// Copyright 2016 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" ) // PageCollections contains the page collections for a site. type PageCollections struct { // Includes only pages of all types, and only pages in the current language. Pages Pages // Includes all pages in all languages, including the current one. // Inlcudes pages of all types. AllPages Pages // A convenience cache for the traditional index types, taxonomies, home page etc. // This is for the current language only. indexPages Pages // A convenience cache for the regular pages. // This is for the current language only. // TODO(bep) np consider exporting this regularPages Pages // Includes absolute all pages (of all types), including drafts etc. rawAllPages Pages } func (c *PageCollections) refreshPageCaches() { c.indexPages = c.findPagesByNodeTypeNotIn(KindPage, c.Pages) c.regularPages = c.findPagesByNodeTypeIn(KindPage, c.Pages) // TODO(bep) np remove eventually for _, n := range c.Pages { if n.Kind == kindUnknown { panic(fmt.Sprintf("Got unknown type %s", n.Title)) } } } func newPageCollections() *PageCollections { return &PageCollections{} } func newPageCollectionsFromPages(pages Pages) *PageCollections { return &PageCollections{rawAllPages: pages} } // TODO(bep) np clean and remove finders func (c *PageCollections) findPagesByNodeType(n Kind) Pages { return c.findPagesByNodeTypeIn(n, c.Pages) } func (c *PageCollections) getPage(typ Kind, path ...string) *Page { pages := c.findPagesByNodeTypeIn(typ, c.Pages) if len(pages) == 0 { return nil } if len(path) == 0 && len(pages) == 1 { return pages[0] } for _, p := range pages { match := false for i := 0; i < len(path); i++ { if len(p.sections) > i && path[i] == p.sections[i] { match = true } else { match = false break } } if match { return p } } return nil } func (c *PageCollections) findIndexNodesByNodeType(n Kind) Pages { return c.findPagesByNodeTypeIn(n, c.indexPages) } func (*PageCollections) findPagesByNodeTypeIn(n Kind, inPages Pages) Pages { var pages Pages for _, p := range inPages { if p.Kind == n { pages = append(pages, p) } } return pages } func (*PageCollections) findPagesByNodeTypeNotIn(n Kind, inPages Pages) Pages { var pages Pages for _, p := range inPages { if p.Kind != n { pages = append(pages, p) } } return pages } func (c *PageCollections) findAllPagesByNodeType(n Kind) Pages { return c.findPagesByNodeTypeIn(n, c.Pages) } func (c *PageCollections) findRawAllPagesByNodeType(n Kind) Pages { return c.findPagesByNodeTypeIn(n, c.rawAllPages) } func (c *PageCollections) addPage(page *Page) { c.rawAllPages = append(c.rawAllPages, page) } func (c *PageCollections) removePageByPath(path string) { if i := c.rawAllPages.FindPagePosByFilePath(path); i >= 0 { c.rawAllPages = append(c.rawAllPages[:i], c.rawAllPages[i+1:]...) } } func (c *PageCollections) removePage(page *Page) { if i := c.rawAllPages.FindPagePos(page); i >= 0 { c.rawAllPages = append(c.rawAllPages[:i], c.rawAllPages[i+1:]...) } } func (c *PageCollections) replacePage(page *Page) { // will find existing page that matches filepath and remove it c.removePage(page) c.addPage(page) }