Function process_task

ocrflux/client.py:147–201

(args, task_name, task_args)

                pass

async def process_task(args, task_name, task_args):
    COMPLETION_URL = f"{args.url}:{args.port}/v1/chat/completions"
    MAX_RETRIES = args.max_page_retries
    attempt = 0
    while attempt < MAX_RETRIES:        
        if task_name == 'page_to_markdown':
            query = build_page_to_markdown_query(args, *task_args)
        elif task_name == 'element_merge_detect':
            query = build_element_merge_detect_query(args, *task_args)
        elif task_name == 'html_table_merge':
            query = build_html_table_merge_query(args, *task_args)
        
        query["temperature"] = 0.1 * attempt

        try:
            status_code, response_body = await apost(COMPLETION_URL, json_data=query)

            if status_code != 200:
                raise ValueError(f"Error http status {status_code}")

            base_response_data = json.loads(response_body)
            response_content = base_response_data["choices"][0]["message"]["content"]

            if task_name == 'page_to_markdown':
                model_response_json = json.loads(response_content)
                page_response = PageResponse(**model_response_json)
                natural_text = page_response.natural_text
                markdown_element_list = []
                for text in natural_text.split('\n\n'):
                    if text.startswith("<Image>") and text.endswith("</Image>"):
                        pass
                    elif text.startswith("<table>") and text.endswith("</table>"):
                        try:
                            new_text = table_matrix2html(text)
                        except:
                            new_text = text.replace("<t>","").replace("<l>","").replace("<lt>","")
                        markdown_element_list.append(new_text)
                    else:
                        markdown_element_list.append(text)
                return_data = markdown_element_list
                   
            elif task_name == 'element_merge_detect':
                return_data = eval(response_content)

            elif task_name == 'html_table_merge':
                if not (response_content.startswith("<table>") and response_content.endswith("</table>")):
                    raise ValueError("Response is not a table")
                return_data = response_content

            return return_data
        
        except Exception as e:
            traceback.print_exc()
            attempt += 1
    return None

def build_document_text(page_to_markdown_result, element_merge_detect_result, html_table_merge_result):
    page_to_markdown_keys = list(page_to_markdown_result.keys())