apps/summarization: Restructure files, summarise modules one by one

apps/projects/utils.py

@@ -14,7 +14,8 @@
 )
 from apps.summarization.export_utils.core import generate_full_export
 from apps.summarization.pydantic_models import ProjectSummaryResponse
-from apps.summarization.services import AIService
+from apps.summarization.services import DocumentProcessor
+from apps.summarization.services import ProjectSummarizer
 
 logger = logging.getLogger(__name__)
 
@@ -44,7 +45,7 @@ def generate_project_summary(
         )
         if documents_dict:
             try:
-                service = AIService()
+                service = DocumentProcessor()
                 document_response = service.request_vision_dict(
                     documents_dict=documents_dict
                 )
@@ -64,7 +65,7 @@ def generate_project_summary(
 
     json_text = json.dumps(export_data, indent=2)
     prompt = Settings.get_value("project_summary_prompt")
-    service = AIService()
+    service = ProjectSummarizer()
     response = service.project_summarize(
         project=project,
         text=json_text,

apps/summarization/pydantic_models.py

@@ -208,6 +208,31 @@ def is_document(self) -> bool:
         return any(url_lower.endswith(ext) for ext in document_extensions)
 
 
+# Add these new models for module-by-module summarization
+class ModuleSummaryResponse(BaseModel):
+    """Response model for a single module summary."""
+
+    summary: str = Field(
+        description="A 2-3 sentence overview of what happened in this module"
+    )
+    bullets: List[str] = Field(
+        default_factory=list,
+        description="Key points about contributions, ideas, or outcomes",
+    )
+
+
+class GeneralInfoResponse(BaseModel):
+    """Response model for project general information."""
+
+    summary: str = Field(
+        description="A 3-4 sentence overview of the entire project's participation journey"
+    )
+    goals: List[str] = Field(
+        default_factory=list,
+        description="Main goals or themes that emerged from the participation",
+    )
+
+
 class DocumentSummaryItem(BaseModel):
     """Response model for a single document summary with handle."""
 

apps/summarization/requests/__init__.py

@@ -0,0 +1,13 @@
+"""AI requests package."""
+
+from .base import AIRequest
+from .document import DocumentRequest
+from .document import MultimodalSummaryRequest
+from .project import SummaryRequest
+
+__all__ = [
+    "AIRequest",
+    "SummaryRequest",
+    "MultimodalSummaryRequest",
+    "DocumentRequest",
+]

apps/summarization/requests/base.py

@@ -0,0 +1,15 @@
+"""Base classes for AI requests."""
+
+from abc import ABC
+from abc import abstractmethod
+
+
+class AIRequest(ABC):
+    """Base class for all AI requests."""
+
+    vision_support = False
+
+    @abstractmethod
+    def prompt(self) -> str:
+        """Return the prompt text for this request."""
+        pass

apps/summarization/requests/document.py

@@ -0,0 +1,37 @@
+"""Document and image processing requests."""
+
+from .base import AIRequest
+
+
+class MultimodalSummaryRequest(AIRequest):
+    """Request model for multimodal document summarization."""
+
+    vision_support = True
+    PROMPT = "Summarize this image/document. Return JSON with summary field."
+
+    def __init__(
+        self, image_urls: list[str], text: str | None = None, prompt: str | None = None
+    ):
+        super().__init__()
+        self.image_urls = image_urls
+        self.prompt_text = prompt or self.PROMPT
+        self.text = text
+
+    def prompt(self) -> str:
+        base = self.prompt_text
+        return f"{base}\n\nText:\n{self.text}" if self.text else base
+
+
+class DocumentRequest(AIRequest):
+    """Request model for document summarization."""
+
+    vision_support = True
+    PROMPT = "Summarize this document. Return JSON with summary field."
+
+    def __init__(self, url: str, prompt: str | None = None):
+        super().__init__()
+        self.image_urls = [url]
+        self.prompt_text = prompt or self.PROMPT
+
+    def prompt(self) -> str:
+        return self.prompt_text

apps/summarization/requests/project.py

@@ -0,0 +1,121 @@
+"""Project-specific AI requests."""
+
+import json
+from typing import Any
+from typing import Dict
+from typing import List
+
+from .base import AIRequest
+
+
+class ModuleSummaryRequest(AIRequest):
+    """Request for summarizing a single module."""
+
+    PROMPT_TEMPLATE = """
+    Summarize this participation module:
+
+    Module: {module_name}
+    Phase: {phase}
+    Description: {description}
+    Data: {content}
+
+    Return ONLY valid JSON with EXACTLY this format:
+    {{
+      "summary": "A 2-3 sentence overview of what happened in this module",
+      "bullets": [
+        "Key point 1 about specific contributions",
+        "Key point 2 about ideas or proposals",
+        "Key point 3 about engagement or outcomes"
+      ]
+    }}
+
+  The response MUST include BOTH "summary" and "bullets" fields.
+  "bullets" MUST be an array of strings, never empty.
+  """
+
+    def __init__(self, module_data: Dict[str, Any], phase: str):
+        super().__init__()
+        self.module_data = module_data
+        self.phase = phase
+
+    def prompt(self) -> str:
+        """Generate the prompt for this module."""
+        return self.PROMPT_TEMPLATE.format(
+            module_name=self.module_data.get("module_name", "Unknown"),
+            phase=self.phase,
+            description=self.module_data.get("description", "No description"),
+            content=json.dumps(self.module_data.get("content", {})),
+        )
+
+
+class GeneralInfoRequest(AIRequest):
+    """Request for project-level summary."""
+
+    PROMPT_TEMPLATE = """
+    Summarize this entire project:
+
+    Project: {project_name}
+    Description: {description}
+    Module Summaries: {module_summaries}
+
+    Return ONLY valid JSON with EXACTLY this format:
+    {{
+      "summary": "A 3-4 sentence overview of the entire project's participation",
+      "goals": [
+        "First main goal or theme",
+        "Second main goal or theme",
+        "Third main goal or theme"
+      ]
+    }}
+
+    The response MUST include BOTH "summary" and "goals" fields.
+    "goals" MUST be an array of strings, at least 2-3 items.
+    """
+
+    def __init__(
+        self, project_data: Dict[str, Any], module_summaries: List[Dict[str, Any]]
+    ):
+        super().__init__()
+        self.project_data = project_data
+        self.module_summaries = module_summaries
+
+    def prompt(self) -> str:
+        """Generate the prompt for project summary."""
+        project = self.project_data.get("project", {})
+        return self.PROMPT_TEMPLATE.format(
+            project_name=project.get("name", "Unknown"),
+            description=project.get("information", "No description"),
+            module_summaries=json.dumps(self.module_summaries),
+        )
+
+
+class SummaryRequest(AIRequest):
+    """Legacy request model for text summarization."""
+
+    DEFAULT_PROMPT = """
+    You are a JSON generator. Return ONLY valid JSON.
+
+    Schema:
+    {
+      "title": "Summary of participation",
+      "general_info": {"summary": "string", "goals": ["string"]},
+      "phases": {
+        "past": {"modules": [{"module_id": "number", "module_name": "string", "status": "past", "final": {"summary": "string", "bullets": ["string"]}}]},
+        "current": {"modules": [{"module_id": "number", "module_name": "string", "status": "current", "final": {"summary": "string", "bullets": ["string"]}}]},
+        "upcoming": {"modules": [{"module_id": "number", "module_name": "string", "status": "upcoming", "final": {"summary": "string", "bullets": ["string"]}}]}
+      }
+    }
+
+    NOTE: module_id in the output should match the given module_id of the input for each module
+
+    Extract real data from the project export.
+    Respond with ONLY the JSON object.
+    """
+
+    def __init__(self, text: str, prompt: str | None = None):
+        super().__init__()
+        self.text = text
+        self.prompt_text = prompt or self.DEFAULT_PROMPT
+
+    def prompt(self) -> str:
+        return f"{self.prompt_text}\n\n{self.text}"