refactoring project structure

internal/fetcher/fetchhtml.go

@@ -0,0 +1,83 @@
+package main
+
+import(
+	"log"
+	"net/http"
+	"github.com/PuerkitoBio/goquery"
+	"strconv"
+	"strings"
+)
+type chapter struct{
+	verses []string 
+}
+
+type book struct{
+	chapters []chapter
+}
+
+func fetchChapter(book, searchedChapter, version string) (chapter, error){
+
+	url := "https://www.biblegateway.com/passage/?search=" + book + "%20" + searchedChapter +"&version=" + version
+
+	res, err := http.Get(url)
+	if err != nil {
+		log.Printf("can’t get %s : %v\n", url, err)
+		return chapter{}, err
+	}
+
+	defer res.Body.Close()
+	if res.StatusCode != http.StatusOK {
+		log.Printf("error code %v: %v\n", res.StatusCode, res.Status)
+		return chapter{}, err
+	}
+
+	doc, err := goquery.NewDocumentFromReader(res.Body)
+	if err != nil {
+		log.Printf("error reading the http response: %v\n", err)
+		return chapter{},  err
+	}
+
+	chapterText := chapter{}
+	replacer := strings.NewReplacer(
+				"L'Éternel", "YAHWEH",
+				"l'Éternel", "YAHWEH",
+				"Éternel", "YAHWEH",
+			)
+
+	 doc.Find(".verse > span").Each(func(i int, s *goquery.Selection){
+	/**********************************************************************************/
+	/**** verses are marked by an .verse class. Inside the text is in a span and  *****/
+	/****    begin either by the chapter number (first verse), or by the verse    *****/
+	/****    (the other verses). With that the code is straight forward.          *****/
+	/**********************************************************************************/
+		rawText := s.Text()
+		verse:= ""
+		if i == 0 {
+			verse, _ = strings.CutPrefix(rawText, searchedChapter )
+		}else{
+
+			verse, _ = strings.CutPrefix(rawText, strconv.Itoa(i+1))
+		}
+		verse = replacer.Replace(verse)
+		chapterText.verses = append(chapterText.verses, strings.TrimSpace(verse))
+	})
+
+	return  chapterText, nil
+}
+
+/*
+func fecthBook (book, version string,  chapterNumbers int) book, error{
+	failedChapter := 0
+	bookChapters := book{}
+
+	for i:= 1 ; i > chapterNumbers ; i++{
+		verses, err := fetchChapter(book, strconv.Itoa(i), version)
+		if err != nil {
+			failedChapter++
+			log.Printf("%s", err)
+			continue
+		}
+		book.chapters = append(book, 
+
+}
+*/

internal/fetcher/textmanipulation.go

@@ -0,0 +1,50 @@
+package main
+
+import(
+	"log"
+	"regexp"
+	"strings"
+	"unicode"
+	"unicode/utf8"
+)
+
+func stripAlternativeVersification(rawChapter chapter) chapter{
+	regExp:=regexp.MustCompile(`\(\d+:\d+\)`)
+
+	var treatedChapter chapter
+	for _, verse := range rawChapter.verses{
+		modifiedVerse := string(regExp.ReplaceAll([]byte(verse), []byte("")))
+		log.Printf("%v\n", modifiedVerse)
+		treatedChapter.verses = append(treatedChapter.verses, modifiedVerse)
+	}
+
+	return treatedChapter
+}
+
+func splitInPart(verse string, properName map[string]struct{})(){
+		
+	currentPart := ""
+	words := strings.Split(strings.TrimSpace(verse), " ")
+	wordsCount := len(words)
+	isTreated := false
+
+	for i :=0 ; i < wordsCount ; i++{
+		word := words[i]
+		r, _ := utf8.DecodeRuneInString(word)
+		trimedWord := strings.TrimRight(word, ".,")
+		_, isProperName := properName[trimedWord]
+		if i != 0 && unicode.IsUpper(r) && !isProperName{
+				log.Printf("%s", currentPart)
+				currentPart = word
+				isTreated = true
+		}else{
+				currentPart += " " + word
+				isTreated = false
+		}
+	}
+
+	if !isTreated {
+		log.Printf("%s", currentPart)
+	}
+
+}

internal/models/yaml.go

@@ -0,0 +1,12 @@
+package models
+
+type yaml struct {
+	Strophes []struct {
+		Lines []struct {
+			Verse   int    `yaml:"verse"`
+			Tet   string `yaml:"text"`
+			Who string `yaml:"who"`
+			To  string `yaml:"to"`
+		} `yaml:"lines"`
+	} `yaml:"strophes"`
+}