GakuNin RDM Search API (`/api/v1/search/`) Investigation Memo

Investigation date: 2026-02-24 Target: GakuNin RDM (GRDM) Search API Source code: RCOSDP/RDM-osf.io (website/search/ directory) Developer guide: RCOSDP/RDM-developer-guide Note: Official documentation for the Search API could not be found. This article is an investigation record based on both the actual API behavior and the source code.

Overview

GakuNin RDM is a fork of OSF (Open Science Framework), and its source code is available on GitHub (RCOSDP/RDM-osf.io). The search functionality implementation is in the website/search/ directory and consists mainly of the following files.

File	Role
`elastic_search.py`	Index mapping definitions, document registration/update
`views.py`	API endpoint handlers
`util.py`	Query construction including `build_private_search_query()`
`search.py`	High-level interface

In Japanese environments, Elasticsearch’s kuromoji_analyzer is used (confirmed in the source code).

Request Format

Parameter	Description	Notes
`api_version`	`vendor: "grdm"`, `version: 2`	`version` supports 1 and 2 (confirmed in source code)
`elasticsearch_dsl.query`	Elasticsearch Query DSL	Uses `filtered` format (ES 2.x syntax)
`from` / `size`	Pagination	Confirmed to work up to `size=100`. `match_all` + `size>50` results in a 500 error
`highlight`	`field_name:character_count` format	GRDM-specific format. Wildcards (`comments.*`) can also be used
`sort`	Sort order	Described below

Sort Options

According to the source code (build_private_search_query in util.py), the following sort targets are defined.

Sort value	Verified	Description
`modified_desc` / `modified_asc`	Verified	By modification date
`created_desc` / `created_asc`	Verified	By creation date
`project_desc` / `project_asc`	Unverified	By project name (defined in source)
`file_desc` / `file_asc`	Unverified	By file name (defined in source)
`wiki_desc` / `wiki_asc`	Unverified	By wiki name (defined in source)
`user_desc` / `user_asc`	Unverified	By user name (defined in source)
`institution_desc` / `institution_asc`	Unverified	By institution name (defined in source)

relevance, title_asc, etc. are not defined in the source and were confirmed to result in 400 errors.

Fields Indexed (From Source Code)

From update_file(), update_node(), and update_user() in the source code elastic_search.py, the fields indexed for each category can be confirmed.

file (`update_file()`)

Field	Source	`_all` search	Notes
`name`	`file_.name`	Hits	Highlight confirmed
`normalized_name`	`unicode_normalize(file_.name)`	–	For kuromoji
`node_title`	`target.title`	Hits	Parent project name
`creator_name`	Obtained from user info	Hits
`modifier_name`	Obtained from user info	Unverified
`tags`	File tags	Hits	Highlight confirmed
`normalized_tags`	Normalized tags	Hits	Highlight confirmed
`extra_search_terms`	`clean_splitters(file_.name)`	Unverified	File name tokenized
`comments`	`comments_to_doc()`	Unverified	No test data available
`node_public`	`target.is_public`	–	For filtering
`node_contributors`	Contributor ID list	–	For permission filtering
`deep_url`	File URL	–	For display
`date_created` / `date_modified`	Datetime	–	For sorting
`category`	Fixed as `"file"`	–

folder_name and parent_title / parent_url are not included in the index. These are dynamically added during response by format_results() (confirmed in source code).

project (`update_node()`)

Field	Source	`_all` search	Notes
`title`	`node.title`	Hits
`normalized_title`	Normalized title	–	For kuromoji
`description`	`node.description`	Hits	Highlight confirmed
`normalized_description`	Normalized description	–	For kuromoji
`tags`	Project tags	Hits
`normalized_tags`	Normalized tags	–
`contributors`	Contributor info	–
`creator_name`	Creator name	Hits
`comments`	Comments	Unverified
`wikis`	Wiki content	Unverified	Mapped via dynamic template
`license`	License info	–	For display
`affiliated_institutions`	Affiliated institutions	–
`boost`	Boost value	–

user (`update_user()`)

Field	Source	Notes
`user`	`user.fullname`
`normalized_user` / `normalized_names`	Normalized names
`job` / `job_title`	Job info
`ongoing_job` / `ongoing_job_department` / `ongoing_job_title`	Current workplace
`school` / `ongoing_school*`	Education history
`emails`	Email addresses
`social`	SNS links
`boost`	Fixed at `2`	Set to boost user search scores

Other Categories (Defined in Source Code)

According to the source code, the following categories exist in addition to file, project, and user.

component – Project sub-components
registration – Registrations (snapshots)
preprint – Preprints
wiki – Wiki pages (text field contains the body)
comment – Comments
institution – Institutions
collectionsubmission – Collection submissions

The text highlight field is most likely for Wiki page content (not file body text).

`_all` Field Search Targets

Elasticsearch Mapping (From Source Code)

The _all field is analyzed with kuromoji_analyzer. In create_index() of the source code, analyzers are configured for each field, and fields with analyzers configured are included in _all.

Search Targets Confirmed by Experiment

Search term	Matched field	Category	Confirmed via highlight
`"2507"`	`name` (file name)	file	`highlight[name]`
`"dmp-project-aaa"`	`node_title` (project name)	file	–
`"Nakamura"`	`creator_name` (creator name)	file	–
`"blockchain"`	`tags` (tags)	file	`highlight[tags]`
`"arxiv"`	`tags` / `normalized_tags`	file	`highlight[tags]`, `highlight[normalized_tags]`
`"アーカイブズ学"`	`description` (project description)	project	`highlight[description]`
`"digital-preservation"`	`tags` (project tags)	project	–

Confirmed Not Searchable

Search term	Target field	Category	Reason
`"NII Storage"`	`folder_name`	file	Not included in index (dynamically added by `format_results()`)
`"digital preservation framework"`	`file_description`	file	DataCite metadata is not searchable
`"Clio-X"`	`.txt` file body	file	File body text is not indexed
`"Victoria Lemieux"`	`.txt` file body	file	Same as above

Filters

Filter format	Result
`{"term": {"category": "file"}}`	Works
`{"and": [{"term": ...}, {"term": ...}]}`	Works
`{"bool": {"should": [...]}}`	Works
`{"bool": {"must": [...]}}`	Results in 500 error

The reason bool + must fails is unknown. Since bool + must is used within build_private_search_query() in the source code, this is presumed to be a limitation on the API wrapper side. As a workaround, and filters or query_string AND syntax can be used.

Highlights

Confirmed

Field	Category	Search term used for confirmation
`name`	file	`"ip2"`
`tags`	file	`"blockchain"`, `"arxiv"`
`normalized_tags`	file	`"arxiv"`
`description`	project	`"アーカイブズ学"`

Unverified

text – According to the source code, Wiki page body text goes into this field. It should be testable with a project that has a Wiki.
comments.* – Dynamic field for comments. Unverified because there were no comments in the test data.
title – Project title. Can be tested with search terms that match the title.
user – User name.

DataCite Metadata

Editable Fields

Managed via /v2/files/{id}/metadata_records/. However, the fields that OSF allows editing are limited to the following four (confirmed in the validation schema).

Field	Type	Description
`resource_type`	enum	`Audio/Video`, `Dataset`, `Image`, `Model`, `Software`, `Book`, `Funding Submission`, `Journal Article`, `Lesson`, `Poster`, `Preprint`, `Presentation`, `Research Tool`, `Thesis`, `Other`
`file_description`	string	File description text
`related_publication_doi`	string	DOI of related publication (`10.xxxx/yyyy` format)
`funders`	array	`[{"funding_agency": "...", "grant_number": "..."}]`

All fields of the DataCite v4.0 schema (titles, creators, subjects, descriptions, etc.) are defined, but due to OSF’s input validation, fields other than the above four result in Additional properties are not allowed errors.

Relationship to Search

I set a value in file_description and searched for it, but it did not hit. There is also no processing to read DataCite metadata in update_file() in the source code, and DataCite metadata is not included in the search index.

Summary

What Was Confirmed

API behavior (experiments):

/api/v1/search/ is an Elasticsearch Query DSL-based search API
It can perform cross-category searches across file, project, and user
_all full-text search hits on name, node_title, creator_name, tags, and description
Field-specific queries (name:, tags:, category:), wildcards, and AND/OR operators can be used
term / and / bool+should filters work
sort was confirmed to work with 4 types: modified_desc/asc and created_desc/asc
size was confirmed to work up to 100
Highlights were confirmed to work for name, tags, normalized_tags, and description

Confirmed from source code:

kuromoji_analyzer is used in Japanese environments
folder_name, parent_title, and parent_url are not included in the index and are dynamically added during response
The text highlight field is for Wiki page content
sort has definitions for each direction of project, file, wiki, user, institution, created, and modified
api_version supports version 1 and 2, and vendor supports "grdm"
In addition to file, project, and user, categories component, registration, preprint, wiki, comment, and institution exist

Confirmed Not Searchable

Text file body (text within .txt files) – Confirmed by both experiment and source code
DataCite metadata (file_description, etc.) – Confirmed by both experiment and source code
folder_name (storage provider name) – Confirmed by both experiment and source code

Unverified

Whether comments hit in _all search
Wiki content (text field) search behavior
Behavior of sort values like project_desc
Exact upper limit of size
Cause of 500 error with bool + must filter

Reference Links

RCOSDP/RDM-osf.io – GakuNin RDM source code
RCOSDP/RDM-developer-guide – Developer guide
GakuNin RDM Support Portal Search – User search manual
website/search/elastic_search.py – Index definitions and document registration
website/search/views.py – API endpoints
website/search/util.py – Query construction and sort definitions

Overview#

Request Format#

Sort Options#

Fields Indexed (From Source Code)#

file (update_file())#

project (update_node())#

user (update_user())#

Other Categories (Defined in Source Code)#

_all Field Search Targets#

Elasticsearch Mapping (From Source Code)#

Search Targets Confirmed by Experiment#

Confirmed Not Searchable#

Filters#

Highlights#

Confirmed#

Unverified#

DataCite Metadata#

Editable Fields#

Relationship to Search#

Summary#

What Was Confirmed#

Confirmed Not Searchable#

Unverified#

Reference Links#

Overview

Request Format

Sort Options

Fields Indexed (From Source Code)

file (`update_file()`)

project (`update_node()`)

user (`update_user()`)

Other Categories (Defined in Source Code)

`_all` Field Search Targets

Elasticsearch Mapping (From Source Code)

Search Targets Confirmed by Experiment

Confirmed Not Searchable

Filters

Highlights

Confirmed

Unverified

DataCite Metadata

Editable Fields

Relationship to Search

Summary

What Was Confirmed

Confirmed Not Searchable

Unverified

Reference Links