Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[rustdoc] Fix type based search #115436

Merged
merged 5 commits into from
Sep 3, 2023
+309 −135
Merged

[rustdoc] Fix type based search #115436

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
5399571
Improve `search.js` code
GuillaumeGomez Aug 31, 2023
09160b3
[rustdoc] Fix path in type-based search
GuillaumeGomez Sep 1, 2023
05eda41
Add tests for type-based search
GuillaumeGomez Sep 1, 2023
36fa557
Merge all loops into one when generating search index
GuillaumeGomez Sep 1, 2023
e161fa1
Correctly handle paths from foreign items
GuillaumeGomez Sep 2, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
220 changes: 115 additions & 105 deletions src/librustdoc/html/render/search_index.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,7 @@ use std::collections::BTreeMap;
use rustc_data_structures::fx::FxHashMap;
use rustc_middle::ty::TyCtxt;
use rustc_span::symbol::Symbol;
use serde::ser::{Serialize, SerializeStruct, Serializer};
use serde::ser::{Serialize, SerializeSeq, SerializeStruct, Serializer};

use crate::clean;
use crate::clean::types::{Function, Generics, ItemId, Type, WherePredicate};
Expand Down Expand Up @@ -78,17 +78,17 @@ pub(crate) fn build_index<'tcx>(
map: &mut FxHashMap<F, usize>,
itemid: F,
lastpathid: &mut usize,
crate_paths: &mut Vec<(ItemType, Symbol)>,
crate_paths: &mut Vec<(ItemType, Vec<Symbol>)>,
item_type: ItemType,
path: Symbol,
path: &[Symbol],
) {
match map.entry(itemid) {
Entry::Occupied(entry) => ty.id = Some(RenderTypeId::Index(*entry.get())),
Entry::Vacant(entry) => {
let pathid = *lastpathid;
entry.insert(pathid);
*lastpathid += 1;
crate_paths.push((item_type, path));
crate_paths.push((item_type, path.to_vec()));
ty.id = Some(RenderTypeId::Index(pathid));
}
}
Expand All @@ -100,7 +100,7 @@ pub(crate) fn build_index<'tcx>(
itemid_to_pathid: &mut FxHashMap<ItemId, usize>,
primitives: &mut FxHashMap<Symbol, usize>,
lastpathid: &mut usize,
crate_paths: &mut Vec<(ItemType, Symbol)>,
crate_paths: &mut Vec<(ItemType, Vec<Symbol>)>,
) {
if let Some(generics) = &mut ty.generics {
for item in generics {
Expand Down Expand Up @@ -131,7 +131,7 @@ pub(crate) fn build_index<'tcx>(
lastpathid,
crate_paths,
item_type,
*fqp.last().unwrap(),
fqp,
);
} else {
ty.id = None;
Expand All @@ -146,7 +146,7 @@ pub(crate) fn build_index<'tcx>(
lastpathid,
crate_paths,
ItemType::Primitive,
sym,
&[sym],
);
}
RenderTypeId::Index(_) => {}
Expand Down Expand Up @@ -191,7 +191,7 @@ pub(crate) fn build_index<'tcx>(
lastpathid += 1;

if let Some(&(ref fqp, short)) = paths.get(&defid) {
crate_paths.push((short, *fqp.last().unwrap()));
crate_paths.push((short, fqp.clone()));
Some(pathid)
} else {
None
Expand All @@ -213,118 +213,128 @@ pub(crate) fn build_index<'tcx>(
struct CrateData<'a> {
doc: String,
items: Vec<&'a IndexItem>,
paths: Vec<(ItemType, Symbol)>,
paths: Vec<(ItemType, Vec<Symbol>)>,
// The String is alias name and the vec is the list of the elements with this alias.
//
// To be noted: the `usize` elements are indexes to `items`.
aliases: &'a BTreeMap<String, Vec<usize>>,
}

struct Paths {
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I needed to create this type because in case there is no path (for primitive types for example), we don't want the array to finish with a null element.

ty: ItemType,
name: Symbol,
path: Option<usize>,
}

impl Serialize for Paths {
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: Serializer,
{
let mut seq = serializer.serialize_seq(None)?;
seq.serialize_element(&self.ty)?;
seq.serialize_element(self.name.as_str())?;
if let Some(ref path) = self.path {
seq.serialize_element(path)?;
}
seq.end()
}
}

impl<'a> Serialize for CrateData<'a> {
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: Serializer,
{
let mut mod_paths = FxHashMap::default();
for (index, item) in self.items.iter().enumerate() {
if item.path.is_empty() {
continue;
}
mod_paths.insert(&item.path, index);
}
let paths = self
.paths
.iter()
.map(|(ty, path)| {
if path.len() < 2 {
return Paths { ty: *ty, name: path[0], path: None };
}
let index = mod_paths.get(&join_with_double_colon(&path[..path.len() - 1]));
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

How does this work when a function signature contains a type that’s defined in another crate and not inlined into this one anywhere?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It doesn't, good catch. We only get the paths from the cache. We can get the path with def_path_str. Going to take a look.

Paths { ty: *ty, name: *path.last().unwrap(), path: index.copied() }
})
.collect::<Vec<_>>();

let mut names = Vec::with_capacity(self.items.len());
let mut types = String::with_capacity(self.items.len());
let mut full_paths = Vec::with_capacity(self.items.len());
let mut descriptions = Vec::with_capacity(self.items.len());
let mut parents = Vec::with_capacity(self.items.len());
let mut functions = Vec::with_capacity(self.items.len());
let mut deprecated = Vec::with_capacity(self.items.len());

for (index, item) in self.items.iter().enumerate() {
let n = item.ty as u8;
let c = char::try_from(n + b'A').expect("item types must fit in ASCII");
assert!(c <= 'z', "item types must fit within ASCII printables");
types.push(c);

assert_eq!(
item.parent.is_some(),
item.parent_idx.is_some(),
"`{}` is missing idx",
item.name
);
// 0 is a sentinel, everything else is one-indexed
parents.push(item.parent_idx.map(|x| x + 1).unwrap_or(0));

names.push(item.name.as_str());
descriptions.push(&item.desc);

if !item.path.is_empty() {
full_paths.push((index, &item.path));
}

// Fake option to get `0` out as a sentinel instead of `null`.
// We want to use `0` because it's three less bytes.
enum FunctionOption<'a> {
Function(&'a IndexItemFunctionType),
None,
}
impl<'a> Serialize for FunctionOption<'a> {
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: Serializer,
{
match self {
FunctionOption::None => 0.serialize(serializer),
FunctionOption::Function(ty) => ty.serialize(serializer),
}
}
}
functions.push(match &item.search_type {
Some(ty) => FunctionOption::Function(ty),
None => FunctionOption::None,
});

if item.deprecation.is_some() {
deprecated.push(index);
}
}

let has_aliases = !self.aliases.is_empty();
let mut crate_data =
serializer.serialize_struct("CrateData", if has_aliases { 9 } else { 8 })?;
crate_data.serialize_field("doc", &self.doc)?;
crate_data.serialize_field(
"t",
&self
.items
.iter()
.map(|item| {
let n = item.ty as u8;
let c = char::try_from(n + b'A').expect("item types must fit in ASCII");
assert!(c <= 'z', "item types must fit within ASCII printables");
c
})
.collect::<String>(),
)?;
crate_data.serialize_field(
"n",
&self.items.iter().map(|item| item.name.as_str()).collect::<Vec<_>>(),
)?;
crate_data.serialize_field(
"q",
&self
.items
.iter()
.enumerate()
// Serialize as an array of item indices and full paths
.filter_map(
|(index, item)| {
if item.path.is_empty() { None } else { Some((index, &item.path)) }
},
)
.collect::<Vec<_>>(),
)?;
crate_data.serialize_field(
"d",
&self.items.iter().map(|item| &item.desc).collect::<Vec<_>>(),
)?;
crate_data.serialize_field(
"i",
&self
.items
.iter()
.map(|item| {
assert_eq!(
item.parent.is_some(),
item.parent_idx.is_some(),
"`{}` is missing idx",
item.name
);
// 0 is a sentinel, everything else is one-indexed
item.parent_idx.map(|x| x + 1).unwrap_or(0)
})
.collect::<Vec<_>>(),
)?;
crate_data.serialize_field(
"f",
&self
.items
.iter()
.map(|item| {
// Fake option to get `0` out as a sentinel instead of `null`.
// We want to use `0` because it's three less bytes.
enum FunctionOption<'a> {
Function(&'a IndexItemFunctionType),
None,
}
impl<'a> Serialize for FunctionOption<'a> {
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: Serializer,
{
match self {
FunctionOption::None => 0.serialize(serializer),
FunctionOption::Function(ty) => ty.serialize(serializer),
}
}
}
match &item.search_type {
Some(ty) => FunctionOption::Function(ty),
None => FunctionOption::None,
}
})
.collect::<Vec<_>>(),
)?;
crate_data.serialize_field(
"c",
&self
.items
.iter()
.enumerate()
// Serialize as an array of deprecated item indices
.filter_map(|(index, item)| item.deprecation.map(|_| index))
.collect::<Vec<_>>(),
)?;
crate_data.serialize_field(
"p",
&self.paths.iter().map(|(it, s)| (it, s.as_str())).collect::<Vec<_>>(),
)?;
crate_data.serialize_field("t", &types)?;
crate_data.serialize_field("n", &names)?;
// Serialize as an array of item indices and full paths
crate_data.serialize_field("q", &full_paths)?;
crate_data.serialize_field("d", &descriptions)?;
crate_data.serialize_field("i", &parents)?;
crate_data.serialize_field("f", &functions)?;
crate_data.serialize_field("c", &deprecated)?;
crate_data.serialize_field("p", &paths)?;
if has_aliases {
crate_data.serialize_field("a", &self.aliases)?;
}
Expand Down
Loading
Loading