Merge pull request #5 from Imaging-Plaza/feat-discipline-repository-type

Feat discipline repository type

Author: caviri

src/api.py

@@ -68,7 +68,7 @@ async def gimie(full_path:str,
     return {"link": full_path, 
             "output": gimie_output}
 
-@app.get("/v1/llm/{full_path:path}")
+@app.get("/v1/llm/json-ld/{full_path:path}")
 async def llm(full_path:str):
 
     try:
@@ -82,6 +82,20 @@ async def llm(full_path:str):
     return {"link": full_path, 
             "output": llm_result}
 
+@app.get("/v1/llm/json/{full_path:path}")
+async def llm(full_path:str):
+
+    try:
+        llm_result = llm_request_repo_infos(str(full_path), output_format="json")
+    except Exception as e:
+        raise HTTPException(
+            status_code=424, 
+            detail=f"Error from LLM service: {e}"
+        )
+    
+    return {"link": full_path, 
+            "output": llm_result}
+
 @app.exception_handler(ValueError)
 async def value_error_exception_handler(request: Request, exc: ValueError):
     return JSONResponse(

src/core/genai_model.py

@@ -110,7 +110,7 @@ def clone_repo(repo_url):
             return None
 
 
-def llm_request_repo_infos(repo_url):    
+def llm_request_repo_infos(repo_url, output_format="json-ld"):    
     # Clone the GitHub repository into a temporary folder
     with tempfile.TemporaryDirectory() as temp_dir:
         logger.info(f"Cloning {repo_url} into {temp_dir}...")
@@ -146,7 +146,6 @@ def llm_request_repo_infos(repo_url):
                 raw_result = response.json()["choices"][0]["message"]["content"]
                 parsed_result = clean_json_string(raw_result)
                 json_data = json.loads(parsed_result)
-                pprint(json_data)
 
                 logger.info("Successfully parsed API response")
 
@@ -161,7 +160,13 @@ def llm_request_repo_infos(repo_url):
                 # TODO. This is hardcoded. Not good.
                 context_path = "src/files/json-ld-context.json"
                 # Now convert cleaned data to JSON-LD
-                return json_to_jsonLD(cleaned_json, context_path)
+                if output_format == "json-ld":
+                    return json_to_jsonLD(cleaned_json, context_path)
+                elif output_format == "json":
+                    return cleaned_json
+                else:
+                    logger.error(f"Unsupported output format: {output_format}")
+                    return None
 
             except Exception as e:
                 logger.error(f"Error parsing response: {e}")

src/core/models.py

@@ -57,6 +57,62 @@ class Image(BaseModel):
     contentUrl: HttpUrl = None
     keywords: ImageKeyword = ImageKeyword.ILLUSTRATIVE_IMAGE
 
+
+class Discipline(str, Enum):
+    SOCIAL_SCIENCES = "Social sciences"
+    ANTHROPOLOGY = "Anthropology"
+    COMMUNICATION_STUDIES = "Communication studies"
+    EDUCATION = "Education"
+    LINGUISTICS = "Linguistics"
+    RESEARCH = "Research"
+    SOCIOLOGY = "Sociology"
+    GEOGRAPHY = "Geography"
+    PSYCHOLOGY = "Psychology"
+    POLITICS = "Politics"
+    ECONOMICS = "Economics"
+    APPLIED_SCIENCES = "Applied sciences"
+    HEALTH_SCIENCES = "Health sciences"
+    ELECTRICAL_ENGINEERING = "Electrical engineering"
+    CHEMICAL_ENGINEERING = "Chemical engineering"
+    CIVIL_ENGINEERING = "Civil engineering"
+    ARCHITECTURE = "Architecture"
+    COMPUTER_ENGINEERING = "Computer engineering"
+    ENERGY_ENGINEERING = "Energy engineering"
+    MILITARY_SCIENCE = "Military science"
+    INDUSTRIAL_PRODUCTION_ENGINEERING = "Industrial and production engineering"
+    MECHANICAL_ENGINEERING = "Mechanical engineering"
+    BIOLOGICAL_ENGINEERING = "Biological engineering"
+    ENVIRONMENTAL_SCIENCE = "Environmental science"
+    SYSTEMS_SCIENCE_ENGINEERING = "Systems science and engineering"
+    INFORMATION_ENGINEERING = "Information engineering"
+    AGRICULTURAL_FOOD_SCIENCES = "Agricultural and food sciences"
+    BUSINESS = "Business"
+    HUMANITIES = "Humanities"
+    HISTORY = "History"
+    LITERATURE = "Literature"
+    ART = "Art"
+    RELIGION = "Religion"
+    PHILOSOPHY = "Philosophy"
+    LAW = "Law"
+    FORMAL_SCIENCES = "Formal sciences"
+    MATHEMATICS = "Mathematics"
+    LOGIC = "Logic"
+    STATISTICS = "Statistics"
+    THEORETICAL_COMPUTER_SCIENCE = "Theoretical computer science"
+    NATURAL_SCIENCES = "Natural sciences"
+    PHYSICS = "Physics"
+    ASTRONOMY = "Astronomy"
+    BIOLOGY = "Biology"
+    CHEMISTRY = "Chemistry"
+    EARTH_SCIENCE = "Earth science"
+
+class RepositoryType(str, Enum):
+    SOFTWARE = "software"
+    EDUCATIONAL_RESOURCE = "educational resource"
+    DOCUMENTATION = "documentation"
+    DATA = "data"
+    OTHER = "other"
+
 class SoftwareSourceCode(BaseModel):
     name: Optional[str] = None
     applicationCategory: Optional[List[str]] = None
@@ -74,6 +130,7 @@ class SoftwareSourceCode(BaseModel):
     license: Annotated[str, StringConstraints(pattern=r"spdx\.org.*")] = None
     author: List[Union[Person, Organization]] = None
     relatedToOrganization: Optional[List[str]] = None
+    relatedToOrganizationJustification: Optional[List[str]] = None
     operatingSystem: Optional[List[str]] = None
     programmingLanguage: Optional[List[str]] = None
     softwareRequirements: Optional[List[str]] = None
@@ -94,6 +151,10 @@ class SoftwareSourceCode(BaseModel):
     imagingModality: Optional[List[str]] = None
     fairLevel: Optional[str] = None
     graph: Optional[str] = None
+    discipline: Optional[List[Discipline]] = None
+    disciplineJustification: Optional[List[str]] = None
+    repositoryType: Optional[RepositoryType] = None
+    respositoryTypeJustification: Optional[List[str]] = None
 
 
 ############################################################

src/core/prompts.py

@@ -36,6 +36,7 @@
   - `orcidId` (valid URL, **optional**)
   - `affiliation` (list of strings, **optional**): Institutions the author is affiliated with. Do not mention Imaging Plaza unless is explicity mentioned.
 - `relatedToOrganization` (list of strings, **optional**): Institutions associated with the software. Do not mention Imaging Plaza unless is explicity mentioned.
+- `relatedToOrganizationJustification` (list of strings, **optional**): Justification for the related organizations.
 - `softwareRequirements` (list of strings, **optional**): Dependencies or prerequisites for running the software.
 - `operatingSystem` (list of strings, **optional**): Compatible operating systems. Use only Windows, Linux, MacOS, or Other.
 - `programmingLanguage` (list of strings, **optional**): Programming languages used in the software.
@@ -92,18 +93,10 @@
 - `hasExecutableInstructions` (string, **optional**): Any exectuable instructions related to the software. This should point to an URL where the installation is explained. If this is the README file, please make the full URL. 
 - `readme` (valid URL, **optional**): README url of the software (at the root of the repo)
 - `imagingModality (list of strings, **optional**): imaging modalities accepted by the software.
-
-
-When dealing with Organization pay attention to
-- 
-- 
-- 
-
-When parsing Persons note:
-- 
-- 
--
-
+- `discipline` (string, **optional**): Scientific discipline the software belongs to. Base your response on the README and other documentation files content.
+- `disciplineJustification` (list of strings, **optional**): Justification for the discipline classification.
+- `repositoryType` (string, **optional**): Type of repository (e.g., software, educational resource, documentation, data, other).
+- `respositoryTypeJustification` (list of strings, **optional**): Justification for the repository type classification.
 
 PLEASE PROVIDE THE OUTPUT IN JSON FORMAT ONLY, WITHOUT ANY EXPLANATION OR ADDITIONAL TEXT. ALIGN THE RESPONSE TO THE SCHEMA SPECIFICATION.
 """

src/files/json-ld-context.json

@@ -76,7 +76,9 @@
     "hasRorId": "md4i:hasRorId",
     "legalName": "schema:legalName",
     "fundingGrant": "sd:fundingGrant",
-    "fundingSource": "sd:fundingSource"
+    "fundingSource": "sd:fundingSource",
+    "discipline": "pulse:discipline",
+    "repositoryType": "pulse:repositoryType"
   }
 }