Handle unpublished / Goitein in citations (#1502)

geniza/common/utils.py

@@ -80,6 +80,12 @@ def simplify_quotes(text):
     return text.translate(smart_quote_conversion)
 
 
+def list_to_string(lst):
+    """Helper function to join a list with commas, and 'and' before the final
+    entry."""
+    return " and ".join([", ".join(lst[:-1]), lst[-1]] if len(lst) > 2 else lst)
+
+
 ## adapted from tabular_export.admin
 class Echo(object):
     # See https://docs.djangoproject.com/en/1.8/howto/outputting-csv/#streaming-csv-files

geniza/corpus/templates/corpus/snippets/document_transcription.html

@@ -267,7 +267,7 @@ <h2>{{ image_info.shelfmark }} {{ image_info.label }}</h2>
                     {% else %}
                         {% if forloop.first %}
                             <span data-transcription-target="transcriptionFullLabel" class="current-transcription{% if document.digital_editions.count > 1 %} multiple{% endif %}">
-                                {{ document.digital_editions.0.source.formatted_display|safe }}
+                                {{ document.digital_editions.0.display|safe }}
                             </span>
                         {% endif %}
                         {% if not image_info.excluded %}
@@ -276,7 +276,7 @@ <h2>{{ image_info.shelfmark }} {{ image_info.label }}</h2>
                                 {% for edition in document.digital_editions.all %}
                                     <div class="transcription ed-{{ edition.pk }}"
                                          data-ittpanel-target="transcription"
-                                         data-label="{{ edition.source.formatted_display }}"
+                                         data-label="{{ edition.display }}"
                                          lang="{{ document.primary_lang_code|default:"" }}"
                                          dir="rtl"
                                          {% if document.primary_script %}
@@ -305,7 +305,7 @@ <h2>{{ image_info.shelfmark }} {{ image_info.label }}</h2>
                     {% else %}
                         {% if forloop.first %}
                             <span data-transcription-target="translationFullLabel" class="current-translation{% if document.digital_editions.count > 1 %} multiple{% endif %}">
-                                {{ document.default_translation.source.formatted_display|safe }}
+                                {{ document.default_translation.display|safe }}
                             </span>
                         {% endif %}
                         {% if not image_info.excluded %}
@@ -314,7 +314,7 @@ <h2>{{ image_info.shelfmark }} {{ image_info.label }}</h2>
                                 {% for translation in document.digital_translations.all %}
                                     <div class="translation tr-{{ translation.pk }}"
                                          data-ittpanel-target="translation"
-                                         data-label="{{ translation.source.formatted_display }}"
+                                         data-label="{{ translation.display }}"
                                          {% if translation.source.languages.exists %}lang="{{ translation.source.languages.all.0.code }}" dir="{{ translation.source.languages.all.0.direction }}"{% else %}dir="ltr"{% endif %}
                                     >
                                         {% for html_section in translation.content_html|dict_item:canvas_uri %}

geniza/corpus/templatetags/corpus_extras.py

@@ -201,56 +201,8 @@ def is_index_cards(source):
     )
 
 
-def list_to_string(lst):
-    """Helper function to join a list with commas, and 'and' before the final
-    entry."""
-    return " and ".join([", ".join(lst[:-1]), lst[-1]] if len(lst) > 2 else lst)
-
-
 @register.filter
 def process_citation(source):
-    """For scholarship records list: handle Goitein index cards by including
-    URLs and index card numbers when available, and by adding the attribution
-    to the PGL."""
-    citation = source.grouper.formatted_display(format_index_cards=True)
-    if "unpublished index cards" in citation:
-        all_cards = []
-        # add card numbers
-        for fn in source.list:
-            # remove "Card" or "card" from the location field first, we just want #NNN
-            loc = re.sub("[Cc]ard ", "", fn.location)
-            card_str = ""
-            # add URL if present
-            if loc and fn.url:
-                card_str = f'<a href="{fn.url}">{loc}</a>'
-            elif loc:
-                card_str = str(loc)
-            if card_str:
-                all_cards.append(card_str)
-        if all_cards:
-            # if card #s are present, include them as a list in the citation
-            joined = list_to_string(all_cards)
-            citation = citation[0:-1] + f", {joined}."
-        # add PGL attribution
-        citation += " Princeton Geniza Lab, Princeton University."
-    elif source.grouper.source_type.type == "Unpublished":
-        relations = [r for rs in [fn.doc_relation for fn in source.list] for r in rs]
-        if (
-            Footnote.DIGITAL_EDITION in relations
-            or Footnote.DIGITAL_TRANSLATION in relations
-        ):
-            authors = [
-                c.creator.firstname_lastname()
-                for c in source.grouper.authorship_set.order_by("creator__last_name")
-            ]
-            citation = list_to_string(authors)
-            relation_display = list_to_string(
-                [fn.get_doc_relation_display() for fn in source.list]
-            )
-            citation += f"'s {relation_display.lower()}"
-            if source.grouper.year:
-                citation += f" ({source.grouper.year})"
-            citation += ", available online through the Princeton Geniza Project at "
-            permalink = source.list[0].content_object.permalink
-            citation += f'<a href="{permalink}">{permalink}</a>.'
-    return mark_safe(citation)
+    """For scholarship records list: handle grouped citations by passing to
+    Footnote.display_multiple class method."""
+    return mark_safe(Footnote.display_multiple(source.list))

geniza/corpus/tests/test_corpus_views.py

@@ -305,7 +305,7 @@ def test_old_pgp_edition():
     doc.footnotes.add(fn)
 
     edition_str = old_pgp_edition(doc.editions())
-    assert edition_str == f"Ed. {fn.display()}"
+    assert edition_str == f"Ed. {fn.display(old_pgp=True)}"
 
     source2 = Source.objects.create(title_en="Arabic dictionary", source_type=book)
     fn2 = Footnote.objects.create(

geniza/corpus/views.py

@@ -1135,7 +1135,7 @@ def old_pgp_edition(editions):
             "%s%s%s"
             % (
                 "and trans. " if Footnote.TRANSLATION in fn.doc_relation else "",
-                fn.display().strip("."),
+                fn.display(old_pgp=True).strip("."),
                 " %s" % fn.url if fn.url else "",
             )
             for fn in editions

geniza/footnotes/conftest.py

@@ -78,11 +78,27 @@ def unpublished_editions(db):
 @pytest.fixture
 def index_cards(db):
     # fixture for unpublished index cards
-    unpub = SourceType.objects.get(type="Unpublished")
+    (unpub, _) = SourceType.objects.get_or_create(type="Unpublished")
     source = Source.objects.create(
         source_type=unpub, title_en="index cards", volume="CUL"
     )
-    author = Creator.objects.create(last_name_en="Goitein", first_name_en="S. D.")
+    (author, _) = Creator.objects.get_or_create(
+        last_name_en="Goitein", first_name_en="S. D."
+    )
+    Authorship.objects.create(creator=author, source=source)
+    return source
+
+
+@pytest.fixture
+def goitein_editions(db):
+    # fixture for Goitein unpublished editions
+    (unpub, _) = SourceType.objects.get_or_create(type="Unpublished")
+    source = Source.objects.create(
+        source_type=unpub, title_en="unpublished editions", volume="CUL"
+    )
+    (author, _) = Creator.objects.get_or_create(
+        last_name_en="Goitein", first_name_en="S. D."
+    )
     Authorship.objects.create(creator=author, source=source)
     return source
 

geniza/footnotes/models.py

@@ -13,16 +13,15 @@
 from django.db.models import Count, Q
 from django.db.models.functions import NullIf
 from django.db.models.query import Prefetch
-from django.urls import reverse
 from django.utils.html import strip_tags
-from django.utils.translation import gettext
 from django.utils.translation import gettext_lazy as _
 from gfklookupwidget.fields import GfkLookupField
 from modeltranslation.manager import MultilingualManager, MultilingualQuerySet
 from multiselectfield import MultiSelectField
 
 from geniza.common.fields import NaturalSortField
 from geniza.common.models import TrackChangesModel
+from geniza.common.utils import list_to_string
 from geniza.footnotes.utils import HTMLLineNumberParser
 
 
@@ -266,7 +265,7 @@ def formatted_display(self, extra_fields=True, format_index_cards=False):
             # only when extra_fields enabled, or there is no author
 
             # Translators: Placeholder for when a work has no title available
-            work_title = gettext("[digital geniza document edition]")
+            work_title = self.source_type.type.lower()
 
         # Wrap title in link to URL
         if self.url and work_title:
@@ -398,7 +397,6 @@ def formatted_display(self, extra_fields=True, format_index_cards=False):
         #   L. B. Yarbrough (in Hebrew)             (no comma)
         #   Author (1964)                           (no comma)
         #   Author, Journal 6 (1964)                (comma)
-        #   Author, [digital geniza document edition]                      (comma)
         use_comma = (
             extra_fields
             or self.title
@@ -591,20 +589,123 @@ def __str__(self):
         )
         return f"{rel} of {self.content_object}"
 
-    def display(self):
-        """format footnote for display; used on document detail page
-        and metadata export for old pgp site"""
-        # source, location. notes.
-        # source. notes.
-        # source, location.
-        parts = [self.source.display()]
-        if self.notes:
-            # uppercase first letter of notes if not capitalized
-            notes = self.notes[0].upper() + self.notes[1:]
-            # append period to notes if not present
-            parts.extend([" ", notes.strip("."), "."])
+    def display(self, old_pgp=False):
+        """format footnote for display; used on document detail page"""
+        # normally, just use source display.
+        # for unpublished records:
+        # source author (year), with minor emendations by person.
+        parts = [self.source.display() if old_pgp else self.source.formatted_display()]
+        if self.source.source_type.type == "Unpublished":
+            authors = [
+                c.creator.firstname_lastname()
+                for c in self.source.authorship_set.order_by("creator__last_name")
+            ]
+            parts = [list_to_string(authors)]
+            relation_display = self.get_doc_relation_display()
+            if "Goitein" in parts[0]:
+                # handle goitein, "unpublished" not "digital"
+                relation_display = relation_display.replace("Digital", "unpublished")
+            if relation_display:
+                parts.append(f"'s {relation_display.lower()}")
+            if self.source.year:
+                parts.append(f" ({self.source.year})")
+            elif "Goitein" in parts[0]:
+                # handle goitein years
+                parts.append(f" (1950–85)")
+            # add emendations
+            if self.emendations:
+                parts.append(f", with minor emendations by {self.emendations}")
+            parts.append(".")
         return "".join(parts)
 
+    @classmethod
+    def display_multiple(cls, footnotes):
+        """Display method for multiple footnotes, grouped by source, on the scholarship
+        records/bibliography page. Normally uses source formatted display, but requires
+        special handling for unpublished source records with multiple footnotes.
+        Handles Goitein index cards by including URLs and index card numbers when
+        available, and by adding the attribution to the PGL."""
+        if not footnotes:
+            # bail out with empty string on no footnotes provided
+            return ""
+        source = footnotes[0].source
+        # start with source formatted display
+        citation = source.formatted_display(format_index_cards=True)
+        # handle Goitein index cards
+        if "unpublished index cards" in citation:
+            all_cards = []
+            # add card numbers
+            for fn in footnotes:
+                # remove "Card" or "card" from the location field first, we just want #NNN
+                loc = re.sub("[Cc]ard ", "", fn.location)
+                card_str = ""
+                # add URL if present
+                if loc and fn.url:
+                    card_str = f'<a href="{fn.url}">{loc}</a>'
+                elif loc:
+                    card_str = str(loc)
+                if card_str:
+                    all_cards.append(card_str)
+            if all_cards:
+                # if card #s are present, include them as a list in the citation
+                joined = list_to_string(all_cards)
+                citation = citation[0:-1] + f", {joined}."
+            # add PGL attribution
+            citation += " Princeton Geniza Lab, Princeton University."
+        elif source.source_type.type == "Unpublished":
+            # handle other unpublished sources
+            authors = [
+                c.creator.firstname_lastname()
+                for c in source.authorship_set.order_by("creator__last_name")
+            ]
+            citation = list_to_string(authors)
+            # get doc relations across all footnotes
+            if "Goitein" in citation:
+                # handle goitein unpublished work (non-index cards):
+                # should say "unpublished edition" or "unpublished translation"
+                relation_set = set(
+                    [
+                        re.sub(
+                            r"([dD]igital )?([Ee]dition|[Tt]ranslation)",
+                            r"unpublished \2",
+                            fn.get_doc_relation_display(),
+                        )
+                        for fn in footnotes
+                    ]
+                )
+            else:
+                relation_set = set([fn.get_doc_relation_display() for fn in footnotes])
+            relation_display = list_to_string(sorted(list(relation_set)))
+            if relation_display:
+                citation += f"'s {relation_display.lower()}"
+            # handle languages (some unpublished records have this)
+            if source.languages.exists():
+                for lang in source.languages.all():
+                    if "English" not in str(lang) and "Unspecified" not in str(lang):
+                        citation += " (in %s)" % lang
+            # year if specified, and special range for Goitein
+            if source.year:
+                citation += f" ({source.year})"
+            elif "Goitein" in citation:
+                citation += f" (1950–85)"
+            # handle minor emendations if specified
+            emendations = list_to_string(
+                [fn.emendations for fn in footnotes if fn.emendations]
+            )
+            if emendations:
+                citation += f", with minor emendations by {emendations}"
+            # include information about PGP for digital edition/translation
+            relations = [r for rs in [fn.doc_relation for fn in footnotes] for r in rs]
+            if cls.DIGITAL_EDITION in relations or cls.DIGITAL_TRANSLATION in relations:
+                citation += (
+                    ", available online through the Princeton Geniza Project at "
+                )
+                permalink = footnotes[0].content_object.permalink
+                citation += f'<a href="{permalink}">{permalink}</a>'
+            citation += "."
+
+        return citation
+
     def has_url(self):
         """Admin display field indicating if footnote has a url."""
         return bool(self.url)