import os import re import base64 import json from typing import Dict, List, Optional, Tuple import gradio as gr from loguru import logger import sambanova_gradio from sambanova_gradio import get_fn import modelscope_studio.components.base as ms import modelscope_studio.components.legacy as legacy import modelscope_studio.components.antd as antd from config import DEMO_LIST, SystemPrompt YOUR_API_TOKEN = os.getenv('SAMBANOVA_API_KEY') History = List[Tuple[str, str]] Messages = List[Dict[str, str]] from tree_sitter_languages import get_parser def check_code_error(code, language='python'): language_ext = {'go': 'go', 'html': 'html', 'python': 'python', 'cpp': 'cpp', # 'sql': 'sql', 'php': 'php', 'javascript': 'javascript', # 'R': 'r', 'typescript': 'typescript', 'ruby': 'ruby', 'bash': 'bash', 'rust': 'rust', # 'D': 'd', 'scala': 'scala', 'csharp': 'c_sharp', 'java': 'java', 'json': 'json', 'julia': 'julia', 'css': 'css', 'jsdoc': 'jsdoc', 'swift': 'swift', 'cli': 'cli', } if language not in language_ext.keys(): return {'result': True, 'error': 'no language in tree_sitter_languages'} try: parser = get_parser(language_ext[language]) code_bytes = code.encode('utf-8') tree = parser.parse(code_bytes) root_node = tree.root_node visit_list = [root_node] ans = [] while visit_list: node = visit_list.pop() if node.type == 'ERROR': return {'result': True, 'error': True, 'root_node': root_node} is_inner = True if node.children else False ans.append([node, is_inner]) for child in node.children: visit_list.append(child) return {'result': True, 'error': False, 'root_node': root_node} except Exception as e: return {'result': False, 'error': f"error:{str(e)}"} def history_to_messages(history: History, system: str) -> Messages: messages = [{'role': 'system', 'content': system}] for h in history: messages.append({'role': 'user', 'content': h[0]}) messages.append({'role': 'assistant', 'content': h[1]}) return messages def messages_to_history(messages: Messages) -> History: assert messages[0]['role'] == 'system' history = [] for q, r in zip(messages[1::2], messages[2::2]): history.append([q['content'], r['content']]) return history def remove_code_block(text, language='html'): pattern = r'```' + language.lower() + '\n(.+?)\n```' match = re.search(pattern, text, re.DOTALL) if match: return match.group(1).strip() else: return text.strip() def history_render(history: History): return gr.update(open=True), history def clear_history(): return [] def send_to_sandbox(code): encoded_html = base64.b64encode(code.encode('utf-8')).decode('utf-8') data_uri = f"data:text/html;charset=utf-8;base64,{encoded_html}" return f"" def demo_card_click(e: gr.EventData): index = e._data['component']['index'] return DEMO_LIST[index]['description'] def read_jsonl(fn): with open(fn, encoding="utf-8") as f: return [json.loads(l) for l in f.readlines()] def get_item_by_index(item_list, index, val=None): if val is not None and val in item_list: return val if len(item_list) > index: return item_list[index] return "" def get_item_list(locale="zh", programming_language="", category="", difficulty="", debugInfo=""): samples = read_jsonl(f'./FullStackBench_data/fsb_{locale}_20241204.jsonl') item_list_count = {} item_samples = [] for sample in samples: labels = sample['labels'] content = [sample['content']] sample_dict = [labels['programming_language'], f"/{labels['category']}/{labels['difficulty']}/", content] if (locale != "" and labels['locale'] == locale): if programming_language == "": item = labels['programming_language'] item_list_count[item] = item_list_count.get(item, 0) + 1 item_samples.append(sample_dict) elif category == "" and labels['programming_language'] == programming_language: item = labels['category'] item_list_count[item] = item_list_count.get(item, 0) + 1 item_samples.append(sample_dict) elif difficulty == "" and labels['programming_language'] == programming_language and labels['category'] == category: item = labels['difficulty'] item_list_count[item] = item_list_count.get(item, 0) + 1 item_samples.append(sample_dict) elif labels['programming_language'] == programming_language and labels['category'] == category and labels['difficulty'] == difficulty: item = labels['difficulty'] item_list_count[item] = item_list_count.get(item, 0) + 1 item_samples.append(sample_dict) item_list = list(item_list_count.keys()) item_list_dropdown = [] all_count = 0 for item in item_list: all_count += item_list_count[item] item_list_dropdown.append((f"{item}[{item_list_count[item]}]",item)) item_list_dropdown.insert(0,(f"All[{all_count}]", "")) return item_list, item_list_dropdown, item_samples item_list_locale = ["en", "zh"] item_list_dropdown_locale = [("en[1687]", "en"), ("zh[1687]", "zh")] item_list_programming_language, item_list_dropdown_programming_language, item_samples = get_item_list(locale=item_list_locale[0], programming_language="", category="", difficulty="") item_list_category, item_list_dropdown_category, item_samples = get_item_list(locale=item_list_locale[0], programming_language=get_item_by_index(item_list_programming_language, 0, "python"), category="", difficulty="") item_list_difficulty, item_list_dropdown_difficulty, item_samples = get_item_list(locale=item_list_locale[0], programming_language=get_item_by_index(item_list_programming_language, 0, "python"), category=get_item_by_index(item_list_category, 0, "Basic Programming"), difficulty="") item_list_select, item_list_select_dropdown, item_samples = get_item_list(locale=item_list_locale[0], programming_language=get_item_by_index(item_list_programming_language, 0, "python"), category=get_item_by_index(item_list_category, 0, "Basic Programming"), difficulty=get_item_by_index(item_list_difficulty, 0)) def get_input_examples(): input_examples = [] for demo in DEMO_LIST: input_examples.append(['html', demo['title'], demo['description']]) return input_examples def code_syntax_check_fun(code, language='html'): check_result = check_code_error(code, language=language.lower()) if check_result['result'] == True: if check_result['error'] == True: return 'No pass' elif check_result['error'] == False: return 'Pass' else: return 'No check [no language in check_languages]' else: return check_result['error'] with gr.Blocks(css_paths="app.css", title="Qwen2.5 Coder Artifacts (+SambaNova+FullStackBench)") as demo: history_state = gr.State([]) setting = gr.State({ "system": SystemPrompt, }) with gr.Row(): code_language_example = gr.Textbox(show_label=False, lines=1, container=False, visible=False) input_text_title= gr.TextArea(show_label=False, container=False, visible=False) input_text_example = gr.TextArea(show_label=False, container=False, visible=False) with gr.Tab("Examples_HTML"): input_examples = get_input_examples() gr.Examples( examples=input_examples, inputs=[code_language_example, input_text_title, input_text_example], outputs=None, label='', examples_per_page=10, cache_examples=False, ) with gr.Tab("Examples_from_FullStackBench"): with gr.Row(): with gr.Column(): locale = gr.Dropdown( choices=item_list_dropdown_locale, label="locale", value=get_item_by_index(item_list_locale, 0), ) with gr.Column(): programming_language = gr.Dropdown( choices=item_list_dropdown_programming_language, label="programming_language", value=get_item_by_index(item_list_programming_language, 0, "python"), ) with gr.Column(): category = gr.Dropdown( choices=item_list_dropdown_category, label="category", value=get_item_by_index(item_list_category, 0, "Basic Programming"), ) with gr.Column(): difficulty = gr.Dropdown( choices=item_list_dropdown_difficulty, label="difficulty", value=get_item_by_index(item_list_difficulty, 0), ) with gr.Row(): test_examples = gr.Examples( examples=item_samples, inputs=[code_language_example, input_text_title, input_text_example], outputs=None, label='Examples:', examples_per_page=5, cache_examples=False, ) def locale_select(locale): item_list_programming_language, item_list_dropdown_programming_language, item_samples = get_item_list(locale=locale, programming_language="", category="", difficulty="", debugInfo="locale_1_") item_list_category, item_list_dropdown_category, item_samples = get_item_list(locale=locale, programming_language=get_item_by_index(item_list_programming_language, 0), category="", difficulty="", debugInfo="locale_2_") item_list_difficulty, item_list_dropdown_difficulty, item_samples = get_item_list(locale=locale, programming_language=get_item_by_index(item_list_programming_language, 0), category=get_item_by_index(item_list_category, 0), difficulty="", debugInfo="locale_3_") _, _, item_samples = get_item_list(locale=locale, programming_language=get_item_by_index(item_list_programming_language, 0), category=get_item_by_index(item_list_category, 0), difficulty=get_item_by_index(item_list_difficulty, 0), debugInfo="locale_4_") return (gr.Dropdown(choices=item_list_dropdown_programming_language, value=item_list_dropdown_programming_language[0][1]), gr.Dropdown(choices=item_list_dropdown_category, value=item_list_dropdown_category[0][1]), gr.Dropdown(choices=item_list_dropdown_difficulty, value=item_list_dropdown_difficulty[0][1]), gr.Dataset(samples=item_samples)) locale.select( fn=locale_select, inputs=[locale], outputs=[programming_language, category, difficulty, test_examples.dataset] ) def programming_language_select(locale, programming_language): item_list_category, item_list_dropdown_category, item_samples = get_item_list(locale=locale, programming_language=programming_language, category="", difficulty="", debugInfo="programming_language_1_") item_list_difficulty, item_list_dropdown_difficulty, item_samples = get_item_list(locale=locale, programming_language=programming_language, category=get_item_by_index(item_list_category, 0), difficulty="", debugInfo="programming_language_2_") _, _, item_samples = get_item_list(locale=locale, programming_language=programming_language, category=get_item_by_index(item_list_category, 0), difficulty=get_item_by_index(item_list_difficulty, 0), debugInfo="programming_language_3_") return (gr.Dropdown(choices=item_list_dropdown_category, value=item_list_dropdown_category[0][1]), gr.Dropdown(choices=item_list_dropdown_difficulty, value=item_list_dropdown_difficulty[0][1]), gr.Dataset(samples=item_samples)) programming_language.select( fn=programming_language_select, inputs=[locale, programming_language], outputs=[category, difficulty, test_examples.dataset] ) def category_select(locale, programming_language, category): item_list_difficulty, item_list_dropdown_difficulty, item_samples = get_item_list(locale=locale, programming_language=programming_language, category=category, difficulty="", debugInfo="category_1_") _, _, item_samples = get_item_list(locale=locale, programming_language=programming_language, category=category, difficulty=get_item_by_index(item_list_difficulty, 0), debugInfo="category_2_") return (gr.Dropdown(choices=item_list_dropdown_difficulty, value=item_list_dropdown_difficulty[0][1]), gr.Dataset(samples=item_samples)) category.select( fn=category_select, inputs=[locale, programming_language, category], outputs=[difficulty, test_examples.dataset] ) def difficulty_select(locale, programming_language, category, difficulty): _, _, item_samples = get_item_list(locale=locale, programming_language=programming_language, category=category, difficulty=difficulty, debugInfo="difficulty_1_") return gr.Dataset(samples=item_samples) difficulty.select( fn=difficulty_select, inputs=[locale, programming_language, category, difficulty], outputs=[test_examples.dataset] ) with ms.Application() as app: with antd.ConfigProvider(): with antd.Row(gutter=[32, 12]) as layout: with antd.Col(span=24, md=8): with antd.Flex(vertical=True, gap="middle", wrap=True): header = gr.HTML("""