kunthawat/ALwrity
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

AI blog writer & blog metadata updates & improvements

Browse Source
This commit is contained in:
ajaysi
2025-05-01 10:38:01 +05:30
parent a6e3ac2f8b
commit a27522d32e
6 changed files with 586 additions and 63 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
lib/ai_writers/gpt_blog_sections.py
View File

@@ -2,8 +2,8 @@ import sys
import os
import json
from ..gpt_providers.openai_chat_completion import openai_chatgpt
from ..gpt_providers.gemini_pro_text import gemini_text_response
from ..gpt_providers.text_generation.openai_text_gen import openai_text_generation
from ..gpt_providers.text_generation.gemini_pro_text import gemini_text_generation
from loguru import logger
logger.remove()

540
lib/ai_writers/keywords_to_blog_streamlit.py
View File

@@ -25,25 +25,12 @@ from ..ai_web_researcher.gpt_online_researcher import (
do_tavily_ai_search as gpt_do_tavily_ai_search,
do_metaphor_ai_research, do_google_pytrends_analysis)
from .blog_from_google_serp import write_blog_google_serp, blog_with_research
from ..ai_web_researcher.you_web_reseacher import get_rag_results, search_ydc_index
from ..blog_metadata.get_blog_metadata import blog_metadata
from ..blog_postprocessing.save_blog_to_file import save_blog_to_file
from ..gpt_providers.text_to_image_generation.main_generate_image_from_prompt import generate_image
# Function to convert text to speech and save as an audio file
def text_to_speech(text, lang='en'):
tts = gTTS(text=text, lang=lang)
tts.save("output.mp3")
return "output.mp3"
# Function to get audio file as a downloadable link
def get_audio_file(audio_file):
with open(audio_file, "rb") as file:
data = file.read()
b64_data = base64.b64encode(data).decode()
return f'<a href="data:audio/mp3;base64,{b64_data}" download="output.mp3">Download audio file</a>'
from ..ai_seo_tools.content_title_generator import generate_blog_titles
from ..ai_seo_tools.meta_desc_generator import generate_blog_metadesc
from ..ai_seo_tools.seo_structured_data import ai_structured_data
def initialize_parameters(search_params=None, blog_params=None):
@@ -285,21 +272,22 @@ def generate_blog_metadata(blog_markdown_str, search_keywords, status):
status: Streamlit status object
Returns:
tuple: (blog_title, blog_meta_desc, blog_tags, blog_categories)
tuple: (blog_title, blog_meta_desc, blog_tags, blog_categories, blog_hashtags, blog_slug)
"""
status.update(label="🔍 Generating title, meta description, tags, and categories...")
status.update(label="🔍 Generating title, meta description, tags, categories, hashtags, and slug...")
try:
blog_title, blog_meta_desc, blog_tags, blog_categories = asyncio.run(blog_metadata(blog_markdown_str))
# Get all 6 metadata values from blog_metadata
blog_title, blog_meta_desc, blog_tags, blog_categories, blog_hashtags, blog_slug = asyncio.run(blog_metadata(blog_markdown_str))
status.update(label="✅ Generated blog metadata successfully")
return blog_title, blog_meta_desc, blog_tags, blog_categories
return blog_title, blog_meta_desc, blog_tags, blog_categories, blog_hashtags, blog_slug
except Exception as err:
st.error(f"Failed to get blog metadata: {err}")
logger.error(f"Failed to get blog metadata: {err}")
status.update(label="❌ Failed to get blog metadata", state="error")
return None, None, None, None
return None, None, None, None, None, None
def generate_blog_image(blog_title, blog_meta_desc, blog_markdown_str, status):
def generate_blog_image(blog_title, blog_meta_desc, blog_markdown_str, status, blog_tags=None):
"""
Generate a featured image for the blog.
@@ -308,6 +296,7 @@ def generate_blog_image(blog_title, blog_meta_desc, blog_markdown_str, status):
blog_meta_desc (str): Blog meta description
blog_markdown_str (str): Blog content
status: Streamlit status object
blog_tags (list, optional): Blog tags to use for image prompt enhancement
Returns:
str: Path to the generated image or None if generation failed
@@ -337,8 +326,17 @@ def generate_blog_image(blog_title, blog_meta_desc, blog_markdown_str, status):
logger.info(f"Generating image with prompt: {text_to_image}")
status.update(label=f"🖼️ Creating image with prompt: \"{text_to_image[:50]}...\"")
# Attempt image generation
generated_image_filepath = generate_image(text_to_image)
# Extract blog tags if available
blog_tags_list = blog_tags if isinstance(blog_tags, list) else []
# Attempt image generation with all available parameters
generated_image_filepath = generate_image(
user_prompt=text_to_image,
title=blog_title,
description=blog_meta_desc,
tags=blog_tags_list,
content=blog_markdown_str[:2000] # Limit content length to avoid too large payloads
)
# If first attempt failed, try with a simplified prompt
if not generated_image_filepath:
@@ -347,7 +345,13 @@ def generate_blog_image(blog_title, blog_meta_desc, blog_markdown_str, status):
# Create a simpler prompt
simplified_prompt = " ".join(text_to_image.split()[:10])
generated_image_filepath = generate_image(simplified_prompt)
generated_image_filepath = generate_image(
user_prompt=simplified_prompt,
title=blog_title,
description=blog_meta_desc,
tags=blog_tags_list,
content=blog_markdown_str[:1000] # Use even shorter content for the retry
)
if generated_image_filepath:
status.update(label="✅ Successfully generated featured image")
@@ -363,7 +367,7 @@ def generate_blog_image(blog_title, blog_meta_desc, blog_markdown_str, status):
return None
def regenerate_blog_image(blog_title, blog_meta_desc, blog_markdown_str):
def regenerate_blog_image(blog_title, blog_meta_desc, blog_markdown_str, blog_tags=None):
"""
Regenerate a blog image on demand.
@@ -371,6 +375,7 @@ def regenerate_blog_image(blog_title, blog_meta_desc, blog_markdown_str):
blog_title (str): Blog title
blog_meta_desc (str): Blog meta description
blog_markdown_str (str): Blog content
blog_tags (list, optional): Blog tags to use for image prompt enhancement
Returns:
str: Path to the generated image or None if generation failed
@@ -390,8 +395,17 @@ def regenerate_blog_image(blog_title, blog_meta_desc, blog_markdown_str):
status.update(label=f"🖼️ Generating new image with prompt: \"{prompt}\"")
# Generate the image
generated_image_filepath = generate_image(prompt)
# Extract any tags if available - will be passed as empty list otherwise
blog_tags_list = blog_tags if isinstance(blog_tags, list) else []
# Generate the image with all parameters
generated_image_filepath = generate_image(
user_prompt=prompt,
title=blog_title,
description=blog_meta_desc,
tags=blog_tags_list,
content=blog_markdown_str[:2000] # Limit content length to avoid too large payloads
)
if generated_image_filepath:
status.update(label="✅ Successfully generated new image", state="complete")
@@ -407,7 +421,7 @@ def regenerate_blog_image(blog_title, blog_meta_desc, blog_markdown_str):
return None
def save_blog_content(blog_markdown_str, blog_title, blog_meta_desc, blog_tags, blog_categories, generated_image_filepath, status):
def save_blog_content(blog_markdown_str, blog_title, blog_meta_desc, blog_tags, blog_categories, generated_image_filepath, status, blog_hashtags=None, blog_slug=None):
"""
Save the blog content to a file.
@@ -419,6 +433,8 @@ def save_blog_content(blog_markdown_str, blog_title, blog_meta_desc, blog_tags,
blog_categories (list): Blog categories
generated_image_filepath (str): Path to the generated image
status: Streamlit status object
blog_hashtags (str, optional): Social media hashtags
blog_slug (str, optional): SEO-friendly URL slug
Returns:
str: Path to the saved file or None if saving failed
@@ -632,51 +648,172 @@ def write_blog_from_keywords(search_keywords, url=None, search_params=None, blog
with st.status("Enhancing content...", expanded=True) as status:
# Generate metadata
status.update(label="🏷️ Generating SEO metadata (title, description, tags)...")
blog_title, blog_meta_desc, blog_tags, blog_categories = generate_blog_metadata(
blog_title, blog_meta_desc, blog_tags, blog_categories, blog_hashtags, blog_slug = generate_blog_metadata(
blog_markdown_str, search_keywords, status
)
# Check if there are updated values in session state
if 'blog_title' in st.session_state:
blog_title = st.session_state.blog_title
status.update(label=f"✅ Using refined title: \"{blog_title}\"")
if 'blog_meta_desc' in st.session_state:
blog_meta_desc = st.session_state.blog_meta_desc
status.update(label=f"✅ Using refined meta description")
if blog_title and blog_meta_desc:
status.update(label=f"✅ Generated metadata: \"{blog_title}\"")
# Generate featured image
status.update(label="🖼️ Creating featured image...")
generated_image_filepath = generate_blog_image(
blog_title, blog_meta_desc, blog_markdown_str, status
blog_title, blog_meta_desc, blog_markdown_str, status, blog_tags
)
# Save blog content to file
status.update(label="💾 Saving blog content...")
saved_blog_to_file = save_blog_content(
blog_markdown_str, blog_title, blog_meta_desc, blog_tags,
blog_categories, generated_image_filepath, status
blog_categories, generated_image_filepath, status, blog_hashtags, blog_slug
)
status.update(label="✅ Content enhancement complete", state="complete")
else:
status.update(label="⚠️ Metadata generation had issues, using simplified format", state="warning")
# STEP 4: Final presentation
update_progress(4, 5, "Preparing final blog presentation")
# Now display the final blog content
with final_content_placeholder.container():
st.markdown("---")
st.header("📝 Generated Blog Content")
# Display metadata
if blog_title:
st.title(blog_title)
if blog_meta_desc:
st.markdown(f"*{blog_meta_desc}*")
if blog_tags:
st.markdown(f"**Tags:** {', '.join(blog_tags)}")
# Add buttons in columns for refining metadata
col1, col2 = st.columns(2)
with col1:
refine_title_button = st.button("🔄 Refine Blog Title", use_container_width=True)
with col2:
refine_meta_button = st.button("🔄 Refine Meta Description", use_container_width=True)
if blog_categories:
st.markdown(f"**Categories:** {', '.join(blog_categories)}")
# Add a row for structured data
st.markdown("---")
structured_data_col1, structured_data_col2 = st.columns([3, 1])
with structured_data_col1:
# Educational popover explaining why rich snippets are important
with st.expander("ℹ️ Why Rich Snippets Are Important for SEO"):
st.markdown("""
### Rich Snippets: Boosting Your SEO and Click-Through Rates
**What are Rich Snippets?**
Rich snippets are enhanced search results that display additional information directly in search engine results pages (SERPs). They're created using structured data markup (JSON-LD) that helps search engines understand your content better.
**Why are they important?**
1. **Increased Visibility**: Rich snippets stand out in search results with stars, images, and additional information
2. **Higher Click-Through Rates (CTR)**: Studies show rich snippets can increase CTR by 30-150%
3. **Improved SEO**: They help search engines understand your content better, potentially improving rankings
4. **Enhanced User Experience**: Users can see key information before clicking, leading to more qualified traffic
5. **Mobile-Friendly**: Rich snippets are especially effective on mobile searches
**Common types of rich snippets include:**
- Articles/Blogs (with author, date, image)
- Products (with ratings, price, availability)
- Recipes (with cooking time, ratings, calories)
- Events (with date, location, ticket info)
- Local Business (with address, hours, ratings)
Adding structured data to your content is a powerful SEO technique that requires minimal effort but provides significant benefits.
""")
with structured_data_col2:
# Button to generate rich snippet
generate_snippet_button = st.button("📊 Generate Rich Snippet", use_container_width=True)
# Dialog for generating structured data
if generate_snippet_button:
with st.expander("Structured Data Generation Tool", expanded=True):
st.subheader("Generate Structured Data (Rich Snippets)")
# Simplified blog URL input
blog_url = st.text_input(
"Blog URL:",
placeholder="https://yourblog.com/your-article",
help="Enter the URL where this blog will be published"
)
# Auto-fill content type to "Article" since we're working with a blog
content_type = "Article"
st.info(f"Content Type: {content_type} (Auto-selected for blog content)")
# Create details dictionary with blog metadata
today = datetime.now().strftime("%Y-%m-%d")
# Form for additional article details
with st.form(key="structured_data_form"):
st.markdown("#### Article Details")
# Pre-fill with blog title and other metadata
article_title = st.text_input("Headline:", value=blog_title if blog_title else "")
article_author = st.text_input("Author:", value="")
article_date = st.date_input("Date Published:", value=datetime.now())
article_keywords = st.text_input("Keywords:", value=blog_tags if blog_tags else "")
submit_structured_data = st.form_submit_button("Generate JSON-LD")
if submit_structured_data:
if not blog_url:
st.error("Please enter a blog URL to generate structured data.")
else:
# Create details dictionary
details = {
"Headline": article_title,
"Author": article_author,
"Date Published": article_date,
"Keywords": article_keywords
}
# Call the imported ai_structured_data function or recreate its functionality
with st.spinner("Generating structured data..."):
# Import and use the function from the module directly
from ..ai_seo_tools.seo_structured_data import generate_json_data
# Generate the structured data
structured_data = generate_json_data(content_type, details, blog_url)
if structured_data:
st.success("✅ Structured data generated successfully!")
st.markdown("### Generated JSON-LD Code")
st.code(structured_data, language="json")
# Download button
st.download_button(
label="📥 Download JSON-LD",
data=structured_data,
file_name=f"{content_type}_structured_data.json",
mime="application/json",
)
# Implementation instructions
with st.expander("How to Implement This Code"):
st.markdown("""
### Adding this JSON-LD to your website:
1. **Copy the generated JSON-LD code** above
2. **Add it to the `<head>` section of your HTML** like this:
```html
<script type="application/ld+json">
[PASTE YOUR JSON-LD CODE HERE]
</script>
```
3. **Verify the implementation** using Google's Rich Results Test tool:
[https://search.google.com/test/rich-results](https://search.google.com/test/rich-results)
4. **Monitor your search appearance** in Google Search Console
""")
else:
st.error("Failed to generate structured data. Please check your inputs and try again.")
# Image section with regeneration option
st.subheader("🖼️ Featured Image")
image_container = st.container()
@@ -688,14 +825,14 @@ def write_blog_from_keywords(search_keywords, url=None, search_params=None, blog
# Add regenerate button
if st.button("🔄 Regenerate Image", key="regenerate_image"):
new_image_path = regenerate_blog_image(blog_title, blog_meta_desc, blog_markdown_str)
new_image_path = regenerate_blog_image(blog_title, blog_meta_desc, blog_markdown_str, blog_tags)
if new_image_path:
generated_image_filepath = new_image_path
st.rerun() # Refresh the page to show the new image
else:
st.info("No featured image was generated. Click below to generate one.")
if st.button("🖼️ Generate Image", key="generate_image"):
new_image_path = regenerate_blog_image(blog_title, blog_meta_desc, blog_markdown_str)
new_image_path = regenerate_blog_image(blog_title, blog_meta_desc, blog_markdown_str, blog_tags)
if new_image_path:
generated_image_filepath = new_image_path
st.rerun() # Refresh the page to show the new image
@@ -718,6 +855,303 @@ def write_blog_from_keywords(search_keywords, url=None, search_params=None, blog
if generate_audio_button:
generate_audio_version(blog_markdown_str)
# STEP 4: Final presentation
update_progress(4, 5, "Preparing final blog presentation")
# Now display the final blog content
with final_content_placeholder.container():
st.markdown("---")
# Display tabular data of metadata
st.subheader("🏷️ Metadata")
metadata_container = st.container()
with metadata_container:
st.table({
"Metadata": ["Blog Title", "Meta Description", "Tags", "Categories", "Hashtags", "Slug"],
"Value": [blog_title, blog_meta_desc, blog_tags, blog_categories, blog_hashtags, blog_slug]
})
# Add buttons in columns for refining metadata
col1, col2 = st.columns(2)
with col1:
refine_title_button = st.button("🔄 Refine Blog Title", use_container_width=True)
with col2:
refine_meta_button = st.button("🔄 Refine Meta Description", use_container_width=True)
# Add a row for structured data with a "Generate Rich Snippet" button
st.markdown("---")
st.markdown("### Get Structured Data")
structured_data_col1, structured_data_col2 = st.columns([3, 1])
with structured_data_col1:
# Educational popover explaining why rich snippets are important
with st.expander("ℹ️ Why Rich Snippets Are Important for SEO"):
st.markdown("""
### Rich Snippets: Boosting Your SEO and Click-Through Rates
**What are Rich Snippets?**
Rich snippets are enhanced search results that display additional information directly in search engine results pages (SERPs). They're created using structured data markup (JSON-LD) that helps search engines understand your content better.
**Why are they important?**
1. **Increased Visibility**: Rich snippets stand out in search results with stars, images, and additional information
2. **Higher Click-Through Rates (CTR)**: Studies show rich snippets can increase CTR by 30-150%
3. **Improved SEO**: They help search engines understand your content better, potentially improving rankings
4. **Enhanced User Experience**: Users can see key information before clicking, leading to more qualified traffic
5. **Mobile-Friendly**: Rich snippets are especially effective on mobile searches
**Common types of rich snippets include:**
- Articles/Blogs (with author, date, image)
- Products (with ratings, price, availability)
- Recipes (with cooking time, ratings, calories)
- Events (with date, location, ticket info)
- Local Business (with address, hours, ratings)
Adding structured data to your content is a powerful SEO technique that requires minimal effort but provides significant benefits.
""")
with structured_data_col2:
# Button to generate rich snippet
generate_snippet_button = st.button("📊 Generate Rich Snippet", use_container_width=True)
# Dialog for generating structured data
if generate_snippet_button:
with st.expander("Structured Data Generation Tool", expanded=True):
st.subheader("Generate Structured Data (Rich Snippets)")
# Simplified blog URL input
blog_url = st.text_input(
"Blog URL:",
placeholder="https://yourblog.com/your-article",
help="Enter the URL where this blog will be published"
)
# Auto-fill content type to "Article" since we're working with a blog
content_type = "Article"
st.info(f"Content Type: {content_type} (Auto-selected for blog content)")
# Form for additional article details
with st.form(key="structured_data_form"):
st.markdown("#### Article Details")
# Pre-fill with blog title and other metadata
article_title = st.text_input("Headline:", value=blog_title if blog_title else "")
article_author = st.text_input("Author:", value="")
article_date = st.date_input("Date Published:", value=datetime.now())
article_keywords = st.text_input("Keywords:", value=blog_tags if blog_tags else "")
submit_structured_data = st.form_submit_button("Generate JSON-LD")
if submit_structured_data:
if not blog_url:
st.error("Please enter a blog URL to generate structured data.")
else:
# Create details dictionary
details = {
"Headline": article_title,
"Author": article_author,
"Date Published": article_date,
"Keywords": article_keywords
}
# Call the imported ai_structured_data function or recreate its functionality
with st.spinner("Generating structured data..."):
# Import and use the function from the module directly
from ..ai_seo_tools.seo_structured_data import generate_json_data
# Generate the structured data
structured_data = generate_json_data(content_type, details, blog_url)
if structured_data:
st.success("✅ Structured data generated successfully!")
st.markdown("### Generated JSON-LD Code")
st.code(structured_data, language="json")
# Download button
st.download_button(
label="📥 Download JSON-LD",
data=structured_data,
file_name=f"{content_type}_structured_data.json",
mime="application/json",
)
# Implementation instructions
with st.expander("How to Implement This Code"):
st.markdown("""
### Adding this JSON-LD to your website:
1. **Copy the generated JSON-LD code** above
2. **Add it to the `<head>` section of your HTML** like this:
```html
<script type="application/ld+json">
[PASTE YOUR JSON-LD CODE HERE]
</script>
```
3. **Verify the implementation** using Google's Rich Results Test tool:
[https://search.google.com/test/rich-results](https://search.google.com/test/rich-results)
4. **Monitor your search appearance** in Google Search Console
""")
else:
st.error("Failed to generate structured data. Please check your inputs and try again.")
# Dialog for refining blog title
if refine_title_button:
with st.expander("Blog Title Refinement Tool", expanded=True):
st.subheader("Refine Your Blog Title")
# Store the current title in session state for later reference
if 'current_title' not in st.session_state:
st.session_state.current_title = blog_title
# Extract keywords from tags and content
keywords_from_tags = blog_tags if blog_tags else ""
blog_content_sample = blog_markdown_str[:3000] if blog_markdown_str else ""
# Title generation form
with st.form(key="title_form"):
st.markdown("#### Provide information to generate new title suggestions")
title_keywords = st.text_input(
"Main Keywords:",
value=keywords_from_tags,
help="Enter main keywords separated by commas"
)
title_type = st.selectbox(
"Blog Type:",
options=['General', 'How-to Guides', 'Tutorials', 'Listicles', 'Newsworthy Posts', 'FAQs', 'Checklists/Cheat Sheets'],
index=0
)
intent_type = st.selectbox(
"Search Intent:",
options=['Informational Intent', 'Commercial Intent', 'Transactional Intent', 'Navigational Intent'],
index=0
)
language = st.selectbox(
"Language:",
options=["English", "Spanish", "French", "German", "Chinese", "Japanese", "Other"],
index=0
)
if language == "Other":
language = st.text_input("Specify Language:", placeholder="e.g., Italian, Dutch")
submit_title = st.form_submit_button("Generate Title Suggestions")
if submit_title:
with st.spinner("Generating title suggestions..."):
# Use the imported generate_blog_titles function
title_suggestions = generate_blog_titles(
title_keywords,
blog_content_sample,
title_type,
intent_type,
language
)
if title_suggestions:
st.success("✅ Title suggestions generated!")
st.markdown("### Title Suggestions")
st.markdown(title_suggestions)
# Allow selecting a title
st.markdown("#### Select or enter a new title")
new_title = st.text_input("New Blog Title", value=st.session_state.current_title)
if st.button("Apply New Title"):
# Store the new title in the session state
st.session_state.blog_title = new_title
st.success(f"✅ Title updated to: {new_title}")
# Return to main page with updated title
st.experimental_rerun()
else:
st.error("Failed to generate title suggestions.")
# Dialog for refining meta description
if refine_meta_button:
with st.expander("Meta Description Refinement Tool", expanded=True):
st.subheader("Refine Your Meta Description")
# Store the current meta description in session state
if 'current_meta_desc' not in st.session_state:
st.session_state.current_meta_desc = blog_meta_desc
# Extract keywords from tags and content
keywords_from_tags = blog_tags if blog_tags else ""
# Meta description generation form
with st.form(key="meta_desc_form"):
st.markdown("#### Provide information to generate new meta description suggestions")
meta_keywords = st.text_input(
"Target Keywords:",
value=keywords_from_tags,
help="Enter target keywords separated by commas"
)
tone_options = ["General", "Informative", "Engaging", "Humorous", "Intriguing", "Playful"]
tone = st.selectbox(
"Desired Tone:",
options=tone_options,
index=0
)
search_type = st.selectbox(
"Search Intent:",
options=['Informational Intent', 'Commercial Intent', 'Transactional Intent', 'Navigational Intent'],
index=0
)
language_options = ["English", "Spanish", "French", "German", "Other"]
language_choice = st.selectbox(
"Preferred Language:",
options=language_options,
index=0
)
if language_choice == "Other":
language = st.text_input("Specify Language:", placeholder="e.g., Italian, Chinese")
else:
language = language_choice
submit_meta = st.form_submit_button("Generate Meta Description Suggestions")
if submit_meta:
with st.spinner("Generating meta description suggestions..."):
# Use the imported generate_blog_metadesc function
meta_suggestions = generate_blog_metadesc(
meta_keywords,
tone,
search_type,
language
)
if meta_suggestions:
st.success("✅ Meta description suggestions generated!")
st.markdown("### Meta Description Suggestions")
st.markdown(meta_suggestions)
# Allow selecting a meta description
st.markdown("#### Select or enter a new meta description")
new_meta_desc = st.text_area("New Meta Description", value=st.session_state.current_meta_desc)
if st.button("Apply New Meta Description"):
# Store the new meta description in the session state
st.session_state.blog_meta_desc = new_meta_desc
st.success(f"✅ Meta description updated!")
# Return to main page with updated meta description
st.experimental_rerun()
else:
st.error("Failed to generate meta description suggestions.")
# Final progress update
update_progress(5, 5, "Blog generation complete!")

8
lib/blog_metadata/get_blog_metadata.py
View File

@@ -72,10 +72,10 @@ async def blog_metadata(blog_article):
# Present the result in a table format
status_container.success("✅ Metadata generation complete")
st.table({
"Metadata": ["Blog Title", "Meta Description", "Tags", "Categories", "Social Hashtags", "URL Slug"],
"Value": [blog_title, blog_meta_desc, blog_tags, blog_categories, blog_hashtags, blog_slug]
})
#st.table({
# "Metadata": ["Blog Title", "Meta Description", "Tags", "Categories", "Social Hashtags", "URL Slug"],
# "Value": [blog_title, blog_meta_desc, blog_tags, blog_categories, blog_hashtags, blog_slug]
#})
return blog_title, blog_meta_desc, blog_tags, blog_categories, blog_hashtags, blog_slug

10
lib/gpt_providers/text_to_image_generation/gen_gemini_images.py
View File

@@ -190,6 +190,14 @@ def generate_gemini_image(prompt, keywords=None, style=None, focus=None, enhance
"""
logger.info(f"Generating image with prompt: '{prompt[:100]}...'")
# Check if the GEMINI_API_KEY is available
api_key = os.getenv("GEMINI_API_KEY")
if not api_key:
error_msg = "GEMINI_API_KEY is missing. Please set it in your environment variables."
logger.error(error_msg)
st.error(f"🔑 {error_msg}")
return None
# Enhance the prompt if requested
if enhance_prompt and keywords:
prompt_generator = AIPromptGenerator()
@@ -209,7 +217,7 @@ def generate_gemini_image(prompt, keywords=None, style=None, focus=None, enhance
while retry_count <= max_retries:
try:
client = genai.Client(api_key=os.getenv("GEMINI_API_KEY"))
client = genai.Client(api_key=api_key)
contents = (prompt)
logger.info("Sending request to Gemini API")

74
lib/gpt_providers/text_to_image_generation/main_generate_image_from_prompt.py
View File

@@ -27,7 +27,7 @@ from .gen_stabl_diff_img import generate_stable_diffusion_image
from ..text_generation.main_text_generation import llm_text_gen
from .gen_gemini_images import generate_gemini_image
def generate_image(user_prompt):
def generate_image(user_prompt, title=None, description=None, tags=None, content=None):
"""
The generation API endpoint creates an image based on a text prompt.
@@ -49,7 +49,8 @@ def generate_image(user_prompt):
if user_prompt:
try:
img_prompt = generate_img_prompt(user_prompt)
# Use enhanced prompt generator with all available parameters
img_prompt = generate_enhanced_img_prompt(user_prompt, title, description, tags, content)
if 'Dalle3' in image_engine:
logger.info(f"Calling Dalle3 text-to-image with prompt: {img_prompt}")
image_stored_at = generate_dalle3_images(img_prompt)
@@ -85,3 +86,72 @@ def generate_img_prompt(user_prompt):
response = llm_text_gen(prompt)
return response
def generate_enhanced_img_prompt(user_prompt, title=None, description=None, tags=None, content=None):
"""
Given user prompt and additional context (title, description, tags, content),
this function generates an enhanced prompt for better image generation.
Args:
user_prompt (str): Base prompt from the user
title (str, optional): Blog title or content title
description (str, optional): Blog or content description/summary
tags (list, optional): List of tags related to the content
content (str, optional): Actual content or excerpt
Returns:
str: Enhanced prompt for image generation
"""
# Start with the base prompt
context_parts = [user_prompt]
# Add relevant context if available
if title:
context_parts.append(f"Title: {title}")
if description:
context_parts.append(f"Description: {description}")
if tags and len(tags) > 0:
tag_text = ", ".join(tags[:5]) # Limit to 5 tags to avoid too much noise
context_parts.append(f"Tags: {tag_text}")
# Create a combined context
combined_context = "\n".join(context_parts)
# Add some content excerpt if available (limited to avoid token limits)
content_excerpt = ""
if content:
# Just use the first few hundred characters as excerpt
content_excerpt = content[:300] + "..." if len(content) > 300 else content
# Create the prompt for LLM
prompt = f"""
As an expert prompt engineer for AI image generation models, create a detailed, creative prompt
for generating a high-quality, relevant image based on the following context:
{combined_context}
Additional content excerpt:
{content_excerpt}
Your task is to:
1. Analyze the context and content to understand the main theme and subject
2. Create a rich, detailed prompt for image generation (50-75 words)
3. Include specific visual details, art style, mood, lighting, composition
4. Make sure the prompt is highly relevant to the original context
5. Avoid prohibited content or anything that violates image generation guidelines
Reply with ONLY the final prompt. No explanations or other text.
"""
# Generate the enhanced prompt
try:
enhanced_prompt = llm_text_gen(prompt)
logger.info(f"Generated enhanced image prompt: {enhanced_prompt[:100]}...")
return enhanced_prompt
except Exception as e:
logger.error(f"Error generating enhanced prompt: {e}")
# Fall back to the simple prompt generation if enhanced fails
return generate_img_prompt(user_prompt)

13
lib/gpt_providers/text_to_image_generation/save_image.py
View File

@@ -12,8 +12,16 @@ def save_generated_image(img_generation_response):
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
# Get image save directory with fallback to a local directory
image_save_dir = os.getenv('IMG_SAVE_DIR', 'generated_images')
# Create the directory if it doesn't exist
if not os.path.exists(image_save_dir):
logger.info(f"Creating image save directory: {image_save_dir}")
os.makedirs(image_save_dir, exist_ok=True)
generated_image_name = f"generated_image_{datetime.datetime.now():%Y-%m-%d-%H-%M-%S}.webp"
generated_image_filepath = os.path.join(os.getenv('IMG_SAVE_DIR'), generated_image_name)
generated_image_filepath = os.path.join(image_save_dir, generated_image_name)
try:
for i, image in enumerate(img_generation_response["artifacts"]):
@@ -22,6 +30,9 @@ def save_generated_image(img_generation_response):
except requests.exceptions.RequestException as e:
logger.error(f"Failed to get generated image content: {e}")
return None
except Exception as e:
logger.error(f"Error saving image: {e}")
return None
logger.info(f"Saved image at path: {generated_image_filepath}")