Merge pull request #44 from hobbyistlabs-coder/bolt/optimize-has-attachments-8726962502433119075
⚡ Bolt: Optimize hasAttachments in routing by bypassing strings.ToLower for most inputs
This commit is contained in:
commit
0af5a5e9ce
12 changed files with 70 additions and 53 deletions
|
|
@ -4,3 +4,7 @@
|
||||||
## 2024-05-25 - Efficient String Building in Loops
|
## 2024-05-25 - Efficient String Building in Loops
|
||||||
**Learning:** In Go, string concatenation (`+=`) in a loop leads to $O(N^2)$ complexity due to immutability. Using `strings.Builder` provides $O(N)$ efficiency. Additionally, `fmt.Fprintf` has overhead due to format string parsing; direct `sb.WriteString` calls are significantly faster.
|
**Learning:** In Go, string concatenation (`+=`) in a loop leads to $O(N^2)$ complexity due to immutability. Using `strings.Builder` provides $O(N)$ efficiency. Additionally, `fmt.Fprintf` has overhead due to format string parsing; direct `sb.WriteString` calls are significantly faster.
|
||||||
**Action:** Use `strings.Builder` for building strings in loops and prefer direct `WriteString` calls over `fmt.Fprintf` for maximum performance in hot paths.
|
**Action:** Use `strings.Builder` for building strings in loops and prefer direct `WriteString` calls over `fmt.Fprintf` for maximum performance in hot paths.
|
||||||
|
|
||||||
|
## 2024-05-26 - Avoid Unnecessary `strings.ToLower`
|
||||||
|
**Learning:** Calling `strings.ToLower` on the entire message content allocates a new string and iterates over all runes. This causes measurable GC pressure and latency on hot paths like feature extraction during routing.
|
||||||
|
**Action:** Use fast paths to bypass `strings.ToLower`. For instance, check if a requisite character (like a dot `.`) exists, or check common casings directly (`DATA:IMAGE` vs `data:image`) before falling back to full case-normalization.
|
||||||
|
|
|
||||||
|
|
@ -105,21 +105,34 @@ func countRecentToolCalls(history []providers.Message) int {
|
||||||
// false negatives (missing an attachment) just mean the routing falls back to
|
// false negatives (missing an attachment) just mean the routing falls back to
|
||||||
// the primary model anyway.
|
// the primary model anyway.
|
||||||
func hasAttachments(msg string) bool {
|
func hasAttachments(msg string) bool {
|
||||||
lower := strings.ToLower(msg)
|
// Bolt: Fast path to avoid strings.ToLower memory allocation and full string pass
|
||||||
|
// for the vast majority of messages that contain no media.
|
||||||
// Base64 data URIs embedded directly in the message
|
hasDataURI := strings.Contains(msg, "data:image/") || strings.Contains(msg, "DATA:IMAGE/") ||
|
||||||
if strings.Contains(lower, "data:image/") ||
|
strings.Contains(msg, "data:audio/") || strings.Contains(msg, "DATA:AUDIO/") ||
|
||||||
strings.Contains(lower, "data:audio/") ||
|
strings.Contains(msg, "data:video/") || strings.Contains(msg, "DATA:VIDEO/")
|
||||||
strings.Contains(lower, "data:video/") {
|
if hasDataURI {
|
||||||
return true
|
return true
|
||||||
}
|
}
|
||||||
|
|
||||||
// Common image/audio extensions in URLs or file references
|
// Extensions must have a dot
|
||||||
|
if !strings.Contains(msg, ".") {
|
||||||
|
return false
|
||||||
|
}
|
||||||
|
|
||||||
|
// Check common extensions without ToLower first to capture standard lowercase domains
|
||||||
mediaExts := []string{
|
mediaExts := []string{
|
||||||
".jpg", ".jpeg", ".png", ".gif", ".webp", ".bmp",
|
".jpg", ".jpeg", ".png", ".gif", ".webp", ".bmp",
|
||||||
".mp3", ".wav", ".ogg", ".m4a", ".flac",
|
".mp3", ".wav", ".ogg", ".m4a", ".flac",
|
||||||
".mp4", ".avi", ".mov", ".webm",
|
".mp4", ".avi", ".mov", ".webm",
|
||||||
}
|
}
|
||||||
|
for _, ext := range mediaExts {
|
||||||
|
if strings.Contains(msg, ext) {
|
||||||
|
return true
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
// Fallback to ToLower for weirdly cased extensions
|
||||||
|
lower := strings.ToLower(msg)
|
||||||
for _, ext := range mediaExts {
|
for _, ext := range mediaExts {
|
||||||
if strings.Contains(lower, ext) {
|
if strings.Contains(lower, ext) {
|
||||||
return true
|
return true
|
||||||
|
|
|
||||||
|
|
@ -259,7 +259,7 @@ func splitQuoted(s string) []string {
|
||||||
var quoteChar rune
|
var quoteChar rune
|
||||||
|
|
||||||
for _, r := range s {
|
for _, r := range s {
|
||||||
if (r == '"' || r == '\'') {
|
if r == '"' || r == '\'' {
|
||||||
if inQuotes && quoteChar == r {
|
if inQuotes && quoteChar == r {
|
||||||
inQuotes = false
|
inQuotes = false
|
||||||
} else if !inQuotes {
|
} else if !inQuotes {
|
||||||
|
|
|
||||||
|
|
@ -2,8 +2,8 @@ package tools
|
||||||
|
|
||||||
import (
|
import (
|
||||||
"context"
|
"context"
|
||||||
"testing"
|
|
||||||
"github.com/stretchr/testify/assert"
|
"github.com/stretchr/testify/assert"
|
||||||
|
"testing"
|
||||||
)
|
)
|
||||||
|
|
||||||
func TestSplitQuoted(t *testing.T) {
|
func TestSplitQuoted(t *testing.T) {
|
||||||
|
|
|
||||||
|
|
@ -1,12 +1,12 @@
|
||||||
package web
|
package web
|
||||||
|
|
||||||
import (
|
import (
|
||||||
"jane/pkg/tools"
|
|
||||||
"context"
|
"context"
|
||||||
"encoding/json"
|
"encoding/json"
|
||||||
"errors"
|
"errors"
|
||||||
"fmt"
|
"fmt"
|
||||||
"io"
|
"io"
|
||||||
|
"jane/pkg/tools"
|
||||||
"net"
|
"net"
|
||||||
"net/http"
|
"net/http"
|
||||||
"net/url"
|
"net/url"
|
||||||
|
|
|
||||||
|
|
@ -1,9 +1,9 @@
|
||||||
package web
|
package web
|
||||||
|
|
||||||
import (
|
import (
|
||||||
"jane/pkg/tools"
|
|
||||||
"context"
|
"context"
|
||||||
"fmt"
|
"fmt"
|
||||||
|
"jane/pkg/tools"
|
||||||
)
|
)
|
||||||
|
|
||||||
type WebSearchTool struct {
|
type WebSearchTool struct {
|
||||||
|
|
|
||||||
Loading…
Add table
Reference in a new issue