Converti (AetherOS) - Revision history

Converti at 06:37, 15 August 2025

2025-08-15T06:37:29Z

← Older revision		Revision as of 06:37, 15 August 2025
Line 10:		Line 10:

	== The Converti Toolkit ==		== The Converti Toolkit ==
	Converti is comprised of several ~~distinct scripts~~, each serving a specific function in the Collegium's review process.		Converti is comprised of a single command-line interface (main.py) that provides several commands, each serving a specific function in the Collegium's review process.

	=== Stage 1: ~~Dependency Analysis~~ ===		=== Stage 1: Map Dependencies ===
	The first step is to map the full dependency tree of a set of root templates from ~~an external~~ source like Wikipedia.		The first step is to map the full dependency tree of a set of root templates from a source wiki (like Wikipedia) to a named JSON file.
	* '''`dependency_resolver.py`''': Scans a list of templates and recursively maps every sub-template and Lua module they depend on, saving the result to ~~`dependency_map.json`.~~
	* '''`dependency_analyzer.py` (`scriptor_explico.py`)''': Ingests the raw dependency map and generates a ~~human-readable report (`dependency_report.txt`) and an interactive graph (`dependency_graph.html`), highlighting the most foundational and complex components~~.

	~~=== Stage 2: Controlled Import ===~~		'''map command''': Scans a list of templates on a specified source wiki (--source-family) and recursively maps every sub-template and Lua module they depend on. The result is saved to a specified output file (--output).
	~~After the `Custos Animae` and `Custos Structurae` have reviewed the analysis and created a curated `approved_list.txt`, the batch installer imports only the necessary components.~~
	* '''~~`batch_installer.py`~~''': ~~Reads the approved~~ list and ~~the dependency map, and surgically imports only the approved pages~~ and ~~their direct dependencies from Wikipedia, skipping any that already exist~~ on ~~ooda~~.~~wiki~~.

	=== Stage 3: ~~Quality Assurance~~ & ~~Refactoring~~ ===		=== Stage 2: Audit & Analyze ===
	~~Once templates are imported,~~ the ~~Collegium can analyze their quality and complexity~~.		The second step is to analyze the generated dependency map to assess its technical debt.
	* '''~~`scriptor_explico.py`~~''': ~~Calculates~~ a '''[[Wiki Maneuverability Score]]''' for each ~~template~~ based on its complexity, dependency count, and documentation quality.
	* '''~~`refactoring_assistant.py`~~''': (~~Under Development~~) ~~An assistant~~ that ~~uses an LLM to review low~~-~~scoring templates and suggest simplifications or alternative implementations~~.		'''audit command''': Ingests a dependency map (--map-file) and generates a human-readable report (dependency_report.txt) and an interactive graph (dependency_graph.html). This command calculates a '''[[Wiki Maneuverability Score]]''' for each component based on its complexity, dependency count, and documentation quality, highlighting the most foundational and complex pages.

			=== Stage 3: Deploy to Wiki ===
			After the Collegium has reviewed the analysis and created a curated approved_list.txt, this command imports only the necessary components to the target wiki.

			'''deploy command''': Reads the approved list and a dependency map. It then resolves the full dependency tree and surgically imports only the required pages from the source wiki to the target (ooda.wiki), skipping any that already exist. A --dry-run flag allows for simulating the deployment without making any edits.

	== Future Development ==		== Future Development ==
	* '''Active Voice Assistant''': An on-wiki tool to help editors convert passive voice to active voice, further improving the clarity and readability of the knowledge base.
			'''refactor command''': (Under Development) An assistant that uses an LLM to review low-scoring templates and suggest simplifications or alternative implementations.

			'''Active Voice Assistant''': An on-wiki tool to help editors convert passive voice to active voice, further improving the clarity and readability of the knowledge base.

AdminIsidore at 20:13, 10 August 2025

2025-08-10T20:13:54Z

← Older revision		Revision as of 20:13, 10 August 2025
Line 1:		Line 1:
	{{AetherOS_Component}}		{{AetherOS_Component}}
	'''Converti''' is a command-line utility designed to ~~import articles from external wikis, such as Wikipedia,~~ and ~~intelligently prepare them for use~~ within the AetherOS ecosystem. It serves as ~~a gatekeeper~~, ~~ensuring that new information is clean~~, ~~functional~~, and ~~structurally sound before it becomes a source~~ of ~~knowledge for both human users and their [[Wingman AI\|Wingman AIs]]~~.		'''Converti''' is a command-line utility suite designed to manage the lifecycle of templates and modules within the AetherOS ecosystem. It serves as the primary toolkit for the [[Collegium (AetherOS)]], enabling the analysis, controlled importation, and quality assurance of the wiki's structural components.

	The tool was developed by [[User:~~AdminIsidore~~\|Isidore Lands]] and Silas Corvus.		The tool was developed by [[User:Isidore Lands\|Isidore Lands]] and [[User:Silas Corvus\|Silas Corvus]].

	== Purpose in Symbiotic Learning ==		== Purpose in Symbiotic Learning ==
	The core purpose of Converti is to ~~support~~ the symbiotic learning ~~relationship~~ between ~~a user~~ and their Wingman AI~~. Raw, unformatted text from external sources can contain broken templates, irrelevant maintenance tags, and visual media that can confuse a learning model~~.		The core purpose of Converti is to ensure the structural integrity and maneuverability of the wiki's knowledge base. A lean, well-documented, and minimally dependent codebase is essential for the symbiotic learning between users and their [[Wingman AI\|Wingman AIs]].

	~~Converti sanitizes this content, creating a clean textual baseline that is optimized for machine readability~~ and ~~comprehension. By ensuring the structural integrity of imported knowledge~~, it prevents ~~the Wingman from learning from~~ "~~broken~~" ~~or incomplete data~~.		By providing tools to analyze and gracefully import components, Converti prevents "dependency hell" and ensures the wiki does not become cluttered with the unmanaged complexity of its source wikis.

	== ~~Core Functionality~~ ==		== The Converti Toolkit ==
	Converti is a ~~Python script that uses~~ the ~~[[Pywikibot]] framework to perform a multi-stage import and cleaning~~ process.		Converti is comprised of several distinct scripts, each serving a specific function in the Collegium's review process.

	=== Stage 1: ~~Content Cleaning~~ ===		=== Stage 1: Dependency Analysis ===
	~~When an article~~ is ~~imported, Converti first performs two cleaning passes on~~ the ~~wikitext:~~		The first step is to map the full dependency tree of a set of root templates from an external source like Wikipedia.
	# '''~~Image Removal~~''': ~~All file links (e.g.~~, `~~[[File:~~...]]`~~) and gallery tags~~ (`~~<gallery>~~`) ~~are stripped from the text. This focuses the content on pure text, which is the primary medium for the Wingman AI.~~		* '''`dependency_resolver.py`''': Scans a list of templates and recursively maps every sub-template and Lua module they depend on, saving the result to `dependency_map.json`.
	~~# '''Template Reformatting~~''': ~~Common Wikipedia~~-~~specific maintenance templates~~ (e.~~g.,~~ `~~{{cleanup}}~~`~~, `{{unreferenced}}~~`) ~~are removed~~, ~~as they are not relevant to~~ the ~~local wiki's context~~.		* '''`dependency_analyzer.py` (`scriptor_explico.py`)''': Ingests the raw dependency map and generates a human-readable report (`dependency_report.txt`) and an interactive graph (`dependency_graph.html`), highlighting the most foundational and complex components.

	=== Stage 2: ~~Recursive Dependency Resolution~~ ===		=== Stage 2: Controlled Import ===
	After ~~cleaning~~ the ~~text, Converti scans~~ the ~~article for all required [[Templates]]~~ and ~~[[Lua\|Modules]]~~.		After the `Custos Animae` and `Custos Structurae` have reviewed the analysis and created a curated `approved_list.txt`, the batch installer imports only the necessary components.
	~~# It checks if each dependency exists on~~ the ~~local wiki.~~		* '''`batch_installer.py`''': Reads the approved list and the dependency map, and surgically imports only the approved pages and their direct dependencies from Wikipedia, skipping any that already exist on ooda.wiki.
	~~# If a dependency is missing and its corresponding local page is empty, Converti automatically~~ imports ~~it from~~ the ~~source wiki (Wikipedia)~~.
	~~# This process is~~ '''~~recursive~~'''; the ~~script then checks~~ the ~~newly imported~~ dependency ~~for its own~~ dependencies ~~and adds them to an import queue. This continues until all required components are present.~~
	~~# To ensure safety~~, ~~Converti will '''not''' overwrite~~ any ~~template or module~~ that already ~~has content, preserving any local customizations~~.

	=== Stage 3: ~~Reporting~~ ===		=== Stage 3: Quality Assurance & Refactoring ===
	~~Upon completion~~, the ~~script generates a detailed report listing all the dependencies it successfully imported~~ and ~~any it skipped because they already existed~~.		Once templates are imported, the Collegium can analyze their quality and complexity.
			* '''`scriptor_explico.py`''': Calculates a '''[[Wiki Maneuverability Score]]''' for each template based on its complexity, dependency count, and documentation quality.
	~~== Usage ==~~		* '''`refactoring_assistant.py`''': (Under Development) An assistant that uses an LLM to review low-scoring templates and suggest simplifications or alternative implementations.
	~~Converti is run from the command line~~. ~~The user specifies the title of the page to import~~.
	~~<syntaxhighlight lang="bash">~~
	~~# Example: Import the "OODA loop" article from Wikipedia~~
	~~python scripts/wiki_importer~~.py -~~-page "OODA_loop"~~
	~~</syntaxhighlight>~~

	== Future Development ==		== Future Development ==
	* '''Active Voice Assistant''': ~~Functionality is being developed to create a human-in-the~~-~~loop~~ tool ~~that will identify~~ passive voice ~~sentences in an article and allow an editor to convert them~~ to active voice, further improving clarity and readability.		* '''Active Voice Assistant''': An on-wiki tool to help editors convert passive voice to active voice, further improving the clarity and readability of the knowledge base.

AdminIsidore: Created page with "{{AetherOS_Component}} '''Converti''' is a command-line utility designed to import articles from external wikis, such as Wikipedia, and intelligently prepare them for use within the AetherOS ecosystem. It serves as a gatekeeper, ensuring that new information is clean, functional, and structurally sound before it becomes a source of knowledge for both human users and their Wingman AIs. The tool was developed by Isidore Lands and Silas..."

2025-08-09T03:09:30Z

Created page with "{{AetherOS_Component}} '''Converti''' is a command-line utility designed to import articles from external wikis, such as Wikipedia, and intelligently prepare them for use within the AetherOS ecosystem. It serves as a gatekeeper, ensuring that new information is clean, functional, and structurally sound before it becomes a source of knowledge for both human users and their Wingman AIs. The tool was developed by Isidore Lands and Silas..."

New page

{{AetherOS_Component}}
'''Converti''' is a command-line utility designed to import articles from external wikis, such as Wikipedia, and intelligently prepare them for use within the AetherOS ecosystem. It serves as a gatekeeper, ensuring that new information is clean, functional, and structurally sound before it becomes a source of knowledge for both human users and their [[Wingman AI|Wingman AIs]].

The tool was developed by [[User:AdminIsidore|Isidore Lands]] and Silas Corvus.

== Purpose in Symbiotic Learning ==
The core purpose of Converti is to support the symbiotic learning relationship between a user and their Wingman AI. Raw, unformatted text from external sources can contain broken templates, irrelevant maintenance tags, and visual media that can confuse a learning model.

Converti sanitizes this content, creating a clean textual baseline that is optimized for machine readability and comprehension. By ensuring the structural integrity of imported knowledge, it prevents the Wingman from learning from "broken" or incomplete data.

== Core Functionality ==
Converti is a Python script that uses the [[Pywikibot]] framework to perform a multi-stage import and cleaning process.

=== Stage 1: Content Cleaning ===
When an article is imported, Converti first performs two cleaning passes on the wikitext:
# '''Image Removal''': All file links (e.g., `[[File:...]]`) and gallery tags (`<gallery>`) are stripped from the text. This focuses the content on pure text, which is the primary medium for the Wingman AI.
# '''Template Reformatting''': Common Wikipedia-specific maintenance templates (e.g., `{{cleanup}}`, `{{unreferenced}}`) are removed, as they are not relevant to the local wiki's context.

=== Stage 2: Recursive Dependency Resolution ===
After cleaning the text, Converti scans the article for all required [[Templates]] and [[Lua|Modules]].
# It checks if each dependency exists on the local wiki.
# If a dependency is missing and its corresponding local page is empty, Converti automatically imports it from the source wiki (Wikipedia).
# This process is '''recursive'''; the script then checks the newly imported dependency for its own dependencies and adds them to an import queue. This continues until all required components are present.
# To ensure safety, Converti will '''not''' overwrite any template or module that already has content, preserving any local customizations.

=== Stage 3: Reporting ===
Upon completion, the script generates a detailed report listing all the dependencies it successfully imported and any it skipped because they already existed.

== Usage ==
Converti is run from the command line. The user specifies the title of the page to import.
<syntaxhighlight lang="bash">
# Example: Import the "OODA loop" article from Wikipedia
python scripts/wiki_importer.py --page "OODA_loop"
</syntaxhighlight>

== Future Development ==
* '''Active Voice Assistant''': Functionality is being developed to create a human-in-the-loop tool that will identify passive voice sentences in an article and allow an editor to convert them to active voice, further improving clarity and readability.