mirror of
https://github.com/coleam00/Archon.git
synced 2025-12-24 02:39:17 -05:00
Fixes: crawl code storage issue with <think> tags for ollama models. (#775)
* Fixes: crawl code storage issue with <think> tags for ollama models. * updates from code rabbit review
This commit is contained in:
@@ -782,6 +782,12 @@ CREATE POLICY "Allow public read access to archon_code_examples"
|
||||
TO public
|
||||
USING (true);
|
||||
|
||||
CREATE POLICY "Allow public read access to archon_page_metadata"
|
||||
ON archon_page_metadata
|
||||
FOR SELECT
|
||||
TO public
|
||||
USING (true);
|
||||
|
||||
-- =====================================================
|
||||
-- SECTION 7: PROJECTS AND TASKS MODULE
|
||||
-- =====================================================
|
||||
@@ -954,6 +960,73 @@ COMMENT ON COLUMN archon_document_versions.change_type IS 'Type of change: creat
|
||||
COMMENT ON COLUMN archon_document_versions.document_id IS 'For docs arrays, the specific document ID that was changed';
|
||||
COMMENT ON COLUMN archon_document_versions.task_id IS 'DEPRECATED: No longer used for new versions, kept for historical task version data';
|
||||
|
||||
-- =====================================================
|
||||
-- SECTION 6.5: PAGE METADATA FOR PAGE-BASED RAG
|
||||
-- =====================================================
|
||||
|
||||
-- Create archon_page_metadata table
|
||||
-- This table stores complete documentation pages alongside chunks for improved agent context retrieval
|
||||
CREATE TABLE IF NOT EXISTS archon_page_metadata (
|
||||
-- Primary identification
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
|
||||
source_id TEXT NOT NULL,
|
||||
url TEXT NOT NULL,
|
||||
|
||||
-- Content
|
||||
full_content TEXT NOT NULL,
|
||||
|
||||
-- Section metadata (for llms-full.txt H1 sections)
|
||||
section_title TEXT,
|
||||
section_order INT DEFAULT 0,
|
||||
|
||||
-- Statistics
|
||||
word_count INT NOT NULL,
|
||||
char_count INT NOT NULL,
|
||||
chunk_count INT NOT NULL DEFAULT 0,
|
||||
|
||||
-- Timestamps
|
||||
created_at TIMESTAMPTZ DEFAULT NOW(),
|
||||
updated_at TIMESTAMPTZ DEFAULT NOW(),
|
||||
|
||||
-- Flexible metadata storage
|
||||
metadata JSONB DEFAULT '{}'::jsonb,
|
||||
|
||||
-- Constraints
|
||||
CONSTRAINT archon_page_metadata_url_unique UNIQUE(url),
|
||||
CONSTRAINT archon_page_metadata_source_fk FOREIGN KEY (source_id)
|
||||
REFERENCES archon_sources(source_id) ON DELETE CASCADE
|
||||
);
|
||||
|
||||
-- Add page_id foreign key to archon_crawled_pages
|
||||
-- This links chunks back to their parent page
|
||||
-- NULLABLE because existing chunks won't have a page_id yet
|
||||
ALTER TABLE archon_crawled_pages
|
||||
ADD COLUMN IF NOT EXISTS page_id UUID REFERENCES archon_page_metadata(id) ON DELETE SET NULL;
|
||||
|
||||
-- Create indexes for query performance
|
||||
CREATE INDEX IF NOT EXISTS idx_archon_page_metadata_source_id ON archon_page_metadata(source_id);
|
||||
CREATE INDEX IF NOT EXISTS idx_archon_page_metadata_url ON archon_page_metadata(url);
|
||||
CREATE INDEX IF NOT EXISTS idx_archon_page_metadata_section ON archon_page_metadata(source_id, section_title, section_order);
|
||||
CREATE INDEX IF NOT EXISTS idx_archon_page_metadata_created_at ON archon_page_metadata(created_at);
|
||||
CREATE INDEX IF NOT EXISTS idx_archon_page_metadata_metadata ON archon_page_metadata USING GIN(metadata);
|
||||
CREATE INDEX IF NOT EXISTS idx_archon_crawled_pages_page_id ON archon_crawled_pages(page_id);
|
||||
|
||||
-- Add comments to document the table structure
|
||||
COMMENT ON TABLE archon_page_metadata IS 'Stores complete documentation pages for agent retrieval';
|
||||
COMMENT ON COLUMN archon_page_metadata.source_id IS 'References the source this page belongs to';
|
||||
COMMENT ON COLUMN archon_page_metadata.url IS 'Unique URL of the page (synthetic for llms-full.txt sections with #anchor)';
|
||||
COMMENT ON COLUMN archon_page_metadata.full_content IS 'Complete markdown/text content of the page';
|
||||
COMMENT ON COLUMN archon_page_metadata.section_title IS 'H1 section title for llms-full.txt pages';
|
||||
COMMENT ON COLUMN archon_page_metadata.section_order IS 'Order of section in llms-full.txt file (0-based)';
|
||||
COMMENT ON COLUMN archon_page_metadata.word_count IS 'Number of words in full_content';
|
||||
COMMENT ON COLUMN archon_page_metadata.char_count IS 'Number of characters in full_content';
|
||||
COMMENT ON COLUMN archon_page_metadata.chunk_count IS 'Number of chunks created from this page';
|
||||
COMMENT ON COLUMN archon_page_metadata.metadata IS 'Flexible JSON metadata (page_type, knowledge_type, tags, etc)';
|
||||
COMMENT ON COLUMN archon_crawled_pages.page_id IS 'Foreign key linking chunk to parent page';
|
||||
|
||||
-- Enable RLS on archon_page_metadata
|
||||
ALTER TABLE archon_page_metadata ENABLE ROW LEVEL SECURITY;
|
||||
|
||||
-- =====================================================
|
||||
-- SECTION 7: MIGRATION TRACKING
|
||||
-- =====================================================
|
||||
@@ -991,7 +1064,9 @@ VALUES
|
||||
('0.1.0', '006_ollama_create_indexes_optional'),
|
||||
('0.1.0', '007_add_priority_column_to_tasks'),
|
||||
('0.1.0', '008_add_migration_tracking'),
|
||||
('0.1.0', '009_add_cascade_delete_constraints')
|
||||
('0.1.0', '009_add_cascade_delete_constraints'),
|
||||
('0.1.0', '009_add_provider_placeholders'),
|
||||
('0.1.0', '010_add_page_metadata_table')
|
||||
ON CONFLICT (version, migration_name) DO NOTHING;
|
||||
|
||||
-- Enable Row Level Security on migrations table
|
||||
|
||||
Reference in New Issue
Block a user