src/llm/llm_tools.py

"""
TODO: add a boolean to switch llms
"""


import json
import string
import openai

import wikipedia
from langchain.text_splitter import CharacterTextSplitter
from langchain.prompts import PromptTemplate
from langchain.chains import LLMChain
from src.domain.block import Block
from src.llm.llms import openai_llm
from src.tools.wiki import Wiki


# async def get_wikilist_open_source(task: {}) -> str:
#     """
#     get the titles of wiki pages interesting for solving the given task
#     """

#     template = ("<s>[INST] Your task consists in finding the list of wikipedia page titles which provide useful content "
#                 "    for a paragraph whose description is delimited by triple backticks.\n"
#                 "    Make sure that you provide no more than 10 elements and that the list is actually finished."
#                 "    Format your response as a valid JSON list of strings separated by commas.[/INST]</s>"
#                 "    Description: ```{description}```")

#     prompt = PromptTemplate(template=template, input_variables=['description'])
#     llm_chain = LLMChain(llm=opensource_llm, prompt=prompt)
#     response = llm_chain.run({'description': task['description']})
#     llm_list = response.choices[0].message.content
#     try:
#         wikilist = json.loads(llm_list)
#     except:
#         print("json loads failed with" + llm_list)
#         wikilist = list(llm_list.split(','))

#     expanded_wikilist = []

#     expand_factor = 2

#     for wikipage in wikilist:
#         expanded_wikilist += wikipedia.search(wikipage, expand_factor)

#     wikilist = list(set(expanded_wikilist))

#     return wikilist



async def get_wikilist(task: {}) -> str:
    """
    get the titles of wiki pages interesting for solving the given task
    """

    llm = openai_llm
    template = (f"\n"
                f"    Your task consists in finding the list of wikipedia page titles which provide useful content "
                f"    for a paragraph whose description is delimited by triple backticks: ```{task['description']}```\n"
                f"    "
                f"    Make sure that you provide no more than 10 elements and that the list is actually finished."
                f"    Format your response as a valid JSON list of strings separated by commas.\n"
                f"  \n"
                f"    ")

    #wikilist = LLMChain(llm=openai_llm, prompt=prompt).run()
    llm_list = llm.invoke(template)
    try:
        wikilist = json.loads(llm_list)
    except:
        print("json loads failed with" + llm_list)
        wikilist = list(llm_list.split(','))

    expanded_wikilist = []

    expand_factor = 2

    for wikipage in wikilist:
        expanded_wikilist += wikipedia.search(wikipage, expand_factor)

    wikilist = list(set(expanded_wikilist))

    return wikilist


def extract_list(llm_list: str):

    def filter_(el: str):
        resp = 2 < len(el)
        usable_length = len([c for c in el if c in string.ascii_letters])
        resp = resp and len(el)*3/4 < usable_length
        return resp

    try:
        wikilist = llm_list[1:-1].split('"')
        wikilist = [el for el in wikilist if filter_(el)]
        print(wikilist)
    except:
        wikilist = []
        print('issues with the wikilist')
    return wikilist


# def get_public_paragraph_open_source(task: {}) -> str:
#     """returns the task directly performed by chat GPT"""

#     template = ("<s>[INST] Your task consists in generating a paragraph whose description is delimited by triple "
#                 "backticks.\n"
#                 "    The paragraph belongs at the top level of the hierarchy to a document"
#                 "    whose doc_description is delimited by triple backticks.\n"
#                 "    Make sure that the paragraph relates the top level of the document\n"
#                 "    The paragraph belongs to a higher paragraph in the hierarchy whose description (above) is delimited by "
#                 "    triple backticks."
#                 "    Make sure that the paragraph relates with the paragraph in the hierarchy of the document\n"
#                 "    The paragraphs comes after previous paragraphs whose description (before) is delimited by triple "
#                 "    backticks.\n"
#                 "    Make sure that the paragraph relates with previous paragraph without any repetition\n"
#                 "    The paragraphs comes before next paragraphs whose description (after) is delimited by triple backticks.\n"
#                 "    Make sure that the paragraph prepares the transition to the next paragraph without any "
#                 "    repetition. [/INST]</s>"
#                 "    Description: ```{description}```"
#                 "    Doc description: ```{doc_description}```"
#                 "    Above: ```{above}```"
#                 "    Before: ```{before}```"
#                 "    After: ```{after}```"
#                 )
    
#     prompt = PromptTemplate(template=template, input_variables=['description', 'doc_description', 'above', 'before', 'after'])
#     llm_chain = LLMChain(llm=opensource_llm, prompt=prompt)
#     response = llm_chain.run({'description': task['description'], 'doc_description': task['doc_description'],
#                               'above': task['above'], 'before': task['before'], 'after': task['after']})
#     p = response.choices[0].message.content
#     return p

def get_public_paragraph(task: {}) -> str:
    """returns the task directly performed by chat GPT"""
    print(task)
    llm = openai_llm
    template = (f"\n"
                f"    Your task consists in generating a paragraph\\n"
                f"    whose description is delimited by triple backticks: ```{task['description']}```\n"
                f"\n"
                f"    The paragraph belongs at the top level of the hierarchy to a document \\n"
                f"    whose description is delimited by triple backticks: ``` {task['doc_description']}```\n"
                f"    Make sure that the paragraph relates the top level of the document\n"
                f"    \n"
                f"    The paragraph belongs to a higher paragraph in the hierarchy \\n"
                f"    whose description is delimited by triple backticks: ``` {task['above']}```\n"
                f"    Make sure that the paragraph relates with the paragraph in the hierarchy of the document\n"
                f"        \n"
                f"    The paragraphs comes after previous paragraphs \\n"
                f"    whose description is delimited by triple backticks: ``` {task['before']}```\n"
                f"    Make sure that the paragraph relates with previous paragraph without any repetition\n"
                f"    \n"
                f"    The paragraphs comes before next paragraphs \\n"
                f"    whose description is delimited by triple backticks: ``` {task['after']}```\n"
                f"    Make sure that the paragraph prepares the transition to the next paragraph without any repetition\n"
                f"    \n"
                f"  \n"
                f"\n"
                f"    ")

    p = llm.invoke(template)

    return p


def create_index(wikilist: [str]):
    """
    useful for creating the index of wikipages
    """
    fetch = Wiki().fetch

    pages = [(title, fetch(title)) for title in wikilist if type(fetch(title)) != str]
    texts = []
    chunk = 800
    for title, page in pages:
        texts.append(WikiPage(title=title, fulltext=page.page_content))

    doc_splitter = CharacterTextSplitter(
        separator=".",
        chunk_size=chunk,
        chunk_overlap=100,
        length_function=len,
    )

    paragraphs = texts[0].get_paragraphs(chunk=800)

    split_texts = []
    for p in paragraphs:
        split_texts += doc_splitter.split_text(p)

    for split_text in split_texts:
        assert type(split_text) == str
        assert 0 < len(split_text) < 2 * 500

    wiki_index = Chroma.from_texts(split_texts)

    return wiki_index


def get_wiki_paragraph(wiki_index, task: {}) -> str:
    """useful to get a summary in one line from wiki index"""

    task_description = get_public_paragraph(task)
    wiki_paragraphs = semantic_search(wiki_index, task_description)
    text_content = ""
    for p in wiki_paragraphs:
        text_content += p.page_content + "/n/n"

    template = (f"\n"
                f"    Your task consists in generating a paragraph\\n"
                f"    whose description is delimited by triple backticks: ```{task['description']}```\n"
                f"\n"
                f"    The text generation is based in the documents provided in these sections \n"
                f"    delimited by by triple backticks: ``` {text_content}``` \n"
                f"    The paragraph belongs at the top level of the hierarchy to a document \\n"
                f"    whose description is delimited by triple backticks: ``` {task['doc_description']}```\n"
                f"    Make sure that the paragraph relates the top level of the document\n"
                f"    \n"
                f"    The paragraph belongs to a higher paragraph in the hierarchy \\n"
                f"    whose description is delimited by triple backticks: ``` {task['above']}```\n"
                f"    Make sure that the paragraph relates with the paragraph in the hierarchy of the document\n"
                f"        \n"
                f"    The paragraphs comes after previous paragraphs \\n"
                f"    whose description is delimited by triple backticks: ``` {task['before']}```\n"
                f"    Make sure that the paragraph relates with previous paragraph without any repetition\n"
                f"    \n"
                f"    The paragraphs comes before next paragraphs \\n"
                f"    whose description is delimited by triple backticks: ``` {task['after']}```\n"
                f"    Make sure that the paragraph prepares the transition to the next paragraph without any repetition\n"
                f"    \n"
                f"  \n"
                f"\n"
                f"    ")

    llm = openai_llm
    p = llm(template)

    return p


# def get_private_paragraph_open_source(texts, task: {}) -> str:
#     """useful to get a summary in one line from wiki index"""

#     text_content = ""
#     for t in texts:
#         text_content += t + "/n/n"

#     template = ("\n"
#                 "    Your task consists in generating a paragraph"
#                 "    whose description is delimited by triple backticks\n"
#                 "    The text generation is based in the documents provided in these sections \n"
#                 "    delimited by by triple backticks (text_content)\n"
#                 "    The paragraph belongs at the top level of the hierarchy to a document"
#                 "    whose description is delimited by triple backticks (doc_decription)\n"
#                 "    Make sure that the paragraph relates the top level of the document\n"
#                 "    \n"
#                 "    The paragraph belongs to a higher paragraph in the hierarchy"
#                 "    whose description is delimited by triple backticks (above)\n"
#                 "    Make sure that the paragraph relates with the paragraph in the hierarchy of the document\n"
#                 "        \n"
#                 "    The paragraphs comes after previous paragraphs"
#                 "    whose description is delimited by triple backticks (before)\n"
#                 "    Make sure that the paragraph relates with previous paragraph without any repetition\n"
#                 "    \n"
#                 "    The paragraphs comes before next paragraphs"
#                 "    whose description is delimited by triple backticks (after)\n"
#                 "    Make sure that the paragraph prepares the transition to the next paragraph without any repetition\n"
#                 "    description: ```{description}```"
#                 "    text_content: ```{text_content}```"
#                 "    doc_description: ```{doc_description}```"
#                 "    above: ```{above}```"
#                 "    before: ```{before}```"
#                 "    after: ```{after}```")

#     prompt = PromptTemplate(template=template, input_variables=['description', 'text_content', 'doc_description', 'above', 'before', 'after'])
#     llm_chain = LLMChain(llm=opensource_llm, prompt=prompt)
#     response = llm_chain.run({'description': task['description'], 'text_content': text_content, 'doc_description': task['doc_description'],
#                                 'above': task['above'], 'before': task['before'], 'after': task['after']})
#     p = response.choices[0].message.content


def get_private_paragraph(texts, task: {}) -> str:
    """useful to get a summary in one line from wiki index"""

    text_content = ""
    for t in texts:
        text_content += t + "/n/n"

    template = (f"\n"
                f"    Your task consists in generating a paragraph\\n"
                f"    whose description is delimited by triple backticks: ```{task['description']}```\n"
                f"\n"
                f"    The text generation is based in the documents provided in these sections \n"
                f"    delimited by by triple backticks: ``` {text_content}``` \n"
                f"    The paragraph belongs at the top level of the hierarchy to a document \\n"
                f"    whose description is delimited by triple backticks: ``` {task['doc_description']}```\n"
                f"    Make sure that the paragraph relates the top level of the document\n"
                f"    \n"
                f"    The paragraph belongs to a higher paragraph in the hierarchy \\n"
                f"    whose description is delimited by triple backticks: ``` {task['above']}```\n"
                f"    Make sure that the paragraph relates with the paragraph in the hierarchy of the document\n"
                f"        \n"
                f"    The paragraphs comes after previous paragraphs \\n"
                f"    whose description is delimited by triple backticks: ``` {task['before']}```\n"
                f"    Make sure that the paragraph relates with previous paragraph without any repetition\n"
                f"    \n"
                f"    The paragraphs comes before next paragraphs \\n"
                f"    whose description is delimited by triple backticks: ``` {task['after']}```\n"
                f"    Make sure that the paragraph prepares the transition to the next paragraph without any repetition\n"
                f"    \n"
                f"  \n"
                f"\n"
                f"    ")

    llm = openai_llm
    p = llm.invoke(template)

    return p

def summarize_paragraph_v2(prompt : str, title_doc : str = '', title_para : str = ''):
    max_tokens = 850
    location_of_the_paragraph = prompt.split(" :")[0]
    """summarizes the paragraph"""
    task = (f"Your task consists in summarizing in English the paragraph of the document untitled ```{title_doc}``` located in the ```{location_of_the_paragraph}``` section of the document."
                f"The paragraph title is ```{title_para}```."
                f"Your response shall be concise and shall respect the following format:"
                f"<summary>"
                f"If you see that the summary that you are creating will not respect ```{max_tokens}``` tokens, find a way to make it shorter.")
    generation = openai.chat.completions.create(model="gpt-3.5-turbo-16k", messages=[{"role":"system","content":task},{"role":"user","content":prompt}])
    res = generation.choices[0].message.content
    print("****************")
    print(res)
    print("----")
    return str(res).strip()

def generate_response_to_exigence(exigence : str, titre_exigence : str, content : str):
    """
    Generates a response to an exigence depending on the context of the exigence and the blocks of the document.
    """
    task = (f"Your task consists in generating a response to a requirement in a tender for Orange, a telecommunication operator."
            f"The requirement dealing with {titre_exigence} is expressed below between triple backquotes:"
            f"```{exigence}```"
            f"Your answer should be precise, consistent and as concise as possible with no politeness formulas and strictly be based on the following text delimited by triple backquotes : ```{content}```"
            )
    llm = openai_llm
    generation = llm.invoke(task)
    return generation