This repository has been archived on 2025-03-01. You can view files and clone it, but cannot push or open issues or pull requests.
trantor/parser/parser.go

101 lines
2.3 KiB
Go
Raw Normal View History

package parser
import (
"regexp"
"strings"
"git.gitorious.org/go-pkg/epubgo.git"
)
type MetaData map[string]interface{}
func EpubMetadata(epub *epubgo.Epub) MetaData {
metadata := MetaData{}
for _, m := range epub.MetadataFields() {
data, err := epub.Metadata(m)
if err != nil {
continue
}
switch m {
case "creator":
metadata["author"] = parseAuthr(data)
case "description":
metadata[m] = parseDescription(data)
case "subject":
metadata[m] = parseSubject(data)
case "date":
metadata[m] = parseDate(data)
case "language":
metadata["lang"] = GuessLang(epub, data)
case "title", "contributor", "publisher":
metadata[m] = cleanStr(strings.Join(data, ", "))
case "identifier":
attr, _ := epub.MetadataAttr(m)
for i, d := range data {
if attr[i]["scheme"] == "ISBN" {
metadata["isbn"] = d
}
}
default:
metadata[m] = strings.Join(data, ", ")
}
}
return metadata
}
func cleanStr(str string) string {
str = strings.Replace(str, "'", "'", -1)
exp, _ := regexp.Compile("&[^;]*;")
str = exp.ReplaceAllString(str, "")
exp, _ = regexp.Compile("[ ,]*$")
str = exp.ReplaceAllString(str, "")
return str
}
func parseAuthr(creator []string) []string {
exp1, _ := regexp.Compile("^(.*\\( *([^\\)]*) *\\))*$")
exp2, _ := regexp.Compile("^[^:]*: *(.*)$")
res := make([]string, len(creator))
for i, s := range creator {
auth := exp1.FindStringSubmatch(s)
if auth != nil {
res[i] = cleanStr(strings.Join(auth[2:], ", "))
} else {
auth := exp2.FindStringSubmatch(s)
if auth != nil {
res[i] = cleanStr(auth[1])
} else {
res[i] = cleanStr(s)
}
}
}
return res
}
func parseDescription(description []string) string {
str := cleanStr(strings.Join(description, "\n"))
str = strings.Replace(str, "</p>", "\n", -1)
exp, _ := regexp.Compile("<[^>]*>")
str = exp.ReplaceAllString(str, "")
str = strings.Replace(str, "&amp;", "&", -1)
str = strings.Replace(str, "&lt;", "<", -1)
str = strings.Replace(str, "&gt;", ">", -1)
str = strings.Replace(str, "\\n", "\n", -1)
return str
}
func parseSubject(subject []string) []string {
var res []string
for _, s := range subject {
res = append(res, strings.Split(s, " / ")...)
}
return res
}
func parseDate(date []string) string {
if len(date) == 0 {
return ""
}
return strings.Replace(date[0], "Unspecified: ", "", -1)
}