feat: use globs in includes field

CHANGELOG.md

@@ -9,6 +9,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Added
 - Print that no updates are available in `outdated` command by @daimond113
 - Support negated globs in `workspace_members` field by @daimond113
+- Make `includes` use glob patterns by @daimond113
 
 ## [0.5.0-rc.12] - 2024-11-22
 ### Added

Cargo.lock

@@ -2376,10 +2376,18 @@ dependencies = [
 ]
 
 [[package]]
-name = "glob"
+name = "globset"
-version = "0.3.1"
+version = "0.4.15"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d2fabcfbdc87f4758337ca535fb41a6d701b65693ce38287d856d1674551ec9b"
+checksum = "15f1ce686646e7f1e19bf7d5533fe443a45dbfb990e00629110797578b42fb19"
+dependencies = [
+ "aho-corasick",
+ "bstr",
+ "log",
+ "regex-automata",
+ "regex-syntax",
+ "serde",
+]
 
 [[package]]
 name = "governor"
@@ -3654,7 +3662,7 @@ dependencies = [
  "futures",
  "git2",
  "gix",
- "glob",
+ "globset",
  "indicatif",
  "indicatif-log-bridge",
  "inquire",

Cargo.toml

@@ -65,7 +65,7 @@ url = { version = "2.5.3", features = ["serde"] }
 chrono = { version = "0.4.38", features = ["serde"] }
 sha2 = "0.10.8"
 tempfile = "3.14.0"
-glob = "0.3.1"
+globset = { version = "0.4.15", features = ["serde1"] }
 fs-err = { version = "3.0.0", features = ["tokio"] }
 
 # TODO: remove this when gitoxide adds support for: committing, pushing, adding

src/cli/commands/publish.rs

@@ -6,6 +6,7 @@ use colored::Colorize;
 use fs_err::tokio as fs;
 use pesde::{
     manifest::{target::Target, DependencyType},
+    matching_globs,
     scripts::ScriptName,
     source::{
         pesde::{specifier::PesdeDependencySpecifier, PesdePackageSource},
@@ -21,7 +22,7 @@ use pesde::{
 };
 use reqwest::{header::AUTHORIZATION, StatusCode};
 use semver::VersionReq;
-use std::{collections::HashSet, path::Component};
+use std::{collections::HashSet, path::PathBuf};
 use tempfile::Builder;
 use tokio::io::{AsyncSeekExt, AsyncWriteExt};
 
@@ -41,7 +42,12 @@ pub struct PublishCommand {
 }
 
 impl PublishCommand {
-    async fn run_impl(self, project: &Project, reqwest: reqwest::Client) -> anyhow::Result<()> {
+    async fn run_impl(
+        self,
+        project: &Project,
+        reqwest: reqwest::Client,
+        is_root: bool,
+    ) -> anyhow::Result<()> {
         let mut manifest = project
             .deser_manifest()
             .await
@@ -55,7 +61,9 @@ impl PublishCommand {
         );
 
         if manifest.private {
+            if !is_root {
                 println!("{}", "package is private, cannot publish".red().bold());
+            }
 
             return Ok(());
         }
@@ -97,7 +105,6 @@ impl PublishCommand {
             async_compression::tokio::write::GzipEncoder::with_quality(vec![], Level::Best),
         );
 
-        let mut display_includes: Vec<String> = vec![MANIFEST_FILE_NAME.to_string()];
         let mut display_build_files: Vec<String> = vec![];
 
         let (lib_path, bin_path, target_kind) = (
@@ -112,64 +119,75 @@ impl PublishCommand {
             _ => None,
         };
 
-        if manifest.includes.insert(MANIFEST_FILE_NAME.to_string()) {
+        let mut paths = matching_globs(project.package_dir(), manifest.includes.clone(), true)
+            .await
+            .context("failed to get included files")?;
+
+        if paths.insert(PathBuf::from(MANIFEST_FILE_NAME)) {
             println!(
-                "{}: {MANIFEST_FILE_NAME} was not in includes, adding it",
+                "{}: {MANIFEST_FILE_NAME} was not included, adding it",
                 "warn".yellow().bold()
             );
         }
 
-        if manifest.includes.remove(".git") {
+        if paths.iter().any(|p| p.starts_with(".git")) {
-            println!(
+            anyhow::bail!("git directory was included, please remove it");
-                "{}: .git was in includes, removing it",
-                "warn".yellow().bold()
-            );
         }
 
-        if !manifest.includes.iter().any(|f| {
+        if !paths.iter().any(|f| {
             matches!(
-                f.to_lowercase().as_str(),
+                f.to_str().unwrap().to_lowercase().as_str(),
                 "readme" | "readme.md" | "readme.txt"
             )
         }) {
             println!(
-                "{}: no README file in includes, consider adding one",
+                "{}: no README file included, consider adding one",
                 "warn".yellow().bold()
             );
         }
 
-        if !manifest.includes.iter().any(|f| f == "docs") {
+        if !paths.iter().any(|p| p.starts_with("docs")) {
             println!(
-                "{}: no docs directory in includes, consider adding one",
+                "{}: docs directory not included, consider adding one",
                 "warn".yellow().bold()
             );
         }
 
-        if manifest.includes.remove("default.project.json") {
+        for path in &paths {
-            println!(
+            if path
-                "{}: default.project.json was in includes, this should be generated by the {} script upon dependants installation",
+                .file_name()
-                "warn".yellow().bold(),
+                .is_some_and(|n| n == "default.project.json")
+            {
+                anyhow::bail!(
+                    "default.project.json was included at `{}`, this should be generated by the {} script upon dependants installation",
+                    path.display(),
                     ScriptName::RobloxSyncConfigGenerator
                 );
             }
+        }
 
         for ignored_path in IGNORED_FILES.iter().chain(IGNORED_DIRS.iter()) {
-            if manifest.includes.remove(*ignored_path) {
+            if paths.iter().any(|p| {
-                println!(
+                p.components()
-                    r#"{}: {ignored_path} was in includes, removing it.
+                    .any(|ct| ct == std::path::Component::Normal(ignored_path.as_ref()))
-{}: if this was a toolchain manager's manifest file, do not include it due to it possibly messing with user scripts
+            }) {
-{}: otherwise, the file was deemed unnecessary, if you don't understand why, please contact the maintainers"#,
+                anyhow::bail!(
-                    "warn".yellow().bold(),
+                    r#"forbidden file {ignored_path} was included.
-                    "info".blue().bold(),
+info: if this was a toolchain manager's manifest file, do not include it due to it possibly messing with user scripts
-                    "info".blue().bold()
+info: otherwise, the file was deemed unnecessary, if you don't understand why, please contact the maintainers"#,
                 );
             }
         }
 
         for (name, path) in [("lib path", lib_path), ("bin path", bin_path)] {
-            let Some(export_path) = path else { continue };
+            let Some(relative_export_path) = path else {
+                continue;
+            };
 
-            let export_path = export_path.to_path(project.package_dir());
+            let export_path = relative_export_path
+                .to_path(project.package_dir())
+                .canonicalize()
+                .context(format!("failed to canonicalize {name}"))?;
             if !export_path.exists() {
                 anyhow::bail!("{name} points to non-existent file");
             }
@@ -191,71 +209,33 @@ impl PublishCommand {
                 anyhow::bail!("{name} is not a valid Luau file: {err}");
             }
 
-            let first_part = export_path
+            let first_part = relative_export_path
-                .strip_prefix(project.package_dir())
-                .context(format!("{name} not within project directory"))?
                 .components()
                 .next()
                 .context(format!("{name} must contain at least one part"))?;
 
             let first_part = match first_part {
-                Component::Normal(part) => part,
+                relative_path::Component::Normal(part) => part,
                 _ => anyhow::bail!("{name} must be within project directory"),
             };
 
-            let first_part_str = first_part.to_string_lossy();
+            if paths.insert(PathBuf::from(relative_export_path.as_str())) {
-
-            if manifest.includes.insert(first_part_str.to_string()) {
                 println!(
-                    "{}: {name} was not in includes, adding {first_part_str}",
+                    "{}: {name} was not included, adding {relative_export_path}",
                     "warn".yellow().bold()
                 );
             }
 
             if roblox_target.as_mut().map_or(false, |build_files| {
-                build_files.insert(first_part_str.to_string())
+                build_files.insert(first_part.to_string())
             }) {
                 println!(
-                    "{}: {name} was not in build files, adding {first_part_str}",
+                    "{}: {name} was not in build files, adding {first_part}",
                     "warn".yellow().bold()
                 );
             }
         }
 
-        for included_name in &manifest.includes {
-            let included_path = project.package_dir().join(included_name);
-
-            if !included_path.exists() {
-                anyhow::bail!("included file {included_name} does not exist");
-            }
-
-            // it's already included, and guaranteed to be a file
-            if included_name.eq_ignore_ascii_case(MANIFEST_FILE_NAME) {
-                continue;
-            }
-
-            if included_path.is_file() {
-                display_includes.push(included_name.clone());
-
-                archive
-                    .append_file(
-                        included_name,
-                        fs::File::open(&included_path)
-                            .await
-                            .context(format!("failed to read {included_name}"))?
-                            .file_mut(),
-                    )
-                    .await?;
-            } else {
-                display_includes.push(format!("{included_name}/*"));
-
-                archive
-                    .append_dir_all(included_name, &included_path)
-                    .await
-                    .context(format!("failed to include directory {included_name}"))?;
-            }
-        }
-
         if let Some(build_files) = &roblox_target {
             for build_file in build_files.iter() {
                 if build_file.eq_ignore_ascii_case(MANIFEST_FILE_NAME) {
@@ -273,8 +253,8 @@ impl PublishCommand {
                     anyhow::bail!("build file {build_file} does not exist");
                 }
 
-                if !manifest.includes.contains(build_file) {
+                if !paths.iter().any(|p| p.starts_with(build_file)) {
-                    anyhow::bail!("build file {build_file} is not in includes, please add it");
+                    anyhow::bail!("build file {build_file} is not included, please add it");
                 }
 
                 if build_file_path.is_file() {
@@ -285,6 +265,42 @@ impl PublishCommand {
             }
         }
 
+        for relative_path in &paths {
+            let path = project.package_dir().join(relative_path);
+
+            if !path.exists() {
+                anyhow::bail!("included file `{}` does not exist", path.display());
+            }
+
+            let file_name = relative_path
+                .file_name()
+                .context("failed to get file name")?
+                .to_string_lossy()
+                .to_string();
+
+            // it'll be included later after transformations, and is guaranteed to be a file
+            if file_name.eq_ignore_ascii_case(MANIFEST_FILE_NAME) {
+                continue;
+            }
+
+            if path.is_dir() {
+                archive.append_dir(file_name, &path).await.context(format!(
+                    "failed to include directory `{}`",
+                    relative_path.display()
+                ))?;
+            } else {
+                archive
+                    .append_file(
+                        &relative_path,
+                        fs::File::open(&path)
+                            .await
+                            .context(format!("failed to read `{}`", relative_path.display()))?
+                            .file_mut(),
+                    )
+                    .await?;
+            }
+        }
+
         #[cfg(feature = "wally-compat")]
         let mut has_wally = false;
         let mut has_git = false;
@@ -433,7 +449,11 @@ impl PublishCommand {
 
             println!(
                 "includes: {}",
-                display_includes.into_iter().collect::<Vec<_>>().join(", ")
+                paths
+                    .into_iter()
+                    .map(|p| p.to_string_lossy().to_string())
+                    .collect::<Vec<_>>()
+                    .join(", ")
             );
 
             if !self.dry_run
@@ -569,7 +589,7 @@ impl PublishCommand {
     }
 
     pub async fn run(self, project: Project, reqwest: reqwest::Client) -> anyhow::Result<()> {
-        let result = self.clone().run_impl(&project, reqwest.clone()).await;
+        let result = self.clone().run_impl(&project, reqwest.clone(), true).await;
         if project.workspace_dir().is_some() {
             return result;
         } else {
@@ -579,7 +599,7 @@ impl PublishCommand {
         run_on_workspace_members(&project, |project| {
             let reqwest = reqwest.clone();
             let this = self.clone();
-            async move { this.run_impl(&project, reqwest).await }
+            async move { this.run_impl(&project, reqwest, false).await }
         })
         .await
         .map(|_| ())

src/lib.rs

@@ -16,7 +16,6 @@ use std::{
     collections::{HashMap, HashSet},
     path::{Path, PathBuf},
 };
-use tokio::task::spawn_blocking;
 
 /// Downloading packages
 pub mod download;
@@ -193,29 +192,7 @@ impl Project {
             errors::WorkspaceMembersError::ManifestDeser(dir.to_path_buf(), Box::new(e))
         })?;
 
-        let mut members = HashSet::new();
+        let members = matching_globs(dir, manifest.workspace_members, false).await?;
-
-        for glob in &manifest.workspace_members {
-            let is_removal = glob.starts_with('!');
-            let glob = if is_removal { &glob[1..] } else { glob };
-
-            let path = dir.join(glob);
-            let paths = spawn_blocking(move || {
-                glob::glob(&path.as_os_str().to_string_lossy())?
-                    .collect::<Result<Vec<_>, _>>()
-                    .map_err(errors::WorkspaceMembersError::Globbing)
-            })
-            .await
-            .unwrap()?;
-
-            if is_removal {
-                for path in paths {
-                    members.remove(&path);
-                }
-            } else {
-                members.extend(paths);
-            }
-        }
 
         Ok(stream! {
             for path in members {
@@ -232,6 +209,53 @@ impl Project {
     }
 }
 
+/// Gets all matching paths in a directory
+pub async fn matching_globs<P: AsRef<Path>>(
+    dir: P,
+    members: Vec<globset::Glob>,
+    relative: bool,
+) -> Result<HashSet<PathBuf>, errors::MatchingGlobsError> {
+    let mut positive_globset = globset::GlobSetBuilder::new();
+    let mut negative_globset = globset::GlobSetBuilder::new();
+
+    for pattern in members {
+        match pattern.glob().strip_prefix('!') {
+            Some(pattern) => negative_globset.add(globset::Glob::new(pattern)?),
+            None => positive_globset.add(pattern),
+        };
+    }
+
+    let positive_globset = positive_globset.build()?;
+    let negative_globset = negative_globset.build()?;
+
+    let mut read_dirs = vec![fs::read_dir(dir.as_ref().to_path_buf())];
+    let mut paths = HashSet::new();
+
+    while let Some(read_dir) = read_dirs.pop() {
+        let mut read_dir = read_dir.await?;
+        while let Some(entry) = read_dir.next_entry().await? {
+            let path = entry.path();
+            if entry.file_type().await?.is_dir() {
+                read_dirs.push(fs::read_dir(path));
+                continue;
+            }
+
+            let relative_path = path.strip_prefix(dir.as_ref()).unwrap();
+
+            if positive_globset.is_match(relative_path) && !negative_globset.is_match(relative_path)
+            {
+                paths.insert(if relative {
+                    relative_path.to_path_buf()
+                } else {
+                    path.to_path_buf()
+                });
+            }
+        }
+    }
+
+    Ok(paths)
+}
+
 /// Refreshes the sources asynchronously
 pub async fn refresh_sources<I: Iterator<Item = PackageSources>>(
     project: &Project,
@@ -312,12 +336,21 @@ pub mod errors {
         #[error("error interacting with the filesystem")]
         Io(#[from] std::io::Error),
 
-        /// An invalid glob pattern was found
+        /// An error occurred while globbing
-        #[error("invalid glob pattern")]
+        #[error("error globbing")]
-        Glob(#[from] glob::PatternError),
+        Globbing(#[from] MatchingGlobsError),
+    }
+
+    /// Errors that can occur when finding matching globs
+    #[derive(Debug, Error)]
+    #[non_exhaustive]
+    pub enum MatchingGlobsError {
+        /// An error occurred interacting with the filesystem
+        #[error("error interacting with the filesystem")]
+        Io(#[from] std::io::Error),
 
         /// An error occurred while globbing
         #[error("error globbing")]
-        Globbing(#[from] glob::GlobError),
+        Globbing(#[from] globset::Error),
     }
 }

src/main.rs

@@ -6,7 +6,7 @@ use clap::Parser;
 use fs_err::tokio as fs;
 use indicatif::MultiProgress;
 use indicatif_log_bridge::LogWrapper;
-use pesde::{AuthConfig, Project, MANIFEST_FILE_NAME};
+use pesde::{matching_globs, AuthConfig, Project, MANIFEST_FILE_NAME};
 use std::{
     collections::HashSet,
     path::{Path, PathBuf},
@@ -133,17 +133,9 @@ async fn run() -> anyhow::Result<()> {
                 return Ok(HashSet::new());
             }
 
-            manifest
+            matching_globs(path, manifest.workspace_members, false)
-                .workspace_members
+                .await
-                .iter()
+                .context("failed to get workspace members")
-                .map(|member| path.join(member))
-                .map(|p| glob::glob(&p.to_string_lossy()))
-                .collect::<Result<Vec<_>, _>>()
-                .context("invalid glob patterns")?
-                .into_iter()
-                .flat_map(|paths| paths.into_iter())
-                .collect::<Result<HashSet<_>, _>>()
-                .context("failed to expand glob patterns")
         }
 
         while let Some(path) = current_path {

src/manifest/mod.rs

@@ -1,8 +1,7 @@
-use std::collections::{BTreeMap, BTreeSet};
-
 use relative_path::RelativePathBuf;
 use semver::Version;
 use serde::{Deserialize, Serialize};
+use std::collections::{BTreeMap, HashMap};
 
 use crate::{
     manifest::{overrides::OverrideKey, target::Target},
@@ -45,7 +44,7 @@ pub struct Manifest {
     /// The indices to use for the package
     #[serde(
         default,
-        serialize_with = "crate::util::serialize_gix_url_map",
+        skip_serializing,
         deserialize_with = "crate::util::deserialize_gix_url_map"
     )]
     pub indices: BTreeMap<String, gix::Url>,
@@ -53,8 +52,7 @@ pub struct Manifest {
     #[cfg(feature = "wally-compat")]
     #[serde(
         default,
-        skip_serializing_if = "BTreeMap::is_empty",
+        skip_serializing,
-        serialize_with = "crate::util::serialize_gix_url_map",
         deserialize_with = "crate::util::deserialize_gix_url_map"
     )]
     pub wally_indices: BTreeMap<String, gix::Url>,
@@ -63,7 +61,7 @@ pub struct Manifest {
     pub overrides: BTreeMap<OverrideKey, DependencySpecifiers>,
     /// The files to include in the package
     #[serde(default)]
-    pub includes: BTreeSet<String>,
+    pub includes: Vec<globset::Glob>,
     /// The patches to apply to packages
     #[cfg(feature = "patches")]
     #[serde(default, skip_serializing)]
@@ -76,7 +74,7 @@ pub struct Manifest {
     pub pesde_version: Option<Version>,
     /// A list of globs pointing to workspace members' directories
     #[serde(default, skip_serializing_if = "Vec::is_empty")]
-    pub workspace_members: Vec<String>,
+    pub workspace_members: Vec<globset::Glob>,
     /// The Roblox place of this project
     #[serde(default, skip_serializing)]
     pub place: BTreeMap<target::RobloxPlaceKind, String>,
@@ -90,6 +88,9 @@ pub struct Manifest {
     /// The dev dependencies of the package
     #[serde(default, skip_serializing_if = "BTreeMap::is_empty")]
     pub dev_dependencies: BTreeMap<String, DependencySpecifiers>,
+    /// The user-defined fields of the package
+    #[serde(flatten)]
+    pub user_defined_fields: HashMap<String, toml::Value>,
 }
 
 /// A dependency type

src/util.rs

@@ -1,6 +1,6 @@
 use crate::AuthConfig;
 use gix::bstr::BStr;
-use serde::{ser::SerializeMap, Deserialize, Deserializer, Serializer};
+use serde::{Deserialize, Deserializer, Serializer};
 use sha2::{Digest, Sha256};
 use std::collections::BTreeMap;
 
@@ -37,17 +37,6 @@ pub fn deserialize_gix_url<'de, D: Deserializer<'de>>(
     gix::Url::from_bytes(BStr::new(&s)).map_err(serde::de::Error::custom)
 }
 
-pub fn serialize_gix_url_map<S: Serializer>(
-    url: &BTreeMap<String, gix::Url>,
-    serializer: S,
-) -> Result<S::Ok, S::Error> {
-    let mut map = serializer.serialize_map(Some(url.len()))?;
-    for (k, v) in url {
-        map.serialize_entry(k, &v.to_bstring().to_string())?;
-    }
-    map.end()
-}
-
 pub fn deserialize_gix_url_map<'de, D: Deserializer<'de>>(
     deserializer: D,
 ) -> Result<BTreeMap<String, gix::Url>, D::Error> {