AdGuardHome/internal/querylog/qlogreader.go
Dimitry Kolyshev cbc7985e75 Pull request: querylog imp code
Merge in DNS/adguard-home from querylog-imp-code to master

Squashed commit of the following:

commit a58ad36508a2355b686d314dec51ac0b5e357281
Merge: df5494f2c 941eb1dd7
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Wed May 24 15:26:55 2023 +0300

    Merge remote-tracking branch 'origin/master' into querylog-imp-code

commit df5494f2c337736690a3c2a547c2d71858d0378f
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Wed May 24 15:24:43 2023 +0300

    querylog: imp code

commit 8c3c2b76dd5858e7b107f222c112e9cde2477fb3
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Wed May 24 12:14:15 2023 +0300

    all: lint script

commit be04a4decfaf20a1649d32ecaab3c1c6bb205ffd
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Wed May 24 12:03:12 2023 +0300

    querylog: imp code

commit fe7beacff3a5cfcf2332c4998b9c65820284eaf7
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Wed May 24 11:57:33 2023 +0300

    querylog: imp docs

commit 2ae239c57d12524fbc092f582842af2ad726c1d0
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Wed May 24 11:46:54 2023 +0300

    querylog: imp code

commit 417216cefbf154fa870f8f43468f35e0e345971f
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Wed May 24 11:25:44 2023 +0300

    querylog: imp code

commit 514b6ee99113844a4e0dad30dc53703e3220c289
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Wed May 24 11:14:13 2023 +0300

    querylog: imp docs

commit 321351a3abb524208daacd5a3a7fbf5f07ab259d
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Mon May 22 16:38:31 2023 +0300

    querylog: imp code

commit ee91de5c43210b5bc213f933d411adb894d2e586
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Mon May 22 16:01:32 2023 +0300

    querylog: imp code

commit 862ff12177fb769d5cb2ec250eaee538dc91d70a
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Mon May 22 15:07:24 2023 +0300

    querylog: imp code

commit cc62c1c4ae8b813d03ccf51b596ba1ebf44d9a1f
Merge: 37ace34e9 24b41100c
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Mon May 22 13:09:10 2023 +0300

    Merge remote-tracking branch 'origin/master' into querylog-imp-code

commit 37ace34e91e5189bef6e774db960f40cdaa18270
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Mon May 22 11:23:08 2023 +0300

    querylog: imp code

commit 8417815a6349f10b5dbad410ce28aab98bc479fa
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Mon May 22 11:08:29 2023 +0300

    querylog: imp docs

commit 4e5cde74d25713f78675aa3e18083b4fb5e619f3
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Fri May 19 16:41:34 2023 +0300

    querylog: imp code

commit 3494eab7006240f652a0217d305ac916bd6c3c83
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Fri May 19 16:13:08 2023 +0300

    all: lint script

commit 704534ce6278e7d9b1bef30a3acc4e59f25693bc
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Fri May 19 16:12:04 2023 +0300

    querylog: imp code

commit 48510102a2fa5187f78067d2b9157dac62f8bb56
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Fri May 19 15:52:57 2023 +0300

    querylog: imp code

commit 89c273aea0e6758eb749a2d3bbaf1bc385a57797
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Fri May 19 15:40:50 2023 +0300

    querylog: imp code

commit 0057fe64553ad38de0fda10efb9d3512c9a00e45
Author: Dimitry Kolyshev <dkolyshev@adguard.com>
Date:   Fri May 19 13:54:46 2023 +0300

    querylog: imp code

... and 1 more commit
2023-05-24 16:33:15 +03:00

170 lines
4.1 KiB
Go

package querylog
import (
"fmt"
"io"
"os"
"github.com/AdguardTeam/golibs/errors"
"github.com/AdguardTeam/golibs/log"
)
// qLogReader allows reading from multiple query log files in the reverse
// order.
//
// Please note that this is a stateful object. Internally, it contains a
// pointer to a particular query log file, and to a specific position in this
// file, and it reads lines in reverse order starting from that position.
type qLogReader struct {
// qFiles is an array with the query log files. The order is from oldest
// to newest.
qFiles []*qLogFile
// currentFile is the index of the current file.
currentFile int
}
// newQLogReader initializes a qLogReader instance with the specified files.
func newQLogReader(files []string) (*qLogReader, error) {
qFiles := make([]*qLogFile, 0)
for _, f := range files {
q, err := newQLogFile(f)
if err != nil {
if errors.Is(err, os.ErrNotExist) {
continue
}
// Close what we've already opened.
cErr := closeQFiles(qFiles)
if cErr != nil {
log.Debug("querylog: closing files: %s", cErr)
}
return nil, err
}
qFiles = append(qFiles, q)
}
return &qLogReader{qFiles: qFiles, currentFile: len(qFiles) - 1}, nil
}
// seekTS performs binary search of a query log record with the specified
// timestamp. If the record is found, it sets qLogReader's position to point
// to that line, so that the next ReadNext call returned this line.
func (r *qLogReader) seekTS(timestamp int64) (err error) {
for i := len(r.qFiles) - 1; i >= 0; i-- {
q := r.qFiles[i]
_, _, err = q.seekTS(timestamp)
if err != nil {
if errors.Is(err, errTSTooEarly) {
// Look at the next file, since we've reached the end of this
// one. If there is no next file, it's not found.
err = errTSNotFound
continue
} else if errors.Is(err, errTSTooLate) {
// Just seek to the start then. timestamp is probably between
// the end of the previous one and the start of this one.
return r.SeekStart()
} else if errors.Is(err, errTSNotFound) {
return err
} else {
return fmt.Errorf("seekts: file at index %d: %w", i, err)
}
}
// The search is finished, and the searched element has been found.
// Update currentFile only, position is already set properly in
// qLogFile.
r.currentFile = i
return nil
}
if err != nil {
return fmt.Errorf("seekts: %w", err)
}
return nil
}
// SeekStart changes the current position to the end of the newest file.
// Please note that we're reading query log in the reverse order and that's why
// the log starts actually at the end of file.
//
// Returns nil if we were able to change the current position. Returns error
// in any other cases.
func (r *qLogReader) SeekStart() error {
if len(r.qFiles) == 0 {
return nil
}
r.currentFile = len(r.qFiles) - 1
_, err := r.qFiles[r.currentFile].SeekStart()
return err
}
// ReadNext reads the next line (in the reverse order) from the query log
// files. Then shifts the current position left to the next (actually prev)
// line (or the next file).
//
// Returns io.EOF if there is nothing more to read.
func (r *qLogReader) ReadNext() (string, error) {
if len(r.qFiles) == 0 {
return "", io.EOF
}
for r.currentFile >= 0 {
q := r.qFiles[r.currentFile]
line, err := q.ReadNext()
if err != nil {
// Shift to the older file.
r.currentFile--
if r.currentFile < 0 {
break
}
q = r.qFiles[r.currentFile]
// Set its position to the start right away.
_, err = q.SeekStart()
// This is unexpected, return an error right away.
if err != nil {
return "", err
}
} else {
return line, nil
}
}
// Nothing to read anymore.
return "", io.EOF
}
// Close closes the qLogReader.
func (r *qLogReader) Close() error {
return closeQFiles(r.qFiles)
}
// closeQFiles is a helper method to close multiple qLogFile instances.
func closeQFiles(qFiles []*qLogFile) error {
var errs []error
for _, q := range qFiles {
err := q.Close()
if err != nil {
errs = append(errs, err)
}
}
if len(errs) > 0 {
return errors.List("error while closing qLogReader", errs...)
}
return nil
}