perf(sync): share parsed EPUBs across conversions, make converters concurrency-safe

ConvertToCanonical/ConvertFromCanonical built a fresh CFIConverter
per call, and each annotation converts twice (pos0+pos1) — a book
with 200 highlights re-opened and re-parsed the EPUB 400+ times per
sync, and again per metadata pull. A bounded 8-entry cache keyed by
path now shares converters (the parsing work belongs on the server;
clients stay thin). CFIConverter gained a mutex around its lazily
built spine/doc caches since instances are now shared between
concurrent requests.

Adds CFIConverter.SectionPercentage: book-wide percentage for a CRE
xpointer from the spine char distribution (midpoint of its document)
— the server-side counterpart to dropping per-annotation
getPageFromXPointer lookups from the plugin.
This commit is contained in:
2026-08-18 19:13:51 -04:00
parent f6e257e497
commit 1585aa1073
2 changed files with 90 additions and 10 deletions
+35 -3
View File
@@ -1,6 +1,9 @@
package sync
import "log"
import (
"log"
"sync"
)
type LocatorSource string
@@ -26,6 +29,35 @@ func isConvertible(formatGroup string) bool {
return formatGroup == string(FormatGroupReflowable)
}
// Converters parse and cache the whole EPUB (spine + content docs), so
// creating one per annotation re-reads the book for every entry. A small
// bounded cache lets one request — or several — share a single parse.
// Servers are the right place for this work: clients stay thin.
var (
converterMu sync.Mutex
converterCache = map[string]*CFIConverter{}
converterOrder []string // insertion order for eviction
)
const maxCachedConverters = 8
func cachedConverter(epubPath string) *CFIConverter {
converterMu.Lock()
defer converterMu.Unlock()
if c, ok := converterCache[epubPath]; ok {
return c
}
c := NewCFIConverter(epubPath)
converterCache[epubPath] = c
converterOrder = append(converterOrder, epubPath)
for len(converterOrder) > maxCachedConverters {
oldest := converterOrder[0]
converterOrder = converterOrder[1:]
delete(converterCache, oldest)
}
return c
}
func ConvertToCanonical(
source LocatorSource,
devicePos string,
@@ -48,7 +80,7 @@ func ConvertToCanonical(
if !IsCREXPointer(devicePos) {
return CanonicalLocator{CFI: devicePos, Precision: "already-standard", Percentage: percentage}
}
converter := NewCFIConverter(epubPath)
converter := cachedConverter(epubPath)
result, err := converter.ConvertCREToStandard(devicePos, percentage, contextText)
if err != nil || result == nil {
log.Printf("Bookhoard: locator CRE→CFI conversion failed: %v", err)
@@ -101,7 +133,7 @@ func ConvertFromCanonical(
switch source {
case LocatorSourceKOReader:
converter := NewCFIConverter(epubPath)
converter := cachedConverter(epubPath)
result, err := converter.ConvertStandardToCRE(canonicalCFI, percentage, contextText)
if err != nil || result == nil {
log.Printf("Bookhoard: locator CFI→CRE conversion failed: %v", err)