func createExceptionNoun() {
name := "verb"
buf := bytes.NewBuffer(nil)
f, _ := os.Open("/home/serega/Copy/database/dict/" + name + ".exc") // Error handling elided for brevity.
io.Copy(buf, f) // Error handling elided for brevity.
f.Close()
sentense := (buf.String())
buf.Reset()
rubbx = frub.NewRubAr()
sentense = strings.ToLower(sentense)
lin := regexp.MustCompile(`\n`).Split(sentense, -1)
//word := make(ByLength, 2000)
//wordArray := make([]string, 2000)
findIsh := make(map[string]string)
//findForm := make(map[string]map[string]string)
//fmt.Println(sentense)
for _, s := range lin {
words := regexp.MustCompile(` `).Split(s, -1)
if len(words) == 2 {
findIsh[words[0]] = words[1]
}
}
fmt.Println(len(findIsh))
frub.CreateByZn(findIsh, "file/englishLemm/find"+name+".gob")
wodfFile, _ := os.Create("file/englishLemm/find" + name)
frub.Encode(findIsh, wodfFile)
//frub.CreateByZn(findForm, "file/englishLemm/form" + name + ".gob")
//wodfFile, _ = os.Create("file/englishLemm/form" + name)
//frub.Encode(findForm, wodfFile)
}
func (nn *MorfE) InitEngl() {
nn.Graf.Init()
nn.Unions = make(map[string]map[string]string, 0)
frub.Decode(&nn.Unions, "file/englishLemm/noFounf")
nn.frequence = make(map[string]int, 0)
frub.GetByZn(&nn.frequence, "file/englishLemm/frequence1.gob")
nn.fraz = make(map[string][][]string)
frub.Decode(&nn.fraz, "file/englishLemm/fraz")
nn.expVerb = make(map[string][]string, 0)
frub.GetByZn(&nn.expVerb, "file/englishLemm/findverb1.gob")
nn.expVerb1 = make(map[string]string, 0)
frub.GetByZn(&nn.expVerb, "file/englishLemm/findverb.gob")
nn.expAdj = make(map[string][]string, 0)
frub.GetByZn(&nn.expAdj, "file/englishLemm/findadj1.gob")
nn.expNoun = make(map[string]string, 0)
frub.Decode(&nn.expNoun, "file/englishLemm/findnoun")
nn.Verb = make(map[string]string, 0)
frub.Decode(&nn.Verb, "file/englishLemm/verb")
nn.Noun = make(map[string]string, 0)
frub.Decode(&nn.Noun, "file/englishLemm/noun")
nn.Adj = make(map[string]string, 0)
frub.Decode(&nn.Adj, "file/englishLemm/adj")
nn.Adv = make(map[string]string, 0)
frub.Decode(&nn.Adv, "file/englishLemm/adv")
nn.rubbx = frub.NewRubAr()
}
func translateSentense() {
dbinfo := fmt.Sprintf("user=%s password=%s dbname=%s sslmode=disable",
DB_USER, DB_PASSWORD, DB_NAME)
db, err := sql.Open("postgres", dbinfo)
if err != nil {
log.Fatal(err)
}
defer db.Close()
rows, err := db.Query(`SELECT id, text, count FROM sentenses1 WHERE trans = '' ORDER BY count ASC `)
if err != nil {
log.Fatal(err)
}
var text string
var countElement, id, i, i1, j int
buf := bytes.NewBuffer(nil)
rubbx = f.NewRubAr()
for rows.Next() {
rows.Scan(&id, &text, &countElement)
buf.WriteString(text)
//fmt.Println(text)
if i >= 0 {
respons, _ := getTranslation("en", "ru", buf.String())
respons = rubbx.ReplaseRub(respons, `"`, `'`)
mapResult := (spltOnDimiter(respons))
//sql := `UPDATE sentenses1 SET trans = '`+ respons +`' WHERE id = ` + strconv.Itoa(id) + ` `
//_, err = db.Exec(sql)
//if err != nil {
// fmt.Println(sql)
// log.Fatal(err)
//}
for k, v := range mapResult {
sql := `UPDATE sentenses1 SET trans = '` + v + `' WHERE id = ` + strconv.Itoa(k) + ` `
_, err = db.Exec(sql)
if err != nil {
fmt.Println(sql)
log.Fatal(err)
}
}
fmt.Println(buf.String())
fmt.Println(j)
buf.Reset()
i = 0
if i1 >= 0 {
time.Sleep(0 * time.Second)
i1 = 0
} else {
i1++
}
//break
} else {
buf.WriteString(` || ` + strconv.Itoa(id) + ` || `)
i++
}
j++
}
}
func createEngWord() {
name := "verb"
buf := bytes.NewBuffer(nil)
f, _ := os.Open("/home/serega/Copy/database/dict/data." + name) // Error handling elided for brevity.
io.Copy(buf, f) // Error handling elided for brevity.
f.Close()
sentense := (buf.String())
buf.Reset()
rubbx = frub.NewRubAr()
sentense = strings.ToLower(sentense)
lin := regexp.MustCompile(`\n`).Split(sentense, -1)
word := make(map[string]string, 20000)
for k, s := range lin {
if k < 29 {
continue
}
words := regexp.MustCompile(` `).Split(s, -1)
cicl := true
key := 4
for cicl {
if len(words) > key {
pr := rubbx.IsTru1(words[key], `^[a-zA-Z]+$`)
if pr {
word[words[key]] = name
} else {
if rubbx.IsTru1(words[key], `^[a-zA-Z_]+$`) {
words[key] = rubbx.ReplaseRub(words[key], ` `, `_`)
word[words[key]] = "fraz"
}
//break
}
}
if len(words) > key+2 {
if words[key+1] != "0" {
cicl = false
}
key = key + 2
} else {
cicl = false
}
}
}
frub.CreateByZn(word, "file/englishLemm/"+name+".gob")
wodfFile, _ := os.Create("file/englishLemm/" + name)
frub.Encode(word, wodfFile)
fmt.Println(len(word))
}
func removeExcess() {
dbinfo := fmt.Sprintf("user=%s password=%s dbname=%s sslmode=disable",
DB_USER, DB_PASSWORD, DB_NAME)
db, err := sql.Open("postgres", dbinfo)
if err != nil {
log.Fatal(err)
}
defer db.Close()
rows, err := db.Query(`SELECT id, text, count FROM sentenses ORDER BY count ASC `)
if err != nil {
log.Fatal(err)
}
var text string
var countElement, id, i, j, chet int
rubbx = f.NewRubAr()
var gref Graf.Graf
gref.Init()
morfEng.InitEngl()
sqlStart := `DELETE FROM sentenses WHERE 1=1 AND (`
buf := bytes.NewBuffer(nil)
for rows.Next() {
rows.Scan(&id, &text, &countElement)
_, prov := morfEng.getWordWithoutOmonemii(text)
if !prov {
i++
//fmt.Println(i)
buf.WriteString(`id = ` + strconv.Itoa(id))
if j >= 300 {
_, err = db.Exec(sqlStart + buf.String() + ")")
if err != nil {
fmt.Println(sqlStart + buf.String())
log.Fatal(err)
}
j = 0
buf.Reset()
} else {
buf.WriteString(` or `)
j++
}
}
chet++
if chet%10000 == 0 {
fmt.Printf("%d - %d - %d\r\n", i, chet, j)
}
}
fmt.Println(i)
}
// инициализация словарей
func (nn *Morf) Init() {
nn.ItogOk = make([]map[string][]map[string]string, 20503)
nn.Search = make(map[string]map[string][]uint16, 234457)
var wg sync.WaitGroup
wg.Add(2)
go func() {
defer wg.Done()
f.GetByZn(&nn.ItogOk, "file/itogOkI")
}()
go func() {
defer wg.Done()
f.GetByZn(&nn.Search, "file/search1")
}()
wg.Wait()
nn.rubbx = f.NewRubAr()
}
func extractFromQz() {
soursStart := "/media/serega/c48f1bd8-a939-4630-ab12-9787df1f1fa0/home/paralel/OpenSubtitles2016/xml/en/"
allGzFile := dirHod(soursStart, "txt", []string{})
rubbx = f.NewRubAr()
fmt.Println(len(allGzFile))
sours := "/media/serega/c48f1bd8-a939-4630-ab12-9787df1f1fa0/home/paralel/OpenSubtitles2016/xml/en1/"
for k, v := range allGzFile {
//path := rubbx.ReplaseRub(v, ``, `[0-9]*\.xml.gz`)
//name := strings.Replace(v, path, "", -1)
ungzip(v, sours)
//fmt.Println(name)
if k%100 == 0 {
fmt.Println(k)
}
//break
}
//fmt.Println(allGzFile)
}
func markerOnEnglish() {
dbinfo := fmt.Sprintf("user=%s password=%s dbname=%s sslmode=disable",
DB_USER, DB_PASSWORD, DB_NAME)
db, err := sql.Open("postgres", dbinfo)
if err != nil {
log.Fatal(err)
}
defer db.Close()
rows, err := db.Query(`SELECT id, trans, count FROM sentenses1 ORDER BY count ASC `)
if err != nil {
log.Fatal(err)
}
var text string
var countElement, id, i int
rubbx = f.NewRubAr()
for rows.Next() {
rows.Scan(&id, &text, &countElement)
if !rubbx.IsTru1(text, `^[а-яА-Я ,.!?&\-0-9\:\;]*$`) {
sql := `UPDATE sentenses1 SET engword = 'e' WHERE id = ` + strconv.Itoa(id) + ` `
_, err = db.Exec(sql)
if err != nil {
fmt.Println(sql)
log.Fatal(err)
}
//fmt.Println(text)
//fmt.Println("en")
}
if i > 2000 {
//break
}
if i%1000 == 0 {
fmt.Println(i)
}
i++
}
}
func createException() {
name := "verb1"
buf := bytes.NewBuffer(nil)
f, _ := os.Open("/home/serega/Copy/database/dict/" + name + ".exc") // Error handling elided for brevity.
io.Copy(buf, f) // Error handling elided for brevity.
f.Close()
sentense := (buf.String())
buf.Reset()
rubbx = frub.NewRubAr()
sentense = strings.ToLower(sentense)
lin := regexp.MustCompile(`\n`).Split(sentense, -1)
//word := make(ByLength, 2000)
//wordArray := make([]string, 2000)
findIsh := make(map[string][]string)
//findForm := make(map[string]map[string]string)
//fmt.Println(sentense)
for _, s := range lin {
words := regexp.MustCompile(` `).Split(s, -1)
if len(words) > 1 {
//findForm[words[0]] = map[string]string{"v1":words[0]}
words1 := regexp.MustCompile(`\/`).Split(words[1], -1)
words2 := regexp.MustCompile(`\/`).Split(words[2], -1)
//fmt.Println(words1)
//findForm[words[0]]["v2"] = words1[0]
for _, s1 := range words1 {
types := "v2"
if s1 == words[0] {
types = "u1"
}
for _, s2 := range words2 {
if s1 == s2 {
if types == "v2" {
types = "u2"
} else {
types = "u"
}
}
}
findIsh[s1] = []string{words[0], types}
}
//fmt.Println(words1)
//findForm[words[0]]["v3"] = words1[0]
for _, s1 := range words2 {
types := "v3"
if s1 == words[0] {
types = "u3"
}
for _, s2 := range words1 {
if s1 == s2 {
if types == "v3" {
types = "u2"
} else {
types = "u"
}
}
}
findIsh[s1] = []string{words[0], types}
}
}
}
frub.CreateByZn(findIsh, "file/englishLemm/find"+name+".gob")
wodfFile, _ := os.Create("file/englishLemm/find" + name)
frub.Encode(findIsh, wodfFile)
//frub.CreateByZn(findForm, "file/englishLemm/form" + name + ".gob")
//wodfFile, _ = os.Create("file/englishLemm/form" + name)
//frub.Encode(findForm, wodfFile)
}
func spliteonPredl() {
dbinfo := fmt.Sprintf("user=%s password=%s dbname=%s sslmode=disable",
DB_USER, DB_PASSWORD, DB_NAME)
db, err := sql.Open("postgres", dbinfo)
if err != nil {
log.Fatal(err)
}
defer db.Close()
//allName = dirHod("/home/serega/Copy/code/text/book/",, allName)
allName := dirHod("/media/serega/c48f1bd8-a939-4630-ab12-9787df1f1fa0/home/paralel/OpenSubtitles2016/xml/txt/", "", []string{})
//fmt.Println(allName)
//echoArray(allName)
var gref Graf.Graf
gref.Init()
rubbx = f.NewRubAr()
morfEng.InitEngl()
buf := bytes.NewBuffer(nil)
for k, f1 := range allName {
if k < 30002 {
// for /media/serega/c48f1bd8-a939-4630-ab12-9787df1f1fa0/home/paralel/OpenSubtitles2016/xml/txt/
continue
}
buf.Reset()
f, _ := os.Open(f1) // Error handling elided for brevity.
io.Copy(buf, f) // Error handling elided for brevity.
f.Close()
//text, _ := html2text.FromString(buf.String())
sentense := gref.PrepereText(buf.String())
//echoArray(sentense)
//fmt.Println(f1)
//buf.Reset()
i := 0
sqlStart := `INSERT INTO sentenses(trans, text, count) VALUES `
buf1 := bytes.NewBuffer(nil)
for _, s := range sentense {
_ = s
if rubbx.IsTru1(s, `.*[-"].*`) {
continue
}
sentenseItem, prov := morfEng.getWordWithoutOmonemii(s)
if !prov || len(sentenseItem) < 4 || len(sentenseItem) > 20 {
continue
}
s = rubbx.ReplaseRub(s, `''`, `'`)
rows, err := db.Query(`SELECT count(*) as cou FROM sentenses WHERE text = '` + s + `' `)
if err != nil {
log.Fatal(err)
}
var cou string
rows.Next()
rows.Scan(&cou)
rows.Close()
// //fmt.Println(cou)
if cou == `0` && len(sentenseItem) > 4 && len(sentenseItem) < 20 {
//fmt.Println(s)
buf1.WriteString(`('','` + s + `',` + strconv.Itoa(len(sentenseItem)) + `)`)
if i >= 300 {
_, err = db.Exec(sqlStart + buf1.String())
if err != nil {
fmt.Println(sqlStart + buf1.String())
log.Fatal(err)
}
i = 0
buf1.Reset()
} else {
buf1.WriteString(`,`)
i++
}
}
// //
// // ////fmt.Println(" ---------- ")
}
if i != 0 {
str := buf1.String()
str = string([]byte(str)[:len(str)-1])
_, err = db.Exec(sqlStart + str)
if err != nil {
fmt.Println(sqlStart + buf1.String())
log.Fatal(err)
}
}
//
if k%1 == 0 {
fmt.Println(f1)
fmt.Println(k)
}
if k > 10 {
//break
}
////fmt.Println(f)
//break
}
}
func fillTableWord() {
dbinfo := fmt.Sprintf("user=%s password=%s dbname=%s sslmode=disable",
DB_USER, DB_PASSWORD, DB_NAME)
db, err := sql.Open("postgres", dbinfo)
if err != nil {
log.Fatal(err)
}
defer db.Close()
rows, err := db.Query(`SELECT id, text, count FROM sentenses ORDER BY count ASC `)
if err != nil {
log.Fatal(err)
}
var text string
var countElement, id, i int
rubbx = f.NewRubAr()
var gref Graf.Graf
gref.Init()
type words struct {
result string
types string
forma string
count int
}
type read struct {
text string
countElement int
id int
}
word := make(map[string]words, 0)
word_use := make(map[string][]int, 0)
readderArray := make([]read, 6762636)
morfEng.InitEngl()
localCou := 0
for rows.Next() {
rows.Scan(&id, &text, &countElement)
readderArray[localCou] = read{text, countElement, id}
localCou++
}
for _, rowFrom := range readderArray {
i++
id = rowFrom.id
text = rowFrom.text
countElement = rowFrom.countElement
sentenseItem, prov := morfEng.getWordWithoutOmonemii(text)
if prov {
fraz := []int{}
for k, s := range sentenseItem {
if len(fraz) > 0 {
//fraz--
if fraz[0] == k {
fraz = fraz[1:]
continue
}
}
if s.types != "number" && s.types != "punctuation" {
if zn, found := morfEng.fraz[s.result]; found {
k1 := 0
shift := 1
for {
fraz2 := false
find := false
for _, v1 := range zn {
if k+shift+k1 < len(sentenseItem) {
val := sentenseItem[k+shift+k1].result
if len(v1) > k1 && val == v1[k1] {
fraz = append(fraz, k+shift+k1)
find = true
if len(v1)-1 == k1 {
for jc := 0; jc < len(v1); jc++ {
s.result += " " + sentenseItem[k+jc+shift].result
}
s.types = "phras"
if shift == 1 {
s.forma = ""
} else {
s.forma = "stm"
}
shift = 2
break
} else {
fraz2 = true
}
}
}
}
if !find {
}
if !fraz2 && shift == 2 {
break
}
if !fraz2 {
shift++
}
k1++
}
}
s.result = rubbx.ReplaseRub((s.result), `''`, `'`)
keyWord := s.result + s.types + s.forma
if _, found := word_use[keyWord]; found {
word_use[keyWord] = append(word_use[keyWord], id)
} else {
word_use[keyWord] = []int{id}
}
if _, found := word[keyWord]; found {
word[keyWord] = words{s.result, s.types, s.forma, word[keyWord].count + 1}
} else {
word[keyWord] = words{s.result, s.types, s.forma, 1}
}
}
}
}
if i%1000 == 0 {
fmt.Println(i)
//break
}
if i > 100000 {
//break
}
}
//for k, v := range word {
// if v.forma == "phras"{
// fmt.Printf("%s - %v+\n", k, v)
// }
//}
//fmt.Println(word)
//fmt.Println(word_use)
i = 0
jj := 0
sqlStart := `INSERT INTO words (word_key, word, forma, type, frequensy) VALUES `
buf := bytes.NewBuffer(nil)
for k, s := range word {
buf.WriteString(`('` + k + `','` + s.result + `','` + (s.forma) + `','` + (s.types) + `',` + strconv.Itoa(s.count) + `)`)
if i >= 300 {
_, err = db.Exec(sqlStart + buf.String())
if err != nil {
fmt.Println(sqlStart + buf.String())
log.Fatal(err)
}
i = 0
buf.Reset()
} else {
buf.WriteString(`,`)
i++
}
jj++
if jj%10000 == 0 {
fmt.Println(jj)
//break
}
}
if i != 0 {
str := buf.String()
str = string([]byte(str)[:len(str)-1])
_, err = db.Exec(sqlStart + str)
if err != nil {
fmt.Println(sqlStart + buf.String())
log.Fatal(err)
}
}
i = 0
jj = 0
sqlStart = `INSERT INTO word_use (id_word, id_sentense) VALUES `
buf = bytes.NewBuffer(nil)
var id_wodr int
for k, s := range word_use {
rows1, _ := db.Query(`SELECT id FROM words WHERE word_key = '` + k + `'`)
rows1.Next()
rows1.Scan(&id_wodr)
rows1.Close()
for _, s1 := range s {
buf.WriteString(`('` + strconv.Itoa(id_wodr) + `',` + strconv.Itoa(s1) + `)`)
if i >= 300 {
_, err = db.Exec(sqlStart + buf.String())
if err != nil {
fmt.Println(sqlStart + buf.String())
log.Fatal(err)
}
i = 0
buf.Reset()
} else {
buf.WriteString(`,`)
i++
}
}
jj++
if jj%10000 == 0 {
fmt.Println(jj)
//break
}
}
if i != 0 {
str := buf.String()
str = string([]byte(str)[:len(str)-1])
_, err = db.Exec(sqlStart + str)
if err != nil {
fmt.Println(sqlStart + buf.String())
log.Fatal(err)
}
}
}