api breakage: rework the subtitle dl code, added usage example

.gitignore

@@ -0,0 +1 @@
+conformance-files/

README.md

@@ -16,6 +16,18 @@ go get github.com/martinlindhe/subtitles
 ```
 
 
+# Example
+
+Fetch subtitle from thesubdb.com:
+```go
+f, _ := os.Open(fileName)
+
+finder := NewSubFinder(f, fileName, "en")
+
+text, err := finder.TheSubDb()
+```
+
+
 # See also
 
 - [subber](https://github.com/martinlindhe/subber) command line tool for subtitles

dl-conformance-files

@@ -0,0 +1,7 @@
+#!/bin/sh
+
+mkdir -p conformance-files/thesubdb
+
+# fetch conformance check videos from http://thesubdb.com/api/
+curl http://thesubdb.com/api/samples/dexter.mp4    -o conformance-files/thesubdb/dexter.mp4
+curl http://thesubdb.com/api/samples/justified.mp4 -o conformance-files/thesubdb/justified.mp4

finder.go

@@ -0,0 +1,21 @@
+package subtitles
+
+import "os"
+
+// SubFinder represents a video being queried for subtitles
+type SubFinder struct {
+	FileName  string
+	Language  string
+	VideoFile *os.File
+	Quiet     bool
+}
+
+// NewSubFinder creates a SubFilePair object used to download subs for a video
+func NewSubFinder(video *os.File, fileName string, language string) *SubFinder {
+
+	return &SubFinder{
+		FileName:  fileName,
+		Language:  language,
+		VideoFile: video,
+	}
+}

testextras.go

@@ -9,17 +9,20 @@ import (
 const tempFilePrefix = "moviehash-temp"
 
 func check(e error) {
+
 	if e != nil {
 		fmt.Println(e)
 		panic(e)
 	}
 }
 
 func makeTime(h int, m int, s int, ms int) time.Time {
+
 	return time.Date(0, 1, 1, h, m, s, ms*1000*1000, time.UTC)
 }
 
 func createTempFile(byteSize int) string {
+
 	data := make([]byte, byteSize)
 
 	cnt := uint8(0)
@@ -39,6 +42,7 @@ func createTempFile(byteSize int) string {
 }
 
 func createZeroedTempFile(byteSize int) string {
+
 	data := make([]byte, byteSize)
 
 	f, err := ioutil.TempFile("/tmp", tempFilePrefix)

thesubdb.go

@@ -6,72 +6,80 @@ import (
 	"io/ioutil"
 	"net/http"
 	"os"
-
-	log "github.com/Sirupsen/logrus"
 )
 
-// FindSub finds subtitle online, returns untouched data
-func FindSub(videoFileName string, language string) ([]byte, error) {
-	if !exists(videoFileName) {
-		return nil, fmt.Errorf("%s not found", videoFileName)
-	}
+// TheSubDb downloads a subtitle from thesubdb.com
+func (f SubFinder) TheSubDb(args ...string) ([]byte, error) {
 
-	if isDirectory(videoFileName) {
-		return nil, fmt.Errorf("%s is not a file", videoFileName)
+	apiHost := ""
+	if len(args) > 0 {
+		apiHost = args[0]
+	} else {
+		apiHost = "api.thesubdb.com"
 	}
 
-	text, err := fromTheSubDb(videoFileName, language)
+	hash, err := SubDbHashFromFile(f.VideoFile)
 	if err != nil {
 		return nil, err
 	}
 
-	return text, nil
-}
+	client := &http.Client{}
 
-// FromTheSubDb downloads a subtitle from thesubdb.com
-func fromTheSubDb(videoFileName string, language string, optional ...string) ([]byte, error) {
+	query := "http://" + apiHost +
+		"/?action=download" +
+		"&hash=" + hash +
+		"&language=" + f.Language
 
-	_apiHost := "api.thesubdb.com"
-	if len(optional) > 0 {
-		_apiHost = optional[0]
+	if !f.Quiet {
+		fmt.Println("Fetching", query, "...")
 	}
 
-	hash, err := createMovieHashFromMovieFile(videoFileName)
+	req, err := http.NewRequest("GET", query, nil)
 	if err != nil {
 		return nil, err
 	}
 
-	actualText, err := downloadSubtitleByHash(hash, language, _apiHost)
+	req.Header.Set("User-Agent",
+		"SubDB/1.0 (GoSubber/1.0; https://github.com/martinlindhe/subber)")
+
+	resp, err := client.Do(req)
 	if err != nil {
 		return nil, err
 	}
 
-	return actualText, nil
-}
+	if resp.StatusCode == 404 {
+		return nil, fmt.Errorf("Subtitle not found")
+	}
 
-// returns a md5-sum in hex-string representation
-func createMovieHashFromMovieFile(fileName string) (string, error) {
+	if resp.StatusCode != 200 {
+		return nil, fmt.Errorf("Server error %s", resp.Status)
+	}
 
-	if !exists(fileName) {
-		return "", fmt.Errorf("File %s not found", fileName)
+	slurp, err := ioutil.ReadAll(resp.Body)
+	if err != nil {
+		return nil, fmt.Errorf("Server reading request body: %v", err)
 	}
 
+	return slurp, nil
+}
+
+// SubDbHashFromFile returns a checksum in hex-string representation
+// conforming to http://trac.opensubtitles.org/projects/opensubtitles/wiki/HashSourceCodes
+func SubDbHashFromFile(f *os.File) (string, error) {
+
+	// rewind
+	f.Seek(0, 0)
+
 	// block size which is required for the API call
 	readSize := int64(64 * 1024)
 
-	f, err := os.Open(fileName)
-	if err != nil {
-		return "", err
-	}
-	defer f.Close()
-
 	fi, err := f.Stat()
 	if err != nil {
 		return "", err
 	}
 
 	if fi.Size() < readSize {
-		return "", fmt.Errorf("File is too small: %s", fileName)
+		return "", fmt.Errorf("Stream is too small: %d", fi.Size())
 	}
 
 	// read first part
@@ -99,40 +107,3 @@ func createMovieHashFromMovieFile(fileName string) (string, error) {
 
 	return fmt.Sprintf("%x", md5.Sum(combined)), nil
 }
-
-func downloadSubtitleByHash(hash string, language string, apiHost string) ([]byte, error) {
-
-	client := &http.Client{}
-
-	query := "http://" + apiHost + "/?action=download&hash=" + hash + "&language=" + language
-
-	log.Printf("Fetching %s ...\n", query)
-
-	req, err := http.NewRequest("GET", query, nil)
-	if err != nil {
-		return nil, err
-	}
-
-	req.Header.Set("User-Agent",
-		"SubDB/1.0 (GoSubber/1.0; https://github.com/martinlindhe/subber)")
-
-	resp, err := client.Do(req)
-	if err != nil {
-		return nil, err
-	}
-
-	if resp.StatusCode == 404 {
-		return nil, fmt.Errorf("Subtitle not found")
-	}
-
-	if resp.StatusCode != 200 {
-		return nil, fmt.Errorf("Server error %s", resp.Status)
-	}
-
-	slurp, err := ioutil.ReadAll(resp.Body)
-	if err != nil {
-		return nil, fmt.Errorf("Server reading request body: %v", err)
-	}
-
-	return slurp, nil
-}

thesubdb_test.go

@@ -1,30 +1,53 @@
 package subtitles
 
 import (
+	"fmt"
 	"os"
 	"testing"
 
 	"github.com/stretchr/testify/assert"
 )
 
-func TestCreateMovieHashFromMovieFile(t *testing.T) {
+func TestDownloadFromTheSubDb(t *testing.T) {
 
-	fileName := createTempFile(1024 * 1024 * 2)
+	fileName := createZeroedTempFile(1024 * 1024 * 4)
+	defer os.Remove(fileName)
 
-	hash, err := createMovieHashFromMovieFile(fileName)
+	f, err := os.Open(fileName)
+	assert.Equal(t, nil, err)
 
+	hash, err := SubDbHashFromFile(f)
 	assert.Equal(t, nil, err)
-	assert.Equal(t, "38a503307786991a982f8ded498b90e0", hash)
+	assert.Equal(t, "0dfbe8aa4c20b52e1b8bf3cb6cbdf193", hash)
 
-	os.Remove(fileName)
+	finder := NewSubFinder(f, fileName, "en")
+
+	text, err := finder.TheSubDb("sandbox.thesubdb.com")
+	assert.Equal(t, nil, err)
+	assert.True(t, len(text) > 1000)
 }
 
-func TestDownloadFromTheSubDb(t *testing.T) {
-	fileName := createZeroedTempFile(1024 * 1024 * 2)
+func subDbConformTest(t *testing.T, fileName string, expectedHash string) {
+	if !exists(fileName) {
+		fmt.Println("ERROR thesubdb.com conformance tests missing, run ./hash-conformance-deps if you want to run these tests")
+		return
+	}
 
-	text, err := fromTheSubDb(fileName, "en", "sandbox.thesubdb.com")
+	f, err := os.Open(fileName)
 	assert.Equal(t, nil, err)
-	assert.True(t, len(text) > 1000)
 
-	os.Remove(fileName)
+	hash, err := SubDbHashFromFile(f)
+	assert.Equal(t, nil, err)
+	assert.Equal(t, expectedHash, hash)
+}
+
+func TestSubDbHashFromFile(t *testing.T) {
+
+	// NOTE for this to work, run "./hash-conformance-deps" to fetch needed files
+
+	// http://thesubdb.com/api/samples/dexter.mp4
+	subDbConformTest(t, "conformance-files/thesubdb/dexter.mp4", "ffd8d4aa68033dc03d1c8ef373b9028c")
+
+	// http://thesubdb.com/api/samples/justified.mp4
+	subDbConformTest(t, "conformance-files/thesubdb/justified.mp4", "edc1981d6459c6111fe36205b4aff6c2")
 }