pkh/src/test/expand.rs

//! DEP-8 `Depends:` substitution: `@`, `@builddeps@` and `@recommends@`
//! expanded into concrete package specs for `apt-get install`.
//!
//! The built binaries themselves are installed separately (from the
//! staged `.deb` files, so tests exercise exactly what pkh built);
//! expansion substitutes bare binary *names*, which apt then resolves
//! against the installed binaries first and the archive second.

use crate::debian::control::ControlInfo;
use crate::debian::deps::{Deps, ParseOpts};

/// Expand a test stanza's `Depends:` value into apt package specs.
///
/// `raw` is the value as written (`None` = absent, the DEP-8 default
/// `@`). The substitutions, per DEP-8:
///
/// - `@` in an or-clause replicates the clause once per binary package
///   built from the source, with the name substituted (other
///   alternatives and version constraints are kept);
/// - `@builddeps@` becomes the source's Build-Depends,
///   Build-Depends-Indep and Build-Depends-Arch plus `build-essential`;
/// - `@recommends@` becomes the Recommends of every binary package,
///   with `${...}` substvar references dropped (autopkgtest's
///   "variables stripped") but versions kept.
///
/// Returns one apt spec per comma-separated clause, in order; an empty
/// result means nothing to install.
pub fn expand_depends(
    raw: Option<&str>,
    control: &ControlInfo,
    binaries: &[String],
    host_arch: &str,
) -> Result<Vec<String>, String> {
    let raw = raw.unwrap_or("@");
    let mut specs = Vec::new();
    for token in split_top_level_commas(raw) {
        match token.as_str() {
            "@builddeps@" => {
                for field in ["Build-Depends", "Build-Depends-Indep", "Build-Depends-Arch"] {
                    if let Some(value) = non_empty(control.source.get(field)) {
                        for clause in split_top_level_commas(value) {
                            if !clause.is_empty() {
                                specs.push(clause);
                            }
                        }
                    }
                }
                specs.push("build-essential".to_string());
            }
            "@recommends@" => {
                for paragraph in &control.binaries {
                    if let Some(value) = non_empty(paragraph.get("Recommends")) {
                        specs.extend(name_only_clauses(value, host_arch)?);
                    }
                }
            }
            token if token.split_whitespace().any(|word| word == "@") => {
                // Textual substitution: DEP-8 @ is a word-level
                // placeholder the deps grammar does not know. The clause
                // replicates once per binary with the name substituted,
                // constraints and alternatives kept verbatim.
                for binary in binaries {
                    specs.push(
                        token
                            .split_whitespace()
                            .map(|word| {
                                if word == "@" {
                                    binary.clone()
                                } else {
                                    word.to_string()
                                }
                            })
                            .collect::<Vec<_>>()
                            .join(" "),
                    );
                }
            }
            token => specs.push(token.to_string()),
        }
    }
    Ok(specs)
}

/// Split on commas that sit outside any parentheses: or-clauses carry
/// their own commas inside `(...)` constraints, top-level commas separate
/// dependencies.
fn split_top_level_commas(input: &str) -> Vec<String> {
    let mut parts = Vec::new();
    let mut depth = 0usize;
    let mut current = String::new();
    for character in input.chars() {
        match character {
            '(' => {
                depth += 1;
                current.push(character);
            }
            ')' => {
                depth = depth.saturating_sub(1);
                current.push(character);
            }
            ',' if depth == 0 => {
                parts.push(current.trim().to_string());
                current.clear();
            }
            _ => current.push(character),
        }
    }
    parts.push(current.trim().to_string());
    parts
}

/// Render each or-clause of a dependency list, dropping alternatives
/// that are `${...}` substvar references (autopkgtest's "variables
/// stripped") and keeping everything else verbatim.
fn name_only_clauses(input: &str, host_arch: &str) -> Result<Vec<String>, String> {
    let opts = ParseOpts::build_deps(host_arch.to_string(), Vec::new());
    let deps = Deps::parse(input, &opts).map_err(|e| e.to_string())?;
    let mut clauses = Vec::new();
    for clause in deps.clauses() {
        let kept: Vec<String> = clause
            .iter()
            .filter(|relation| !relation.package.contains('$'))
            .map(|relation| relation.output())
            .collect();
        if !kept.is_empty() {
            clauses.push(kept.join(" | "));
        }
    }
    Ok(clauses)
}

/// A field value with its surrounding whitespace trimmed; `None` when the
/// field is absent or whitespace-only.
fn non_empty(value: Option<&str>) -> Option<&str> {
    value.map(str::trim).filter(|value| !value.is_empty())
}

/// The binary package names declared by a control file, in stanza order.
pub fn binary_names(control: &ControlInfo) -> Vec<String> {
    control
        .binaries
        .iter()
        .filter_map(|paragraph| paragraph.get("Package"))
        .map(|name| name.trim().to_string())
        .collect()
}

#[cfg(test)]
mod tests {
    use super::*;
    use std::str::FromStr;

    fn control() -> ControlInfo {
        ControlInfo::from_str(
            "Source: hello\n\
             Build-Depends: debhelper-compat (= 13),\n meson\n\
             Build-Depends-Indep: dh-python\n\
             \n\
             Package: hello\n\
             Architecture: any\n\
             Recommends: hello-doc, pulseaudio (>= 15)\n\
             \n\
             Package: hello-doc\n\
             Architecture: all\n\
             Recommends: hello\n",
        )
        .unwrap()
    }

    fn binaries() -> Vec<String> {
        vec!["hello".to_string(), "hello-doc".to_string()]
    }

    /// The default (absent) Depends is `@`: one spec per binary.
    #[test]
    fn default_is_at() {
        let specs = expand_depends(None, &control(), &binaries(), "amd64").unwrap();
        assert_eq!(specs, ["hello", "hello-doc"]);
    }

    /// An or-clause containing @ replicates once per binary, keeping the
    /// other alternatives and version constraints.
    #[test]
    fn or_clauses_replicate_per_binary() {
        let specs = expand_depends(
            Some("@ (>= 1.0) | fallback, jq"),
            &control(),
            &binaries(),
            "amd64",
        )
        .unwrap();
        assert_eq!(
            specs,
            [
                "hello (>= 1.0) | fallback",
                "hello-doc (>= 1.0) | fallback",
                "jq"
            ]
        );
    }

    /// @builddeps@ pulls the three Build-* fields plus build-essential.
    #[test]
    fn builddeps_expansion() {
        let specs =
            expand_depends(Some("@builddeps@, @"), &control(), &binaries(), "amd64").unwrap();
        assert_eq!(
            specs,
            [
                "debhelper-compat (= 13)",
                "meson",
                "dh-python",
                "build-essential",
                "hello",
                "hello-doc"
            ]
        );
    }

    /// @recommends@ collects every binary's Recommends, stripped to names.
    #[test]
    fn recommends_expansion() {
        let specs = expand_depends(Some("@recommends@"), &control(), &binaries(), "amd64").unwrap();
        assert_eq!(specs, ["hello-doc", "pulseaudio (>= 15)", "hello"]);
    }

    /// Plain dependencies pass through untouched; a token mentioning @ in
    /// a version constraint is still a substitution.
    #[test]
    fn plain_tokens_pass_through() {
        let specs = expand_depends(
            Some("ca-certificates, python3:any"),
            &control(),
            &binaries(),
            "amd64",
        )
        .unwrap();
        assert_eq!(specs, ["ca-certificates", "python3:any"]);
    }

    /// Commas inside version constraints do not split clauses.
    #[test]
    fn parenthesised_commas_stay_together() {
        let specs =
            expand_depends(Some("foo (>= 1) , bar"), &control(), &binaries(), "amd64").unwrap();
        assert_eq!(specs, ["foo (>= 1)", "bar"]);
    }
}