Implements URI guessing

bd.go

@@ -1,6 +1,7 @@
 package main
 
 import (
+	"fmt"
 	"regexp"
 	"strings"
 	"time"
@@ -13,11 +14,16 @@ import (
 type AlbumState int
 
 const (
-	NEW     AlbumState = iota // 0
-	MINT                      // 1
-	GOOD                      // 2
-	AVERAGE                   // 3
-	BAD                       // 4
+	// NEW is "État neuf" state
+	NEW AlbumState = iota // 0
+	// MINT is "Très bon état" state
+	MINT // 1
+	// GOOD is "Bon état" state
+	GOOD // 2
+	// AVERAGE is "État moyen" state
+	AVERAGE // 3
+	// BAD is "Mauvais état" state
+	BAD // 4
 )
 
 // An Album is the core object in our system
@@ -40,13 +46,14 @@ type Album struct {
 
 	LegalDeposit time.Time
 	PrintDate    time.Time
+	PurchaseDate time.Time
 }
 
 var endDelim = regexp.MustCompile(` \(.*\)\z`)
 var wordBoundaries = regexp.MustCompile(`[^[:alnum:]]+`)
 var punctuation = regexp.MustCompile(`[!?\.:;,]`)
 
-func sanitizeTitleString(title string, removeEndParent bool) string {
+func sanitizeTitleString(title string) string {
 	// first sanitize accuented characters.
 	isOk := func(r rune) bool {
 		return r < 32 || r >= 127
@@ -57,16 +64,52 @@ func sanitizeTitleString(title string, removeEndParent bool) string {
 	// or io.Writer filter to automatically do such filtering when reading
 	// or writing data anywhere.
 	title, _, _ = transform.String(t, title)
-	//Now we remove all punctuation
-	if removeEndParent == true {
-		title = endDelim.ReplaceAllString(title, "")
-	}
-
+	//Now we remove all punctuat
 	return strings.Trim(wordBoundaries.ReplaceAllString(punctuation.ReplaceAllString(title, ""), "-"), "-")
 }
 
-func (*Album) GetBedethequeComURI() string {
-	return ""
+// GetBedethequeComURI tries to guess the URI used by bedetheque.com to reference an album, using reverse-engineered euristics
+func (a *Album) GetBedethequeComURI() string {
+	// we check for determinant
+	matches := endDelim.FindString(a.Series)
+	series := a.Series
+	titleMatch := false
+	if len(matches) != 0 {
+		series = strings.TrimSuffix(series, matches)
+		det := strings.Trim(matches, " ()")
+		if det[len(det)-1] != '\'' {
+			det = det + " "
+		}
+		titleCompare := det + strings.ToLower(series[:1]) + series[1:]
+		titleMatch = (titleCompare == a.Title)
+	} else {
+		titleMatch = (a.Series == a.Title)
+	}
+
+	series = sanitizeTitleString(series)
+	title := sanitizeTitleString(a.Title)
+	//first we test if we have a tome identifier
+	tomeIdent := ""
+	if a.Num < 0 {
+		tomeIdent = a.NumA
+	} else {
+		tomeIdent = fmt.Sprintf("Tome-%d%s", a.Num, a.NumA)
+	}
+
+	if titleMatch {
+		if len(tomeIdent) == 0 {
+			return fmt.Sprintf("BD-%s-%d.html", series, a.ID)
+		}
+		return fmt.Sprintf("BD-%s-%s-%d.html", series, tomeIdent, a.ID)
+	}
+	if len(tomeIdent) == 0 {
+		return fmt.Sprintf("BD-%s-%s-%d.html", series, title, a.ID)
+	}
+	return fmt.Sprintf("BD-%s-%s-%s-%d.html",
+		series,
+		tomeIdent,
+		title,
+		a.ID)
 }
 
 // An AlbumDescription is a more complete BD description

bd_test.go

@@ -14,19 +14,19 @@ var _ = Suite(&AlbumSuite{})
 
 func (s *AlbumSuite) TestSanitazation(c *C) {
 	data := map[string]string{
-		"Le cycle de l'eau - I": "Le-cycle-de-l-eau-I",
-		"Nef des fous (La)":     "Nef-des-fous",
-		"Oiseau noir (L')":      "Oiseau-noir",
-		"Foo":                   "Foo",
-		"Nuit de l'étoile (La)":         "Nuit-de-l-etoile",
+		"Le cycle de l'eau - I":         "Le-cycle-de-l-eau-I",
+		"Nef des fous":                  "Nef-des-fous",
+		"Oiseau noir":                   "Oiseau-noir",
+		"Foo":                           "Foo",
+		"Nuit de l'étoile":              "Nuit-de-l-etoile",
 		"Mon Père saigne l'Histoire":    "Mon-Pere-saigne-l-Histoire",
 		"Les disparus d'apostrophes !":  "Les-disparus-d-apostrophes",
 		"Eden - It's an Endless World!": "Eden-It-s-an-Endless-World",
-		"100.000 femmes (Les)":          "100000-femmes",
+		"100.000 femmes":                "100000-femmes",
 	}
 
 	for title, expected := range data {
-		san := sanitizeTitleString(title, true)
+		san := sanitizeTitleString(title)
 		c.Check(san, Equals, expected, Commentf("Processing '%s'", title))
 	}
 }
@@ -66,7 +66,7 @@ func (s *AlbumSuite) TestBedecomURI(c *C) {
 			NumA:   "",
 		},
 		"BD-Nef-des-fous-HS03-Le-petit-Roy-1387.html": Album{
-			ID:     1287,
+			ID:     1387,
 			ISBN:   "2-84055-142-X",
 			Series: "Nef des fous (La)",
 			Title:  "Le petit Roy",
@@ -113,10 +113,18 @@ func (s *AlbumSuite) TestBedecomURI(c *C) {
 			Num:    1,
 			NumA:   "",
 		},
+		"BD-Catalogues-Expositions-Brassens-ou-la-liberte-124218.html": Album{
+			ID:     124218,
+			ISBN:   "9782205066975",
+			Series: "(Catalogues) Expositions",
+			Title:  "Brassens ou la liberté",
+			Num:    -1,
+			NumA:   "",
+		},
 	}
 
-	for expectedUrl, album := range data {
+	for expectedURL, album := range data {
 		url := album.GetBedethequeComURI()
-		c.Check(url, Equals, expectedUrl)
+		c.Check(url, Equals, expectedURL)
 	}
 }