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::{ArchiveTypePlugin, ExtensionRow, ExtensionValue, HandlerCommand, HandlerMeta};
25use arrow::datatypes::{DataType, Field};
26use std::collections::HashMap;
27
28/// Native conda plugin. `name`/`version`/`build`/`subdir` come from
29/// `info/index.json` for `.tar.bz2` packages when parseable, else from the
30/// filename.
31pub struct CondaPlugin;
32
33/// Authoritative conda coords parsed from `info/index.json`.
34struct CondaIndex {
35    name: String,
36    version: String,
37    build: String,
38    subdir: Option<String>,
39}
40
41impl CondaPlugin {
42    /// Best-effort `(name, version)` from a conda filename like
43    /// `{name}-{version}-{build}.tar.bz2` / `.conda`: strip the extension, then
44    /// split off the trailing `-{build}` and the `-{version}` before it. The build
45    /// string is NOT a distinct field here — that's exactly why `info/index.json`
46    /// is preferred.
47    fn parse_filename(path: &str) -> (String, Option<String>) {
48        let filename = path.rsplit('/').next().unwrap_or(path);
49        let stem = filename
50            .strip_suffix(".tar.bz2")
51            .or_else(|| filename.strip_suffix(".conda"))
52            .unwrap_or(filename);
53        // `name-version-build`: peel the last two `-`-separated segments as
54        // build + version, leaving the (possibly hyphenated) name.
55        let mut parts: Vec<&str> = stem.rsplitn(3, '-').collect();
56        // rsplitn yields [build, version, name] reversed.
57        if parts.len() == 3 {
58            let name = parts.pop().unwrap();
59            let version = parts.pop().unwrap();
60            (name.to_string(), Some(version.to_string()))
61        } else {
62            (stem.to_string(), None)
63        }
64    }
65
66    /// Parse `info/index.json` out of a `.tar.bz2` conda package: bunzip2 the
67    /// outer layer, untar, find `info/index.json`, read the fields. `None` on any
68    /// failure (caller falls back to the filename). Only compiled when host
69    /// decompressors are available.
70    #[cfg(feature = "host-decompressors")]
71    fn parse_tar_bz2(data: &[u8]) -> Option<CondaIndex> {
72        use crate::plugins::npm_native::MAX_INGEST_DECOMPRESS;
73        use std::io::Read;
74        // 1. bunzip2 the outer layer — capped so a bzip2 compression bomb errors
75        //    here (→ filename fallback) instead of expanding to GB and OOM-aborting.
76        let tar_bytes = lbzip2::stream::decompress_capped(data, MAX_INGEST_DECOMPRESS).ok()?;
77        // 2. untar and find info/index.json.
78        let mut archive = tar::Archive::new(&tar_bytes[..]);
79        let mut index_json: Option<Vec<u8>> = None;
80        for entry in archive.entries().ok()? {
81            let mut entry = entry.ok()?;
82            let path = entry.path().ok()?.to_string_lossy().to_string();
83            if path == "info/index.json" || path.ends_with("/info/index.json") {
84                // Cap the per-tar-entry read so a huge index member can't OOM.
85                let mut buf = Vec::new();
86                let read = entry
87                    .by_ref()
88                    .take(MAX_INGEST_DECOMPRESS as u64)
89                    .read_to_end(&mut buf)
90                    .ok()?;
91                if read >= MAX_INGEST_DECOMPRESS {
92                    return None;
93                }
94                index_json = Some(buf);
95                break;
96            }
97        }
98        let index_json = index_json?;
99        let v: serde_json::Value = serde_json::from_slice(&index_json).ok()?;
100        let name = v.get("name")?.as_str()?.to_string();
101        let version = v.get("version")?.as_str()?.to_string();
102        let build = v
103            .get("build")
104            .and_then(|b| b.as_str())
105            .map(|s| s.to_string())
106            .unwrap_or_default();
107        let subdir = v.get("subdir").and_then(|s| s.as_str()).map(|s| s.to_string());
108        if name.is_empty() || version.is_empty() {
109            return None;
110        }
111        Some(CondaIndex { name, version, build, subdir })
112    }
113
114    /// Parse the authoritative index for a conda package, if possible. `.tar.bz2`
115    /// is fully supported; `.conda` is recognized but not yet parsed (documented
116    /// follow-up) — returns `None` so the caller uses the filename.
117    #[cfg(feature = "host-decompressors")]
118    fn parse_index(path: &str, data: &[u8]) -> Option<CondaIndex> {
119        let filename = path.rsplit('/').next().unwrap_or(path);
120        if filename.ends_with(".tar.bz2") {
121            return Self::parse_tar_bz2(data);
122        }
123        if filename.ends_with(".conda") {
124            log::info!(
125                "conda: .conda (zip+zstd) index parsing is a follow-up; \
126                 falling back to filename coords for {filename}"
127            );
128        }
129        None
130    }
131
132    /// Resolve `(name, version)`: authoritative `info/index.json` first
133    /// (`.tar.bz2`), filename fallback otherwise.
134    fn resolve_coords(path: &str, _data: &[u8]) -> (String, Option<String>) {
135        #[cfg(feature = "host-decompressors")]
136        if let Some(idx) = Self::parse_index(path, _data) {
137            return (idx.name, Some(idx.version));
138        }
139        Self::parse_filename(path)
140    }
141
142    /// Resolve the conda build string (empty when unknown / filename mode).
143    fn resolve_build(_path: &str, _data: &[u8]) -> String {
144        #[cfg(feature = "host-decompressors")]
145        if let Some(idx) = Self::parse_index(_path, _data) {
146            return idx.build;
147        }
148        String::new()
149    }
150
151    /// Resolve the conda subdir (platform, e.g. `linux-64`). `None` when unknown.
152    fn resolve_subdir(_path: &str, _data: &[u8]) -> Option<String> {
153        #[cfg(feature = "host-decompressors")]
154        if let Some(idx) = Self::parse_index(_path, _data) {
155            return idx.subdir;
156        }
157        None
158    }
159}
160
161impl ArchiveTypePlugin for CondaPlugin {
162    fn name(&self) -> &str {
163        "conda"
164    }
165
166    fn type_id(&self) -> i8 {
167        14
168    }
169
170    fn meta(&self) -> HandlerMeta {
171        HandlerMeta {
172            name: "conda".into(),
173            aliases: vec!["anaconda".into(), "mamba".into()],
174            type_id: 14,
175            ecosystem: "Conda packages (Anaconda / conda-forge)".into(),
176            extensions: vec![".conda".into(), ".tar.bz2".into()],
177            description:
178                "Conda packages — authoritative name/version/build/subdir from info/index.json (.tar.bz2)"
179                    .into(),
180            commands: vec![HandlerCommand::new(
181                "coords",
182                "Print conda package name + version (info/index.json if readable, else filename)",
183            )],
184        }
185    }
186
187    fn run_command(&self, cmd: &str, args: &[String]) -> anyhow::Result<()> {
188        match cmd {
189            "coords" => {
190                let path = args
191                    .first()
192                    .ok_or_else(|| anyhow::anyhow!("usage: conda coords <file.tar.bz2|.conda>"))?;
193                let (name, version) = Self::parse_filename(path);
194                match version {
195                    Some(v) => println!("{} {}", name, v),
196                    None => println!("{}", name),
197                }
198                Ok(())
199            }
200            other => anyhow::bail!("conda: unknown subcommand '{}'", other),
201        }
202    }
203
204    fn matches_path(&self, path: &str) -> bool {
205        path.ends_with(".tar.bz2") || path.ends_with(".conda")
206    }
207
208    /// Columns this handler contributes — the conda coords the read-side
209    /// [`CondaView`](crate::views::CondaView) resolves on, plus the authoritative
210    /// `build` and `subdir` (platform).
211    fn schema_fields(&self) -> Vec<Field> {
212        vec![
213            Field::new("name", DataType::Utf8, true),
214            Field::new("version", DataType::Utf8, true),
215            Field::new("build", DataType::Utf8, true),
216            Field::new("subdir", DataType::Utf8, true),
217        ]
218    }
219
220    fn extract_metadata(&self, path: &str, data: &[u8]) -> Option<ExtensionRow> {
221        let (name, version) = Self::resolve_coords(path, data);
222        let build = Self::resolve_build(path, data);
223        let subdir = Self::resolve_subdir(path, data);
224        let mut fields = HashMap::new();
225        fields.insert("name".into(), ExtensionValue::Str(name));
226        fields.insert("version".into(), ExtensionValue::OptStr(version));
227        fields.insert("build".into(), ExtensionValue::Str(build));
228        fields.insert("subdir".into(), ExtensionValue::OptStr(subdir));
229        Some(ExtensionRow { fields })
230    }
231}
232
233#[cfg(test)]
234mod tests {
235    use super::*;
236
237    #[test]
238    fn matches_conda_extensions() {
239        let p = CondaPlugin;
240        assert!(p.matches_path("linux-64/numpy-1.26.0-py311h1234567_0.tar.bz2"));
241        assert!(p.matches_path("linux-64/numpy-1.26.0-py311h1234567_0.conda"));
242        assert!(!p.matches_path("foo.tgz"));
243    }
244
245    #[test]
246    fn filename_fallback_splits_name_version() {
247        let (n, v) = CondaPlugin::parse_filename("linux-64/numpy-1.26.0-py311h1234567_0.tar.bz2");
248        assert_eq!(n, "numpy");
249        assert_eq!(v.as_deref(), Some("1.26.0"));
250    }
251
252    #[test]
253    fn schema_has_name_version_build_subdir() {
254        let f = CondaPlugin.schema_fields();
255        assert_eq!(f.len(), 4);
256        assert_eq!(f[0].name(), "name");
257        assert_eq!(f[1].name(), "version");
258        assert_eq!(f[2].name(), "build");
259        assert_eq!(f[3].name(), "subdir");
260    }
261}
znippy_common/plugins/native/conda_native.rs

znippy_common/plugins/native/
conda_native.rs
