Error - A Parser can not resolve classes

[jorisfayard]

Context I am using an Open API specs to serve as the single source of truth in my application across my frontend written in Typescript and my backend written in Python.

Describe the bug I get a "A Parser can not resolve classes" error if I don't reference some models from my openapi.yaml file. I checked multiple times that there were no issues but couldn't find anything wrong.

To Reproduce *Create the following files: ./openapi.yaml ./schemas/feedback.yaml

*Run datamodel-codegen --input ./schemas/openapi.yaml --input-file-type openapi --output ../../apps/backend/app/generated/models.py --output-model-type pydantic_v2.BaseModel --use-schema-description --use-field-description --field-constraints

*Get the following error: Exception: A Parser can not resolve classes: [class: openapi.yaml#/components/schemas/FeedbackItem references: frozenset({'schemas/feedback.yaml#/FeedbackItem'})], [class: schemas/feedback.yaml#/FeedbackItem references: frozenset({'schemas/feedback.yaml#/FeedbackItemBase'})], [class: openapi.yaml#/components/schemas/FeedbackItemCreate references: frozenset({'schemas/feedback.yaml#/FeedbackItemCreate'})], [class: schemas/feedback.yaml#/FeedbackItemCreate references: frozenset({'schemas/feedback.yaml#/FeedbackItemBase'})].

*Add FeedbackItemBase to openapi.yaml. Run the command again, it works.

Example schema:

./openapi.yaml

openapi: 3.0.0
info:
  title: Workshop Feedback API
  version: 1.0.0
  description: API for managing workshop feedback and summaries.
  contact:
    name: test
    email: [email protected]
servers:
  - url: https://api.yourdomain.com/v1 # Replace with your Production URL
    description: Production server
  - url: https://staging-api.yourdomain.com/v1 # Replace with your Staging URL
    description: Staging server
  - url: http://localhost:8080 # Replace with your local development URL/port
    description: Local development server
tags:
  - name: Feedback Summary
    description: Operations related to generating and managing feedback summaries.

paths:
  /workshops/{workshopId}/feedbacks/summary:
    $ref: "./paths/feedback_path.yaml"
components:
  schemas:
    Workshop:
      $ref: "./schemas/workshops.yaml#/Workshop"
    WorkshopCreate:
      $ref: "./schemas/workshops.yaml#/WorkshopCreate"
    WorkshopBase:
      $ref: "./schemas/workshops.yaml#/WorkshopBase"
    FeedbackItem:
      $ref: "./schemas/feedback.yaml#/FeedbackItem"
    FeedbackItemCreate:
      $ref: "./schemas/feedback.yaml#/FeedbackItemCreate"

./schemas/feedback.yaml

# 1. Base Schema with Common/Client-Provided Fields
FeedbackItemBase:
  type: object
  properties:
    content:
      type: string
      minLength: 1
      description: The content of the feedback. Cannot be empty.
    author:
      type: string
      description: Optional name or identifier of the person providing feedback.
      example: "Jane Doe"
  required:
    # Fields required when SUBMITTING feedback
    - content

# 2. Schema for CREATING a feedback item (Client -> Server POST Body)
FeedbackItemCreate:
  description: Data required to create a new feedback item.
  allOf: # Inherit all properties and requirements from FeedbackItemBase
    - $ref: "#/FeedbackItemBase"

# 3. Schema for REPRESENTING an existing feedback item (Server -> Client Response)
FeedbackItem:
  description: Represents an existing feedback item resource.
  allOf: # Inherit properties and requirements from FeedbackItemBase
    - $ref: "#/FeedbackItemBase"
  # Define properties UNIQUE to the full FeedbackItem object
  properties:
    id:
      type: string
      description: Unique identifier for the feedback item (server-generated).
      readOnly: true # Indicate server-generated
    createdAt:
      type: string
      format: date-time
      description: Timestamp when the feedback was submitted (server-generated).
      example: "2025-03-29T13:01:15Z"
      readOnly: true # Indicate server-generated
  # List ONLY the ADDITIONAL required fields for the full object.
  # 'content' is implicitly required via WorkshopBase. 'author' remains optional.
  required:
    - id
    - timestamp

FeedbackCategory:
  type: object
  properties:
    name:
      type: string
      minLength: 1
      description: Name of the feedback category. Cannot be empty.
      example: "Clarity"
      readOnly: true
    content:
      type: string
      minLength: 1
      description: Content or description for the feedback category. Cannot be empty.
      example: "Feedback related to how clear the topic was explained."
      readOnly: true
    imagePrompt:
      type: string
      description: The prompt used to generate the category image (if applicable).
      example: "A lightbulb turning on over a confused student's head, cartoon style"
      readOnly: true
    imageUrl:
      type: string
      format: url
      description: Optional URL of an image representing the category. Must be a valid URL format.
      example: "https://example.com/images/clarity.png"
      readOnly: true
  required:
    - name
    - content
    - imagePrompt # Mark required as per zod schema (not optional)

Summary:
  type: object
  properties:
    categories:
      type: array
      items:
        $ref: "#/FeedbackCategory"
      minItems: 1 # Corresponds to .nonempty()
      description: An array containing feedback categories. Must contain at least one category.
    shared:
      type: boolean
      default: false # Corresponds to .default(false)
      description: Indicates if the summary is shared. Defaults to false.
  required:
    - categories # `shared` is not required because it has a default

Used commandline:

$ datamodel-codegen --input ./schemas/openapi.yaml --input-file-type openapi --output ../../apps/backend/app/generated/models.py --output-model-type pydantic_v2.BaseModel --use-schema-description --use-field-description --field-constraints

Expected behavior Generate Pydantic types successfully even if FeebackItemBase is not included in the openapi.yaml file since it's already referenced by FeedbackItem and FeedbackItemCreate that are already included it it.

Version:

• OS: Ubuntu
• Python version: 3.12.3
• datamodel-code-generator version: 0.28.5

Additional context Add any other context about the problem here.