To be clear, this is absolutely possible with Zotero, it's just a bit cumbersome. I don't see this happening as built in functionality any time soon though

It's easy enough to do manually: make each section a separate document, then manually combine them to produce a combined bibliography and combine all of those for the final report. If you need it to be automatic, that won't work, but it is possible.

(In principle, you might also be able to write a complicated Word macro that would do this: create separate documents for each section, then import them into one main document [is that possible in a macro? or you could use a Master Document], then the macro would need to search through it for Zotero bibliography fields, which can be identified by their metadata, then take all of the references in each of those, combine them into a long list and resort and remove duplicates. I'm certain someone who knows much more about Word macros than I do could create that somehow. If you're doing this for a company, you could look into hiring someone to do it. It might not be that hard for an expert.)

Alternatively, what about a footnote format that does have the references throughout the document, but then also adding a bibliography at the end? Maybe that's a reasonable compromise for your organization?

Even closer, you could set Word to create endnotes at the end of each section (rather than footnotes throughout), and then also add a bibliography at the end. That's very close to what you're after. You could even set a custom mark for the endnotes so they look less like endnotes. I haven't tried all of that, but I think it could look very close. (Possible limitations: 1. what happens if you cite the same item repeatedly, it would have a separate entry in the endnotes each time; they'd also be organized by order of citation, not alphabetically; 2. you might get some interference if you're also looking to add some other unrelated footnotes or endnotes.)

If you typically have so few citations per section, it seems like simply using footnotes plus a final bibliography would be the easiest solution, within the standard features of Zotero. Maybe focus on lobbying for that.

PS: if the development work is the issue and if it takes a lot of money, I will be delighted to financially support a crowd-funding campaign. I can promise that a lot of academics I know would be delighted to pay for it, too. We really don't want to go back to Endnote...

One reason is that our guest lecturers and tutors change all the time, so their names need to be added in different sections of the doc. If all is in one place, searching and replacing names is very easy, but I can promise that no over-worked academic is going to open 14 individual files to update names. Also, the different sections are highly interwoven in terms of content. We give an overall course outline at the beginning and reference assignments that are later on explained down below. So if, for instance, the word count for an essay changes, that also needs to be fixed in several places. The teaching handbook is regularly exported as a PDF with hyperlinks to be provided to students and staff alongside the virtual teaching platform. This is different from editing a book which at some point has a "final" version.

I had a bit of spare time so looked to see if this could be done as a macro.
There are a couple of things to consider when using the attached python script.
1) This is for libreoffice (a linux distribution version so that the standard python could be accessed).
2) citeproc and lxml need to be loaded into python (for a debian based distro this is probably python-citeproc)
3) Headings need to use the basic format of Heading1 etc. This can be changed in the script if more esoteric formatting is used.
4) If there are no citations in a chapter (aka Heading1 format) then nothing is entered.
6) citations before the first Heading1 format are ignored.
7) This is really only for author date reference formats. I could not see a way to maintain a link between numbers for numbered reference formats. However, you can get a list of references even if the numbers are wrong. I would suggest to change sorting to False (note the capital letter).
8) The python is not state of the art. Mainly so that the ideas could be transferred more easily for vba for MSoffice.
9) Formatting marks such as ttt are ignored but this should be simple enough to search and replace.
10) The line starting with style can be simply the full path name and file name of a citation style from the Zotero directory.


import uno
import json
import ast

# Import the citeproc-py classes we"ll use below.
from citeproc import CitationStylesStyle, CitationStylesBibliography
from citeproc import Citation, CitationItem
from citeproc import formatter
from citeproc.source.json import CiteProcJSON

style = "harvard1"

sorting = True # True or False

document = XSCRIPTCONTEXT.getDocument()


def delete_previously_inserted_sections():
vSections = document.getTextSections()
sEventNames = vSections.getElementNames()
for sEventName in sEventNames:
if "Zotero_references_chapter" in sEventName:
oSection = document.getTextSections().getByName(sEventName)
oSectionTextRange = oSection.getAnchor()
oSectionTextRange.setString("")
oSection.dispose()


def find_headings():
"""Use the search function to locate the headings and
return a list of cursor start"""
heading_list = []
search_object = document.createSearchDescriptor()
search_object.SearchRegularExpression = False
search_object.SearchStyles = True
search_object.SearchString = "Heading 1"
found = document.findFirst(search_object)
while found:
text_cursor = found.Text.createTextCursorByRange(found)
# m = document.Text.compareRegionEnds(text_cursor.getEnd(), text_cursor.getStart()) # less is 1, more is -1, same is 0
heading_list.append(text_cursor.getStart())
found = document.findNext(found, search_object)
return heading_list



def get_reference_marks_data(start, stop):
reference_marks = document.ReferenceMarks
reference_list = []
ci = ""

for reference_mark in reference_marks:
m = document.Text.compareRegionEnds(reference_mark.getAnchor(), start)
n = document.Text.compareRegionEnds(reference_mark.getAnchor(), stop)

if m == -1 and n == 1:
reference_mark_content = reference_mark.Name
reference_mark_content, _ = reference_mark_content.rsplit(" ", 1)
_, _, reference_mark_data = reference_mark_content.split(" ", 2)
data_dictionary = json.loads(reference_mark_data, strict=False)

for citation_items in data_dictionary["citationItems"]:
ci = (ci + str(citation_items["itemData"]) + ",")

ci = ci[:-1]
ci = ("[" + ci + "]")
return ci


def create_bibliography(json_data):

bib_style = CitationStylesStyle(style, validate=False)
bib_source = CiteProcJSON(json_data)

bibliography = CitationStylesBibliography(bib_style, bib_source, formatter.html)

for key,value in bib_source.items():
citation = Citation([CitationItem(key)])
bibliography.register(citation)

reference_list = "Chapter references"

if sorting == True:
for item in sorted(bibliography.bibliography()):
item = str(item)
reference_list = reference_list + "\r\n" + item
if sorting == False:
for item in bibliography.bibliography():
item = str(item)
reference_list = reference_list + "\r\n" + item

return reference_list


def insert_text_section(bib, chapter, location):
text_cursor = document.getText().createTextCursorByRange(location)
text_cursor.goLeft(1, False)
name = "Zotero_references_chapter " + chapter
doc_text = document.getText()
section = document.createInstance("com.sun.star.text.TextSection")
section.setName(name)
doc_text.insertTextContent(text_cursor, section, True)
oSection = document.getTextSections().getByName(name)
oSectionTextRange = oSection.getAnchor()
doc_text.insertString(oSectionTextRange, bib, True)


def author_date_chapter_reference_tables():
delete_previously_inserted_sections()
heading_list = find_headings()
items = len(heading_list) - 1
for item in range(items, 0, -1):
ci = ast.literal_eval(get_reference_marks_data(heading_list[item-1], heading_list[item]))
json_data = json.loads(json.dumps(ci))
reference_list = create_bibliography(json_data)
if reference_list != "Chapter references":
insert_text_section(reference_list, str(item), heading_list[item])


g_exportedScripts = author_date_chapter_reference_tables,





I, too, am in a situation of strongly needing this feature in Word! Any guidance would help if someone has found a workaround...

I will also throw in my vote for Zotero to implement this feature in the future.

In my case, it's not for a book, but for a scientific publication in which the publisher explicitly requires separate bibliographies for two different sections in a single word document, where the second bibliography continues (numerically) from the first.

Since this doesn't seem to be an option with Zotero, I guess my solution will be to wait until the document is fully complete, then make a "citations unlinked" version of the document where everything is converted to text, then manually cut/copy/paste sections of the full bibliography to the individual sections.