wiki/search.go

package main

import (
	"encoding/json"
	"fmt"
	"log"
	"net/http"
	"os"
	"strings"

	"github.com/blevesearch/bleve"
	"github.com/blevesearch/bleve/mapping"
	"github.com/iancoleman/strcase"
)

// TODO: http handler
// TODO: index all pages on start
// TODO: reindex all command
// TODO: search(query) command

type searchHandler struct {
	indexMapping mapping.IndexMapping
	searchIndex  bleve.Index
}

type nameLine struct {
	Name  string `json:"name"`
	Title string `json:"title"`
	Line  string `json:"line"`
}

type searchObject struct {
	Title  string            `json:"title"`
	Blocks []string          `json:"blocks"`
	Refs   []nameLine        `json:"refs"`
	Meta   map[string]string `json:"meta"`
	Links  []ParsedLink      `json:"links"`
}

func NewSearchHandler(searchIndex bleve.Index) (http.Handler, error) {
	return &searchHandler{
		searchIndex: searchIndex,
	}, nil
}

func (s *searchHandler) ServeHTTP(w http.ResponseWriter, r *http.Request) {
	defer r.Body.Close()

	sess, err := NewSession(w, r)
	if err != nil {
		http.Error(w, err.Error(), 500)
		return
	}
	defer func() {
		if err := sess.Flush(); err != nil {
			log.Println(err)
		}
	}()
	if !sess.LoggedIn {
		fmt.Fprint(w, "{}")
		return
	}

	if r.URL.Query().Get("fields") == "1" {
		fields, err := s.searchIndex.Fields()
		if err != nil {
			http.Error(w, err.Error(), 500)
			return
		}
		enc := json.NewEncoder(w)
		enc.SetIndent("", "    ")
		err = enc.Encode(&fields)
		if err != nil {
			http.Error(w, err.Error(), 500)
			return
		}
		return
	} else if r.Method == "GET" && r.URL.Query().Get("reset") == "1" {
		w.Header().Add("Content-Type", "text/html")
		fmt.Fprint(w, `<form action="/search/" method="post"><input type=hidden name=reset value=1><input type=submit value="Reset Index"></form>`)
		return
	} else if r.Method == "POST" {
		err = r.ParseForm()
		if err != nil {
			http.Error(w, err.Error(), 500)
			return
		}
		if r.PostForm.Get("reset") == "1" {

			refs := make(Refs)
			mp := NewFilePages("data", nil)

			pages, err := mp.AllPages()
			if err != nil {
				http.Error(w, err.Error(), 500)
				return
			}

			for _, page := range pages {
				err = saveBlocksFromPage("data", page)
				if err != nil {
					log.Printf("error while processing blocks from page %s: %w", page.Name, err)
					continue
				}
			}

			// Reload all pages
			pages, err = mp.AllPages()
			if err != nil {
				http.Error(w, err.Error(), 500)
				return
			}

			for _, page := range pages {
				log.Println("processing ", page.Title)
				err = processBackrefsForPage(page, refs)
				if err != nil {
					log.Println("error while processing backrefs: ", err)
					continue
				}
			}

			log.Println("saveLinks")
			err = saveLinks(mp)
			if err != nil {
				log.Printf("error while saving links %w", err)
				http.Error(w, err.Error(), 500)
				return
			}

			log.Println("saveBackrefs")
			err = saveBackrefs("data/backrefs.json", refs)
			if err != nil {
				log.Printf("error while saving backrefs %w", err)
				http.Error(w, err.Error(), 500)
				return
			}

			err = os.RemoveAll("data/_tmp_index")
			if err != nil {
				log.Printf("error while remove old index %w", err)
				http.Error(w, err.Error(), 500)
				return
			}

			index, err := createSearchIndex("data", "_tmp_index")
			if err != nil {
				http.Error(w, err.Error(), 500)
				return
			}

			for _, page := range pages {
				searchObjects, err := createSearchObjects(page.Name)
				if err != nil {
					log.Printf("error while creating search object %s: %w", page.Title, err)
					continue
				}

				for _, so := range searchObjects {
					err = index.Index(so.ID, so)
					if err != nil {
						log.Printf("error while indexing %s: %w", page.Title, err)
						continue
					}
				}
			}

			err = os.Rename("data/_page-index", "data/_page-index-old")
			if err != nil {
				log.Printf("error while resetting index: %w", err)
				http.Error(w, err.Error(), 500)
				return
			}
			err = os.Rename("data/_tmp_index", "data/_page-index")
			if err != nil {
				log.Printf("error while putthing new index in place: %w", err)
				http.Error(w, err.Error(), 500)
				return
			}
			err = os.RemoveAll("data/_page-index-old")
			if err != nil {
				log.Printf("error while remove old index %w", err)
				http.Error(w, err.Error(), 500)
				return
			}

			enc := json.NewEncoder(w)
			enc.SetIndent("", "    ")
			err = enc.Encode(struct {
				Ok bool `json:"ok"`
			}{Ok: true})
			if err != nil {
				http.Error(w, err.Error(), 500)
				return
			}
		}
		return
	}

	q := bleve.NewQueryStringQuery(r.URL.Query().Get("q"))
	sr := bleve.NewSearchRequest(q)
	sr.IncludeLocations = false
	sr.Size = 25
	sr.Fields = []string{"page", "title", "text"}
	sr.Highlight = bleve.NewHighlightWithStyle("html")
	sr.Highlight.AddField("text")
	results, err := s.searchIndex.Search(sr)
	if err != nil {
		http.Error(w, err.Error(), 500)
	}
	enc := json.NewEncoder(w)
	enc.SetIndent("", "    ")
	err = enc.Encode(&results)
	if err != nil {
		http.Error(w, err.Error(), 500)
	}
}

type pageBlock struct {
	ID    string `json:"id"`
	Title string `json:"title"`
	Page  string `json:"page"`
	Text  string `json:"text"`
	Link  string `json:"link"`
}

func (p pageBlock) Type() string {
	return "block"
}

func createSearchObjects(rootBlockID string) ([]pageBlock, error) {
	blocks, err := loadBlocks("data", rootBlockID)
	if err != nil {
		return nil, err
	}

	var pageBlocks []pageBlock

	queue := []string{blocks.PageID}

	for len(queue) > 0 {
		current := queue[0]
		queue = queue[1:]

		links, err := ParseLinks(current, blocks.Texts[current])
		if err != nil {
			continue
		}

		if len(links) == 0 {
			pageBlocks = append(pageBlocks, pageBlock{
				ID:    current,
				Title: blocks.Texts[blocks.PageID],
				Page:  blocks.PageID,
				Text:  blocks.Texts[current],
				Link:  "",
			})
		} else {
			for _, link := range links {
				pageBlocks = append(pageBlocks, pageBlock{
					ID:    current,
					Title: blocks.Texts[blocks.PageID],
					Page:  blocks.PageID,
					Text:  blocks.Texts[current],
					Link:  link.Name,
				})
			}
		}

		queue = append(queue, blocks.Children[current]...)
	}

	return pageBlocks, nil
}

func createStructuredFormat(page Page) (searchObject, error) {
	so := searchObject{}
	so.Title = page.Title
	so.Meta = make(map[string]string)

	type simpleListItem struct {
		Text string
		ID   string
	}

	var listItems []simpleListItem
	if err := json.NewDecoder(strings.NewReader(page.Content)).Decode(&listItems); err != nil {
		so.Blocks = append(so.Blocks, page.Content)
	} else {
		for _, li := range listItems {
			meta := strings.SplitN(li.Text, "::", 2)
			if len(meta) == 2 {
				key := strcase.ToSnake(strings.TrimSpace(meta[0]))
				value := strings.TrimSpace(meta[1])
				if key == "title" {
					so.Title = value
				}
				so.Meta[key] = value
			}

			so.Blocks = append(so.Blocks, li.Text)

			links, err := ParseLinks(li.ID, li.Text)
			if err != nil {
				continue
			}

			for i, link := range links {
				links[i].Href = fmt.Sprintf("%s%s", *baseurl, link.PageName)
			}

			so.Links = append(so.Links, links...)
		}
	}

	for _, refs := range page.Refs {
		for _, ref := range refs {
			so.Refs = append(so.Refs, nameLine{
				ref.Name,
				ref.Title,
				strings.TrimSpace(ref.Line),
			})
		}
	}

	return so, nil
}