T2I / app.py
VirtualOasis's picture
Update app.py
30a1cae verified
raw
history blame
8.33 kB
import os
import tempfile
import base64
import gradio as gr
from PIL import Image, ImageDraw, ImageFont
# --- Image Generation Logic (from coverter.py) ---
def draw_horizontal_lines(draw, width, height, spacing=60, color=(230, 230, 230)):
"""Draws horizontal lines on the image."""
for y in range(0, height, spacing):
draw.line([(0, y), (width, y)], fill=color, width=2)
def draw_dot_grid(draw, width, height, spacing=50, color=(220, 220, 220)):
"""Draws a dot grid on the image."""
for x in range(0, width, spacing):
for y in range(0, height, spacing):
draw.ellipse([(x-2, y-2), (x+2, y+2)], fill=color)
def draw_lattice_grid(draw, width, height, spacing=100, color=(235, 235, 235)):
"""Draws a lattice/graph paper grid on the image."""
# Draw vertical lines
for x in range(0, width, spacing):
draw.line([(x, 0), (x, height)], fill=color, width=2)
# Draw horizontal lines
for y in range(0, height, spacing):
draw.line([(0, y), (width, y)], fill=color, width=2)
def text_to_images_generator(text_content, style='lines'):
"""
Converts a given string of text into a single combined image and returns the file path.
This is compatible with both UI and MCP.
Args:
text_content (str): The text to be converted.
style (str): The background style ('plain', 'lines', 'dots', 'grid').
Returns:
str: Message with the path to the generated combined image file.
"""
if not text_content or not text_content.strip():
return "Error: Input text is empty. Please enter some text to generate images."
# --- Configuration ---
IMG_WIDTH = 1080
IMG_HEIGHT = 1080
BACKGROUND_COLOR = (255, 255, 255)
TEXT_COLOR = (10, 10, 10)
STYLE_COLOR = (225, 225, 225) # Color for lines/dots/grid
PADDING_X = 80
PADDING_Y = 80
FONT_SIZE = 48
LINE_SPACING = 20
# --- Font Loading ---
font = None
try:
font_paths_to_try = [
"Arial.ttf", "arial.ttf", "DejaVuSans.ttf",
"/System/Library/Fonts/Supplemental/Arial.ttf",
"/usr/share/fonts/truetype/liberation/LiberationSans-Regular.ttf",
"/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf"
]
for f_path in font_paths_to_try:
try:
font = ImageFont.truetype(f_path, FONT_SIZE)
break
except IOError:
continue
if not font:
font = ImageFont.load_default()
except Exception as e:
print(f"An unexpected error occurred during font loading: {e}")
font = ImageFont.load_default()
# --- Text Wrapping Logic ---
drawable_width = IMG_WIDTH - 2 * PADDING_X
paragraphs = [p.strip() for p in text_content.strip().split('\n') if p.strip()]
all_lines_and_breaks = []
for i, paragraph in enumerate(paragraphs):
words = paragraph.split()
current_line = ""
for word in words:
if font.getlength(word) > drawable_width:
temp_word = ""
for char in word:
if font.getlength(temp_word + char) > drawable_width:
all_lines_and_breaks.append(temp_word)
temp_word = char
else:
temp_word += char
word = temp_word
if font.getlength(current_line + " " + word) <= drawable_width:
current_line += " " + word
else:
all_lines_and_breaks.append(current_line.strip())
current_line = word
all_lines_and_breaks.append(current_line.strip())
if i < len(paragraphs) - 1:
all_lines_and_breaks.append(None)
# --- Image Generation ---
try:
line_height = font.getbbox("A")[3] - font.getbbox("A")[1]
except AttributeError:
line_height = 12
PARAGRAPH_SPACING = line_height
# Calculate pages and total height needed
pages_content = []
current_page = []
y_text = PADDING_Y
for item in all_lines_and_breaks:
is_break = item is None
item_height = PARAGRAPH_SPACING if is_break else line_height
if y_text + item_height > IMG_HEIGHT - PADDING_Y:
pages_content.append(current_page)
current_page = [item]
y_text = PADDING_Y + item_height + (0 if is_break else LINE_SPACING)
else:
current_page.append(item)
y_text += item_height + (0 if is_break else LINE_SPACING)
if current_page:
pages_content.append(current_page)
# Create a single combined image with all pages
total_height = len(pages_content) * IMG_HEIGHT
combined_img = Image.new('RGB', (IMG_WIDTH, total_height), color=BACKGROUND_COLOR)
for page_idx, page_content in enumerate(pages_content):
# Create individual page
page_img = Image.new('RGB', (IMG_WIDTH, IMG_HEIGHT), color=BACKGROUND_COLOR)
draw = ImageDraw.Draw(page_img)
if style == 'lines':
line_style_spacing = line_height + LINE_SPACING
draw_horizontal_lines(draw, IMG_WIDTH, IMG_HEIGHT, spacing=line_style_spacing, color=STYLE_COLOR)
elif style == 'dots':
draw_dot_grid(draw, IMG_WIDTH, IMG_HEIGHT, color=STYLE_COLOR)
elif style == 'grid':
draw_lattice_grid(draw, IMG_WIDTH, IMG_HEIGHT, color=STYLE_COLOR)
current_y = PADDING_Y
for page_item in page_content:
if page_item is not None:
draw.text((PADDING_X, current_y), page_item, font=font, fill=TEXT_COLOR)
current_y += line_height + LINE_SPACING
else:
current_y += PARAGRAPH_SPACING
# Paste this page onto the combined image
combined_img.paste(page_img, (0, page_idx * IMG_HEIGHT))
# Save combined image to temporary file and also encode as base64
temp_file = tempfile.NamedTemporaryFile(delete=False, suffix='.png')
combined_img.save(temp_file.name, format='PNG')
temp_file.close()
# Convert to base64 for inline viewing
import io
img_buffer = io.BytesIO()
combined_img.save(img_buffer, format='PNG')
img_buffer.seek(0)
img_base64 = base64.b64encode(img_buffer.getvalue()).decode('utf-8')
return f"""Image successfully generated and saved to: {temp_file.name}
Image details:
- Total pages: {len(pages_content)}
- Image dimensions: {IMG_WIDTH} x {total_height} pixels
- Style: {style}
- Format: PNG
Base64 encoded image (you can copy this and paste into a base64 image viewer):
data:image/png;base64,{img_base64}
To view the image:
1. Copy the entire data:image/png;base64,... string above
2. Paste it into your browser's address bar, or
3. Use an online base64 image viewer like: https://base64.guru/converter/decode/image"""
# --- Gradio Interface ---
example_text = """In the heart of a bustling city, there lived a clockmaker named Alistair. His shop, a quaint corner of tranquility amidst the urban chaos, was filled with the gentle ticking of countless timepieces. Each clock was a masterpiece, a testament to his dedication and skill.
One day, a young girl with eyes as curious as a cat's wandered into his shop. She wasn't interested in the shiny new watches but was drawn to the grandfather clock in the corner. "What's its story?" she asked, her voice soft. Alistair smiled, for he knew he had found the next guardian of the stories. The legacy of the whispering clock would live on."""
demo = gr.Interface(
fn=text_to_images_generator,
inputs=[
gr.Textbox(lines=15, label="Text Content", placeholder="Paste your long-form text here...", value=example_text),
gr.Radio(['lines', 'dots', 'grid', 'plain'], label="Background Style", value='lines')
],
outputs=gr.Textbox(label="Result", show_label=True),
title="Text-to-Image Converter",
description="Transforms long-form text into a single combined image with multiple pages. Paste your text, choose a style, and click 'Submit'. The result will show the file path where your image was saved.",
allow_flagging="never"
)
# --- Main Execution ---
if __name__ == "__main__":
demo.launch(mcp_server=True)