add entries to search index

domain/search/mysql/store.go

@@ -15,14 +15,13 @@ import (
 	"database/sql"
 	"fmt"
 	"strings"
-	"time"
 
 	"github.com/documize/community/core/env"
 	"github.com/documize/community/core/streamutil"
 	"github.com/documize/community/core/stringutil"
 	"github.com/documize/community/domain"
-	"github.com/documize/community/domain/store/mysql"
-	"github.com/documize/community/model"
+	"github.com/documize/community/model/attachment"
+	"github.com/documize/community/model/doc"
 	"github.com/documize/community/model/page"
 	"github.com/documize/community/model/search"
 	"github.com/jmoiron/sqlx"
@@ -34,210 +33,160 @@ type Scope struct {
 	Runtime *env.Runtime
 }
 
-// Add search entry  (legacy name: searchAdd).
-func (s Scope) Add(ctx domain.RequestContext, page page.Page) (err error) {
-	id := page.RefID
-
-	// translate the html into text for the search
-	nonHTML, err := stringutil.HTML(page.Body).Text(false)
+// IndexDocument adds search index entries for document inserting title, tags and attachments as
+// searchable items. Any existing document entries are removed.
+func (s Scope) IndexDocument(ctx domain.RequestContext, doc doc.Document, a []attachment.Attachment) (err error) {
+	// remove previous search entries
+	var stmt1 *sqlx.Stmt
+	stmt1, err = ctx.Transaction.Preparex("DELETE FROM search WHERE orgid=? AND documentid=? AND (itemtype='doc' OR itemtype='file' OR itemtype='tag')")
+	defer streamutil.Close(stmt1)
 	if err != nil {
-		errors.Wrap(err, "search decode body")
+		err = errors.Wrap(err, "prepare delete document index entries")
 		return
 	}
 
-	// insert into the search table, getting the document title along the way
-	var stmt *sqlx.Stmt
-	stmt, err = ctx.Transaction.Preparex(
-		"INSERT INTO search (id, orgid, documentid, level, sequence, documenttitle, slug, pagetitle, body, created, revised) " +
-			" SELECT page.refid,page.orgid,document.refid,page.level,page.sequence,document.title,document.slug,page.title,?,page.created,page.revised " +
-			" FROM document,page WHERE page.refid=? AND document.refid=page.documentid")
-
-	defer streamutil.Close(stmt)
-
+	_, err = stmt1.Exec(ctx.OrgID, doc.RefID)
 	if err != nil {
-		err = errors.Wrap(err, "prepare search insert")
+		err = errors.Wrap(err, "execute delete document index entries")
 		return
 	}
 
-	_, err = stmt.Exec(nonHTML, id)
-
-	if err != nil {
-		err = errors.Wrap(err, "execute search insert")
-		return
-	}
-
-	return nil
-}
-
-// Update search entry (legacy name: searchUpdate).
-func (s Scope) Update(ctx domain.RequestContext, page page.Page) (err error) {
-	// translate the html into text for the search
-	nonHTML, err := stringutil.HTML(page.Body).Text(false)
-	if err != nil {
-		err = errors.Wrap(err, "search decode body")
-		return
-	}
-
-	su, err := ctx.Transaction.Preparex("UPDATE search SET pagetitle=?,body=?,sequence=?,level=?,revised=? WHERE id=?")
-	defer streamutil.Close(su)
-
-	if err != nil {
-		err = errors.Wrap(err, "prepare search update")
-		return err
-	}
-
-	_, err = su.Exec(page.Title, nonHTML, page.Sequence, page.Level, page.Revised, page.RefID)
-
-	if err != nil {
-		err = errors.Wrap(err, "execute search update")
-		return
-	}
-
-	return nil
-}
-
-// UpdateDocument search entries for document (legacy name: searchUpdateDocument).
-func (s Scope) UpdateDocument(ctx domain.RequestContext, page page.Page) (err error) {
-	stmt, err := ctx.Transaction.Preparex("UPDATE search SET documenttitle=?, slug=?, revised=? WHERE documentid=?")
-	defer streamutil.Close(stmt)
-
-	if err != nil {
-		err = errors.Wrap(err, "prepare search document update")
-		return err
-	}
-
-	_, err = stmt.Exec(page.Title, page.Body, time.Now().UTC(), page.DocumentID)
-
-	if err != nil {
-		err = errors.Wrap(err, "execute search document update")
-		return err
-	}
-
-	return nil
-}
-
-// DeleteDocument removes document search entries (legacy name: searchDeleteDocument)
-func (s Scope) DeleteDocument(ctx domain.RequestContext, page page.Page) (err error) {
-	var bm = mysql.BaseQuery{}
-
-	_, err = bm.DeleteWhere(ctx.Transaction, fmt.Sprintf("DELETE from search WHERE documentid='%s'", page.DocumentID))
-
-	if err != nil {
-		err = errors.Wrap(err, "delete document search entries")
-	}
-
-	return nil
-}
-
-// Rebuild ... (legacy name: searchRebuild)
-func (s Scope) Rebuild(ctx domain.RequestContext, p page.Page) (err error) {
-	var bm = mysql.BaseQuery{}
-
-	_, err = bm.DeleteWhere(ctx.Transaction, fmt.Sprintf("DELETE from search WHERE documentid='%s'", p.DocumentID))
-	if err != nil {
-		err = errors.Wrap(err, err.Error())
-		return err
-	}
-
-	var pages []struct{ ID string }
-
-	stmt2, err := ctx.Transaction.Preparex("SELECT refid as id FROM page WHERE documentid=? ")
+	// insert doc title
+	var stmt2 *sqlx.Stmt
+	stmt2, err = ctx.Transaction.Preparex("INSERT INTO search (orgid, documentid, itemid, itemtype, content) VALUES (?, ?, ?, ?, ?)")
 	defer streamutil.Close(stmt2)
-
 	if err != nil {
-		err = errors.Wrap(err, err.Error())
-		return err
+		err = errors.Wrap(err, "prepare insert document title entry")
+		return
 	}
 
-	err = stmt2.Select(&pages, p.DocumentID)
+	_, err = stmt2.Exec(ctx.OrgID, doc.RefID, "", "doc", doc.Title)
 	if err != nil {
-		err = errors.Wrap(err, err.Error())
-		return err
+		err = errors.Wrap(err, "execute insert document title entry")
+		return
 	}
 
-	if len(pages) > 0 {
-		for _, pg := range pages {
-			err = s.Add(ctx, page.Page{BaseEntity: model.BaseEntity{RefID: pg.ID}})
-			if err != nil {
-				err = errors.Wrap(err, err.Error())
-				return err
-			}
+	// insert doc tags
+	tags := strings.Split(doc.Tags, "#")
+	for _, t := range tags {
+		if len(t) == 0 {
+			continue
 		}
 
-		// rebuild doc-level tags & excerpts
-		// get the 0'th page data and rewrite it
-
-		target := page.Page{}
-
-		stmt1, err := ctx.Transaction.Preparex("SELECT * FROM page WHERE refid=?")
-		defer streamutil.Close(stmt1)
-
+		var stmt3 *sqlx.Stmt
+		stmt3, err = ctx.Transaction.Preparex("INSERT INTO search (orgid, documentid, itemid, itemtype, content) VALUES (?, ?, ?, ?, ?)")
+		defer streamutil.Close(stmt3)
 		if err != nil {
-			err = errors.Wrap(err, err.Error())
-			return err
+			err = errors.Wrap(err, "prepare insert document tag entry")
+			return
 		}
 
-		err = stmt1.Get(&target, pages[0].ID)
+		_, err = stmt3.Exec(ctx.OrgID, doc.RefID, "", "tag", t)
 		if err != nil {
-			err = errors.Wrap(err, err.Error())
-			return err
-		}
-
-		err = s.Update(ctx, target) // to rebuild the document-level tags + excerpt
-		if err != nil {
-			err = errors.Wrap(err, err.Error())
-			return err
+			err = errors.Wrap(err, "execute insert document tag entry")
+			return
 		}
 	}
 
-	return
+	for _, file := range a {
+		var stmt4 *sqlx.Stmt
+		stmt4, err = ctx.Transaction.Preparex("INSERT INTO search (orgid, documentid, itemid, itemtype, content) VALUES (?, ?, ?, ?, ?)")
+		defer streamutil.Close(stmt4)
+		if err != nil {
+			err = errors.Wrap(err, "prepare insert document file entry")
+			return
+		}
+
+		_, err = stmt4.Exec(ctx.OrgID, doc.RefID, file.RefID, "file", file.Filename)
+		if err != nil {
+			err = errors.Wrap(err, "execute insert document file entry")
+			return
+		}
+	}
+
+	return nil
 }
 
-// UpdateSequence ... (legacy name: searchUpdateSequence)
-func (s Scope) UpdateSequence(ctx domain.RequestContext, page page.Page) (err error) {
-	supdate, err := ctx.Transaction.Preparex("UPDATE search SET sequence=?,revised=? WHERE id=?")
-	defer streamutil.Close(supdate)
-
+// DeleteDocument removes all search entries for document.
+func (s Scope) DeleteDocument(ctx domain.RequestContext, ID string) (err error) {
+	// remove all search entries
+	var stmt1 *sqlx.Stmt
+	stmt1, err = ctx.Transaction.Preparex("DELETE FROM search WHERE orgid=? AND documentid=?")
+	defer streamutil.Close(stmt1)
 	if err != nil {
-		err = errors.Wrap(err, "prepare search update sequence")
-		return err
+		err = errors.Wrap(err, "prepare delete document entries")
+		return
 	}
 
-	_, err = supdate.Exec(page.Sequence, time.Now().UTC(), page.RefID)
+	_, err = stmt1.Exec(ctx.OrgID, ID)
 	if err != nil {
-		err = errors.Wrap(err, "execute search update sequence")
+		err = errors.Wrap(err, "execute delete document entries")
 		return
 	}
 
 	return
 }
 
-// UpdateLevel ... legacy name: searchUpdateLevel)
-func (s Scope) UpdateLevel(ctx domain.RequestContext, page page.Page) (err error) {
-	pageID := page.RefID
-	level := page.Level
-
-	supdate, err := ctx.Transaction.Preparex("UPDATE search SET level=?,revised=? WHERE id=?")
-	defer streamutil.Close(supdate)
-
+// IndexContent adds search index entry for document context.
+// Any existing document entries are removed.
+func (s Scope) IndexContent(ctx domain.RequestContext, p page.Page) (err error) {
+	// remove previous search entries
+	var stmt1 *sqlx.Stmt
+	stmt1, err = ctx.Transaction.Preparex("DELETE FROM search WHERE orgid=? AND documentid=? AND itemid=? AND itemtype='page' )")
+	defer streamutil.Close(stmt1)
 	if err != nil {
-		err = errors.Wrap(err, "prepare search update level")
-		return err
-	}
-
-	_, err = supdate.Exec(level, time.Now().UTC(), pageID)
-	if err != nil {
-		err = errors.Wrap(err, "execute search update level")
+		err = errors.Wrap(err, "prepare delete document content entry")
 		return
 	}
 
-	return
+	_, err = stmt1.Exec(ctx.OrgID, p.DocumentID)
+	if err != nil {
+		err = errors.Wrap(err, "execute delete document content entry")
+		return
+	}
+
+	// insert doc title
+	var stmt2 *sqlx.Stmt
+	stmt2, err = ctx.Transaction.Preparex("INSERT INTO search (orgid, documentid, itemid, itemtype, content) VALUES (?, ?, ?, ?, ?)")
+	defer streamutil.Close(stmt2)
+	if err != nil {
+		err = errors.Wrap(err, "prepare insert document content entry")
+		return
+	}
+
+	// prepare content
+	content, err := stringutil.HTML(p.Body).Text(false)
+	if err != nil {
+		err = errors.Wrap(err, "search strip HTMl failed")
+		return
+	}
+	content = strings.TrimSpace(content)
+
+	_, err = stmt2.Exec(ctx.OrgID, p.DocumentID, p.RefID, "page", content)
+	if err != nil {
+		err = errors.Wrap(err, "execute insert document content entry")
+		return
+	}
+
+	return nil
 }
 
-// Delete ... (legacy name: searchDelete).
-func (s Scope) Delete(ctx domain.RequestContext, page page.Page) (err error) {
-	var bm = mysql.BaseQuery{}
-	_, err = bm.DeleteConstrainedWithID(ctx.Transaction, "search", ctx.OrgID, page.RefID)
+// DeleteContent removes all search entries for specific document content.
+func (s Scope) DeleteContent(ctx domain.RequestContext, pageID string) (err error) {
+	// remove all search entries
+	var stmt1 *sqlx.Stmt
+	stmt1, err = ctx.Transaction.Preparex("DELETE FROM search WHERE orgid=? AND itemid=? AND itemtype='page'")
+	defer streamutil.Close(stmt1)
+	if err != nil {
+		err = errors.Wrap(err, "prepare delete document content entry")
+		return
+	}
+
+	_, err = stmt1.Exec(ctx.OrgID, pageID)
+	if err != nil {
+		err = errors.Wrap(err, "execute delete document content entry")
+		return
+	}
 
 	return
 }

domain/search/queue.go

@@ -12,131 +12,21 @@
 package search
 
 import (
-	"errors"
-	"fmt"
-	"sync"
-
 	"github.com/documize/community/core/env"
 	"github.com/documize/community/domain"
-	"github.com/documize/community/model/page"
 )
 
-// Indexer type provides the datastructure for the queues of activity to be serialized through a single background goroutine.
-// NOTE if the queue becomes full, the system will trigger the rebuilding entire files in order to clear the backlog.
+// Indexer documents!
 type Indexer struct {
-	queue        chan queueEntry
-	rebuild      map[string]bool
-	rebuildLock  sync.RWMutex
-	givenWarning bool
-	runtime      *env.Runtime
-	store        *domain.Store
+	runtime *env.Runtime
+	store   *domain.Store
 }
 
-type queueEntry struct {
-	action    func(domain.RequestContext, page.Page) error
-	isRebuild bool
-	page.Page
-	ctx domain.RequestContext
-}
-
-const searchQueueLength = 2048 // NOTE the largest 15Mb docx in the test set generates 2142 queue entries, but the queue is constantly emptied
-
 // NewIndexer provides background search indexer
 func NewIndexer(rt *env.Runtime, s *domain.Store) (i Indexer) {
 	i = Indexer{}
-	i.queue = make(chan queueEntry, searchQueueLength) // provide some decoupling
-	i.rebuild = make(map[string]bool)
 	i.runtime = rt
 	i.store = s
 
-	go i.processQueue()
-
 	return
 }
-
-// processQueue is run as a goroutine, it processes the queue of search index update requests.
-func (m *Indexer) processQueue() {
-	for {
-		//fmt.Println("DEBUG queue length=", len(Searches.queue))
-		if len(m.queue) <= searchQueueLength/20 { // on a busy server, the queue may never get to zero - so use 5%
-			m.rebuildLock.Lock()
-			for docid := range m.rebuild {
-				m.queue <- queueEntry{
-					action:    m.store.Search.Rebuild,
-					isRebuild: true,
-					Page:      page.Page{DocumentID: docid},
-				}
-				delete(m.rebuild, docid)
-			}
-			m.rebuildLock.Unlock()
-		}
-
-		qe := <-m.queue
-		doit := true
-
-		if len(qe.DocumentID) > 0 {
-			m.rebuildLock.RLock()
-			if m.rebuild[qe.DocumentID] {
-				doit = false // don't execute an action on a document queued to be rebuilt
-			}
-			m.rebuildLock.RUnlock()
-		}
-
-		if doit {
-			tx, err := m.runtime.Db.Beginx()
-			if err != nil {
-			} else {
-				ctx := qe.ctx
-				ctx.Transaction = tx
-				err = qe.action(ctx, qe.Page)
-				if err != nil {
-					tx.Rollback()
-					// This action has failed, so re-build indexes for the entire document,
-					// provided it was not a re-build command that failed and we know the documentId.
-					if !qe.isRebuild && len(qe.DocumentID) > 0 {
-						m.rebuildLock.Lock()
-						m.rebuild[qe.DocumentID] = true
-						m.rebuildLock.Unlock()
-					}
-				} else {
-					tx.Commit()
-				}
-			}
-		}
-	}
-}
-
-func (m *Indexer) addQueue(qe queueEntry) error {
-	lsq := len(m.queue)
-
-	if lsq >= (searchQueueLength - 1) {
-		if qe.DocumentID != "" {
-			m.rebuildLock.Lock()
-			if !m.rebuild[qe.DocumentID] {
-				m.runtime.Log.Info(fmt.Sprintf("WARNING: Search Queue Has No Space! Marked rebuild index for document id %s", qe.DocumentID))
-			}
-			m.rebuild[qe.DocumentID] = true
-			m.rebuildLock.Unlock()
-		} else {
-			m.runtime.Log.Error("addQueue", errors.New("WARNING: Search Queue Has No Space! But unable to index unknown document id"))
-		}
-
-		return nil
-	}
-
-	if lsq > ((8 * searchQueueLength) / 10) {
-		if !m.givenWarning {
-			m.runtime.Log.Info(fmt.Sprintf("WARNING: Searches.queue length %d exceeds 80%% of capacity", lsq))
-			m.givenWarning = true
-		}
-	} else {
-		if m.givenWarning {
-			m.runtime.Log.Info(fmt.Sprintf("INFO: Searches.queue length %d now below 80%% of capacity", lsq))
-			m.givenWarning = false
-		}
-	}
-
-	m.queue <- qe
-
-	return nil
-}

domain/search/search.go

@@ -13,102 +13,93 @@ package search
 
 import (
 	"github.com/documize/community/domain"
-	"github.com/documize/community/model"
+	"github.com/documize/community/model/attachment"
 	"github.com/documize/community/model/doc"
 	"github.com/documize/community/model/page"
 )
 
-// Add should be called when a new page is added to a document.
-func (m *Indexer) Add(ctx domain.RequestContext, page page.Page, id string) (err error) {
-	page.RefID = id
+// IndexDocument adds search indesd entries for document inserting title, tags and attachments as
+// searchable items. Any existing document entries are removed.
+func (m *Indexer) IndexDocument(ctx domain.RequestContext, d doc.Document, a []attachment.Attachment) {
+	method := "search.IndexDocument"
+	var err error
 
-	err = m.addQueue(queueEntry{
-		action: m.store.Search.Add,
-		Page:   page,
-		ctx:    ctx,
-	})
-
-	return
-}
-
-// Update should be called after a page record has been updated.
-func (m *Indexer) Update(ctx domain.RequestContext, page page.Page) (err error) {
-	err = m.addQueue(queueEntry{
-		action: m.store.Search.Update,
-		Page:   page,
-		ctx:    ctx,
-	})
-
-	return
-}
-
-// UpdateDocument should be called after a document record has been updated.
-func (m *Indexer) UpdateDocument(ctx domain.RequestContext, document doc.Document) (err error) {
-	err = m.addQueue(queueEntry{
-		action: m.store.Search.UpdateDocument,
-		Page: page.Page{
-			DocumentID: document.RefID,
-			Title:      document.Title,
-			Body:       document.Slug, // NOTE body==slug in this context
-		},
-		ctx: ctx,
-	})
-
-	return
-}
-
-// DeleteDocument should be called after a document record has been deleted.
-func (m *Indexer) DeleteDocument(ctx domain.RequestContext, documentID string) (err error) {
-	if len(documentID) > 0 {
-		m.queue <- queueEntry{
-			action: m.store.Search.DeleteDocument,
-			Page:   page.Page{DocumentID: documentID},
-			ctx:    ctx,
-		}
+	ctx.Transaction, err = m.runtime.Db.Beginx()
+	if err != nil {
+		m.runtime.Log.Error(method, err)
+		return
 	}
-	return
+
+	err = m.store.Search.IndexDocument(ctx, d, a)
+	if err != nil {
+		ctx.Transaction.Rollback()
+		m.runtime.Log.Error(method, err)
+		return
+	}
+
+	ctx.Transaction.Commit()
 }
 
-// UpdateSequence should be called after a page record has been resequenced.
-func (m *Indexer) UpdateSequence(ctx domain.RequestContext, documentID, pageID string, sequence float64) (err error) {
-	err = m.addQueue(queueEntry{
-		action: m.store.Search.UpdateSequence,
-		Page: page.Page{
-			BaseEntity: model.BaseEntity{RefID: pageID},
-			Sequence:   sequence,
-			DocumentID: documentID,
-		},
-		ctx: ctx,
-	})
+// DeleteDocument removes all search entries for document.
+func (m *Indexer) DeleteDocument(ctx domain.RequestContext, ID string) {
+	method := "search.DeleteDocument"
+	var err error
 
-	return
+	ctx.Transaction, err = m.runtime.Db.Beginx()
+	if err != nil {
+		m.runtime.Log.Error(method, err)
+		return
+	}
+
+	err = m.store.Search.DeleteDocument(ctx, ID)
+	if err != nil {
+		ctx.Transaction.Rollback()
+		m.runtime.Log.Error(method, err)
+		return
+	}
+
+	ctx.Transaction.Commit()
 }
 
-// UpdateLevel should be called after the level of a page has been changed.
-func (m *Indexer) UpdateLevel(ctx domain.RequestContext, documentID, pageID string, level int) (err error) {
-	err = m.addQueue(queueEntry{
-		action: m.store.Search.UpdateLevel,
-		Page: page.Page{
-			BaseEntity: model.BaseEntity{RefID: pageID},
-			Level:      uint64(level),
-			DocumentID: documentID,
-		},
-		ctx: ctx,
-	})
+// IndexContent adds search index entry for document context.
+// Any existing document entries are removed.
+func (m *Indexer) IndexContent(ctx domain.RequestContext, p page.Page) {
+	method := "search.IndexContent"
+	var err error
 
-	return
+	ctx.Transaction, err = m.runtime.Db.Beginx()
+	if err != nil {
+		m.runtime.Log.Error(method, err)
+		return
+	}
+
+	err = m.store.Search.IndexContent(ctx, p)
+	if err != nil {
+		ctx.Transaction.Rollback()
+		m.runtime.Log.Error(method, err)
+		return
+	}
+
+	ctx.Transaction.Commit()
 }
 
-// Delete should be called after a page has been deleted.
-func (m *Indexer) Delete(ctx domain.RequestContext, documentID, pageID string) (rows int64, err error) {
-	err = m.addQueue(queueEntry{
-		action: m.store.Search.Delete,
-		Page: page.Page{
-			BaseEntity: model.BaseEntity{RefID: pageID},
-			DocumentID: documentID,
-		},
-		ctx: ctx,
-	})
+// DeleteContent removes all search entries for specific document content.
+func (m *Indexer) DeleteContent(ctx domain.RequestContext, pageID string) {
+	method := "search.DeleteContent"
+	var err error
 
-	return
-}
+	ctx.Transaction, err = m.runtime.Db.Beginx()
+	if err != nil {
+		m.runtime.Log.Error(method, err)
+		return
+	}
+
+	err = m.store.Search.DeleteContent(ctx, pageID)
+	if err != nil {
+		ctx.Transaction.Rollback()
+		m.runtime.Log.Error(method, err)
+		return
+	}
+
+	ctx.Transaction.Commit()
+}