full detail, but instead attempts to compress TZif data into
+//! a functional, data driven equivalent.
+
+use parse_zoneinfo::table::Table;
+use std::{
+    borrow::Cow,
+    collections::{BTreeMap, BTreeSet},
+    path::Path,
+};
+use zerotrie::{ZeroTrieBuildError, ZeroTrieSimpleAscii};
+use zerovec::{VarZeroVec, ZeroVec};
+
+use crate::tzdb::TzdbDataProvider;
+
+#[derive(Debug, Clone, yoke::Yokeable, databake::Bake, serde::Serialize)]
+#[databake(path = temporal_provider::tzif)]
+pub struct ZeroZoneInfo<'data> {
+    // Why u16? It would suck to have to refactor because there are > 256 TZifs
+    ids: ZeroTrieSimpleAscii<ZeroVec<'data, u8>>,
+
+    tzifs: VarZeroVec<'data, ZeroTzifULE>,
+}
+
+#[zerovec::make_varule(ZeroTzifULE)]
+#[derive(PartialEq, Debug, Clone, yoke::Yokeable, serde::Serialize, databake::Bake)]
+#[zerovec::skip_derive(Ord)]
+#[zerovec::derive(Debug, Serialize)]
+#[databake(path = temporal_provider::tzif)]
+pub struct ZeroTzif<'data> {
+    transitions: ZeroVec<'data, i64>,
+    // NOTE: zoneinfo64 does a fun little bitmap str
+    transition_types: ZeroVec<'data, u32>,
+    types: ZeroVec<'data, i64>,
+    posix: Cow<'data, str>,
+}
+
+pub enum ZoneInfoDataError {
+    Build(ZeroTrieBuildError),
+}
+
+impl ZeroZoneInfo<'_> {
+    pub fn build(tzdata: &Path) -> Result<Self, ZoneInfoDataError> {
+        let provider = TzdbDataProvider::new(tzdata).unwrap();
+        let mut identifiers = BTreeMap::default();
+        let mut zones_set = BTreeSet::default();
+        for zoneset_id in provider.data.zonesets.keys() {
+            let _ = zones_set.insert(zoneset_id.clone());
+            identifiers.insert(zoneset_id.clone(), zoneset_id.clone());
+        }
+        for (link, zone) in provider.data.links.iter() {
+            let _ = zones_set.insert(zone.clone());
+            identifiers.insert(link.clone(), zone.clone());
+        }
+
+        let zones: Vec<String> = zones_set.iter().cloned().collect();
+
+        let identier_map: BTreeMap<Vec<u8>, usize> = identifiers
+            .iter()
+            .map(|(id, zoneid)| (id.as_bytes().to_vec(), zones.binary_search(zoneid).unwrap()))
+            .collect();
+
+        let tzifs: Vec<ZeroTzif<'_>> = zones
+            .iter()
+            .map(|id| ZeroTzif::build(&provider.data, id))
+            .collect();
+
+        let tzifs_zerovec: VarZeroVec<'static, ZeroTzifULE> = tzifs.as_slice().into();
+
+        let ids = ZeroTrieSimpleAscii::try_from(&identier_map)
+            .map_err(ZoneInfoDataError::Build)?
+            .convert_store();
+
+        Ok(ZeroZoneInfo {
+            ids,
+            tzifs: tzifs_zerovec,
+        })
+    }
+}
+
+impl ZeroTzif<'_> {
+    fn build(data: &Table, id: &str) -> Self {
+        todo!()
+    }
+}

provider/src/tzif.rs

@@ -0,0 +1,19 @@
+[package]
+name = "zoneinfo"
+edition.workspace = true
+version.workspace = true
+rust-version.workspace = true
+authors.workspace = true
+license.workspace = true
+repository.workspace = true
+readme.workspace = true
+exclude.workspace = true
+
+[dependencies]
+hashbrown = "0.15.2"
+
+[dev-dependencies]
+tzif = { workspace = true }
+
+[features]
+std = []