Merge pull request #2921 from axiomizer/isbn-hyphenation

Isbn hyphenation

bookwyrm/isbn/isbn.py

@@ -0,0 +1,78 @@
+""" Use the range message from isbn-international to hyphenate ISBNs """
+import os
+from xml.etree import ElementTree
+import requests
+
+from bookwyrm import settings
+
+
+class IsbnHyphenator:
+    """Class to manage the range message xml file and use it to hyphenate ISBNs"""
+
+    __range_message_url = "https://www.isbn-international.org/export_rangemessage.xml"
+    __range_file_path = os.path.join(
+        settings.BASE_DIR, "bookwyrm", "isbn", "RangeMessage.xml"
+    )
+    __element_tree = None
+
+    def update_range_message(self):
+        """Download the range message xml file and save it locally"""
+        response = requests.get(self.__range_message_url)
+        with open(self.__range_file_path, "w", encoding="utf-8") as file:
+            file.write(response.text)
+        self.__element_tree = None
+
+    def hyphenate(self, isbn_13):
+        """hyphenate the given ISBN-13 number using the range message"""
+        if isbn_13 is None:
+            return None
+        if self.__element_tree is None:
+            self.__element_tree = ElementTree.parse(self.__range_file_path)
+        gs1_prefix = isbn_13[:3]
+        reg_group = self.__find_reg_group(isbn_13, gs1_prefix)
+        if reg_group is None:
+            return isbn_13  # failed to hyphenate
+        registrant = self.__find_registrant(isbn_13, gs1_prefix, reg_group)
+        if registrant is None:
+            return isbn_13  # failed to hyphenate
+        publication = isbn_13[len(gs1_prefix) + len(reg_group) + len(registrant) : -1]
+        check_digit = isbn_13[-1:]
+        return "-".join((gs1_prefix, reg_group, registrant, publication, check_digit))
+
+    def __find_reg_group(self, isbn_13, gs1_prefix):
+        for ean_ucc_el in self.__element_tree.find("EAN.UCCPrefixes").findall(
+            "EAN.UCC"
+        ):
+            if ean_ucc_el.find("Prefix").text == gs1_prefix:
+                for rule_el in ean_ucc_el.find("Rules").findall("Rule"):
+                    length = int(rule_el.find("Length").text)
+                    if length == 0:
+                        continue
+                    reg_grp_range = [
+                        int(x[:length]) for x in rule_el.find("Range").text.split("-")
+                    ]
+                    reg_group = isbn_13[len(gs1_prefix) : len(gs1_prefix) + length]
+                    if reg_grp_range[0] <= int(reg_group) <= reg_grp_range[1]:
+                        return reg_group
+                return None
+        return None
+
+    def __find_registrant(self, isbn_13, gs1_prefix, reg_group):
+        from_ind = len(gs1_prefix) + len(reg_group)
+        for group_el in self.__element_tree.find("RegistrationGroups").findall("Group"):
+            if group_el.find("Prefix").text == "-".join((gs1_prefix, reg_group)):
+                for rule_el in group_el.find("Rules").findall("Rule"):
+                    length = int(rule_el.find("Length").text)
+                    if length == 0:
+                        continue
+                    registrant_range = [
+                        int(x[:length]) for x in rule_el.find("Range").text.split("-")
+                    ]
+                    registrant = isbn_13[from_ind : from_ind + length]
+                    if registrant_range[0] <= int(registrant) <= registrant_range[1]:
+                        return registrant
+                return None
+        return None
+
+
+hyphenator_singleton = IsbnHyphenator()

bookwyrm/models/book.py

@@ -15,6 +15,7 @@ from model_utils.managers import InheritanceManager
 from imagekit.models import ImageSpecField
 
 from bookwyrm import activitypub
+from bookwyrm.isbn.isbn import hyphenator_singleton as hyphenator
 from bookwyrm.preview_images import generate_edition_preview_image_task
 from bookwyrm.settings import (
     DOMAIN,
@@ -321,6 +322,11 @@ class Edition(Book):
     serialize_reverse_fields = [("file_links", "fileLinks", "-created_date")]
     deserialize_reverse_fields = [("file_links", "fileLinks")]
 
+    @property
+    def hyphenated_isbn13(self):
+        """generate the hyphenated version of the ISBN-13"""
+        return hyphenator.hyphenate(self.isbn_13)
+
     def get_rank(self):
         """calculate how complete the data is on this edition"""
         rank = 0

bookwyrm/templates/book/book_identifiers.html

@@ -6,7 +6,7 @@
     {% if book.isbn_13 %}
         <div class="is-flex is-flex-wrap-wrap">
             <dt class="mr-1">{% trans "ISBN:" %}</dt>
-            <dd itemprop="isbn" class="mr-1" id="isbn_content">{{ book.isbn_13 }}</dd>
+            <dd itemprop="isbn" class="mr-1" id="isbn_content">{{ book.hyphenated_isbn13 }}</dd>
             <div>
                 <button class="button is-small" data-copywithtooltip data-content-id="isbn_content" data-tooltip-id="isbn_tooltip">
                     <span class="icon icon-copy" title="{% trans "Copy ISBN" %}">