Now the bootstrapping seems to work again...

This commit is contained in:
kalzu rekku 2024-10-05 15:46:58 +03:00
parent 4dfc81bd44
commit c87a924669

View File

@ -228,16 +228,21 @@ class MarkdownProcessor:
def merge_structures(self, existing: Dict, new: Dict, document_id: int) -> None:
def merge_recursive(existing_node, new_node, parent_uuid):
if not existing_node:
print('#### This seems to be a new stuff...')
# This is a new node, insert it
print(f"##### inserting new heading: {new_node['title']}")
heading_uuid = self.insert_heading(new_node['level'], new_node['title'], parent_uuid, document_id, new_node['path'])
body_uuid = self.insert_body(new_node['content'], heading_uuid, document_id)
##print(f"##### inserting new body with id: {heading_uuid}, :\n {new_node['content']}\n")
if new_node['content']:
body_uuid = self.insert_body(new_node['content'], heading_uuid, document_id)
for child in new_node['children']:
merge_recursive(None, new[child], heading_uuid)
else:
print('#### This seems to known ')
# Update existing node
self.update_heading(existing_node['uuid'], new_node['title'], new_node['level'], parent_uuid, new_node['path'])
self.update_body(existing_node['body_uuid'], new_node['content'], document_id)
print('#### Does these update_body and update_heading methods work?')
# Process children
existing_children = {child['title']: child for child in existing_node['children']}
new_children = {child['title']: child for child in new_node['children']}
@ -253,6 +258,7 @@ class MarkdownProcessor:
self.soft_delete_heading(child['uuid'])
for new_root in new.values():
print('#### Found a chapter!')
existing_root = next((node for node in existing.values() if node['path'] == new_root['path']), None)
merge_recursive(existing_root, new_root, None)
@ -274,10 +280,39 @@ class MarkdownProcessor:
def insert_body(self, content: str, heading_uuid: str, document_id: int) -> str:
body_uuid = str(uuid.uuid4())
md5sum = hashlib.md5(content.encode()).hexdigest()
self.db_manager.cursor.execute('''
INSERT INTO body (uuid, content, heading_uuid, document_id, md5sum)
VALUES (?, ?, ?, ?, ?)
''', (body_uuid, content, heading_uuid, document_id, md5sum))
print(f"###### Trying to insert body text with md5sum of: {md5sum} to uuid: {body_uuid}, with content: \n{content}\n")
# Verify input parameters
if not all([content, heading_uuid, document_id]):
raise ValueError("Missing required parameters for insert_body")
try:
# Check if heading_uuid exists
self.db_manager.cursor.execute("SELECT 1 FROM headings WHERE uuid = ?", (heading_uuid,))
if not self.db_manager.cursor.fetchone():
raise ValueError(f"heading_uuid {heading_uuid} does not exist in headings table")
# Check if document_id exists
self.db_manager.cursor.execute("SELECT 1 FROM documents WHERE id = ?", (document_id,))
if not self.db_manager.cursor.fetchone():
raise ValueError(f"document_id {document_id} does not exist in documents table")
# Insert the body
self.db_manager.cursor.execute('''
INSERT INTO body (uuid, content, heading_uuid, document_id, md5sum)
VALUES (?, ?, ?, ?, ?)
''', (body_uuid, content, heading_uuid, document_id, md5sum))
self.db_manager.conn.commit()
print(f"###### Successfully inserted body with uuid: {body_uuid}")
except sqlite3.Error as e:
print(f"An error occurred while inserting body: {e}")
self.db_manager.conn.rollback()
raise
except ValueError as e:
print(f"Validation error: {e}")
raise
return body_uuid
def update_body(self, body_uuid: str, content: str, document_id: int) -> None: