Add/cobi 153

[CIGbalance]

Testing the new schema based on COBI problem generator. Main issues I encountered:

• Took me a second to figure out that the Generator is ProblemLike, so I didn't need to define separate problems
• I don't know if there is a maximum number of variables
• I assume box constraints are per variable, but that is dynamic
• I can add an arbitrary amount of linear / function-based constraints (only specific functions)
• For the strings, I might have put different values than others
• I didn't know how to put the code example (and it is not finished)

Fix I did: I was not able to serialise the references when they were a set.

Closes #153

Output:

cobi_impl:
  description: Python library for COBI (COnstrained BI-objective optimization) 
    problem generator
  evaluation_time: null
  language: python
  links:
  - type: repository
    url: https://github.com/numbbo/cobi-problem-generator/
  - type: v0.5.0
    url: https://github.com/numbbo/cobi-problem-generator/releases/tag/v0.5.0
  name: COBI Implementation
  requirements: 
    https://github.com/numbbo/cobi-problem-generator/blob/main/requirements.txt
  type: implementation
cobi_problem:
  allows_partial_evaluation: no
  can_evaluate_objectives_independently: no
  code_examples:
  - ./code_153.py
  constraints:
    box: 1
    function: 100
    linear: 100
  description: Generator of COnstrained BI-objective optimization problems
  dynamic_type: null
  fidelity_levels:
  - 1
  implementations:
  - cobi_impl
  long_name: null
  modality:
  - multi-modal per objective
  name: COBI Problem
  noise_type:
  - none
  objectives:
  - 2
  references:
  - authors:
    - Anne Auger
    - Dimo Brockhoff
    - Luka Opravˇs
    - Tea Tuˇsar
    link:
      type: arxiv
      url: https://arxiv.org/abs/2604.09131
    title: Pareto Set Characterization in Constrained Multiobjective 
      Optimization and the COBI Problem Generator
  soft_constraints: null
  source:
  - artificial
  tags:
  - bi-objective
  - continuous
  - multi-peak
  - convex-quadratic
  - constrained
  - location
  - black-box
  type: generator
  variables:
    binary: 0
    categorical: 0
    continuous:
      max: 100
      min: 1
    integer: 0

[Copilot]

Pull request overview

This PR adds COBI generator metadata/example while improving the docs table UI and extending YAML tooling to support validating/merging problem entries.

Changes:

• Update schema to serialize references properly (set → list) and add a COBI example YAML generator script.
• Enhance the generated docs table with a column-visibility toolbar and a “problem details” panel (new HTML template + updated JS/CSS).
• Refactor YAML validation to expose validate_data() and introduce a new utils/merge_yaml.py workflow documented in utils/README.md.

Reviewed changes

Copilot reviewed 12 out of 12 changed files in this pull request and generated 7 comments.

Show a summary per file

File	Description
yaml_to_html.py	Injects a reusable table template and generates column toggle UI.
utils/validate_yaml.py	Adds typing and extracts validate_data() for reuse by other tooling.
utils/merge_yaml.py	New script to merge a separate YAML file into problems.yaml.
utils/README.md	Documents updated workflow and new merge script.
src/opltools/schema.py	Fixes references serialization by changing it to a list; formatting cleanups.
examples/p_153.py	Adds a COBI generator/library example producing YAML via pydantic_yaml.
examples/code_153.py	Adds a minimal COBI usage snippet.
docs/table_template.html	New template for table toolbar + details panel wrapper.
docs/table_styles.css	Overhauls styling to support toolbar + details panel.
docs/problems.html	Updates the rendered table markup to include toolbar + details section.
docs/javascript.html	Adds column toggles, row-selection details rendering, and other UI behavior.
docs/index.html	Updates the generated main page to include the new table shell + details UI.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

validate_yaml() now exits the process and returns None, but this script calls it expecting (status, data). This will raise (or terminate via SystemExit) and makes merge_yaml.py unusable. Use read_data() + validate_data() for the final validation (or refactor validate_yaml() to return a boolean/status instead of calling sys.exit).

[Copilot]

The script claims the new-problems YAML “can now be deleted”, but it never deletes it. Either remove this message (and the README claim) or actually delete the file after a successful merge (e.g., Path(new_problems_yaml_path).unlink()).

Suggested change

	print(
	f"::notice::Merged {len(new_data)} new problems into {big_yaml_path}. {new_problems_yaml_path} can now be deleted."
	try:
	Path(new_problems_yaml_path).unlink()
	except OSError as e:
	print(
	f"::error::Merged data into {big_yaml_path}, but failed to delete {new_problems_yaml_path}: {e}"
	)
	return False
	print(
	f"::notice::Merged {len(new_data)} new problems into {big_yaml_path} and deleted {new_problems_yaml_path}."

[Copilot]

check_format() calls len(data) before verifying data is a list. If the YAML top-level is a scalar (e.g. an int), this will raise TypeError instead of producing a validation error. Check isinstance(data, list) (and handle None) before calling len().

Suggested change

	num_problems = len(data)
	if not isinstance(data, list):
	print("::error::YAML file should contain a list of entries.")
	return False
	if len(data) < 1:
	if data is None:
	print("::error::YAML file should contain a list of entries.")
	return False
	if not isinstance(data, list):
	print("::error::YAML file should contain a list of entries.")
	return False
	num_problems = len(data)
	if num_problems < 1:

[Copilot]

This says the merge script “deletes the separate yaml file after merging”, but utils/merge_yaml.py currently does not delete it. Please either implement deletion (and mention it’s irreversible) or adjust this description to match actual behavior.

Suggested change

	This script merges a new problem description in a separate yaml file into the main [problems.yaml](../problems.yaml) file. It runs the validation checks from the above script before merging and deletes the separate yaml file after merging.
	This script merges a new problem description in a separate yaml file into the main [problems.yaml](../problems.yaml) file. It runs the validation checks from the above script before merging.

[Copilot]

renderRowDetails() builds HTML using rowData[index] and then assigns it via innerHTML. Because table cell content ultimately comes from YAML and yaml_to_html.py renders cells with escape=False, this enables XSS if any YAML string contains HTML/script. Consider sanitizing values before injecting (e.g., escape everything and only allow safe link markup, or use a sanitizer like DOMPurify).

[Copilot]

The details container is a <dl>, but the JS renders each item wrapped in a <div> containing <dt>/<dd>. In HTML, <dt>/<dd> should be direct children of <dl> for correct semantics/screen reader navigation. Either change the container to a regular <div> (and use appropriate roles) or render dt/dd pairs directly without a wrapper element.

Suggested change

	<dl class="details-grid" id="problem-details-content"></dl>
	<div class="details-grid" id="problem-details-content"></div>

[Copilot]

This example constructs ValueRange(min=0, max=None) for constraints. With the current ValueRange validator (if not min and not max), min=0 is treated as empty and raises a validation error, so this example won’t run as written. Either update ValueRange to treat 0 as valid (check is None instead of falsiness) or avoid min=0 in the example.

Suggested change

	box=ValueRange(min=0, max=None),
	linear=ValueRange(min=0, max=None),
	function=ValueRange(min=0, max=None),
	box=ValueRange(min=1, max=None),
	linear=ValueRange(min=1, max=None),
	function=ValueRange(min=1, max=None),