znippy_common/plugins/native/
conda_native.rs

1//! Native Conda plugin.
2//!
3//! Handles the classic **`.tar.bz2`** conda package: a bzip2-compressed tar whose
4//! `info/index.json` carries the **authoritative** `name`, `version`, `build`, and
5//! `subdir`. These are NOT reliably recoverable from the filename: a conda file is
6//! named `{name}-{version}-{build}.tar.bz2`, and `build` itself contains hyphens
7//! and digits (`py311h1234567_0`), so a naive split mis-attributes the version.
8//! Parsing `info/index.json` (under the `host-decompressors` feature, via `lbzip2`
9//! for the bunzip2 + the `tar` crate for the untar) is what lets the read-side
10//! [`CondaView`](crate::views::CondaView) resolve conda coords correctly.
11//!
12//! The newer **`.conda`** format (a zip whose members are zstd-compressed tars
13//! `info-*.tar.zst`) is recognized by extension, but its `info/index.json` is NOT
14//! parsed here: extracting it needs a zip reader plus a zstd-tar path that the
15//! available `host-decompressors` deps (`lgz`/`lbzip2`/`tar`) do not provide. For a
16//! `.conda` file the plugin therefore falls back to the filename split and logs a
17//! one-line note. Real `.conda` index parsing is a documented follow-up — we do
18//! NOT fake the columns.
19//!
20//! When `host-decompressors` is off (or the package can't be parsed), the plugin
21//! falls back to a best-effort `(name, version)` split from the filename — never
22//! panics, always returns a row.
23
24use crate::plugin::{file_name, ArchiveTypePlugin, Col, ExtensionRow, ExtensionValue, Claim, HandlerSpec};
25use std::collections::HashMap;
26
27/// Native conda plugin. `name`/`version`/`build`/`subdir` come from
28/// `info/index.json` for `.tar.bz2` packages when parseable, else from the
29/// filename.
30pub struct CondaPlugin;
31
32// Only ever built by `parse_tar_bz2` / `parse_index`, both of which already
33// carry `#[cfg(feature = "host-decompressors")]`. Without the same gate here
34// the default build warned ``struct `CondaIndex` is never constructed`` — a
35// TRUE statement about that build, and the warning every other item in this
36// file already avoids by carrying the gate. Gated rather than
37// `#[allow(dead_code)]`-ed: the type genuinely is absent from a build without
38// the decompressors, and an `allow` would also have swallowed the day it goes
39// unused in a build WITH them, which is the only version of this warning worth
40// hearing.
41#[cfg(feature = "host-decompressors")]
42/// Authoritative conda coords parsed from `info/index.json`.
43struct CondaIndex {
44    name: String,
45    version: String,
46    build: String,
47    subdir: Option<String>,
48}
49
50impl CondaPlugin {
51    /// Best-effort `(name, version)` from a conda filename like
52    /// `{name}-{version}-{build}.tar.bz2` / `.conda`: strip the extension, then
53    /// split off the trailing `-{build}` and the `-{version}` before it. The build
54    /// string is NOT a distinct field here — that's exactly why `info/index.json`
55    /// is preferred.
56    fn parse_filename(path: &str) -> (String, Option<String>) {
57        let filename = file_name(path);
58        let stem = filename
59            .strip_suffix(".tar.bz2")
60            .or_else(|| filename.strip_suffix(".conda"))
61            .unwrap_or(filename);
62        // `name-version-build`: peel the last two `-`-separated segments as
63        // build + version, leaving the (possibly hyphenated) name.
64        let mut parts: Vec<&str> = stem.rsplitn(3, '-').collect();
65        // rsplitn yields [build, version, name] reversed.
66        if parts.len() == 3 {
67            let name = parts.pop().unwrap();
68            let version = parts.pop().unwrap();
69            (name.to_string(), Some(version.to_string()))
70        } else {
71            (stem.to_string(), None)
72        }
73    }
74
75    /// Parse `info/index.json` out of a `.tar.bz2` conda package: bunzip2 the
76    /// outer layer, untar, find `info/index.json`, read the fields. `None` on any
77    /// failure (caller falls back to the filename). Only compiled when host
78    /// decompressors are available.
79    #[cfg(feature = "host-decompressors")]
80    fn parse_tar_bz2(data: &[u8]) -> Option<CondaIndex> {
81        use crate::plugins::npm_native::MAX_INGEST_DECOMPRESS;
82        use std::io::Read;
83        // 1. bunzip2 the outer layer — capped so a bzip2 compression bomb errors
84        //    here (→ filename fallback) instead of expanding to GB and OOM-aborting.
85        let tar_bytes = lbzip2::stream::decompress_capped(data, MAX_INGEST_DECOMPRESS).ok()?;
86        // 2. untar and find info/index.json.
87        let mut archive = tar::Archive::new(&tar_bytes[..]);
88        let mut index_json: Option<Vec<u8>> = None;
89        for entry in archive.entries().ok()? {
90            let mut entry = entry.ok()?;
91            let path = entry.path().ok()?.to_string_lossy().to_string();
92            if path == "info/index.json" || path.ends_with("/info/index.json") {
93                // Cap the per-tar-entry read so a huge index member can't OOM.
94                let mut buf = Vec::new();
95                let read = entry
96                    .by_ref()
97                    .take(MAX_INGEST_DECOMPRESS as u64)
98                    .read_to_end(&mut buf)
99                    .ok()?;
100                if read >= MAX_INGEST_DECOMPRESS {
101                    return None;
102                }
103                index_json = Some(buf);
104                break;
105            }
106        }
107        let index_json = index_json?;
108        let v: serde_json::Value = serde_json::from_slice(&index_json).ok()?;
109        let name = v.get("name")?.as_str()?.to_string();
110        let version = v.get("version")?.as_str()?.to_string();
111        let build = v
112            .get("build")
113            .and_then(|b| b.as_str())
114            .map(|s| s.to_string())
115            .unwrap_or_default();
116        let subdir = v.get("subdir").and_then(|s| s.as_str()).map(|s| s.to_string());
117        if name.is_empty() || version.is_empty() {
118            return None;
119        }
120        Some(CondaIndex { name, version, build, subdir })
121    }
122
123    /// Parse the authoritative index for a conda package, if possible. `.tar.bz2`
124    /// is fully supported; `.conda` is recognized but not yet parsed (documented
125    /// follow-up) — returns `None` so the caller uses the filename.
126    #[cfg(feature = "host-decompressors")]
127    fn parse_index(path: &str, data: &[u8]) -> Option<CondaIndex> {
128        let filename = file_name(path);
129        if filename.ends_with(".tar.bz2") {
130            return Self::parse_tar_bz2(data);
131        }
132        if filename.ends_with(".conda") {
133            log::info!(
134                "conda: .conda (zip+zstd) index parsing is a follow-up; \
135                 falling back to filename coords for {filename}"
136            );
137        }
138        None
139    }
140
141    /// Resolve `(name, version)`: authoritative `info/index.json` first
142    /// (`.tar.bz2`), filename fallback otherwise.
143    fn resolve_coords(path: &str, _data: &[u8]) -> (String, Option<String>) {
144        #[cfg(feature = "host-decompressors")]
145        if let Some(idx) = Self::parse_index(path, _data) {
146            return (idx.name, Some(idx.version));
147        }
148        Self::parse_filename(path)
149    }
150
151    /// Resolve the conda build string (empty when unknown / filename mode).
152    fn resolve_build(_path: &str, _data: &[u8]) -> String {
153        #[cfg(feature = "host-decompressors")]
154        if let Some(idx) = Self::parse_index(_path, _data) {
155            return idx.build;
156        }
157        String::new()
158    }
159
160    /// Resolve the conda subdir (platform, e.g. `linux-64`). `None` when unknown.
161    fn resolve_subdir(_path: &str, _data: &[u8]) -> Option<String> {
162        #[cfg(feature = "host-decompressors")]
163        if let Some(idx) = Self::parse_index(_path, _data) {
164            return idx.subdir;
165        }
166        None
167    }
168}
169
170/// The conda handler, stated once: the coords the read-side
171/// [`CondaView`](crate::views::CondaView) resolves on, plus `build` and `subdir`.
172pub static CONDA_SPEC: HandlerSpec = HandlerSpec {
173    name: "conda",
174    aliases: &["anaconda", "mamba"],
175    type_id: 14,
176    ecosystem: "Conda packages (Anaconda / conda-forge)",
177    extensions: &[".conda", ".tar.bz2"],
178    claim: Claim::Extensions,
179    description: "Conda packages — authoritative name/version/build/subdir from info/index.json (.tar.bz2)",
180    fields: &[("name", Col::Utf8), ("version", Col::Utf8), ("build", Col::Utf8), ("subdir", Col::Utf8)],
181    commands: &[],
182};
183
184impl ArchiveTypePlugin for CondaPlugin {
185    fn spec(&self) -> &HandlerSpec {
186        &CONDA_SPEC
187    }
188
189    fn extract_metadata(&self, path: &str, data: &[u8]) -> Option<ExtensionRow> {
190        let (name, version) = Self::resolve_coords(path, data);
191        let build = Self::resolve_build(path, data);
192        let subdir = Self::resolve_subdir(path, data);
193        let mut fields = HashMap::new();
194        fields.insert("name".into(), ExtensionValue::Str(name));
195        fields.insert("version".into(), ExtensionValue::OptStr(version));
196        fields.insert("build".into(), ExtensionValue::Str(build));
197        fields.insert("subdir".into(), ExtensionValue::OptStr(subdir));
198        Some(ExtensionRow { fields })
199    }
200}
201
202#[cfg(test)]
203mod tests {
204    use super::*;
205
206    #[test]
207    fn matches_conda_extensions() {
208        let p = CondaPlugin;
209        assert!(p.matches_path("linux-64/numpy-1.26.0-py311h1234567_0.tar.bz2"));
210        assert!(p.matches_path("linux-64/numpy-1.26.0-py311h1234567_0.conda"));
211        assert!(!p.matches_path("foo.tgz"));
212    }
213
214    #[test]
215    fn filename_fallback_splits_name_version() {
216        let (n, v) = CondaPlugin::parse_filename("linux-64/numpy-1.26.0-py311h1234567_0.tar.bz2");
217        assert_eq!(n, "numpy");
218        assert_eq!(v.as_deref(), Some("1.26.0"));
219    }
220
221    #[test]
222    fn schema_has_name_version_build_subdir() {
223        let f = CondaPlugin.schema_fields();
224        assert_eq!(f.len(), 4);
225        assert_eq!(f[0].name(), "name");
226        assert_eq!(f[1].name(), "version");
227        assert_eq!(f[2].name(), "build");
228        assert_eq!(f[3].name(), "subdir");
229    }
230}
znippy_common/plugins/native/conda_native.rs

znippy_common/plugins/native/
conda_native.rs
