"Create a ppt with the given information" import asyncio import json import math import os import time import pptx from dotenv import load_dotenv from langchain.prompts import ChatPromptTemplate from langchain_core.output_parsers import JsonOutputParser from langchain_openai import ChatOpenAI from PIL import ImageFont from pptx.enum.shapes import MSO_SHAPE # pylint: disable=no-name-in-module from pptx.util import Inches, Pt from pydantic import BaseModel from configs.config import OPENAI_MODEL_35 # load services from services.company_profile.company_profile import (CompanyProfileMaker, find_llm_response) # load in data models from services.company_profile.data_classes.company_info import CompanyInfo from services.company_profile.data_classes.llm_results import LLMResults from services.ppt_generator.chart_creator import ChartCreator from services.ppt_generator.data_classes.company_overview import \ CompanyOverview # load database functions from services.ppt_generator.layouts.xcm.layouts import (DefaultSettings, SlideWorkArea) # load slide layouts from services.ppt_generator.slide_layouts_choices import ( slide_layouts_choices, slide_layouts_choices_keys) # load other utilities from utils.download_image import DownloadImage from utils.search_google import GoogleSearch from utils.url_parser import parsed_url # Load the environment variables load_dotenv() GOOGLE_SEARCH_API = os.environ["GOOGLE_SEARCH_API"] PRES_TEMPATE = "ppt_templates/ppt_template.pptx" class NewArticles(BaseModel): "New articles" articles: list[str] articles_summary: list[str] articles_links: list[str] class CompanyProfileSlide: "Create a slide for the company profile" def __init__( self, company_url: str, ppt_object: pptx.Presentation = None, work_area: SlideWorkArea = SlideWorkArea(), default_settings: DefaultSettings = DefaultSettings(), comparable_tickers: list = None, new_deck: bool = False, ): """ Create a ppt with the given information ppt_object: the ppt slide object company_url: the company url work_area: the work area for the slide default_settings: the default settings for the slide comparable_tickers: the comparable tickers for the company """ self.prs = ppt_object if self.prs is None and new_deck: self.prs = pptx.Presentation(PRES_TEMPATE) self.work_area = work_area self.comparable_tickers = comparable_tickers if self.comparable_tickers is None: self.comparable_tickers = ["^DJI", "^SPX", "^IXIC"] self.default_font = default_settings.font self.default_font_size = default_settings.font_size self.default_font_color = default_settings.font_color # parse the URL self.parsed_url = parsed_url(company_url) self.company_url = self.parsed_url.url # get items from dynamodb self.get_data() self.company_overview = CompanyOverview( company_name=self.company_info.company_name, logo=self.company_info.logo, stock_ticker=self.company_info.stock_ticker, customers=find_llm_response( self.llm_results.model_dump()["LLM_results"], "Customers" ), products=find_llm_response( self.llm_results.model_dump()["LLM_results"], "Products" ), management=self.llm_results.leadership["leadership"], hq=find_llm_response( self.llm_results.model_dump()["LLM_results"], "HQ_Founded" ), founded=find_llm_response( self.llm_results.model_dump()["LLM_results"], "HQ_Founded" ), overview=find_llm_response( self.llm_results.model_dump()["LLM_results"], "Overview" ) + "\n" + find_llm_response( self.llm_results.model_dump()["LLM_results"], "Overview_Stats" ), ) # self.relevant_news = self.get_company_news() self.relevant_news = "" def get_data(self): "Get the data from the database" self.company_info = CompanyInfo.get_company_info(self.company_url) self.llm_results = LLMResults.get_LLM_results(self.company_url) ### if there is no logo, pull in the logo if self.company_info.logo is None: self.company_overview = CompanyProfileMaker(self.company_url) self.company_info = asyncio.run( self.company_overview.load_brand( supplied_company_info=self.company_info ) ) if self.company_info.logo: setattr( self.company_info.logo, "img_loc", asyncio.run( self.download_image( self.company_info.logo.logo_url, "logo", self.company_info.logo.logo_dark, ) ), ) ### if there is no leadership, get leadership if self.llm_results.leadership is None: self.company_overview = CompanyProfileMaker(self.company_url) company_leadership_text = find_llm_response( self.llm_results.model_dump()["LLM_results"], "Leadership" ) self.llm_results.leadership = asyncio.run( self.company_overview.map_leadership(company_leadership_text) ) asyncio.run(self.search_management_pictures()) async def download_image( self, image_url: str, file_name: str, dark_background=True ): "Download the logo from the company url" di = DownloadImage() if image_url == "": image_url = di.STOCK_IMAGE_FILENAME _, image_loc = await di.get_image( url=image_url, file_name=f"{self.company_info.company_name}_{file_name}.png", picture_background_dark=dark_background, ) return image_loc async def search_management_pictures(self): "Search for the image of the management person on google" search_tasks = [] for leader in self.llm_results.leadership["leadership"]: google_search = GoogleSearch() search_tasks.append( asyncio.create_task( google_search.search_google( f'"{self.company_info.company_name}" "{leader.name}" headshot square', "images", domain_check=True, domain_name_to_check="linkedin.com", num_of_results=100, ) ) ) results_of_results = await asyncio.gather(*search_tasks) image_results = [] for results in results_of_results: items_to_remove = [] for result in results: if ( "https://static.licdn.com/aero-v1/sc/h/1c5u578iilxfi4m4dvc4q810q" in result["imageUrl"] ): items_to_remove.append(result) for item in items_to_remove: results.remove(item) for i, leader in enumerate(self.llm_results.leadership["leadership"]): # find the image closest to 1 aspect ratio if len(results_of_results[i]) == 0: image_url = "" else: image_url = results_of_results[i][0]["imageUrl"] for result in results_of_results[i]: image_aspect_ratio = result["imageWidth"] / result["imageHeight"] if 0.9 < image_aspect_ratio < 1.2: image_url = result["imageUrl"] break task = asyncio.create_task( self.download_image(image_url, leader.name, False) ) image_results.append(task) image_results = await asyncio.gather(*image_results) for i, leader in enumerate(self.llm_results.leadership["leadership"]): setattr( leader, "img_loc", image_results[i], ) return self.llm_results.leadership @staticmethod def convert_llm_to_ci(llm_response_category: str): "Convert the llm response category to the company info category" llm_to_ci_mapping = { "Products": "products", "Leadership": "management", "Customers": "customers", "Overview_Stats": "overview", "Overview": "overview", "HQ_Founded": "hq", } if llm_response_category not in llm_to_ci_mapping: return None return llm_to_ci_mapping[llm_response_category] def get_slide_layout(self, layout_name): """Get the slide layout by name input: - layout_name: the name of the layout to get output: - layout: the layout - slide_structure: the structure of the slide """ for layout in self.prs.slide_layouts: if layout.name == layout_name: if layout_name in slide_layouts_choices: slide_structure = slide_layouts_choices[layout_name] else: slide_structure = None return layout, slide_structure return None, None def create_slides(self, slide_to_create: slide_layouts_choices_keys): "Create the slides for the ppt" # for slide_to_create in self.slides_to_create: if slide_to_create not in slide_layouts_choices: raise ValueError(f"{slide_to_create} not in slide_layouts_choices") layout, slide_structure = self.get_slide_layout(slide_to_create) try: slide = self.prs.slides.add_slide(layout) if slide_to_create == "private_company_profile": self.create_private_company_profile(slide, slide_structure) elif slide_to_create == "stock_chart_output": # chart_class = ChartCreator() self.add_stock_chart_slide(slide, slide_structure, years_to_show=5) elif slide_to_create == "extra_content": self.extra_content_slide( slide_to_create, self.company_overview.company_name, self.company_overview.logo, slide_structure["header"], slide_structure["content"], ) except Exception as e: # pylint: disable=broad-except print(e) return None return slide def create_private_company_profile(self, slide, slide_structure): "Create the slide" for place_holder in slide.placeholders: item_name = place_holder.name if item_name in slide_structure: # if the placeholder is a content type, then replace the text with the data from the company info if slide_structure[item_name]["field"] == "content": if slide_structure[item_name]["data"] == "news": continue # TODO implement news # if self.relevant_news is None or self.relevant_news == "": # continue # self.write_content( # place_holder, # '\n'.join([f"{title}: {self.relevant_news.articles_summary[i]}" for i, title in enumerate(self.relevant_news.articles)]), # slide_structure[item_name], # ) # slide.notes_slide.notes_text_frame.text = "News Links:\n" + '\n\n'.join( # [ # f"{self.relevant_news.articles[i]}\n{link}" # for i, link in enumerate(self.relevant_news.articles_links) # ] # ) if self.relevant_news is None or self.relevant_news == "": continue self.write_content( place_holder, "\n".join( [ f"{title}: {self.relevant_news.articles_summary[i]}" for i, title in enumerate( self.relevant_news.articles ) ] ), slide_structure[item_name], ) slide.notes_slide.notes_text_frame.text = ( "News Links:\n" + "\n\n".join( [ f"{self.relevant_news.articles[i]}\n{link}" for i, link in enumerate( self.relevant_news.articles_links ) ] ) ) else: self.write_content( place_holder, getattr( self.company_overview, slide_structure[item_name]["data"], ), slide_structure[item_name], ) # if the placeholder is a header type, then replace the text with the data from the structure elif slide_structure[item_name]["field"] == "header": place_holder.text = slide_structure[item_name]["data"] # if the placeholder is a logo_image type, then replace the image with the data from the company info elif slide_structure[item_name]["field"] == "logo_image": self.add_logo(place_holder, self.company_overview.logo, slide) elif slide_structure[item_name]["field"] == "leader_picture": if ( len(self.company_overview.management) > slide_structure[item_name]["data"] ): leader = self.company_overview.management[ slide_structure[item_name]["data"] ] if leader.img_loc is not None: place_holder.insert_picture(leader.img_loc) elif slide_structure[item_name]["field"] == "leader_name": if ( len(self.company_overview.management) > slide_structure[item_name]["data"] ): leader = self.company_overview.management[ slide_structure[item_name]["data"] ] # add the leader name and title to the placeholder # make the leadername bold run = place_holder.text_frame.paragraphs[0].add_run() run.text = leader.name run.font.bold = True # add a new line and add the title run = place_holder.text_frame.paragraphs[0].add_run() run.text = f"\n{leader.title}" if len(self.company_overview.management) > 4: self.add_leadership( "extra_content", self.company_overview.company_name, self.company_overview.logo, "Leadership", self.company_overview.management, ) def add_stock_chart_slide(self, slide, slide_structure, years_to_show=5): "add the stock chart slides" for place_holder in slide.placeholders: item_name = place_holder.name if item_name in slide_structure: # if the placeholder is a content type, then replace the text with the data from the company info if slide_structure[item_name]["field"] == "content": self.write_content( place_holder, getattr( self.company_overview, slide_structure[item_name]["data"] ), slide_structure[item_name], ) # if the placeholder is a header type, then replace the text with the data from the structure elif slide_structure[item_name]["field"] == "header": place_holder.text = slide_structure[item_name]["data"] # if the placeholder is a logo_image type, then replace the image with the data from the company info elif slide_structure[item_name]["field"] == "logo_image": self.add_logo(place_holder, self.company_overview.logo, slide) elif slide_structure[item_name]["field"] == "stock_chart": # get the data chart_creator = ChartCreator() stock_data_df = chart_creator.get_stock_data( self.company_info.stock_ticker, years_to_show, other_tickers=self.comparable_tickers, ) chart_creator.add_line_chart( slide, place_holder.top, place_holder.left, place_holder.width, place_holder.height, stock_data_df, x_axis="date", ) # delete the placeholder place_holder.element.getparent().remove(place_holder.element) def add_logo(self, place_holder, logo_item, slide_obj): "Adds the logo in place of the placeholder and resizes it to fit the placeholder" if logo_item is None: # add text for no logo found place_holder.text = "No Logo Found" return new_pic = slide_obj.shapes.add_picture( logo_item.img_loc, top=place_holder.top, left=place_holder.left, height=place_holder.height, ) place_holder.element.getparent().remove(place_holder.element) if new_pic.width > place_holder.width: rescale = place_holder.width / new_pic.width new_pic.width = int(new_pic.width * rescale) new_pic.height = int(new_pic.height * rescale) def write_content(self, place_holder, content, content_attributes): "Write the content to the placeholder" content = content.replace("- ", "") text_to_write, text_over_height = self.estimate_text_size( content, max_width_px=( (place_holder.width - Pt(0.25) - Pt(0.2)) / Pt(1) ), # subtract the bullet spacing and the padding max_height_px=(place_holder.height / Pt(1)) * content_attributes["height_multiplier"], font_size=content_attributes["font_size"], font=content_attributes["font_type"], ) place_holder.text = text_to_write if text_over_height: self.extra_content_slide( "extra_content", self.company_overview.company_name, self.company_overview.logo, content_attributes["data"], content, ) def extra_content_slide(self, slide_name, company_name, logo, header, content=""): "Create an extra content slide" layout, slide_structure = self.get_slide_layout(slide_name) slide = self.prs.slides.add_slide(layout) slide.name = f"{company_name}_{header}_extra_content" header = header.capitalize() for place_holder in slide.placeholders: item_name = place_holder.name if slide_structure.get(item_name, None) is not None: if slide_structure[item_name]["field"] == "content": place_holder.text = locals()[slide_structure[item_name]["data"]] elif slide_structure[item_name]["field"] == "image": if logo is not None: self.add_logo(place_holder, logo, slide) return slide def add_leadership(self, slide_name, company_name, logo, header, leadership: list): "Add the leadership to the slide" slide = self.extra_content_slide(slide_name, company_name, logo, header) for place_holder in slide.placeholders: if place_holder.name == "Content Placeholder 1": content_left = place_holder.left content_top = place_holder.top for leader in leadership: try: # picture new_pic = slide.shapes.add_picture( leader.img_loc, # auto_shape_type = MSO_SHAPE.OVAL, top=content_top, left=content_left, height=Inches(0.56), width=Inches(0.56), ) new_pic.auto_shape_type = MSO_SHAPE.OVAL # add a text box to the right of the picture left = content_left + new_pic.width + Inches(0.1) top = content_top width = Inches(2.36) height = Inches(0.56) txBox = slide.shapes.add_textbox(left, top, width, height) txBox.text_frame.word_wrap = True run = txBox.text_frame.paragraphs[0].add_run() run.text = leader.name run.font.size = Pt(12) run.font.bold = True # add a new line and add the title run = txBox.text_frame.paragraphs[0].add_run() run.text = f"\n{leader.title}" run.font.size = Pt(12) # move the content top down if content top is more 5 inches from the top, then move the content left to the left if content_top > Inches(5): content_left += Inches(3.5) content_top = place_holder.top else: content_top += Inches(0.56) + Inches(0.25) except Exception as e: # pylint: disable=broad-except print(e) continue place_holder.element.getparent().remove(place_holder.element) return slide def estimate_text_size( self, text: str, max_width_px: int, max_height_px: int, font=None, font_size=None, ): """ estimate the size of a text body input: - text: the text to estimate - max_width_px: the max width of the text in pixels - max_height_px: the max height of the text in pixels - font: the font to use - font_size: the font size to use output: - text_to_draw: the text to draw - text_over_height: if the text is over the height """ if font is None: font = self.default_font if font_size is None: font_size = self.default_font_size # get the font_type fnt = ImageFont.truetype(font, font_size) # split the text into lines text_lines = text.split("\n") # loop through the text lines and see what the max width and height is text_to_draw = [] text_over_height = False for line in text_lines: if line == "" or line is None: continue # estimate the line height line_height = ( math.ceil(fnt.getmask(line).getbbox()[2] / max_width_px) * font_size * 1.5 ) if max_height_px > line_height: text_to_draw.append(line) max_height_px -= line_height if max_height_px < 0: text_over_height = True return "\n".join(text_to_draw), text_over_height def get_company_news(self): "Get the company news" search_google = GoogleSearch() results = asyncio.run( search_google.search_google(f"{self.company_info.company_name}", "news") ) if len(results) == 0: return None # run through an LLM model to filter the news return self.filter_news(results) def filter_news(self, news: list): "Filter the news" # run through an LLM model to filter the news llm = ChatOpenAI( model=OPENAI_MODEL_35, api_key=os.environ["OPENAI_API_KEY"], temperature=0.2, ) prompt = ChatPromptTemplate.from_messages( [ ( "system", """ You are going to be given news articles with the title and snippet. You need to determine if the news is relevant to your client. You need to return the articles that are relevant, with a summary and the link to the article. Make sure the article is relevant to the company and industry you are working with. """, ), ( "human", "The company you are working with is {company_name} in the {industry} industry with an overview of this {overview}", ), ("human", "{articles}"), ( "human", "Output format should be: {json_structure} and give me atleast 5 articles if not more", ), ] ) parser = JsonOutputParser(pydantic_object=NewArticles) chain = prompt | llm | parser output = chain.invoke( { "company_name": self.company_info.company_name, "industry": self.company_info.industry, "overview": self.company_overview.overview, "articles": json.dumps(news), "json_structure": parser.get_format_instructions(), } ) try: output = NewArticles(**output) except Exception: # pylint: disable=broad-exceptfas time.sleep(5) output = chain.invoke( { "company_name": self.company_info.company_name, "industry": ( self.company_info.industry if self.company_info.industry else "" ), "overview": self.company_overview.overview, "articles": json.dumps( [ { "title": article["title"], "snippet": article["snippet"], "link": article["link"], } for article in news ] ), "json_structure": parser.get_format_instructions(), } ) output = NewArticles(**output) return output if __name__ == "__main__": urls = [ "unity.com", ] for url in urls: # prs_object = pptx.Presentation("ppt_templates/ppt_template.pptx") # new_slide = prs_object.slides.add_slide(prs_object.slide_layouts[1]) pptg = CompanyProfileSlide( company_url=url, new_deck=True, ) pptg.create_slides("private_company_profile") if pptg.company_info.stock_ticker is not None: pptg.create_slides("stock_chart_output") pptg.prs.save(f"{pptg.company_info.company_name}_company_profile.pptx") print(f"{pptg.company_info.company_name} PPT created")