1C_Forums / README.md
arefaste's picture
Update README.md
044d858 verified
metadata
license: mit
task_categories:
  - text-generation
  - table-question-answering
  - question-answering
language:
  - ru
  - en
tags:
  - 1с
  - ones
  - 1C
  - 1С
  - один-с
pretty_name: 1C_forums_dataset_by_arefaste
size_categories:
  - 1K<n<10K
configs:
  - config_name: default
    data_files:
      - split: train
        path: result_final_make_dataset.parquet

1C_forums: Dataset based parsed data from two of the most popular forums for 1c

This dataset made of the parsed data from two forums for coders at the 1C languages:

  • Infostart - all threads presents as like think section for model, and marked as the best result for queestion as final answer
  • Fastcode - Only templates

Dataset Overview

All rows prepared as useful columns. All text prepared as markdown text, and code 1c looks as like:

\`\`\`1c
 "ВЫБРАТЬ
        |    Ссылка.Контрагент,
        |    Ссылка.Магазин,
        |    СУММА(КоличествоСпрос) КАК КолвоСпрос
        |ИЗ
        |    Документ.СпросНаТовар.Товары
        |ГДЕ Ссылка.Дата МЕЖДУ ДОБАВИТЬКДАТЕ(&ДатаНачало, ГОД, -10) И &ДатаНачало
        | СГРУППИРОВАТЬ ПО Ссылка.Контрагент, СсылкаМагазин";.
\`\`\`

❗IMPORTANT❗: Escaping special characters

Values in columns prompt, think_process, and solution have escaped special characters like \n, \t, and \\!

The reason for this behavior: the Python's csv module that I'm using in the parser to save rows saves \n as an actual newline character, which makes reading the prepared file with Pandas impossible.

To fix it use simple lambda code:

return_back_special_symbols = lambda x: x.replace("\\n", "\n").replace("\\t", "\t").replace("\\\\", "\\")

Infostart

Choosen data at Infostart are all forums threads with found answers in section Dev.

  • source - for all rows for Infostart forum contains this value forum_infostart
  • in_source_id - id of the thread from that web forum. To get link to this page use this format: https://forum.infostart.ru/forum9/{in_source_id}/
  • prompt - it's a first question(msg) in thread.
  • think_process - its all thread discution presented as like think process that in there each msg between others separeted \n----\n
  • solution - msgs marked in thread as like solution for question
  • is_answer_a_link - if link contains a more then 60-70% of all solution, it's contains value True, else False
  • has_link - Counter links in solution. If solution doesn't contains any link, it value will be a Nan
  • tags - predicted tags

FastCode

All data at Fastcode are only templates, comments for this code made as new section # Примечание that what shows each comment.

  • source - for all rows for Fastcode forum contains this value fastcode_Templates
  • in_source_id - id of the thread from that web forum. To get link to this page use this format: https://fastcode.im/Templates/{in_source_id}/
  • prompt - it's title of the template.
  • think_process - None - it's dont use it at all!
  • solution - it's description(if it's contains there) and code with notes(it's comments, if its are there)
  • is_answer_a_link - if link contains a more then 60-70% of all solution, it's contains value True, else False
  • has_link - Counter links in solution. If solution doesn't contains any link, it value will be a Nan
  • tags - Tags at template separated by comma

Dataset Statistics

  • Total examples: 19041
  • forum infostart: 18299
  • fastcode Template: 742

Parsing scripts

You can find in folder scripts_parsing!

Citation

Please cite this dataset if you use it in your work:

@misc{arefaste2025,
  title={Arefaste: Parsed dataset for 1c from dataset},
  author={arefaste},
  year={2025},
  publisher={Hugging Face},
  howpublished={\url{https://huggingface.co/datasets/arefaste/1C_Forums}}
}

RU DOCS

1C_forums: Набор данных на основе распарсенной информации с двух популярных форумов по 1С

Этот набор данных состоит из распарсенной информации с двух форумов для программистов на языке 1С:

  • Infostart - все ветки представлены как раздел для размышлений модели, а сообщение, отмеченное как лучший ответ на вопрос, помечено как итоговое решение.
  • Fastcode - Только шаблоны кода.

Обзор набора данных

Все строки подготовлены в виде полезных столбцов. Весь текст представлен в формате markdown, а код 1C выглядит следующим образом:

\`\`\`1c
 "ВЫБРАТЬ
        |    Ссылка.Контрагент,
        |    Ссылка.Магазин,
        |    СУММА(КоличествоСпрос) КАК КолвоСпрос
        |ИЗ
        |    Документ.СпросНаТовар.Товары
        |ГДЕ Ссылка.Дата МЕЖДУ ДОБАВИТЬКДАТЕ(&ДатаНачало, ГОД, -10) И &ДатаНачало
        | СГРУППИРОВАТЬ ПО Ссылка.Контрагент, СсылкаМагазин";.
\`\`\`

❗ВАЖНО❗: Экранирование специальных символов

Значения в столбцах prompt, think_process и solution содержат экранированные специальные символы, такие как \n, \t и \\!

Причина такого поведения: модуль csv в Python, который используется в парсере для сохранения строк, сохраняет \n как символ переноса строки, что делает невозможным чтение подготовленного файла с помощью Pandas.

Чтобы это исправить, используйте простой лямбда-код:

return_back_special_symbols = lambda x: x.replace("\\n", "\n").replace("\\t", "\t").replace("\\\\", "\\")

Infostart

Выбранные данные с Infostart - это все ветки форума с найденными ответами в разделе для разработчиков (Dev).

  • source - для всех строк с форума Infostart содержит значение forum_infostart
  • in_source_id - идентификатор ветки с этого веб-форума. Чтобы получить ссылку на страницу, используйте формат: https://forum.infostart.ru/forum9/{in_source_id}/
  • prompt - это первый вопрос (сообщение) в ветке.
  • think_process - это все обсуждение в ветке, представленное как процесс размышления, где каждое сообщение отделено от других с помощью \n----\n
  • solution - сообщения в ветке, отмеченные как решение вопроса
  • is_answer_a_link - если ссылка составляет более 60-70% всего решения, содержит значение True, иначе False
  • has_link - счетчик ссылок в решении. Если решение не содержит ссылок, значение будет NaN
  • tags - предсказанные теги

FastCode

Все данные с Fastcode - это только шаблоны кода. Комментарии к этому коду сделаны в новом разделе # Примечание, который показывает каждый комментарий.

  • source - для всех строк с форума Fastcode содержит значение fastcode_Templates
  • in_source_id - идентификатор шаблона с этого веб-форума. Чтобы получить ссылку на страницу, используйте формат: https://fastcode.im/Templates/{in_source_id}/
  • prompt - это заголовок шаблона.
  • think_process - None - не используется вообще!
  • solution - это описание (если есть) и код с примечаниями (комментариями, если они есть)
  • is_answer_a_link - если ссылка составляет более 60-70% всего решения, содержит значение True, иначе False
  • has_link - счетчик ссылок в решении. Если решение не содержит ссылок, значение будет NaN
  • tags - Теги шаблона, разделенные запятой

Статистика набора данных

  • Общее количество примеров: 19041
    • форум infostart: 18299
    • шаблонов fastcode: 742

Скрипты для парсинга

Их можно найти в папке scripts_parsing!

Цитирование

Пожалуйста, процитируйте этот набор данных, если используете его в своей работе:

@misc{arefaste2025,
  title={Arefaste: Parsed dataset for 1c from dataset},
  author={arefaste},
  year={2025},
  publisher={Hugging Face},
  howpublished={\url{https://huggingface.co/datasets/arefaste/1C_Forums}}
}