// most of the code in here is because NAL's QC sucks...
// sometimes annotations end with a '.'...
// sometimes annotations are duplicated...
func edgesFromRecord(r *gomarc.Reader) (ret []string) {
agricolaId, valid := r.GetField(primaryId[:3], primaryId[3:])
if !valid {
return nil
}
nodupes := make(map[string]bool)
for _, sub := range secondaryIds {
flds, exists := r.GetFields(sub[:3], sub[3:])
if !exists {
continue
}
for _, fld := range flds {
if strings.HasSuffix(fld, ".") {
if strings.HasSuffix(fld, "etc.") {
// skip
} else if fld[len(fld)-2] == ')' || fld[len(fld)-3] != '.' { // acronym? "U.S."
fld = fld[:len(fld)-1]
}
}
if !nodupes[fld] {
ret = append(ret, agricolaId+"\t"+fld)
nodupes[fld] = true
}
}
}
return ret
}
func xmlRecord(r *gomarc.Reader) string {
xmldata := make(map[string][]string)
for name, tagsf := range xmlFieldMap {
tag := tagsf[:3]
sf := tagsf[3:]
flds, hadFlds := r.GetFields(tag, sf)
if hadFlds {
switch name {
case "language":
flds = []string{flds[0][35:38]}
case "published":
flds = []string{strings.Join(flds, " ")}
}
xmldata[name] = append(xmldata[name], flds...)
}
}
xmlout := make(map[string]string)
for tagname, contents := range xmldata {
maintag := ""
indent := " "
if strings.Contains(tagname, "/") {
parts := strings.Split(tagname, "/")
maintag = parts[0]
tagname = parts[1]
indent = " "
}
nodupes := make(map[string]bool)
for _, c := range contents {
if !nodupes[c] {
xmlout[maintag] += fmt.Sprintf("%s<%s>%s</%s>\n", indent, tagname, c, tagname)
nodupes[c] = true
}
}
}
xmlstring := xmlout[""]
for othertag, inner := range xmlout {
if othertag == "" {
continue
}
xmlstring += fmt.Sprintf(" <%s>\n%s </%s>\n", othertag, inner, othertag)
}
return "<document>\n" + xmlstring + "</document>"
}
