mirror of
https://github.com/ralsina/tartrazine.git
synced 2025-06-19 14:43:05 -03:00
Added frequencies.go generation
This commit is contained in:
@ -21,6 +21,14 @@ func FromFile(fileToParse, outPath, tmplPath, tmplName, commit string, generate
|
||||
return err
|
||||
}
|
||||
|
||||
if err := formatedWrite(outPath, source); err != nil {
|
||||
return err
|
||||
}
|
||||
|
||||
return nil
|
||||
}
|
||||
|
||||
func formatedWrite(outPath string, source []byte) error {
|
||||
formatedSource, err := format.Source(source)
|
||||
if err != nil {
|
||||
return err
|
||||
|
@ -4,15 +4,20 @@ import (
|
||||
"fmt"
|
||||
"io/ioutil"
|
||||
"os"
|
||||
"os/exec"
|
||||
"path/filepath"
|
||||
"testing"
|
||||
|
||||
"github.com/stretchr/testify/assert"
|
||||
"github.com/stretchr/testify/suite"
|
||||
)
|
||||
|
||||
const (
|
||||
lingustURL = "https://github.com/github/linguist.git"
|
||||
commitTree = "60f864a138650dd17fafc94814be9ee2d3aaef8c"
|
||||
commitTest = "fe8b44ab8a225b1ffa75b983b916ea22fee5b6f7"
|
||||
|
||||
// Languages test
|
||||
// Extensions test
|
||||
extensionsTestFile = "test_files/extensions.test.yml"
|
||||
extensionsGold = "test_files/extensions.gold"
|
||||
extensionsTestTmplPath = "../assets/extensions.go.tmpl"
|
||||
@ -59,9 +64,48 @@ const (
|
||||
aliasesGold = "test_files/aliases.gold"
|
||||
aliasesTestTmplPath = "../assets/aliases.go.tmpl"
|
||||
aliasesTestTmplName = "aliases.go.tmpl"
|
||||
|
||||
// Frequencies test
|
||||
frequenciesTestDir = "/samples"
|
||||
frequenciesGold = "test_files/frequencies.gold"
|
||||
frequenciesTestTmplPath = "../assets/frequencies.go.tmpl"
|
||||
frequenciesTestTmplName = "frequencies.go.tmpl"
|
||||
)
|
||||
|
||||
func TestFromFile(t *testing.T) {
|
||||
type GeneratorTestSuite struct {
|
||||
suite.Suite
|
||||
tmpLinguist string
|
||||
}
|
||||
|
||||
func (g *GeneratorTestSuite) SetupSuite() {
|
||||
tmpLinguist, err := ioutil.TempDir("", "linguist-")
|
||||
assert.NoError(g.T(), err)
|
||||
g.tmpLinguist = tmpLinguist
|
||||
|
||||
cmd := exec.Command("git", "clone", lingustURL, tmpLinguist)
|
||||
err = cmd.Run()
|
||||
assert.NoError(g.T(), err)
|
||||
|
||||
cwd, err := os.Getwd()
|
||||
assert.NoError(g.T(), err)
|
||||
|
||||
err = os.Chdir(tmpLinguist)
|
||||
assert.NoError(g.T(), err)
|
||||
|
||||
cmd = exec.Command("git", "checkout", commitTree)
|
||||
err = cmd.Run()
|
||||
assert.NoError(g.T(), err)
|
||||
|
||||
err = os.Chdir(cwd)
|
||||
assert.NoError(g.T(), err)
|
||||
}
|
||||
|
||||
func (g *GeneratorTestSuite) TearDownSuite() {
|
||||
err := os.RemoveAll(g.tmpLinguist)
|
||||
assert.NoError(g.T(), err)
|
||||
}
|
||||
|
||||
func (g *GeneratorTestSuite) TestFromFile() {
|
||||
tests := []struct {
|
||||
name string
|
||||
fileToParse string
|
||||
@ -145,20 +189,57 @@ func TestFromFile(t *testing.T) {
|
||||
},
|
||||
}
|
||||
|
||||
for _, tt := range tests {
|
||||
t.Run(tt.name, func(t *testing.T) {
|
||||
gold, err := ioutil.ReadFile(tt.wantOut)
|
||||
assert.NoError(t, err)
|
||||
for _, test := range tests {
|
||||
gold, err := ioutil.ReadFile(test.wantOut)
|
||||
assert.NoError(g.T(), err)
|
||||
|
||||
outPath, err := ioutil.TempFile("/tmp", "generator-test-")
|
||||
assert.NoError(t, err)
|
||||
defer os.Remove(outPath.Name())
|
||||
outPath, err := ioutil.TempFile("/tmp", "generator-test-")
|
||||
assert.NoError(g.T(), err)
|
||||
defer os.Remove(outPath.Name())
|
||||
|
||||
err = FromFile(tt.fileToParse, outPath.Name(), tt.tmplPath, tt.tmplName, tt.commit, tt.generate)
|
||||
assert.NoError(t, err)
|
||||
out, err := ioutil.ReadFile(outPath.Name())
|
||||
assert.NoError(t, err)
|
||||
assert.EqualValues(t, gold, out, fmt.Sprintf("FromFile() = %v, want %v", string(out), string(tt.wantOut)))
|
||||
})
|
||||
err = FromFile(test.fileToParse, outPath.Name(), test.tmplPath, test.tmplName, test.commit, test.generate)
|
||||
assert.NoError(g.T(), err)
|
||||
out, err := ioutil.ReadFile(outPath.Name())
|
||||
assert.NoError(g.T(), err)
|
||||
assert.EqualValues(g.T(), gold, out, fmt.Sprintf("FromFile() = %v, want %v", string(out), string(test.wantOut)))
|
||||
}
|
||||
}
|
||||
|
||||
func (g *GeneratorTestSuite) TestFrequencies() {
|
||||
tests := []struct {
|
||||
name string
|
||||
samplesDir string
|
||||
tmplPath string
|
||||
tmplName string
|
||||
commit string
|
||||
wantOut string
|
||||
}{
|
||||
{
|
||||
name: "Frequencies_1",
|
||||
samplesDir: filepath.Join(g.tmpLinguist, frequenciesTestDir),
|
||||
tmplPath: frequenciesTestTmplPath,
|
||||
tmplName: frequenciesTestTmplName,
|
||||
commit: commitTree,
|
||||
wantOut: frequenciesGold,
|
||||
},
|
||||
}
|
||||
|
||||
for _, test := range tests {
|
||||
gold, err := ioutil.ReadFile(test.wantOut)
|
||||
assert.NoError(g.T(), err)
|
||||
|
||||
outPath, err := ioutil.TempFile("/tmp", "frequencies-test-")
|
||||
assert.NoError(g.T(), err)
|
||||
defer os.Remove(outPath.Name())
|
||||
|
||||
err = Frequencies(test.samplesDir, test.tmplPath, test.tmplName, test.commit, outPath.Name())
|
||||
assert.NoError(g.T(), err)
|
||||
out, err := ioutil.ReadFile(outPath.Name())
|
||||
assert.NoError(g.T(), err)
|
||||
assert.EqualValues(g.T(), gold, out, fmt.Sprintf("Frequencies() = %v, want %v", string(out), string(test.wantOut)))
|
||||
}
|
||||
}
|
||||
|
||||
func TestGeneratorTestSuite(t *testing.T) {
|
||||
suite.Run(t, new(GeneratorTestSuite))
|
||||
}
|
||||
|
202
internal/code-generator/generator/samplesfreq.go
Normal file
202
internal/code-generator/generator/samplesfreq.go
Normal file
@ -0,0 +1,202 @@
|
||||
package generator
|
||||
|
||||
import (
|
||||
"bytes"
|
||||
"fmt"
|
||||
"io"
|
||||
"io/ioutil"
|
||||
"log"
|
||||
"math"
|
||||
"os"
|
||||
"path/filepath"
|
||||
"sort"
|
||||
"strconv"
|
||||
"text/template"
|
||||
|
||||
"gopkg.in/src-d/simple-linguist.v1/internal/tokenizer"
|
||||
)
|
||||
|
||||
type samplesFrequencies struct {
|
||||
LanguageTotal int `json:"language_total,omitempty"`
|
||||
Languages map[string]int `json:"languages,omitempty"`
|
||||
TokensTotal int `json:"tokens_total,omitempty"`
|
||||
Tokens map[string]map[string]int `json:"tokens,omitempty"`
|
||||
LanguageTokens map[string]int `json:"language_tokens,omitempty"`
|
||||
}
|
||||
|
||||
// Frequencies reads directories in samplesDir, retrieves information about frequencies of languages and tokens, and write
|
||||
// the file outPath using frequenciesTmplName as a template.
|
||||
func Frequencies(samplesDir, frequenciesTmplPath, frequenciesTmplName, commit, outPath string) error {
|
||||
freqs, err := getFrequencies(samplesDir)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
|
||||
buf := &bytes.Buffer{}
|
||||
if err := executeFrequenciesTemplate(buf, freqs, frequenciesTmplPath, frequenciesTmplName, commit); err != nil {
|
||||
return err
|
||||
}
|
||||
|
||||
if err := formatedWrite(outPath, buf.Bytes()); err != nil {
|
||||
return err
|
||||
}
|
||||
|
||||
return nil
|
||||
}
|
||||
|
||||
func getFrequencies(samplesDir string) (*samplesFrequencies, error) {
|
||||
entries, err := ioutil.ReadDir(samplesDir)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
|
||||
var languageTotal int
|
||||
var languages = make(map[string]int)
|
||||
var tokensTotal int
|
||||
var tokens = make(map[string]map[string]int)
|
||||
var languageTokens = make(map[string]int)
|
||||
|
||||
for _, entry := range entries {
|
||||
if !entry.IsDir() {
|
||||
log.Println(err)
|
||||
continue
|
||||
}
|
||||
|
||||
samples, err := getSamples(samplesDir, entry)
|
||||
if err != nil {
|
||||
log.Println(err)
|
||||
}
|
||||
|
||||
if len(samples) == 0 {
|
||||
continue
|
||||
}
|
||||
|
||||
samplesTokens, err := getTokens(samples)
|
||||
if err != nil {
|
||||
log.Println(err)
|
||||
continue
|
||||
}
|
||||
|
||||
lang := entry.Name()
|
||||
languageTotal += len(samples)
|
||||
languages[lang] = len(samples)
|
||||
tokensTotal += len(samplesTokens)
|
||||
languageTokens[lang] = len(samplesTokens)
|
||||
tokens[lang] = make(map[string]int)
|
||||
for _, token := range samplesTokens {
|
||||
tokens[lang][token]++
|
||||
}
|
||||
}
|
||||
|
||||
return &samplesFrequencies{
|
||||
TokensTotal: tokensTotal,
|
||||
LanguageTotal: languageTotal,
|
||||
Tokens: tokens,
|
||||
LanguageTokens: languageTokens,
|
||||
Languages: languages,
|
||||
}, nil
|
||||
}
|
||||
|
||||
func getSamples(samplesDir string, langDir os.FileInfo) ([]string, error) {
|
||||
const subDir = "filenames"
|
||||
|
||||
samples := []string{}
|
||||
path := filepath.Join(samplesDir, langDir.Name())
|
||||
entries, err := ioutil.ReadDir(path)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
|
||||
for _, entry := range entries {
|
||||
if entry.Mode().IsRegular() {
|
||||
samples = append(samples, filepath.Join(path, entry.Name()))
|
||||
}
|
||||
|
||||
if entry.IsDir() && entry.Name() == subDir {
|
||||
subSamples, err := getSubSamples(samplesDir, langDir.Name(), entry)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
|
||||
samples = append(samples, subSamples...)
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
return samples, nil
|
||||
}
|
||||
|
||||
func getSubSamples(samplesDir, langDir string, subLangDir os.FileInfo) ([]string, error) {
|
||||
subSamples := []string{}
|
||||
path := filepath.Join(samplesDir, langDir, subLangDir.Name())
|
||||
entries, err := ioutil.ReadDir(path)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
|
||||
for _, entry := range entries {
|
||||
if entry.Mode().IsRegular() {
|
||||
subSamples = append(subSamples, filepath.Join(path, entry.Name()))
|
||||
}
|
||||
}
|
||||
|
||||
return subSamples, nil
|
||||
}
|
||||
|
||||
func getTokens(samples []string) ([]string, error) {
|
||||
tokens := make([]string, 0, 20)
|
||||
var anyError error
|
||||
for _, sample := range samples {
|
||||
content, err := ioutil.ReadFile(sample)
|
||||
if err != nil {
|
||||
anyError = err
|
||||
continue
|
||||
}
|
||||
|
||||
t := tokenizer.Tokenize(content)
|
||||
tokens = append(tokens, t...)
|
||||
}
|
||||
|
||||
return tokens, anyError
|
||||
}
|
||||
|
||||
func executeFrequenciesTemplate(out io.Writer, freqs *samplesFrequencies, frequenciesTmplPath, frequenciesTmpl, commit string) error {
|
||||
fmap := template.FuncMap{
|
||||
"getCommit": func() string { return commit },
|
||||
"toFloat64": func(num int) string { return fmt.Sprintf("%f", float64(num)) },
|
||||
"orderKeys": func(m map[string]int) []string {
|
||||
keys := make([]string, 0, len(m))
|
||||
for key := range m {
|
||||
keys = append(keys, key)
|
||||
}
|
||||
|
||||
sort.Strings(keys)
|
||||
return keys
|
||||
},
|
||||
"languageLogProbability": func(language string) string {
|
||||
num := math.Log(float64(freqs.Languages[language]) / float64(freqs.LanguageTotal))
|
||||
return fmt.Sprintf("%f", num)
|
||||
},
|
||||
"orderMapMapKeys": func(mm map[string]map[string]int) []string {
|
||||
keys := make([]string, 0, len(mm))
|
||||
for key := range mm {
|
||||
keys = append(keys, key)
|
||||
}
|
||||
|
||||
sort.Strings(keys)
|
||||
return keys
|
||||
},
|
||||
"tokenLogProbability": func(language, token string) string {
|
||||
num := math.Log(float64(freqs.Tokens[language][token]) / float64(freqs.LanguageTokens[language]))
|
||||
return fmt.Sprintf("%f", num)
|
||||
},
|
||||
"quote": strconv.Quote,
|
||||
}
|
||||
|
||||
t := template.Must(template.New(frequenciesTmpl).Funcs(fmap).ParseFiles(frequenciesTmplPath))
|
||||
if err := t.Execute(out, freqs); err != nil {
|
||||
return err
|
||||
}
|
||||
|
||||
return nil
|
||||
}
|
128805
internal/code-generator/generator/test_files/frequencies.gold
Normal file
128805
internal/code-generator/generator/test_files/frequencies.gold
Normal file
File diff suppressed because it is too large
Load Diff
Reference in New Issue
Block a user