rustdoc-search: search backend with partitioned suffix tree

Author: notriddle

Cargo.lock

@@ -4812,6 +4812,7 @@ dependencies = [
  "serde_json",
  "sha2",
  "smallvec",
+ "stringdex",
  "tempfile",
  "threadpool",
  "tracing",
@@ -5225,6 +5226,15 @@ dependencies = [
  "quote",
 ]
 
+[[package]]
+name = "stringdex"
+version = "0.0.1-alpha4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2841fd43df5b1ff1b042e167068a1fe9b163dc93041eae56ab2296859013a9a0"
+dependencies = [
+ "stacker",
+]
+
 [[package]]
 name = "strsim"
 version = "0.11.1"

src/ci/docker/host-x86_64/tidy/eslint.version

@@ -1 +1 @@
-8.6.0
+8.57.1

src/etc/htmldocck.py

@@ -15,6 +15,7 @@
 import re
 import shlex
 from collections import namedtuple
+from pathlib import Path
 
 try:
     from html.parser import HTMLParser
@@ -242,6 +243,11 @@ def resolve_path(self, path):
             return self.last_path
 
     def get_absolute_path(self, path):
+        if "*" in path:
+            paths = list(Path(self.root).glob(path))
+            if len(paths) != 1:
+                raise FailedCheck("glob path does not resolve to one file")
+            path = str(paths[0])
         return os.path.join(self.root, path)
 
     def get_file(self, path):

src/librustdoc/Cargo.toml

@@ -21,6 +21,7 @@ rustdoc-json-types = { path = "../rustdoc-json-types" }
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1.0"
 smallvec = "1.8.1"
+stringdex = { version = "0.0.1-alpha4" }
 tempfile = "3"
 threadpool = "1.8.1"
 tracing = "0.1"

src/librustdoc/build.rs

@@ -10,6 +10,7 @@ fn main() {
         "static/css/normalize.css",
         "static/js/main.js",
         "static/js/search.js",
+        "static/js/stringdex.js",
         "static/js/settings.js",
         "static/js/src-script.js",
         "static/js/storage.js",

src/librustdoc/formats/cache.rs

@@ -1,6 +1,5 @@
 use std::mem;
 
-use rustc_ast::join_path_syms;
 use rustc_data_structures::fx::{FxHashMap, FxHashSet, FxIndexMap, FxIndexSet};
 use rustc_hir::StabilityLevel;
 use rustc_hir::def_id::{CrateNum, DefId, DefIdMap, DefIdSet};
@@ -574,7 +573,6 @@ fn add_item_to_search_index(tcx: TyCtxt<'_>, cache: &mut Cache, item: &clean::It
         clean::ItemKind::ImportItem(import) => import.source.did.unwrap_or(item_def_id),
         _ => item_def_id,
     };
-    let path = join_path_syms(parent_path);
     let impl_id = if let Some(ParentStackItem::Impl { item_id, .. }) = cache.parent_stack.last() {
         item_id.as_def_id()
     } else {
@@ -593,11 +591,11 @@ fn add_item_to_search_index(tcx: TyCtxt<'_>, cache: &mut Cache, item: &clean::It
         ty: item.type_(),
         defid: Some(defid),
         name,
-        path,
+        module_path: parent_path.to_vec(),
         desc,
         parent: parent_did,
         parent_idx: None,
-        exact_path: None,
+        exact_module_path: None,
         impl_id,
         search_type,
         aliases,

src/librustdoc/formats/item_type.rs

@@ -4,7 +4,7 @@ use std::fmt;
 
 use rustc_hir::def::{CtorOf, DefKind, MacroKinds};
 use rustc_span::hygiene::MacroKind;
-use serde::{Serialize, Serializer};
+use serde::{Deserialize, Deserializer, Serialize, Serializer, de};
 
 use crate::clean;
 
@@ -68,6 +68,52 @@ impl Serialize for ItemType {
     }
 }
 
+impl<'de> Deserialize<'de> for ItemType {
+    fn deserialize<D>(deserializer: D) -> Result<ItemType, D::Error>
+    where
+        D: Deserializer<'de>,
+    {
+        struct ItemTypeVisitor;
+        impl<'de> de::Visitor<'de> for ItemTypeVisitor {
+            type Value = ItemType;
+            fn expecting(&self, formatter: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+                write!(formatter, "an integer between 0 and 25")
+            }
+            fn visit_u64<E: de::Error>(self, v: u64) -> Result<ItemType, E> {
+                Ok(match v {
+                    0 => ItemType::Keyword,
+                    1 => ItemType::Primitive,
+                    2 => ItemType::Module,
+                    3 => ItemType::ExternCrate,
+                    4 => ItemType::Import,
+                    5 => ItemType::Struct,
+                    6 => ItemType::Enum,
+                    7 => ItemType::Function,
+                    8 => ItemType::TypeAlias,
+                    9 => ItemType::Static,
+                    10 => ItemType::Trait,
+                    11 => ItemType::Impl,
+                    12 => ItemType::TyMethod,
+                    13 => ItemType::Method,
+                    14 => ItemType::StructField,
+                    15 => ItemType::Variant,
+                    16 => ItemType::Macro,
+                    17 => ItemType::AssocType,
+                    18 => ItemType::Constant,
+                    19 => ItemType::AssocConst,
+                    20 => ItemType::Union,
+                    21 => ItemType::ForeignType,
+                    23 => ItemType::ProcAttribute,
+                    24 => ItemType::ProcDerive,
+                    25 => ItemType::TraitAlias,
+                    _ => return Err(E::missing_field("unknown number")),
+                })
+            }
+        }
+        deserializer.deserialize_any(ItemTypeVisitor)
+    }
+}
+
 impl<'a> From<&'a clean::Item> for ItemType {
     fn from(item: &'a clean::Item) -> ItemType {
         let kind = match &item.kind {
@@ -198,6 +244,10 @@ impl ItemType {
     pub(crate) fn is_adt(&self) -> bool {
         matches!(self, ItemType::Struct | ItemType::Union | ItemType::Enum)
     }
+    /// Keep this the same as isFnLikeTy in search.js
+    pub(crate) fn is_fn_like(&self) -> bool {
+        matches!(self, ItemType::Function | ItemType::Method | ItemType::TyMethod)
+    }
 }
 
 impl fmt::Display for ItemType {

src/librustdoc/html/layout.rs

@@ -27,6 +27,7 @@ pub(crate) struct Layout {
 
 pub(crate) struct Page<'a> {
     pub(crate) title: &'a str,
+    pub(crate) short_title: &'a str,
     pub(crate) css_class: &'a str,
     pub(crate) root_path: &'a str,
     pub(crate) static_root_path: Option<&'a str>,

src/librustdoc/html/render/context.rs

@@ -204,6 +204,18 @@ impl<'tcx> Context<'tcx> {
         if !is_module {
             title.push_str(it.name.unwrap().as_str());
         }
+        let short_title;
+        let short_title = if is_module {
+            let module_name = self.current.last().unwrap();
+            short_title = if it.is_crate() {
+                format!("Crate {module_name}")
+            } else {
+                format!("Module {module_name}")
+            };
+            &short_title[..]
+        } else {
+            it.name.as_ref().unwrap().as_str()
+        };
         if !it.is_primitive() && !it.is_keyword() {
             if !is_module {
                 title.push_str(" in ");
@@ -240,6 +252,7 @@ impl<'tcx> Context<'tcx> {
                 root_path: &self.root_path(),
                 static_root_path: self.shared.static_root_path.as_deref(),
                 title: &title,
+                short_title,
                 description: &desc,
                 resource_suffix: &self.shared.resource_suffix,
                 rust_logo: has_doc_flag(self.tcx(), LOCAL_CRATE.as_def_id(), sym::rust_logo),
@@ -617,6 +630,7 @@ impl<'tcx> FormatRenderer<'tcx> for Context<'tcx> {
         let shared = &self.shared;
         let mut page = layout::Page {
             title: "List of all items in this crate",
+            short_title: "All",
             css_class: "mod sys",
             root_path: "../",
             static_root_path: shared.static_root_path.as_deref(),