hugo/related/inverted_index.go
Bjørn Erik Pedersen 7285e74090
all: Rework page store, add a dynacache, improve partial rebuilds, and some general spring cleaning
There are some breaking changes in this commit, see #11455.

Closes #11455
Closes #11549

This fixes a set of bugs (see issue list) and it is also paying some technical debt accumulated over the years. We now build with Staticcheck enabled in the CI build.

The performance should be about the same as before for regular sized Hugo sites, but it should perform and scale much better to larger data sets, as objects that uses lots of memory (e.g. rendered Markdown, big JSON files read into maps with transform.Unmarshal etc.) will now get automatically garbage collected if needed. Performance on partial rebuilds when running the server in fast render mode should be the same, but the change detection should be much more accurate.

A list of the notable new features:

* A new dependency tracker that covers (almost) all of Hugo's API and is used to do fine grained partial rebuilds when running the server.
* A new and simpler tree document store which allows fast lookups and prefix-walking in all dimensions (e.g. language) concurrently.
* You can now configure an upper memory limit allowing for much larger data sets and/or running on lower specced PCs.
We have lifted the "no resources in sub folders" restriction for branch bundles (e.g. sections).
Memory Limit
* Hugos will, by default, set aside a quarter of the total system memory, but you can set this via the OS environment variable HUGO_MEMORYLIMIT (in gigabytes). This is backed by a partitioned LRU cache used throughout Hugo. A cache that gets dynamically resized in low memory situations, allowing Go's Garbage Collector to free the memory.

New Dependency Tracker: Hugo has had a rule based coarse grained approach to server rebuilds that has worked mostly pretty well, but there have been some surprises (e.g. stale content). This is now revamped with a new dependency tracker that can quickly calculate the delta given a changed resource (e.g. a content file, template, JS file etc.). This handles transitive relations, e.g. $page -> js.Build -> JS import, or $page1.Content -> render hook -> site.GetPage -> $page2.Title, or $page1.Content -> shortcode -> partial -> site.RegularPages -> $page2.Content -> shortcode ..., and should also handle changes to aggregated values (e.g. site.Lastmod) effectively.

This covers all of Hugo's API with 2 known exceptions (a list that may not be fully exhaustive):

Changes to files loaded with template func os.ReadFile may not be handled correctly. We recommend loading resources with resources.Get
Changes to Hugo objects (e.g. Page) passed in the template context to lang.Translate may not be detected correctly. We recommend having simple i18n templates without too much data context passed in other than simple types such as strings and numbers.
Note that the cachebuster configuration (when A changes then rebuild B) works well with the above, but we recommend that you revise that configuration, as it in most situations should not be needed. One example where it is still needed is with TailwindCSS and using changes to hugo_stats.json to trigger new CSS rebuilds.

Document Store: Previously, a little simplified, we split the document store (where we store pages and resources) in a tree per language. This worked pretty well, but the structure made some operations harder than they needed to be. We have now restructured it into one Radix tree for all languages. Internally the language is considered to be a dimension of that tree, and the tree can be viewed in all dimensions concurrently. This makes some operations re. language simpler (e.g. finding translations is just a slice range), but the idea is that it should also be relatively inexpensive to add more dimensions if needed (e.g. role).

Fixes #10169
Fixes #10364
Fixes #10482
Fixes #10630
Fixes #10656
Fixes #10694
Fixes #10918
Fixes #11262
Fixes #11439
Fixes #11453
Fixes #11457
Fixes #11466
Fixes #11540
Fixes #11551
Fixes #11556
Fixes #11654
Fixes #11661
Fixes #11663
Fixes #11664
Fixes #11669
Fixes #11671
Fixes #11807
Fixes #11808
Fixes #11809
Fixes #11815
Fixes #11840
Fixes #11853
Fixes #11860
Fixes #11883
Fixes #11904
Fixes #7388
Fixes #7425
Fixes #7436
Fixes #7544
Fixes #7882
Fixes #7960
Fixes #8255
Fixes #8307
Fixes #8863
Fixes #8927
Fixes #9192
Fixes #9324
2024-01-27 16:28:14 +01:00

629 lines
16 KiB
Go

// Copyright 2019 The Hugo Authors. All rights reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
// Package related holds code to help finding related content.
package related
import (
"context"
"errors"
"fmt"
"math"
"sort"
"strings"
"sync"
"time"
xmaps "golang.org/x/exp/maps"
"github.com/gohugoio/hugo/common/collections"
"github.com/gohugoio/hugo/common/maps"
"github.com/gohugoio/hugo/compare"
"github.com/gohugoio/hugo/markup/tableofcontents"
"github.com/spf13/cast"
"github.com/gohugoio/hugo/common/types"
"github.com/mitchellh/mapstructure"
)
const (
TypeBasic = "basic"
TypeFragments = "fragments"
)
var validTypes = map[string]bool{
TypeBasic: true,
TypeFragments: true,
}
var (
_ Keyword = (*StringKeyword)(nil)
zeroDate = time.Time{}
// DefaultConfig is the default related config.
DefaultConfig = Config{
Threshold: 80,
Indices: IndicesConfig{
IndexConfig{Name: "keywords", Weight: 100, Type: TypeBasic},
IndexConfig{Name: "date", Weight: 10, Type: TypeBasic},
},
}
)
// Config is the top level configuration element used to configure how to retrieve
// related content in Hugo.
type Config struct {
// Only include matches >= threshold, a normalized rank between 0 and 100.
Threshold int
// To get stable "See also" sections we, by default, exclude newer related pages.
IncludeNewer bool
// Will lower case all string values and queries to the indices.
// May get better results, but at a slight performance cost.
ToLower bool
Indices IndicesConfig
}
// Add adds a given index.
func (c *Config) Add(index IndexConfig) {
if c.ToLower {
index.ToLower = true
}
c.Indices = append(c.Indices, index)
}
func (c *Config) HasType(s string) bool {
for _, i := range c.Indices {
if i.Type == s {
return true
}
}
return false
}
// IndicesConfig holds a set of index configurations.
type IndicesConfig []IndexConfig
// IndexConfig configures an index.
type IndexConfig struct {
// The index name. This directly maps to a field or Param name.
Name string
// The index type.
Type string
// Enable to apply a type specific filter to the results.
// This is currently only used for the "fragments" type.
ApplyFilter bool
// Contextual pattern used to convert the Param value into a string.
// Currently only used for dates. Can be used to, say, bump posts in the same
// time frame when searching for related documents.
// For dates it follows Go's time.Format patterns, i.e.
// "2006" for YYYY and "200601" for YYYYMM.
Pattern string
// This field's weight when doing multi-index searches. Higher is "better".
Weight int
// A percentage (0-100) used to remove common keywords from the index.
// As an example, setting this to 50 will remove all keywords that are
// used in more than 50% of the documents in the index.
CardinalityThreshold int
// Will lower case all string values in and queries tothis index.
// May get better accurate results, but at a slight performance cost.
ToLower bool
}
// Document is the interface an indexable document in Hugo must fulfill.
type Document interface {
// RelatedKeywords returns a list of keywords for the given index config.
RelatedKeywords(cfg IndexConfig) ([]Keyword, error)
// When this document was or will be published.
PublishDate() time.Time
// Name is used as an tiebreaker if both Weight and PublishDate are
// the same.
Name() string
}
// FragmentProvider is an optional interface that can be implemented by a Document.
type FragmentProvider interface {
Fragments(context.Context) *tableofcontents.Fragments
// For internal use.
ApplyFilterToHeadings(context.Context, func(*tableofcontents.Heading) bool) Document
}
// InvertedIndex holds an inverted index, also sometimes named posting list, which
// lists, for every possible search term, the documents that contain that term.
type InvertedIndex struct {
cfg Config
index map[string]map[Keyword][]Document
// Counts the number of documents added to each index.
indexDocCount map[string]int
minWeight int
maxWeight int
// No modifications after this is set.
finalized bool
}
func (idx *InvertedIndex) getIndexCfg(name string) (IndexConfig, bool) {
for _, conf := range idx.cfg.Indices {
if conf.Name == name {
return conf, true
}
}
return IndexConfig{}, false
}
// NewInvertedIndex creates a new InvertedIndex.
// Documents to index must be added in Add.
func NewInvertedIndex(cfg Config) *InvertedIndex {
idx := &InvertedIndex{index: make(map[string]map[Keyword][]Document), indexDocCount: make(map[string]int), cfg: cfg}
for _, conf := range cfg.Indices {
idx.index[conf.Name] = make(map[Keyword][]Document)
if conf.Weight < idx.minWeight {
// By default, the weight scale starts at 0, but we allow
// negative weights.
idx.minWeight = conf.Weight
}
if conf.Weight > idx.maxWeight {
idx.maxWeight = conf.Weight
}
}
return idx
}
// Add documents to the inverted index.
// The value must support == and !=.
func (idx *InvertedIndex) Add(ctx context.Context, docs ...Document) error {
if idx.finalized {
panic("index is finalized")
}
var err error
for _, config := range idx.cfg.Indices {
if config.Weight == 0 {
// Disabled
continue
}
setm := idx.index[config.Name]
for _, doc := range docs {
var added bool
var words []Keyword
words, err = doc.RelatedKeywords(config)
if err != nil {
continue
}
for _, keyword := range words {
added = true
setm[keyword] = append(setm[keyword], doc)
}
if config.Type == TypeFragments {
if fp, ok := doc.(FragmentProvider); ok {
for _, fragment := range fp.Fragments(ctx).Identifiers {
added = true
setm[FragmentKeyword(fragment)] = append(setm[FragmentKeyword(fragment)], doc)
}
}
}
if added {
idx.indexDocCount[config.Name]++
}
}
}
return err
}
func (idx *InvertedIndex) Finalize(ctx context.Context) error {
if idx.finalized {
return nil
}
for _, config := range idx.cfg.Indices {
if config.CardinalityThreshold == 0 {
continue
}
setm := idx.index[config.Name]
if idx.indexDocCount[config.Name] == 0 {
continue
}
// Remove high cardinality terms.
numDocs := idx.indexDocCount[config.Name]
for k, v := range setm {
percentageWithKeyword := int(math.Ceil(float64(len(v)) / float64(numDocs) * 100))
if percentageWithKeyword > config.CardinalityThreshold {
delete(setm, k)
}
}
}
idx.finalized = true
return nil
}
// queryElement holds the index name and keywords that can be used to compose a
// search for related content.
type queryElement struct {
Index string
Keywords []Keyword
}
func newQueryElement(index string, keywords ...Keyword) queryElement {
return queryElement{Index: index, Keywords: keywords}
}
type ranks []*rank
type rank struct {
Doc Document
Weight int
Matches int
}
func (r *rank) addWeight(w int) {
r.Weight += w
r.Matches++
}
var rankPool = sync.Pool{
New: func() interface{} {
return &rank{}
},
}
func getRank(doc Document, weight int) *rank {
r := rankPool.Get().(*rank)
r.Doc = doc
r.Weight = weight
r.Matches = 1
return r
}
func putRank(r *rank) {
rankPool.Put(r)
}
func (r ranks) Len() int { return len(r) }
func (r ranks) Swap(i, j int) { r[i], r[j] = r[j], r[i] }
func (r ranks) Less(i, j int) bool {
if r[i].Weight == r[j].Weight {
if r[i].Doc.PublishDate() == r[j].Doc.PublishDate() {
return r[i].Doc.Name() < r[j].Doc.Name()
}
return r[i].Doc.PublishDate().After(r[j].Doc.PublishDate())
}
return r[i].Weight > r[j].Weight
}
// SearchOpts holds the options for a related search.
type SearchOpts struct {
// The Document to search for related content for.
Document Document
// The keywords to search for.
NamedSlices []types.KeyValues
// The indices to search in.
Indices []string
// Fragments holds a a list of special keywords that is used
// for indices configured as type "fragments".
// This will match the fragment identifiers of the documents.
Fragments []string
}
// Search finds the documents matching any of the keywords in the given indices
// against query options in opts.
// The resulting document set will be sorted according to number of matches
// and the index weights, and any matches with a rank below the configured
// threshold (normalize to 0..100) will be removed.
// If an index name is provided, only that index will be queried.
func (idx *InvertedIndex) Search(ctx context.Context, opts SearchOpts) ([]Document, error) {
var (
queryElements []queryElement
configs IndicesConfig
)
if len(opts.Indices) == 0 {
configs = idx.cfg.Indices
} else {
configs = make(IndicesConfig, len(opts.Indices))
for i, indexName := range opts.Indices {
cfg, found := idx.getIndexCfg(indexName)
if !found {
return nil, fmt.Errorf("index %q not found", indexName)
}
configs[i] = cfg
}
}
for _, cfg := range configs {
var keywords []Keyword
if opts.Document != nil {
k, err := opts.Document.RelatedKeywords(cfg)
if err != nil {
return nil, err
}
keywords = append(keywords, k...)
}
if cfg.Type == TypeFragments {
for _, fragment := range opts.Fragments {
keywords = append(keywords, FragmentKeyword(fragment))
}
if opts.Document != nil {
if fp, ok := opts.Document.(FragmentProvider); ok {
for _, fragment := range fp.Fragments(ctx).Identifiers {
keywords = append(keywords, FragmentKeyword(fragment))
}
}
}
}
queryElements = append(queryElements, newQueryElement(cfg.Name, keywords...))
}
for _, slice := range opts.NamedSlices {
var keywords []Keyword
key := slice.KeyString()
if key == "" {
return nil, fmt.Errorf("index %q not valid", slice.Key)
}
conf, found := idx.getIndexCfg(key)
if !found {
return nil, fmt.Errorf("index %q not found", key)
}
for _, val := range slice.Values {
k, err := conf.ToKeywords(val)
if err != nil {
return nil, err
}
keywords = append(keywords, k...)
}
queryElements = append(queryElements, newQueryElement(conf.Name, keywords...))
}
if opts.Document != nil {
return idx.searchDate(ctx, opts.Document, opts.Document.PublishDate(), queryElements...)
}
return idx.search(ctx, queryElements...)
}
func (cfg IndexConfig) stringToKeyword(s string) Keyword {
if cfg.ToLower {
s = strings.ToLower(s)
}
if cfg.Type == TypeFragments {
return FragmentKeyword(s)
}
return StringKeyword(s)
}
// ToKeywords returns a Keyword slice of the given input.
func (cfg IndexConfig) ToKeywords(v any) ([]Keyword, error) {
var keywords []Keyword
switch vv := v.(type) {
case string:
keywords = append(keywords, cfg.stringToKeyword(vv))
case []string:
vvv := make([]Keyword, len(vv))
for i := 0; i < len(vvv); i++ {
vvv[i] = cfg.stringToKeyword(vv[i])
}
keywords = append(keywords, vvv...)
case []any:
return cfg.ToKeywords(cast.ToStringSlice(vv))
case time.Time:
layout := "2006"
if cfg.Pattern != "" {
layout = cfg.Pattern
}
keywords = append(keywords, StringKeyword(vv.Format(layout)))
case nil:
return keywords, nil
default:
return keywords, fmt.Errorf("indexing currently not supported for index %q and type %T", cfg.Name, vv)
}
return keywords, nil
}
func (idx *InvertedIndex) search(ctx context.Context, query ...queryElement) ([]Document, error) {
return idx.searchDate(ctx, nil, zeroDate, query...)
}
func (idx *InvertedIndex) searchDate(ctx context.Context, self Document, upperDate time.Time, query ...queryElement) ([]Document, error) {
matchm := make(map[Document]*rank, 200)
defer func() {
for _, r := range matchm {
putRank(r)
}
}()
applyDateFilter := !idx.cfg.IncludeNewer && !upperDate.IsZero()
var fragmentsFilter collections.SortedStringSlice
for _, el := range query {
setm, found := idx.index[el.Index]
if !found {
return []Document{}, fmt.Errorf("index for %q not found", el.Index)
}
config, found := idx.getIndexCfg(el.Index)
if !found {
return []Document{}, fmt.Errorf("index config for %q not found", el.Index)
}
for _, kw := range el.Keywords {
if docs, found := setm[kw]; found {
for _, doc := range docs {
if compare.Eq(doc, self) {
continue
}
if applyDateFilter {
// Exclude newer than the limit given
if doc.PublishDate().After(upperDate) {
continue
}
}
if config.Type == TypeFragments && config.ApplyFilter {
if fkw, ok := kw.(FragmentKeyword); ok {
fragmentsFilter = append(fragmentsFilter, string(fkw))
}
}
r, found := matchm[doc]
if !found {
r = getRank(doc, config.Weight)
matchm[doc] = r
} else {
r.addWeight(config.Weight)
}
}
}
}
}
if len(matchm) == 0 {
return []Document{}, nil
}
matches := make(ranks, 0, 100)
for _, v := range matchm {
avgWeight := v.Weight / v.Matches
weight := norm(avgWeight, idx.minWeight, idx.maxWeight)
threshold := idx.cfg.Threshold / v.Matches
if weight >= threshold {
matches = append(matches, v)
}
}
sort.Stable(matches)
sort.Strings(fragmentsFilter)
result := make([]Document, len(matches))
for i, m := range matches {
result[i] = m.Doc
if len(fragmentsFilter) > 0 {
if dp, ok := result[i].(FragmentProvider); ok {
result[i] = dp.ApplyFilterToHeadings(ctx, func(h *tableofcontents.Heading) bool {
return fragmentsFilter.Contains(h.ID)
})
}
}
}
return result, nil
}
// normalizes num to a number between 0 and 100.
func norm(num, min, max int) int {
if min > max {
panic("min > max")
}
return int(math.Floor((float64(num-min) / float64(max-min) * 100) + 0.5))
}
// DecodeConfig decodes a slice of map into Config.
func DecodeConfig(m maps.Params) (Config, error) {
if m == nil {
return Config{}, errors.New("no related config provided")
}
if len(m) == 0 {
return Config{}, errors.New("empty related config provided")
}
var c Config
if err := mapstructure.WeakDecode(m, &c); err != nil {
return c, err
}
if c.Threshold < 0 || c.Threshold > 100 {
return Config{}, errors.New("related threshold must be between 0 and 100")
}
if c.ToLower {
for i := range c.Indices {
c.Indices[i].ToLower = true
}
}
for i := range c.Indices {
icfg := c.Indices[i]
if icfg.Type == "" {
c.Indices[i].Type = TypeBasic
}
if !validTypes[c.Indices[i].Type] {
return c, fmt.Errorf("invalid index type %q. Must be one of %v", c.Indices[i].Type, xmaps.Keys(validTypes))
}
if icfg.CardinalityThreshold < 0 || icfg.CardinalityThreshold > 100 {
return Config{}, errors.New("cardinalityThreshold threshold must be between 0 and 100")
}
}
return c, nil
}
// StringKeyword is a string search keyword.
type StringKeyword string
func (s StringKeyword) String() string {
return string(s)
}
// FragmentKeyword represents a document fragment.
type FragmentKeyword string
func (f FragmentKeyword) String() string {
return string(f)
}
// Keyword is the interface a keyword in the search index must implement.
type Keyword interface {
String() string
}
// StringsToKeywords converts the given slice of strings to a slice of Keyword.
func (cfg IndexConfig) StringsToKeywords(s ...string) []Keyword {
kw := make([]Keyword, len(s))
for i := 0; i < len(s); i++ {
kw[i] = cfg.stringToKeyword(s[i])
}
return kw
}