Skip to content

Add configurable prompts and stock sentiment analysis tab #1

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Timon0512 wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Add configurable prompts and stock sentiment analysis tab #1

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
142 changes: 113 additions & 29 deletions main.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,6 +9,35 @@

API_KEY = st.secrets["API_KEY"]
json_path = "video_dict.json"

DEFAULT_SUMMARY_PROMPT = """
You are an expert at summarizing YouTube video transcripts.
Your goal is to provide a clear and concise summary that captures the essence of the video.
The summary must be written in {language}.
Please format your response in Markdown.

The summary should have the following structure:
1. A short, catchy title for the summary. Use a heading level 2 (##).
2. A one-paragraph overview of the video's main topic and conclusion.
3. A bulleted list of the 3-5 most important key takeaways or points discussed. Use a heading level 3 (###) for "Key Takeaways".

Here is the transcript:
---
{transcript}
---
""".strip()

DEFAULT_STOCK_PROMPT = """
You are a financial analyst. Identify all stocks or companies that are explicitly discussed in the transcript below.
For each mentioned stock, determine whether the overall sentiment is Positive, Negative, or Neutral.
Respond in {language} using Markdown with a table that has the columns: Unternehmen/Aktie, Sentiment, Begründung.
If no stocks are discussed, explicitly state that no stocks were mentioned.

Transcript:
---
{transcript}
---
""".strip()
def load_video_dict(path):
if os.path.exists(path):
# Datei laden
Expand Down Expand Up @@ -86,25 +115,7 @@ def get_yt_transcript(video_id: str,
full_transcript = " ".join(str_list)
return full_transcript

def summarize_transcript_stream(transcript: str, api_key: str, language: str = "DE"):

prompt = f"""
You are an expert at summarizing YouTube video transcripts.
Your goal is to provide a clear and concise summary that captures the essence of the video.
The summary must be written in {language}.
Please format your response in Markdown.

The summary should have the following structure:
1. A short, catchy title for the summary. Use a heading level 2 (##).
2. A one-paragraph overview of the video's main topic and conclusion.
3. A bulleted list of the 3-5 most important key takeaways or points discussed. Use a heading level 3 (###) for "Key Takeaways".

Here is the transcript:
---
{transcript}
---
"""

def stream_model_response(prompt: str, api_key: str):
client = genai.Client(api_key=api_key)
response = client.models.generate_content_stream(
model='gemma-3n-e2b-it',
Expand All @@ -116,6 +127,16 @@ def summarize_transcript_stream(transcript: str, api_key: str, language: str = "

return ""


def summarize_transcript_stream(transcript: str, api_key: str, language: str, prompt_template: str):
prompt = prompt_template.format(language=language, transcript=transcript)
return stream_model_response(prompt, api_key)


def stock_sentiment_stream(transcript: str, api_key: str, language: str, prompt_template: str):
prompt = prompt_template.format(language=language, transcript=transcript)
return stream_model_response(prompt, api_key)

def get_language(country_iso):
my_dict = {
"DE": "German",
Expand All @@ -126,7 +147,7 @@ def get_language(country_iso):
}
return my_dict.get(country_iso)

def save_stream_to_json(stream, path, language, video_id):
def save_stream_to_json(stream, path, language, video_id, *, store=True, target_key="summary"):
collected = []
# Iteriere über den Stream
for chunk in stream:
Expand All @@ -135,10 +156,13 @@ def save_stream_to_json(stream, path, language, video_id):

# Am Ende zusammenfügen und speichern
result = "".join(collected)
global video_dict
if not store:
return

if language not in video_dict[video_id]["summary"]:
video_dict[video_id]["summary"][language] = result
global video_dict
video_entry = video_dict.setdefault(video_id, {})
target_dict = video_entry.setdefault(target_key, {})
target_dict[language] = result

with open(path, "w", encoding="utf-8") as f:
json.dump(video_dict, f, indent=4, ensure_ascii=False)
Expand All @@ -154,6 +178,9 @@ def my_generator(text: str):
st.markdown("Paste a YouTube link to get a quick AI-powered summary.:notebook:")
video_url = st.text_input(" ", value="Paste your YouTube URL here...")

summary_prompt_template = DEFAULT_SUMMARY_PROMPT
stock_prompt_template = DEFAULT_STOCK_PROMPT

if video_url != "Paste your YouTube URL here...":
video_id = get_video_id(video_url)
if video_id is None:
Expand All @@ -170,11 +197,26 @@ def my_generator(text: str):
default="English",
)

with st.expander("LLM Prompt Einstellungen", expanded=False):
st.caption("Nutze die Platzhalter {language} und {transcript}, um Sprache und Inhalt zu steuern.")
summary_prompt_template = st.text_area(
"Prompt für die Zusammenfassung",
value=DEFAULT_SUMMARY_PROMPT,
height=260,
key="summary_prompt_input",
)
stock_prompt_template = st.text_area(
"Prompt für Aktien-Sentiment",
value=DEFAULT_STOCK_PROMPT,
height=220,
key="stock_prompt_input",
)

else:
st.stop()

if st.button("Summarize"):
tab_summary, tab_transcript = st.tabs(["Ai powered Summary", "Transcript"])
tab_summary, tab_transcript, tab_stocks = st.tabs(["Ai powered Summary", "Transcript", "Besprochene Aktien"])

if key_exists([video_id]):
trans = video_dict.get(video_id)["transcript"]
Expand All @@ -183,7 +225,8 @@ def my_generator(text: str):
# st.write("get trans NOT from dict")
trans = get_yt_transcript(video_id)
video_dict[video_id] = {"transcript": trans,
"summary": {}
"summary": {},
"stock_sentiment": {},
}

with open(json_path, "w", encoding="utf-8") as f:
Expand All @@ -192,8 +235,8 @@ def my_generator(text: str):

with tab_summary:

if key_exists([video_id, "summary", output_language]):
# st.write("key exists")
store_summary = summary_prompt_template.strip() == DEFAULT_SUMMARY_PROMPT
if store_summary and key_exists([video_id, "summary", output_language]):
summary = video_dict.get(video_id)["summary"].get(output_language)
st.write_stream(my_generator(summary))

Expand All @@ -204,16 +247,57 @@ def my_generator(text: str):
try:
st.write_stream(
save_stream_to_json(
summarize_transcript_stream(trans, api_key=API_KEY, language=output_language),
summarize_transcript_stream(
trans,
api_key=API_KEY,
language=output_language,
prompt_template=summary_prompt_template,
),
path=json_path,
language=output_language,
video_id=video_id,
store=store_summary,
target_key="summary",
)
)


except KeyError as e:
missing_key = e.args[0]
st.error(f"Der Prompt muss den Platzhalter {{{missing_key}}} enthalten.")
st.stop()
except genai.errors.ClientError as e:
st.error(e)
st.error(e)

with tab_transcript:
st.write(trans)

with tab_stocks:
store_stock = stock_prompt_template.strip() == DEFAULT_STOCK_PROMPT
if store_stock and key_exists([video_id, "stock_sentiment", output_language]):
stock_summary = video_dict.get(video_id)["stock_sentiment"].get(output_language)
st.write_stream(my_generator(stock_summary))

else:
try:
st.write_stream(
save_stream_to_json(
stock_sentiment_stream(
trans,
api_key=API_KEY,
language=output_language,
prompt_template=stock_prompt_template,
),
path=json_path,
language=output_language,
video_id=video_id,
store=store_stock,
target_key="stock_sentiment",
)
)
except KeyError as e:
missing_key = e.args[0]
st.error(f"Der Prompt muss den Platzhalter {{{missing_key}}} enthalten.")
st.stop()
except genai.errors.ClientError as e:
st.error(e)