optimize: log parser

add: fast text reader to log parser

27-functional-programming/log-parser-exp/groupers.go

@@ -12,8 +12,7 @@ func pageGrouper(r result) string {
 	return r.domain + r.page
 }
 
-// you could have created a noopGrouper as well
-// but it's not necessary i think (map allocation)
+// groupBy allocates map unnecessarily
 func noopGrouper(r result) string {
 	// with something like:
 	// return randomStrings()

27-functional-programming/log-parser-exp/main.go

@@ -15,7 +15,7 @@ func main() {
 	defer recoverErr()
 
 	_, err := newReport().
-		// filterBy(orgDomainsFilter).
+		// from(fastTextReader(os.Stdin)).
 		filterBy(notUsing(domainExtFilter("com", "io"))).
 		groupBy(domainGrouper).
 		start()

27-functional-programming/log-parser-exp/report.go

@@ -17,6 +17,7 @@ type report struct {
 func newReport() *report {
 	return &report{
 		filter: noopFilter,
+		group:  noopGrouper,
 		input:  textReader(os.Stdin),
 		output: textWriter(os.Stdout),
 	}
@@ -43,30 +44,19 @@ func (r *report) groupBy(fn groupFunc) *report {
 }
 
 func (r *report) start() ([]result, error) {
-	if r.input == nil {
-		panic("report input cannot be nil")
-	}
+	//            input      filterBy    groupBy
+	//           scanner  (result) bool  map[string]result
+	//
+	// stdin -> []result -> []results -> []result -> output(stdout)
 
-	results, err := r.input()
+	res, err := r.input()
 	if err != nil {
 		return nil, err
 	}
 
-	// noop if filter is nil
-	results = filterBy(results, r.filter)
+	res = filterBy(res, r.filter)
+	res = groupBy(res, r.group)
+	err = r.output(res)
 
-	// group func is more tricky
-	// you don't want to create an unnecessary map
-	if r.group != nil {
-		results = groupBy(results, r.group)
-	}
-
-	// TODO: prefer: noop writer
-	if r.output != nil {
-		if err := r.output(results); err != nil {
-			return nil, err
-		}
-	}
-
-	return results, nil
+	return res, err
 }

27-functional-programming/log-parser-exp/textreader.go

@@ -9,6 +9,7 @@ package main
 
 import (
 	"bufio"
+	"bytes"
 	"fmt"
 	"io"
 	"strings"
@@ -16,34 +17,51 @@ import (
 
 func textReader(r io.Reader) inputFunc {
 	return func() ([]result, error) {
-		in := bufio.NewScanner(r)
-		return parseText(in)
-	}
-}
-
-func parseText(in *bufio.Scanner) ([]result, error) {
-	var (
-		results []result
-		lines   int
-	)
-
-	results = make([]result, 0, 5000000)
-
-	for in.Scan() {
-		lines++
-
-		res, err := parseFields(strings.Fields(in.Text()))
+		// first: count the lines, so the parseText can create
+		// enough buffer.
+		var buf bytes.Buffer
+		lines, err := countLines(io.TeeReader(r, &buf))
 		if err != nil {
-			// TODO: custom error type for line information
-			return nil, fmt.Errorf("line %d: %v", lines, err)
+			return nil, err
 		}
 
-		results = append(results, res)
+		return parseText(bufio.NewScanner(&buf), lines)
+	}
+}
+
+// TODO: custom error type for line information
+func parseText(in *bufio.Scanner, nlines int) ([]result, error) {
+	res := make([]result, 0, nlines)
+
+	for l := 1; in.Scan(); l++ {
+		fields := strings.Fields(in.Text())
+		r, err := parseFields(fields)
+
+		if err != nil {
+			return nil, fmt.Errorf("line %d: %v", l, err)
+		}
+		res = append(res, r)
+	}
+
+	return res, in.Err()
+}
+
+func countLines(r io.Reader) (int, error) {
+	var (
+		lines int
+		buf   = make([]byte, 1024<<4) // read via 16 KB blocks
+	)
+
+	for {
+		n, err := r.Read(buf)
+		lines += bytes.Count(buf[:n], []byte{'\n'})
+
+		if err == io.EOF {
+			return lines, nil
+		}
+
+		if err != nil {
+			return lines, err
+		}
 	}
-
-	if err := in.Err(); err != nil {
-		return nil, err
-	}
-
-	return results, nil
 }

27-functional-programming/log-parser-exp/textreaderfast.go

@@ -0,0 +1,106 @@
+// For more tutorials: https://blog.learngoprogramming.com
+//
+// Copyright © 2018 Inanc Gumus
+// Learn Go Programming Course
+// License: https://creativecommons.org/licenses/by-nc-sa/4.0/
+//
+
+package main
+
+import (
+	"bufio"
+	"bytes"
+	"errors"
+	"fmt"
+	"io"
+)
+
+// this could be made faster.
+// currently, it's 30-35% faster.
+//
+// so, what's different than the textreader?
+//
+// + creates the buffers specific to the input file/stdin size
+// + manually parses the fields: instead of strings.Fields
+// + gets the lines using scanner's Bytes() method: instead of Text()
+// + uses a manual atoi
+// +
+
+func fastTextReader(r io.Reader) inputFunc {
+	return func() ([]result, error) {
+		// first: count the lines, so the parseText can create
+		// enough buffer.
+		var buf bytes.Buffer
+		l, err := countLines(io.TeeReader(r, &buf))
+		if err != nil {
+			return nil, err
+		}
+
+		return fastParseText(bufio.NewScanner(&buf), l)
+	}
+}
+
+func fastParseText(in *bufio.Scanner, nlines int) ([]result, error) {
+	// needs to know the number of total lines in the file
+	res := make([]result, 0, nlines)
+
+	for l := 0; in.Scan(); l++ {
+		r, err := fastParseFields(in.Bytes())
+
+		if err != nil {
+			return nil, fmt.Errorf("line %d: %v", l, err)
+		}
+		res = append(res, r)
+	}
+
+	return res, in.Err()
+}
+
+func fastParseFields(data []byte) (res result, err error) {
+	var field int
+
+	for i, last := 0, 0; i < len(data); i++ {
+		done := len(data) == i+1
+
+		if c := data[i]; c == ' ' || done {
+			if done {
+				i = len(data)
+			}
+
+			switch field {
+			case 0:
+				res.domain = string(data[last:i])
+			case 1:
+				res.page = string(data[last:i])
+			case 2:
+				res.visits, err = atoi(data[last:i])
+			case 3:
+				res.uniques, err = atoi(data[last:i])
+			}
+
+			if err != nil {
+				return res, err
+			}
+
+			last = i + 1
+			field++
+		}
+	}
+
+	if field != 4 {
+		return result{}, errors.New("wrong number of fields")
+	}
+	return res, nil
+}
+
+func atoi(input []byte) (int, error) {
+	val := 0
+	for i := 0; i < len(input); i++ {
+		char := input[i]
+		if char < '0' || char > '9' {
+			return 0, errors.New("invalid number")
+		}
+		val = val*10 + int(char) - '0'
+	}
+	return val, nil
+}

27-functional-programming/log-parser-exp/textwriter.go

@@ -35,3 +35,9 @@ func textWriter(w io.Writer) outputFunc {
 		return nil
 	}
 }
+
+func noWhere() outputFunc {
+	return func(res []result) error {
+		return nil
+	}
+}