Return 'extra file tags' in report info (#98)

This returns info on ignored files in the `report`. The intent here is
to also make it possible for the consumer to show information on files
that are only used in tests.

change/@good-fences-api-aa31c04e-9fa3-4be0-b954-b56665c76c6d.json

@@ -0,0 +1,7 @@
+{
+  "type": "minor",
+  "comment": "unused_finder: return tagged symbols",
+  "packageName": "@good-fences/api",
+  "email": "mhuan13@gmail.com",
+  "dependentChangeType": "patch"
+}

crates/unused_finder/Cargo.toml

@@ -35,6 +35,8 @@ ahashmap = { path = "../ahashmap" }
 bitflags = "2.6.0"
 ignore = "0.4.23"
 abspath = { version = "0.2.0", path = "../abspath" }
+itertools = "0.13.0"
+debug_print = "1.0.0"
 
 [dev-dependencies]
 stringreader = "0.1.1"

crates/unused_finder/src/cfg/mod.rs

@@ -1,6 +1,8 @@
 use std::fmt::{Display, Formatter};
 
+use itertools::Itertools;
 use package_match_rules::PackageMatchRules;
+use rayon::iter::Either;
 use serde::Deserialize;
 
 pub mod package_match_rules;
@@ -101,13 +103,14 @@ pub struct UnusedFinderJSONConfig {
     /// 3. Otherwise, the item is treated as the name of an individual package, and matched
     ///    literally.
     pub entry_packages: Vec<String>,
-    /// List of globs that will be matched against files in the repository
+    /// List of glob patterns to mark as "tests".
+    /// These files will be marked as used, and all of their transitive
+    /// dependencies will also be marked as used
     ///
-    /// Matches are made against the relative file paths from the repo root.
-    /// A matching file will be tagged as a "test" file, and will be excluded
-    /// from the list of unused files
+    /// glob patterns are matched against the relative file path from the
+    /// root of the repository
     #[serde(default)]
-    pub test_file_patterns: Vec<String>,
+    pub test_files: Vec<String>,
 }
 
 /// Configuration for the unused symbols finder
@@ -134,7 +137,7 @@ pub struct UnusedFinderConfig {
     /// Matches are made against the relative file paths from the repo root.
     /// A matching file will be tagged as a "test" file, and will be excluded
     /// from the list of unused files
-    pub test_file_patterns: Vec<glob::Pattern>,
+    pub test_files: Vec<glob::Pattern>,
 
     /// Globs of individual files & directories to skip during the file walk.
     ///
@@ -146,6 +149,17 @@ pub struct UnusedFinderConfig {
 impl TryFrom<UnusedFinderJSONConfig> for UnusedFinderConfig {
     type Error = ConfigError;
     fn try_from(value: UnusedFinderJSONConfig) -> std::result::Result<Self, Self::Error> {
+        let (test_globs, test_glob_errs): (Vec<glob::Pattern>, Vec<_>) = value
+            .test_files
+            .iter()
+            .partition_map(|pat| match glob::Pattern::new(pat) {
+                Ok(pat) => Either::Left(pat),
+                Err(err) => Either::Right(PatErr(0, GlobInterp::Path, err)),
+            });
+        if !test_glob_errs.is_empty() {
+            return Err(ConfigError::InvalidGlobPatterns(ErrList(test_glob_errs)));
+        }
+
         Ok(UnusedFinderConfig {
             // raw fields that are copied from the JSON config
             report_exported_symbols: value.report_exported_symbols,
@@ -154,14 +168,7 @@ impl TryFrom<UnusedFinderJSONConfig> for UnusedFinderConfig {
             repo_root: value.repo_root,
             // other fields that are processed before use
             entry_packages: value.entry_packages.try_into()?,
-            test_file_patterns: value
-                .test_file_patterns
-                .iter()
-                .map(|p| glob::Pattern::new(p))
-                .collect::<Result<_, _>>()
-                .map_err(|e| {
-                    ConfigError::InvalidGlobPatterns(ErrList(vec![PatErr(0, GlobInterp::Path, e)]))
-                })?,
+            test_files: test_globs,
             skip: value.skip,
         })
     }

crates/unused_finder/src/graph.rs

@@ -1,9 +1,5 @@
-use core::{fmt, option::Option::None};
-use std::{
-    collections::HashSet,
-    fmt::Display,
-    path::{Path, PathBuf},
-};
+use core::option::Option::None;
+use std::{collections::HashSet, path::Path, path::PathBuf};
 
 use ahashmap::{AHashMap, AHashSet};
 use anyhow::Result;
@@ -12,41 +8,10 @@ use rayon::prelude::*;
 use crate::{
     logger::Logger,
     parse::{ExportedSymbol, ResolvedImportExportInfo},
+    tag::UsedTag,
     walked_file::ResolvedSourceFile,
 };
 
-bitflags::bitflags! {
-    #[derive(Clone, Copy, Debug, PartialEq, Eq, Hash, Default)]
-    pub struct UsedTag: u8 {
-        /// True if this file or symbol was used recursively by an
-        /// "entry package" (a package that was passed as an entry point).
-        const FROM_ENTRY = 0x01;
-        /// True if this file or symbol was used recursively by a test file.
-        const FROM_TEST = 0x02;
-        /// True if this file or symbol was used recursively by an
-        /// ignored symbol or file.
-        const FROM_IGNORED = 0x04;
-        // True if this symbol is a type-only symbol
-        const TYPE_ONLY = 0x08;
-    }
-}
-
-impl Display for UsedTag {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        let mut tags = Vec::new();
-        if self.contains(Self::FROM_ENTRY) {
-            tags.push("entry");
-        };
-        if self.contains(Self::FROM_IGNORED) {
-            tags.push("ignored");
-        };
-        if self.contains(Self::TYPE_ONLY) {
-            tags.push("type-only");
-        }
-        write!(f, "{}", tags.join("+"))
-    }
-}
-
 // graph node used to represent a file during the "used file" walk
 #[derive(Debug, Clone, Default)]
 pub struct GraphFile {
@@ -168,13 +133,13 @@ impl Graph {
     pub fn traverse_bfs(
         &mut self,
         logger: impl Logger,
-        initial_frontier_files: Vec<PathBuf>,
-        initial_frontier_symbols: Vec<(PathBuf, Vec<ExportedSymbol>)>,
+        initial_frontier_files: Vec<&Path>,
+        initial_frontier_symbols: Vec<(&Path, Vec<ExportedSymbol>)>,
         tag: UsedTag,
     ) -> Result<()> {
         let initial_file_edges = initial_frontier_files
             .into_iter()
-            .filter_map(|path| match self.path_to_id.get(&path) {
+            .filter_map(|path| match self.path_to_id.get(path) {
                 Some(file_id) => Some(*file_id),
                 None => {
                     logger.log(format!(
@@ -189,8 +154,8 @@ impl Graph {
         let initial_symbol_edges = initial_frontier_symbols
             .into_iter()
             .filter_map(
-                |(path, symbols): (PathBuf, Vec<ExportedSymbol>)| -> Option<Vec<Edge>> {
-                    match self.path_to_id.get(&path).cloned() {
+                |(path, symbols): (&Path, Vec<ExportedSymbol>)| -> Option<Vec<Edge>> {
+                    match self.path_to_id.get(path).cloned() {
                         Some(file_id) => Some(
                             symbols
                                 .into_iter()

crates/unused_finder/src/lib.rs

@@ -18,6 +18,7 @@ mod ignore_file;
 pub mod logger;
 mod parse;
 mod report;
+mod tag;
 #[cfg(test)]
 mod test;
 mod unused_finder;
@@ -26,7 +27,8 @@ mod walked_file;
 
 pub use cfg::{UnusedFinderConfig, UnusedFinderJSONConfig};
 pub use parse::data::ResolvedImportExportInfo;
-pub use report::{SymbolReport, UnusedFinderReport, UsedTagEnum};
+pub use report::{SymbolReport, SymbolReportWithTags, UnusedFinderReport};
+pub use tag::UsedTagEnum;
 pub use unused_finder::{UnusedFinder, UnusedFinderResult};
 
 pub fn find_unused_items(