ref: d6ed17c60fd13c933347110c720afa09e04df1c8
dir: /hugolib/content_map.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 hugolib import ( "fmt" "path" "path/filepath" "strings" "sync" "github.com/gohugoio/hugo/helpers" "github.com/gohugoio/hugo/resources/page" "github.com/pkg/errors" "github.com/gohugoio/hugo/hugofs/files" "github.com/gohugoio/hugo/hugofs" radix "github.com/armon/go-radix" ) // We store the branch nodes in either the `sections` or `taxonomies` tree // with their path as a key; Unix style slashes, a leading and trailing slash. // // E.g. "/blog/" or "/categories/funny/" // // Pages that belongs to a section are stored in the `pages` tree below // the section name and a branch separator, e.g. "/blog/__hb_". A page is // given a key using the path below the section and the base filename with no extension // with a leaf separator added. // // For bundled pages (/mybundle/index.md), we use the folder name. // // An exmple of a full page key would be "/blog/__hb_page1__hl_" // // Bundled resources are stored in the `resources` having their path prefixed // with the bundle they belong to, e.g. // "/blog/__hb_bundle__hl_data.json". // // The weighted taxonomy entries extracted from page front matter are stored in // the `taxonomyEntries` tree below /plural/term/page-key, e.g. // "/categories/funny/blog/__hb_bundle__hl_". const ( cmBranchSeparator = "__hb_" cmLeafSeparator = "__hl_" ) // Used to mark ambigous keys in reverse index lookups. var ambigousContentNode = &contentNode{} func newContentMap(cfg contentMapConfig) *contentMap { m := &contentMap{ cfg: &cfg, pages: &contentTree{Name: "pages", Tree: radix.New()}, sections: &contentTree{Name: "sections", Tree: radix.New()}, taxonomies: &contentTree{Name: "taxonomies", Tree: radix.New()}, taxonomyEntries: &contentTree{Name: "taxonomyEntries", Tree: radix.New()}, resources: &contentTree{Name: "resources", Tree: radix.New()}, } m.pageTrees = []*contentTree{ m.pages, m.sections, m.taxonomies, } m.bundleTrees = []*contentTree{ m.pages, m.sections, m.taxonomies, m.resources, } m.branchTrees = []*contentTree{ m.sections, m.taxonomies, } addToReverseMap := func(k string, n *contentNode, m map[interface{}]*contentNode) { k = strings.ToLower(k) existing, found := m[k] if found && existing != ambigousContentNode { m[k] = ambigousContentNode } else if !found { m[k] = n } } m.pageReverseIndex = &contentTreeReverseIndex{ t: []*contentTree{m.pages, m.sections, m.taxonomies}, initFn: func(t *contentTree, m map[interface{}]*contentNode) { t.Walk(func(s string, v interface{}) bool { n := v.(*contentNode) if n.p != nil && !n.p.File().IsZero() { meta := n.p.File().FileInfo().Meta() if meta.Path() != meta.PathFile() { // Keep track of the original mount source. mountKey := filepath.ToSlash(filepath.Join(meta.Module(), meta.PathFile())) addToReverseMap(mountKey, n, m) } } k := strings.TrimPrefix(strings.TrimSuffix(path.Base(s), cmLeafSeparator), cmBranchSeparator) addToReverseMap(k, n, m) return false }) }, } return m } type cmInsertKeyBuilder struct { m *contentMap err error // Builder state tree *contentTree baseKey string // Section or page key key string } func (b cmInsertKeyBuilder) ForPage(s string) *cmInsertKeyBuilder { //fmt.Println("ForPage:", s, "baseKey:", b.baseKey, "key:", b.key) baseKey := b.baseKey b.baseKey = s if baseKey != "/" { // Don't repeat the section path in the key. s = strings.TrimPrefix(s, baseKey) } s = strings.TrimPrefix(s, "/") switch b.tree { case b.m.sections: b.tree = b.m.pages b.key = baseKey + cmBranchSeparator + s + cmLeafSeparator case b.m.taxonomies: b.key = path.Join(baseKey, s) default: panic("invalid state") } return &b } func (b cmInsertKeyBuilder) ForResource(s string) *cmInsertKeyBuilder { //fmt.Println("ForResource:", s, "baseKey:", b.baseKey, "key:", b.key) baseKey := helpers.AddTrailingSlash(b.baseKey) s = strings.TrimPrefix(s, baseKey) switch b.tree { case b.m.pages: b.key = b.key + s case b.m.sections, b.m.taxonomies: b.key = b.key + cmLeafSeparator + s default: panic(fmt.Sprintf("invalid state: %#v", b.tree)) } b.tree = b.m.resources return &b } func (b *cmInsertKeyBuilder) Insert(n *contentNode) *cmInsertKeyBuilder { if b.err == nil { b.tree.Insert(b.Key(), n) } return b } func (b *cmInsertKeyBuilder) Key() string { switch b.tree { case b.m.sections, b.m.taxonomies: return cleanSectionTreeKey(b.key) default: return cleanTreeKey(b.key) } } func (b *cmInsertKeyBuilder) DeleteAll() *cmInsertKeyBuilder { if b.err == nil { b.tree.DeletePrefix(b.Key()) } return b } func (b *cmInsertKeyBuilder) WithFile(fi hugofs.FileMetaInfo) *cmInsertKeyBuilder { b.newTopLevel() m := b.m meta := fi.Meta() p := cleanTreeKey(meta.Path()) bundlePath := m.getBundleDir(meta) isBundle := meta.Classifier().IsBundle() if isBundle { panic("not implemented") } p, k := b.getBundle(p) if k == "" { b.err = errors.Errorf("no bundle header found for %q", bundlePath) return b } id := k + m.reduceKeyPart(p, fi.Meta().Path()) b.tree = b.m.resources b.key = id b.baseKey = p return b } func (b *cmInsertKeyBuilder) WithSection(s string) *cmInsertKeyBuilder { s = cleanSectionTreeKey(s) b.newTopLevel() b.tree = b.m.sections b.baseKey = s b.key = s return b } func (b *cmInsertKeyBuilder) WithTaxonomy(s string) *cmInsertKeyBuilder { s = cleanSectionTreeKey(s) b.newTopLevel() b.tree = b.m.taxonomies b.baseKey = s b.key = s return b } // getBundle gets both the key to the section and the prefix to where to store // this page bundle and its resources. func (b *cmInsertKeyBuilder) getBundle(s string) (string, string) { m := b.m section, _ := m.getSection(s) p := strings.TrimPrefix(s, section) bundlePathParts := strings.Split(p, "/") basePath := section + cmBranchSeparator // Put it into an existing bundle if found. for i := len(bundlePathParts) - 2; i >= 0; i-- { bundlePath := path.Join(bundlePathParts[:i]...) searchKey := basePath + bundlePath + cmLeafSeparator if _, found := m.pages.Get(searchKey); found { return section + bundlePath, searchKey } } // Put it into the section bundle. return section, section + cmLeafSeparator } func (b *cmInsertKeyBuilder) newTopLevel() { b.key = "" } type contentBundleViewInfo struct { name viewName termKey string termOrigin string weight int ref *contentNode } func (c *contentBundleViewInfo) kind() string { if c.termKey != "" { return page.KindTaxonomy } return page.KindTaxonomyTerm } func (c *contentBundleViewInfo) sections() []string { if c.kind() == page.KindTaxonomyTerm { return []string{c.name.plural} } return []string{c.name.plural, c.termKey} } func (c *contentBundleViewInfo) term() string { if c.termOrigin != "" { return c.termOrigin } return c.termKey } type contentMap struct { cfg *contentMapConfig // View of regular pages, sections, and taxonomies. pageTrees contentTrees // View of pages, sections, taxonomies, and resources. bundleTrees contentTrees // View of sections and taxonomies. branchTrees contentTrees // Stores page bundles keyed by its path's directory or the base filename, // e.g. "blog/post.md" => "/blog/post", "blog/post/index.md" => "/blog/post" // These are the "regular pages" and all of them are bundles. pages *contentTree // A reverse index used as a fallback in GetPage. // There are currently two cases where this is used: // 1. Short name lookups in ref/relRef, e.g. using only "mypage.md" without a path. // 2. Links resolved from a remounted content directory. These are restricted to the same module. // Both of the above cases can result in ambigous lookup errors. pageReverseIndex *contentTreeReverseIndex // Section nodes. sections *contentTree // Taxonomy nodes. taxonomies *contentTree // Pages in a taxonomy. taxonomyEntries *contentTree // Resources stored per bundle below a common prefix, e.g. "/blog/post__hb_". resources *contentTree } func (m *contentMap) AddFiles(fis ...hugofs.FileMetaInfo) error { for _, fi := range fis { if err := m.addFile(fi); err != nil { return err } } return nil } func (m *contentMap) AddFilesBundle(header hugofs.FileMetaInfo, resources ...hugofs.FileMetaInfo) error { var ( meta = header.Meta() classifier = meta.Classifier() isBranch = classifier == files.ContentClassBranch bundlePath = m.getBundleDir(meta) n = m.newContentNodeFromFi(header) b = m.newKeyBuilder() section string ) if isBranch { // Either a section or a taxonomy node. section = bundlePath if tc := m.cfg.getTaxonomyConfig(section); !tc.IsZero() { term := strings.TrimPrefix(strings.TrimPrefix(section, "/"+tc.plural), "/") n.viewInfo = &contentBundleViewInfo{ name: tc, termKey: term, termOrigin: term, } n.viewInfo.ref = n b.WithTaxonomy(section).Insert(n) } else { b.WithSection(section).Insert(n) } } else { // A regular page. Attach it to its section. section, _ = m.getOrCreateSection(n, bundlePath) b = b.WithSection(section).ForPage(bundlePath).Insert(n) } if m.cfg.isRebuild { // The resource owner will be either deleted or overwritten on rebuilds, // but make sure we handle deletion of resources (images etc.) as well. b.ForResource("").DeleteAll() } for _, r := range resources { rb := b.ForResource(cleanTreeKey(r.Meta().Path())) rb.Insert(&contentNode{fi: r}) } return nil } func (m *contentMap) CreateMissingNodes() error { // Create missing home and root sections rootSections := make(map[string]interface{}) trackRootSection := func(s string, b *contentNode) { parts := strings.Split(s, "/") if len(parts) > 2 { root := strings.TrimSuffix(parts[1], cmBranchSeparator) if root != "" { if _, found := rootSections[root]; !found { rootSections[root] = b } } } } m.sections.Walk(func(s string, v interface{}) bool { n := v.(*contentNode) if s == "/" { return false } trackRootSection(s, n) return false }) m.pages.Walk(func(s string, v interface{}) bool { trackRootSection(s, v.(*contentNode)) return false }) if _, found := rootSections["/"]; !found { rootSections["/"] = true } for sect, v := range rootSections { var sectionPath string if n, ok := v.(*contentNode); ok && n.path != "" { sectionPath = n.path firstSlash := strings.Index(sectionPath, "/") if firstSlash != -1 { sectionPath = sectionPath[:firstSlash] } } sect = cleanSectionTreeKey(sect) _, found := m.sections.Get(sect) if !found { m.sections.Insert(sect, &contentNode{path: sectionPath}) } } for _, view := range m.cfg.taxonomyConfig { s := cleanSectionTreeKey(view.plural) _, found := m.taxonomies.Get(s) if !found { b := &contentNode{ viewInfo: &contentBundleViewInfo{ name: view, }, } b.viewInfo.ref = b m.taxonomies.Insert(s, b) } } return nil } func (m *contentMap) getBundleDir(meta hugofs.FileMeta) string { dir := cleanTreeKey(filepath.Dir(meta.Path())) switch meta.Classifier() { case files.ContentClassContent: return path.Join(dir, meta.TranslationBaseName()) default: return dir } } func (m *contentMap) newContentNodeFromFi(fi hugofs.FileMetaInfo) *contentNode { return &contentNode{ fi: fi, path: strings.TrimPrefix(filepath.ToSlash(fi.Meta().Path()), "/"), } } func (m *contentMap) getFirstSection(s string) (string, *contentNode) { s = helpers.AddTrailingSlash(s) for { k, v, found := m.sections.LongestPrefix(s) if !found { return "", nil } if strings.Count(k, "/") <= 2 { return k, v.(*contentNode) } s = helpers.AddTrailingSlash(path.Dir(strings.TrimSuffix(s, "/"))) } } func (m *contentMap) newKeyBuilder() *cmInsertKeyBuilder { return &cmInsertKeyBuilder{m: m} } func (m *contentMap) getOrCreateSection(n *contentNode, s string) (string, *contentNode) { level := strings.Count(s, "/") k, b := m.getSection(s) mustCreate := false if k == "" { mustCreate = true } else if level > 1 && k == "/" { // We found the home section, but this page needs to be placed in // the root, e.g. "/blog", section. mustCreate = true } if mustCreate { k = cleanSectionTreeKey(s[:strings.Index(s[1:], "/")+1]) b = &contentNode{ path: n.rootSection(), } m.sections.Insert(k, b) } return k, b } func (m *contentMap) getPage(section, name string) *contentNode { section = helpers.AddTrailingSlash(section) key := section + cmBranchSeparator + name + cmLeafSeparator v, found := m.pages.Get(key) if found { return v.(*contentNode) } return nil } func (m *contentMap) getSection(s string) (string, *contentNode) { s = helpers.AddTrailingSlash(path.Dir(strings.TrimSuffix(s, "/"))) k, v, found := m.sections.LongestPrefix(s) if found { return k, v.(*contentNode) } return "", nil } func (m *contentMap) getTaxonomyParent(s string) (string, *contentNode) { s = helpers.AddTrailingSlash(path.Dir(strings.TrimSuffix(s, "/"))) k, v, found := m.taxonomies.LongestPrefix(s) if found { return k, v.(*contentNode) } v, found = m.sections.Get("/") if found { return s, v.(*contentNode) } return "", nil } func (m *contentMap) addFile(fi hugofs.FileMetaInfo) error { b := m.newKeyBuilder() return b.WithFile(fi).Insert(m.newContentNodeFromFi(fi)).err } func cleanTreeKey(k string) string { k = "/" + strings.ToLower(strings.Trim(path.Clean(filepath.ToSlash(k)), "./")) return k } func cleanSectionTreeKey(k string) string { k = cleanTreeKey(k) if k != "/" { k += "/" } return k } func (m *contentMap) onSameLevel(s1, s2 string) bool { return strings.Count(s1, "/") == strings.Count(s2, "/") } func (m *contentMap) deleteBundleMatching(matches func(b *contentNode) bool) { // Check sections first s := m.sections.getMatch(matches) if s != "" { m.deleteSectionByPath(s) return } s = m.pages.getMatch(matches) if s != "" { m.deletePage(s) return } s = m.resources.getMatch(matches) if s != "" { m.resources.Delete(s) } } // Deletes any empty root section that's not backed by a content file. func (m *contentMap) deleteOrphanSections() { var sectionsToDelete []string m.sections.Walk(func(s string, v interface{}) bool { n := v.(*contentNode) if n.fi != nil { // Section may be empty, but is backed by a content file. return false } if s == "/" || strings.Count(s, "/") > 2 { return false } prefixBundle := s + cmBranchSeparator if !(m.sections.hasBelow(s) || m.pages.hasBelow(prefixBundle) || m.resources.hasBelow(prefixBundle)) { sectionsToDelete = append(sectionsToDelete, s) } return false }) for _, s := range sectionsToDelete { m.sections.Delete(s) } } func (m *contentMap) deletePage(s string) { m.pages.DeletePrefix(s) m.resources.DeletePrefix(s) } func (m *contentMap) deleteSectionByPath(s string) { if !strings.HasSuffix(s, "/") { panic("section must end with a slash") } if !strings.HasPrefix(s, "/") { panic("section must start with a slash") } m.sections.DeletePrefix(s) m.pages.DeletePrefix(s) m.resources.DeletePrefix(s) } func (m *contentMap) deletePageByPath(s string) { m.pages.Walk(func(s string, v interface{}) bool { fmt.Println("S", s) return false }) } func (m *contentMap) deleteTaxonomy(s string) { m.taxonomies.DeletePrefix(s) } func (m *contentMap) reduceKeyPart(dir, filename string) string { dir, filename = filepath.ToSlash(dir), filepath.ToSlash(filename) dir, filename = strings.TrimPrefix(dir, "/"), strings.TrimPrefix(filename, "/") return strings.TrimPrefix(strings.TrimPrefix(filename, dir), "/") } func (m *contentMap) splitKey(k string) []string { if k == "" || k == "/" { return nil } return strings.Split(k, "/")[1:] } func (m *contentMap) testDump() string { var sb strings.Builder for i, r := range []*contentTree{m.pages, m.sections, m.resources} { sb.WriteString(fmt.Sprintf("Tree %d:\n", i)) r.Walk(func(s string, v interface{}) bool { sb.WriteString("\t" + s + "\n") return false }) } for i, r := range []*contentTree{m.pages, m.sections} { r.Walk(func(s string, v interface{}) bool { c := v.(*contentNode) cpToString := func(c *contentNode) string { var sb strings.Builder if c.p != nil { sb.WriteString("|p:" + c.p.Title()) } if c.fi != nil { sb.WriteString("|f:" + filepath.ToSlash(c.fi.Meta().Path())) } return sb.String() } sb.WriteString(path.Join(m.cfg.lang, r.Name) + s + cpToString(c) + "\n") resourcesPrefix := s if i == 1 { resourcesPrefix += cmLeafSeparator m.pages.WalkPrefix(s+cmBranchSeparator, func(s string, v interface{}) bool { sb.WriteString("\t - P: " + filepath.ToSlash((v.(*contentNode).fi.(hugofs.FileMetaInfo)).Meta().Filename()) + "\n") return false }) } m.resources.WalkPrefix(resourcesPrefix, func(s string, v interface{}) bool { sb.WriteString("\t - R: " + filepath.ToSlash((v.(*contentNode).fi.(hugofs.FileMetaInfo)).Meta().Filename()) + "\n") return false }) return false }) } return sb.String() } type contentMapConfig struct { lang string taxonomyConfig []viewName taxonomyDisabled bool taxonomyTermDisabled bool pageDisabled bool isRebuild bool } func (cfg contentMapConfig) getTaxonomyConfig(s string) (v viewName) { s = strings.TrimPrefix(s, "/") if s == "" { return } for _, n := range cfg.taxonomyConfig { if strings.HasPrefix(s, n.plural) { return n } } return } type contentNode struct { p *pageState // Set for taxonomy nodes. viewInfo *contentBundleViewInfo // Set if source is a file. // We will soon get other sources. fi hugofs.FileMetaInfo // The source path. Unix slashes. No leading slash. path string } func (b *contentNode) rootSection() string { if b.path == "" { return "" } firstSlash := strings.Index(b.path, "/") if firstSlash == -1 { return b.path } return b.path[:firstSlash] } type contentTree struct { Name string *radix.Tree } type contentTrees []*contentTree func (t contentTrees) DeletePrefix(prefix string) int { var count int for _, tree := range t { tree.Walk(func(s string, v interface{}) bool { return false }) count += tree.DeletePrefix(prefix) } return count } type contentTreeNodeCallback func(s string, n *contentNode) bool func newContentTreeFilter(fn func(n *contentNode) bool) contentTreeNodeCallback { return func(s string, n *contentNode) bool { return fn(n) } } var ( contentTreeNoListAlwaysFilter = func(s string, n *contentNode) bool { if n.p == nil { return true } return n.p.m.noListAlways() } contentTreeNoRenderFilter = func(s string, n *contentNode) bool { if n.p == nil { return true } return n.p.m.noRender() } ) func (c *contentTree) WalkQuery(query pageMapQuery, walkFn contentTreeNodeCallback) { filter := query.Filter if filter == nil { filter = contentTreeNoListAlwaysFilter } if query.Prefix != "" { c.WalkBelow(query.Prefix, func(s string, v interface{}) bool { n := v.(*contentNode) if filter != nil && filter(s, n) { return false } return walkFn(s, n) }) return } c.Walk(func(s string, v interface{}) bool { n := v.(*contentNode) if filter != nil && filter(s, n) { return false } return walkFn(s, n) }) } func (c contentTrees) WalkRenderable(fn contentTreeNodeCallback) { query := pageMapQuery{Filter: contentTreeNoRenderFilter} for _, tree := range c { tree.WalkQuery(query, fn) } } func (c contentTrees) Walk(fn contentTreeNodeCallback) { for _, tree := range c { tree.Walk(func(s string, v interface{}) bool { n := v.(*contentNode) return fn(s, n) }) } } func (c contentTrees) WalkPrefix(prefix string, fn contentTreeNodeCallback) { for _, tree := range c { tree.WalkPrefix(prefix, func(s string, v interface{}) bool { n := v.(*contentNode) return fn(s, n) }) } } // WalkBelow walks the tree below the given prefix, i.e. it skips the // node with the given prefix as key. func (c *contentTree) WalkBelow(prefix string, fn radix.WalkFn) { c.Tree.WalkPrefix(prefix, func(s string, v interface{}) bool { if s == prefix { return false } return fn(s, v) }) } func (c *contentTree) getMatch(matches func(b *contentNode) bool) string { var match string c.Walk(func(s string, v interface{}) bool { n, ok := v.(*contentNode) if !ok { return false } if matches(n) { match = s return true } return false }) return match } func (c *contentTree) hasBelow(s1 string) bool { var t bool c.WalkBelow(s1, func(s2 string, v interface{}) bool { t = true return true }) return t } func (c *contentTree) printKeys() { c.Walk(func(s string, v interface{}) bool { fmt.Println(s) return false }) } func (c *contentTree) printKeysPrefix(prefix string) { c.WalkPrefix(prefix, func(s string, v interface{}) bool { fmt.Println(s) return false }) } // contentTreeRef points to a node in the given tree. type contentTreeRef struct { m *pageMap t *contentTree n *contentNode key string } func (c *contentTreeRef) getCurrentSection() (string, *contentNode) { if c.isSection() { return c.key, c.n } return c.getSection() } func (c *contentTreeRef) isSection() bool { return c.t == c.m.sections } func (c *contentTreeRef) getSection() (string, *contentNode) { if c.t == c.m.taxonomies { return c.m.getTaxonomyParent(c.key) } return c.m.getSection(c.key) } func (c *contentTreeRef) getPages() page.Pages { var pas page.Pages c.m.collectPages( pageMapQuery{ Prefix: c.key + cmBranchSeparator, Filter: c.n.p.m.getListFilter(true), }, func(c *contentNode) { pas = append(pas, c.p) }, ) page.SortByDefault(pas) return pas } func (c *contentTreeRef) getPagesRecursive() page.Pages { var pas page.Pages query := pageMapQuery{ Filter: c.n.p.m.getListFilter(true), } query.Prefix = c.key c.m.collectPages(query, func(c *contentNode) { pas = append(pas, c.p) }) page.SortByDefault(pas) return pas } func (c *contentTreeRef) getPagesAndSections() page.Pages { var pas page.Pages query := pageMapQuery{ Filter: c.n.p.m.getListFilter(true), Prefix: c.key, } c.m.collectPagesAndSections(query, func(c *contentNode) { pas = append(pas, c.p) }) page.SortByDefault(pas) return pas } func (c *contentTreeRef) getSections() page.Pages { var pas page.Pages query := pageMapQuery{ Filter: c.n.p.m.getListFilter(true), Prefix: c.key, } c.m.collectSections(query, func(c *contentNode) { pas = append(pas, c.p) }) page.SortByDefault(pas) return pas } type contentTreeReverseIndex struct { t []*contentTree m map[interface{}]*contentNode init sync.Once initFn func(*contentTree, map[interface{}]*contentNode) } func (c *contentTreeReverseIndex) Get(key interface{}) *contentNode { c.init.Do(func() { c.m = make(map[interface{}]*contentNode) for _, tree := range c.t { c.initFn(tree, c.m) } }) return c.m[key] }