bookwyrm/bookwyrm/utils/isni.py

"""ISNI author checking utilities"""
import xml.etree.ElementTree as ET
import requests

from bookwyrm import activitypub, models


def request_isni_data(search_index, search_term, max_records=5):
    """Request data from the ISNI API"""

    search_string = f'{search_index}="{search_term}"'
    query_params = {
        "query": search_string,
        "version": "1.1",
        "operation": "searchRetrieve",
        "recordSchema": "isni-b",
        "maximumRecords": max_records,
        "startRecord": "1",
        "recordPacking": "xml",
        "sortKeys": "RLV,pica,0,,",
    }
    result = requests.get("http://isni.oclc.org/sru/", params=query_params, timeout=15)
    # the OCLC ISNI server asserts the payload is encoded
    # in latin1, but we know better
    result.encoding = "utf-8"
    return result.text


def make_name_string(element):
    """create a string of form 'personal_name surname'"""

    # NOTE: this will often be incorrect, many naming systems
    # list "surname" before personal name
    forename = element.find(".//forename")
    surname = element.find(".//surname")
    if forename is not None:
        return "".join([forename.text, " ", surname.text])
    return surname.text


def get_other_identifier(element, code):
    """Get other identifiers associated with an author from their ISNI record"""

    identifiers = element.findall(".//otherIdentifierOfIdentity")
    for section_head in identifiers:
        if (
            section_head.find(".//type") is not None
            and section_head.find(".//type").text == code
            and section_head.find(".//identifier") is not None
        ):
            return section_head.find(".//identifier").text

    # if we can't find it in otherIdentifierOfIdentity,
    # try sources
    for source in element.findall(".//sources"):
        code_of_source = source.find(".//codeOfSource")
        if code_of_source is not None and code_of_source.text.lower() == code.lower():
            return source.find(".//sourceIdentifier").text

    return ""


def get_external_information_uri(element, match_string):
    """Get URLs associated with an author from their ISNI record"""

    sources = element.findall(".//externalInformation")
    for source in sources:
        information = source.find(".//information")
        uri = source.find(".//URI")
        if (
            uri is not None
            and information is not None
            and information.text.lower() == match_string.lower()
        ):
            return uri.text
    return ""


def find_authors_by_name(name_string, description=False):
    """Query the ISNI database for possible author matches by name"""

    payload = request_isni_data("pica.na", name_string)
    # parse xml
    root = ET.fromstring(payload)
    # build list of possible authors
    possible_authors = []
    for element in root.iter("responseRecord"):
        personal_name = element.find(".//forename/..")
        if not personal_name:
            continue

        author = get_author_from_isni(element.find(".//isniUnformatted").text)

        if bool(description):

            titles = []
            # prefer title records from LoC+ coop, Australia, Ireland, or Singapore
            # in that order
            for source in ["LCNACO", "NLA", "N6I", "NLB"]:
                for parent in element.findall(f'.//titleOfWork/[@source="{source}"]'):
                    titles.append(parent.find(".//title"))
                for parent in element.findall(f'.//titleOfWork[@subsource="{source}"]'):
                    titles.append(parent.find(".//title"))
            # otherwise just grab the first title listing
            titles.append(element.find(".//title"))

            if titles:
                # some of the "titles" in ISNI are a little ...iffy
                # '@' is used by ISNI/OCLC to index the starting point ignoring stop words
                # (e.g. "The @Government of no one")
                title_elements = [
                    e
                    for e in titles
                    if hasattr(e, "text") and not e.text.replace("@", "").isnumeric()
                ]
                if len(title_elements):
                    author.bio = title_elements[0].text.replace("@", "")
                else:
                    author.bio = None

        possible_authors.append(author)

    return possible_authors


def get_author_from_isni(isni):
    """Find data to populate a new author record from their ISNI"""

    payload = request_isni_data("pica.isn", isni)
    # parse xml
    root = ET.fromstring(payload)
    # there should only be a single responseRecord
    # but let's use the first one just in case
    element = root.find(".//responseRecord")
    name = make_name_string(element.find(".//forename/.."))
    viaf = get_other_identifier(element, "viaf")
    # use a set to dedupe aliases in ISNI
    aliases = set()
    aliases_element = element.findall(".//personalNameVariant")
    for entry in aliases_element:
        aliases.add(make_name_string(entry))
    # aliases needs to be list not set
    aliases = list(aliases)
    bio = element.find(".//nameTitle")
    bio = bio.text if bio is not None else ""
    wikipedia = get_external_information_uri(element, "Wikipedia")

    author = activitypub.Author(
        id=element.find(".//isniURI").text,
        name=name,
        isni=isni,
        viafId=viaf,
        aliases=aliases,
        bio=bio,
        wikipediaLink=wikipedia,
    )

    return author


def build_author_from_isni(match_value):
    """Build basic author class object from ISNI URL"""

    # if it is an isni value get the data
    if match_value.startswith("https://isni.org/isni/"):
        isni = match_value.replace("https://isni.org/isni/", "")
        return {"author": get_author_from_isni(isni)}
    # otherwise it's a name string
    return {}


def augment_author_metadata(author, isni):
    """Update any missing author fields from ISNI data"""

    isni_author = get_author_from_isni(isni)
    isni_author.to_model(model=models.Author, instance=author, overwrite=False)

    # we DO want to overwrite aliases because we're adding them to the
    # existing aliases and ISNI will usually have more.
    # We need to dedupe because ISNI records often have lots of dupe aliases
    aliases = set(isni_author.aliases)
    for alias in author.aliases:
        aliases.add(alias)
    author.aliases = list(aliases)
    author.save()