Extract some private submodules from 'bat::assets' (#1850)

src/assets.rs

@@ -1,7 +1,6 @@
-use std::collections::HashMap;
 use std::ffi::OsStr;
 use std::fs;
-use std::path::{Path, PathBuf};
+use std::path::Path;
 
 use lazycell::LazyCell;
 
@@ -15,17 +14,26 @@ use crate::error::*;
 use crate::input::{InputReader, OpenedInput, OpenedInputKind};
 use crate::syntax_mapping::{MappingTarget, SyntaxMapping};
 
+use ignored_suffixes::*;
+use minimal_assets::*;
+use serialized_syntax_set::*;
+
+#[cfg(feature = "build-assets")]
+pub use crate::assets::build_assets::*;
+
+pub(crate) mod assets_metadata;
+#[cfg(feature = "build-assets")]
+mod build_assets;
+mod ignored_suffixes;
+mod minimal_assets;
+mod serialized_syntax_set;
+
 #[derive(Debug)]
 pub struct HighlightingAssets {
     syntax_set_cell: LazyCell<SyntaxSet>,
     serialized_syntax_set: SerializedSyntaxSet,
 
-    minimal_syntaxes: MinimalSyntaxes,
+    minimal_assets: MinimalAssets,
-
-    /// Lazily load serialized [SyntaxSet]s from [Self.minimal_syntaxes]. The
-    /// index in this vec matches the index in
-    /// [Self.minimal_syntaxes.serialized_syntax_sets]
-    deserialized_minimal_syntaxes: Vec<LazyCell<SyntaxSet>>,
 
     theme_set: ThemeSet,
     fallback_theme: Option<&'static str>,
@@ -37,22 +45,6 @@ pub struct SyntaxReferenceInSet<'a> {
     pub syntax_set: &'a SyntaxSet,
 }
 
-/// Stores and allows lookup of minimal [SyntaxSet]s. The [SyntaxSet]s are
-/// stored in serialized form, and are deserialized on-demand. This gives good
-/// startup performance since only the necessary [SyntaxReference]s needs to be
-/// deserialized.
-#[derive(Debug, Clone, serde::Deserialize, serde::Serialize)]
-pub(crate) struct MinimalSyntaxes {
-    /// Lookup the index into `serialized_syntax_sets` of a [SyntaxSet] by the
-    /// name of any [SyntaxReference] inside the [SyntaxSet]
-    /// (We will later add `by_extension`, `by_first_line`, etc.)
-    pub(crate) by_name: HashMap<String, usize>,
-
-    /// Serialized [SyntaxSet]s. Whether or not this data is compressed is
-    /// decided by [COMPRESS_SERIALIZED_MINIMAL_SYNTAXES]
-    pub(crate) serialized_syntax_sets: Vec<Vec<u8>>,
-}
-
 // Compress for size of ~700 kB instead of ~4600 kB at the cost of ~30% longer deserialization time
 pub(crate) const COMPRESS_SYNTAXES: bool = true;
 
@@ -70,41 +62,16 @@ pub(crate) const COMPRESS_SERIALIZED_MINIMAL_SYNTAXES: bool = true;
 // efficient byte-by-byte copy of `serialized_syntax_sets`.
 pub(crate) const COMPRESS_MINIMAL_SYNTAXES: bool = false;
 
-const IGNORED_SUFFIXES: [&str; 13] = [
-    // Editor etc backups
-    "~",
-    ".bak",
-    ".old",
-    ".orig",
-    // Debian and derivatives apt/dpkg/ucf backups
-    ".dpkg-dist",
-    ".dpkg-old",
-    ".ucf-dist",
-    ".ucf-new",
-    ".ucf-old",
-    // Red Hat and derivatives rpm backups
-    ".rpmnew",
-    ".rpmorig",
-    ".rpmsave",
-    // Build system input/template files
-    ".in",
-];
-
 impl HighlightingAssets {
     fn new(
         serialized_syntax_set: SerializedSyntaxSet,
         minimal_syntaxes: MinimalSyntaxes,
         theme_set: ThemeSet,
     ) -> Self {
-        // Prepare so we can lazily load minimal syntaxes without a mut reference
-        let deserialized_minimal_syntaxes =
-            vec![LazyCell::new(); minimal_syntaxes.serialized_syntax_sets.len()];
-
         HighlightingAssets {
             syntax_set_cell: LazyCell::new(),
             serialized_syntax_set,
-            deserialized_minimal_syntaxes,
+            minimal_assets: MinimalAssets::new(minimal_syntaxes),
-            minimal_syntaxes,
             theme_set,
             fallback_theme: None,
         }
@@ -167,37 +134,12 @@ impl HighlightingAssets {
     /// tries to find a minimal [SyntaxSet]. If none is found, returns the
     /// [SyntaxSet] that contains all syntaxes.
     fn get_syntax_set_by_name(&self, name: &str) -> Result<&SyntaxSet> {
-        let minimal_syntax_set = self
+        match self.minimal_assets.get_syntax_set_by_name(name) {
-            .minimal_syntaxes
-            .by_name
-            .get(&name.to_ascii_lowercase())
-            .and_then(|index| self.get_minimal_syntax_set_with_index(*index));
-
-        match minimal_syntax_set {
             Some(syntax_set) => Ok(syntax_set),
             None => self.get_syntax_set(),
         }
     }
 
-    fn load_minimal_syntax_set_with_index(&self, index: usize) -> Result<SyntaxSet> {
-        let serialized_syntax_set = &self.minimal_syntaxes.serialized_syntax_sets[index];
-        asset_from_contents(
-            &serialized_syntax_set[..],
-            &format!("minimal syntax set {}", index),
-            COMPRESS_SERIALIZED_MINIMAL_SYNTAXES,
-        )
-        .map_err(|_| format!("Could not parse minimal syntax set {}", index).into())
-    }
-
-    fn get_minimal_syntax_set_with_index(&self, index: usize) -> Option<&SyntaxSet> {
-        self.deserialized_minimal_syntaxes
-            .get(index)
-            .and_then(|cell| {
-                cell.try_borrow_with(|| self.load_minimal_syntax_set_with_index(index))
-                    .ok()
-            })
-    }
-
     /// Use [Self::get_syntax_for_file_name] instead
     #[deprecated]
     pub fn syntax_for_file_name(
@@ -319,7 +261,9 @@ impl HighlightingAssets {
             syntax = self.find_syntax_by_file_name_extension(file_name)?;
         }
         if syntax.is_none() {
-            syntax = self.get_extension_syntax_with_stripped_suffix(file_name)?;
+            syntax = try_with_stripped_suffix(file_name, |stripped_file_name| {
+                self.get_extension_syntax(stripped_file_name) // Note: recursion
+            })?;
         }
         Ok(syntax)
     }
@@ -340,25 +284,6 @@ impl HighlightingAssets {
         )
     }
 
-    /// If we find an ignored suffix on the file name, e.g. '~', we strip it and
-    /// then try again to find a syntax without it. Note that we do this recursively.
-    fn get_extension_syntax_with_stripped_suffix(
-        &self,
-        file_name: &OsStr,
-    ) -> Result<Option<SyntaxReferenceInSet>> {
-        let file_path = Path::new(file_name);
-        let mut syntax = None;
-        if let Some(file_str) = file_path.to_str() {
-            for suffix in &IGNORED_SUFFIXES {
-                if let Some(stripped_filename) = file_str.strip_suffix(suffix) {
-                    syntax = self.get_extension_syntax(OsStr::new(stripped_filename))?;
-                    break;
-                }
-            }
-        }
-        Ok(syntax)
-    }
-
     fn get_first_line_syntax(
         &self,
         reader: &mut InputReader,
@@ -371,31 +296,6 @@ impl HighlightingAssets {
     }
 }
 
-#[cfg(feature = "build-assets")]
-pub use crate::build_assets::build_assets as build;
-
-/// A SyntaxSet in serialized form, i.e. bincoded and flate2 compressed.
-/// We keep it in this format since we want to load it lazily.
-#[derive(Debug)]
-enum SerializedSyntaxSet {
-    /// The data comes from a user-generated cache file.
-    FromFile(PathBuf),
-
-    /// The data to use is embedded into the bat binary.
-    FromBinary(&'static [u8]),
-}
-
-impl SerializedSyntaxSet {
-    fn deserialize(&self) -> Result<SyntaxSet> {
-        match self {
-            SerializedSyntaxSet::FromBinary(data) => Ok(from_binary(data, COMPRESS_SYNTAXES)),
-            SerializedSyntaxSet::FromFile(ref path) => {
-                asset_from_cache(path, "syntax set", COMPRESS_SYNTAXES)
-            }
-        }
-    }
-}
-
 pub(crate) fn get_serialized_integrated_syntaxset() -> &'static [u8] {
     include_bytes!("../assets/syntaxes.bin")
 }

src/assets/build_assets.rs

@@ -7,7 +7,6 @@ use syntect::parsing::syntax_definition::{
 use syntect::parsing::{Scope, SyntaxSet, SyntaxSetBuilder};
 
 use crate::assets::*;
-use crate::error::*;
 
 type SyntaxName = String;
 
@@ -27,7 +26,7 @@ enum Dependency {
     ByScope(Scope),
 }
 
-pub fn build_assets(
+pub fn build(
     source_dir: &Path,
     include_integrated_assets: bool,
     target_dir: &Path,

src/assets/ignored_suffixes.rs

@@ -0,0 +1,42 @@
+use std::ffi::OsStr;
+use std::path::Path;
+
+use crate::error::*;
+
+const IGNORED_SUFFIXES: [&str; 13] = [
+    // Editor etc backups
+    "~",
+    ".bak",
+    ".old",
+    ".orig",
+    // Debian and derivatives apt/dpkg/ucf backups
+    ".dpkg-dist",
+    ".dpkg-old",
+    ".ucf-dist",
+    ".ucf-new",
+    ".ucf-old",
+    // Red Hat and derivatives rpm backups
+    ".rpmnew",
+    ".rpmorig",
+    ".rpmsave",
+    // Build system input/template files
+    ".in",
+];
+
+/// If we find an ignored suffix on the file name, e.g. '~', we strip it and
+/// then try again without it.
+pub fn try_with_stripped_suffix<T, F>(file_name: &OsStr, func: F) -> Result<Option<T>>
+where
+    F: Fn(&OsStr) -> Result<Option<T>>,
+{
+    let mut from_stripped = None;
+    if let Some(file_str) = Path::new(file_name).to_str() {
+        for suffix in &IGNORED_SUFFIXES {
+            if let Some(stripped_filename) = file_str.strip_suffix(suffix) {
+                from_stripped = func(OsStr::new(stripped_filename))?;
+                break;
+            }
+        }
+    }
+    Ok(from_stripped)
+}

src/assets/minimal_assets.rs

@@ -0,0 +1,72 @@
+use std::collections::HashMap;
+
+use lazycell::LazyCell;
+
+use syntect::parsing::SyntaxSet;
+
+use super::*;
+
+#[derive(Debug)]
+pub(crate) struct MinimalAssets {
+    minimal_syntaxes: MinimalSyntaxes,
+
+    /// Lazily load serialized [SyntaxSet]s from [Self.minimal_syntaxes]. The
+    /// index in this vec matches the index in
+    /// [Self.minimal_syntaxes.serialized_syntax_sets]
+    deserialized_minimal_syntaxes: Vec<LazyCell<SyntaxSet>>,
+}
+
+/// Stores and allows lookup of minimal [SyntaxSet]s. The [SyntaxSet]s are
+/// stored in serialized form, and are deserialized on-demand. This gives good
+/// startup performance since only the necessary [SyntaxReference]s needs to be
+/// deserialized.
+#[derive(Debug, Clone, serde::Deserialize, serde::Serialize)]
+pub(crate) struct MinimalSyntaxes {
+    /// Lookup the index into `serialized_syntax_sets` of a [SyntaxSet] by the
+    /// name of any [SyntaxReference] inside the [SyntaxSet]
+    /// (We will later add `by_extension`, `by_first_line`, etc.)
+    pub(crate) by_name: HashMap<String, usize>,
+
+    /// Serialized [SyntaxSet]s. Whether or not this data is compressed is
+    /// decided by [COMPRESS_SERIALIZED_MINIMAL_SYNTAXES]
+    pub(crate) serialized_syntax_sets: Vec<Vec<u8>>,
+}
+
+impl MinimalAssets {
+    pub(crate) fn new(minimal_syntaxes: MinimalSyntaxes) -> Self {
+        // Prepare so we can lazily load minimal syntaxes without a mut reference
+        let deserialized_minimal_syntaxes =
+            vec![LazyCell::new(); minimal_syntaxes.serialized_syntax_sets.len()];
+
+        Self {
+            minimal_syntaxes,
+            deserialized_minimal_syntaxes,
+        }
+    }
+
+    pub fn get_syntax_set_by_name(&self, name: &str) -> Option<&SyntaxSet> {
+        self.minimal_syntaxes
+            .by_name
+            .get(&name.to_ascii_lowercase())
+            .and_then(|index| self.get_minimal_syntax_set_with_index(*index))
+    }
+
+    fn load_minimal_syntax_set_with_index(&self, index: usize) -> Result<SyntaxSet> {
+        let serialized_syntax_set = &self.minimal_syntaxes.serialized_syntax_sets[index];
+        asset_from_contents(
+            &serialized_syntax_set[..],
+            &format!("minimal syntax set {}", index),
+            COMPRESS_SERIALIZED_MINIMAL_SYNTAXES,
+        )
+        .map_err(|_| format!("Could not parse minimal syntax set {}", index).into())
+    }
+
+    fn get_minimal_syntax_set_with_index(&self, index: usize) -> Option<&SyntaxSet> {
+        self.deserialized_minimal_syntaxes
+            .get(index)
+            .and_then(|cell| {
+                cell.try_borrow_with(|| self.load_minimal_syntax_set_with_index(index))
+                    .ok()
+            })
+    }
+}

src/assets/serialized_syntax_set.rs

@@ -0,0 +1,27 @@
+use std::path::PathBuf;
+
+use syntect::parsing::SyntaxSet;
+
+use super::*;
+
+/// A SyntaxSet in serialized form, i.e. bincoded and flate2 compressed.
+/// We keep it in this format since we want to load it lazily.
+#[derive(Debug)]
+pub enum SerializedSyntaxSet {
+    /// The data comes from a user-generated cache file.
+    FromFile(PathBuf),
+
+    /// The data to use is embedded into the bat binary.
+    FromBinary(&'static [u8]),
+}
+
+impl SerializedSyntaxSet {
+    pub fn deserialize(&self) -> Result<SyntaxSet> {
+        match self {
+            SerializedSyntaxSet::FromBinary(data) => Ok(from_binary(data, COMPRESS_SYNTAXES)),
+            SerializedSyntaxSet::FromFile(ref path) => {
+                asset_from_cache(path, "syntax set", COMPRESS_SYNTAXES)
+            }
+        }
+    }
+}

src/lib.rs

@@ -22,9 +22,9 @@
 mod macros;
 
 pub mod assets;
-pub mod assets_metadata;
+pub mod assets_metadata {
-#[cfg(feature = "build-assets")]
+    pub use super::assets::assets_metadata::*;
-mod build_assets;
+}
 pub mod config;
 pub mod controller;
 mod decorations;