import gradio as gr
from pathlib import Path
import datetime
import re
import os
import shutil
import fitz # PyMuPDF
from PIL import Image
from collections import defaultdict
import io
from pypdf import PdfWriter
import random
# New imports for the SVG Emoji Engine
from lxml import etree
from svglib.svglib import svg2rlg
from reportlab.platypus import SimpleDocTemplate, Paragraph, Spacer, Table, TableStyle, PageBreak, BaseDocTemplate, Frame, PageTemplate, Image as ReportLabImage, Flowable
from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle
from reportlab.lib.pagesizes import letter, A4, legal, landscape
from reportlab.lib.units import inch
from reportlab.lib import colors
from reportlab.pdfbase import pdfmetrics
from reportlab.pdfbase.ttfonts import TTFont
from functools import partial
# --- Configuration & Setup ---
CWD = Path.cwd()
LAYOUTS = {
"A4 Portrait": {"size": A4},
"A4 Landscape": {"size": landscape(A4)},
"Letter Portrait": {"size": letter},
"Letter Landscape": {"size": landscape(letter)},
"Legal Portrait": {"size": legal},
"Legal Landscape": {"size": landscape(legal)},
}
OUTPUT_DIR = CWD / "generated_pdfs"
PREVIEW_DIR = CWD / "previews"
FONT_DIR = CWD # Assumes fonts are in the same directory as the script
# Create necessary directories
OUTPUT_DIR.mkdir(exist_ok=True)
PREVIEW_DIR.mkdir(exist_ok=True)
# --- Font & Emoji Handling ---
# Global cache for rendered emoji images to avoid re-rendering the same emoji
EMOJI_IMAGE_CACHE = {}
EMOJI_SVG_CACHE = {} # New cache for the SVG engine
EMOJI_FONT_PATH = None # Will be set in register_local_fonts
def register_local_fonts():
"""Finds and registers all .ttf files from the application's base directory."""
global EMOJI_FONT_PATH
print("--- Font Registration Process Starting ---")
text_font_names = []
emoji_font_name = None
noto_emoji_path = FONT_DIR / "NotoColorEmoji-Regular.ttf"
if not noto_emoji_path.exists():
print(f"CRITICAL: Color Emoji font not found at {noto_emoji_path}.")
print("Please download 'NotoColorEmoji-Regular.ttf' and place it in the application directory for color emojis to work.")
else:
EMOJI_FONT_PATH = str(noto_emoji_path)
print(f"Scanning for fonts in: {FONT_DIR.absolute()}")
font_files = list(FONT_DIR.glob("*.ttf"))
print(f"Found {len(font_files)} .ttf files: {[f.name for f in font_files]}")
for font_path in font_files:
try:
font_name = font_path.stem
pdfmetrics.registerFont(TTFont(font_name, str(font_path)))
pdfmetrics.registerFont(TTFont(f"{font_name}-Bold", str(font_path)))
pdfmetrics.registerFont(TTFont(f"{font_name}-Italic", str(font_path)))
pdfmetrics.registerFont(TTFont(f"{font_name}-BoldItalic", str(font_path)))
pdfmetrics.registerFontFamily(font_name, normal=font_name, bold=f"{font_name}-Bold", italic=f"{font_name}-Italic", boldItalic=f"{font_name}-BoldItalic")
if "notocoloremoji-regular" in font_name.lower():
emoji_font_name = font_name
elif "notoemoji" not in font_name.lower():
text_font_names.append(font_name)
except Exception as e:
print(f"Could not register font {font_path.name}: {e}")
if not text_font_names:
print("WARNING: No text fonts found. Adding 'Helvetica' as a default.")
text_font_names.append('Helvetica')
print(f"Successfully registered user-selectable fonts: {text_font_names}")
print(f"Emoji font set to: {emoji_font_name}")
print("--- Font Registration Process Finished ---")
return sorted(text_font_names), emoji_font_name
def render_emoji_as_image(emoji_char, size_pt):
"""
Renders a single emoji character as a transparent PNG image in memory using PyMuPDF.
This is the original raster image engine.
"""
if not EMOJI_FONT_PATH: return None
if (emoji_char, size_pt) in EMOJI_IMAGE_CACHE: return EMOJI_IMAGE_CACHE[(emoji_char, size_pt)]
try:
rect = fitz.Rect(0, 0, size_pt * 1.5, size_pt * 1.5)
doc = fitz.open()
page = doc.new_page(width=rect.width, height=rect.height)
page.insert_font(fontname="emoji", fontfile=EMOJI_FONT_PATH)
page.insert_text(fitz.Point(0, size_pt * 1.1), emoji_char, fontname="emoji", fontsize=size_pt)
pix = page.get_pixmap(alpha=True, dpi=300)
doc.close()
img_buffer = io.BytesIO(pix.tobytes("png"))
img_buffer.seek(0)
EMOJI_IMAGE_CACHE[(emoji_char, size_pt)] = img_buffer
return img_buffer
except Exception as e:
print(f"Could not render emoji {emoji_char} as image: {e}")
return None
def render_emoji_as_svg(emoji_char, size_pt):
"""
Renders a single emoji character as an SVG drawing object.
This is the new vector graphics engine.
"""
if not EMOJI_FONT_PATH: return None
if (emoji_char, size_pt) in EMOJI_SVG_CACHE: return EMOJI_SVG_CACHE[(emoji_char, size_pt)]
try:
doc = fitz.open()
page = doc.new_page(width=1000, height=1000) # Large canvas
page.insert_font(fontname="emoji", fontfile=EMOJI_FONT_PATH)
# Get the SVG image for the character
svg_image = page.get_svg_image(page.rect, text=emoji_char, fontname="emoji")
doc.close()
# Convert the SVG XML string to a ReportLab Graphic
drawing = svg2rlg(io.BytesIO(svg_image.encode('utf-8')))
# Scale the drawing to match the font size
scale_factor = (size_pt * 1.2) / drawing.height
drawing.width *= scale_factor
drawing.height *= scale_factor
drawing.scale(scale_factor, scale_factor)
EMOJI_SVG_CACHE[(emoji_char, size_pt)] = drawing
return drawing
except Exception as e:
print(f"Could not render emoji {emoji_char} as SVG: {e}")
return None
# --- AI Content Generation (Simulation) ---
def generate_ai_content_api(prompt):
"""
Simulates a call to an LLM to generate markdown content.
"""
if not prompt:
return "# The Golem awaits your command!\n\nPlease enter a prompt in the box above and click '🧠 Animate Golem!' to get started. I can help you write reports, stories, poems, and more! ✨"
import time
time.sleep(1.5)
sample_story = f"""
# The Quest for the Sunstone ☀️
Long ago, in the mystical land of Aerthos, a creeping shadow began to fall across the valleys. The once-vibrant flowers 🌸 drooped, and the rivers ran slow. The elders knew the cause: the light of the Sunstone, hidden deep within Mount Cinder, was fading.
## The Prophecy 📜
An ancient prophecy spoke of a hero who would rekindle the stone. It read:
> When darkness drapes the verdant ground,
> A soul of courage shall be found.
> Through trials of fire 🔥, wit, and might,
> They'll bring once more the sacred light.
## The Chosen One 🦸♀️
A young woman named Elara, known for her kindness and bravery, was chosen. She accepted the quest, her only companions a loyal wolf 🐺 and a map gifted by the village shaman.
### The Journey Begins
Elara's journey was fraught with peril. She navigated enchanted forests and crossed treacherous chasms.
| Trial | Location | Challenge |
|------------------|-------------------|----------------------------------------|
| The Whispering | The Gloomwood | Resist maddening whispers of despair |
| The Riddle of | Sphinx Gate | Answer three impossible questions |
| The Fiery Path | The Magma Caverns | Walk barefoot across burning embers |
Finally, she reached the heart of the mountain. There, resting on a pedestal, was the dim Sunstone. Pouring her own hope and courage into it, the stone blazed with renewed life, banishing the shadows from Aerthos forever. The people rejoiced, and Elara became a legend.
"""
return f"# Golem's Vision for: '{prompt}'\n\n{sample_story}"
# --- PDF Generation & Handling ---
def _draw_header_footer(canvas, doc, header_text, footer_text, title):
"""Draws the header and footer on each page."""
canvas.saveState()
page_num = canvas.getPageNumber()
final_footer_text = footer_text.replace("[Page #]", str(page_num)).replace("[Total Pages]", str(doc.page))
final_header_text = header_text.replace("[Page #]", str(page_num)).replace("[Title]", title)
if final_header_text:
canvas.setFont('Helvetica', 9)
canvas.setFillColor(colors.grey)
canvas.drawRightString(doc.width + doc.leftMargin, doc.height + doc.topMargin + 0.25*inch, final_header_text)
if final_footer_text:
canvas.setFont('Helvetica', 9)
canvas.setFillColor(colors.grey)
canvas.drawString(doc.leftMargin, doc.bottomMargin - 0.25*inch, final_footer_text)
canvas.restoreState()
def markdown_to_story(markdown_text: str, font_name: str, emoji_font: str, font_size_body: int, font_size_h1: int, font_size_h2: int, font_size_h3: int, use_svg_engine: bool):
"""Converts markdown to a ReportLab story, with a switch for emoji rendering engines."""
styles = getSampleStyleSheet()
leading_body = font_size_body * 1.4
style_normal = ParagraphStyle('BodyText', fontName=font_name, fontSize=font_size_body, leading=leading_body, spaceAfter=6)
style_h1 = ParagraphStyle('h1', parent=styles['h1'], fontName=font_name, fontSize=font_size_h1, leading=font_size_h1*1.2, spaceBefore=12, textColor=colors.HexColor("#a855f7"))
style_h2 = ParagraphStyle('h2', parent=styles['h2'], fontName=font_name, fontSize=font_size_h2, leading=font_size_h2*1.2, spaceBefore=10, textColor=colors.HexColor("#6366f1"))
style_h3 = ParagraphStyle('h3', parent=styles['h3'], fontName=font_name, fontSize=font_size_h3, leading=font_size_h3*1.2, spaceBefore=8, textColor=colors.HexColor("#3b82f6"))
style_code = ParagraphStyle('Code', fontName='Courier', backColor=colors.HexColor("#333333"), textColor=colors.HexColor("#f472b6"), borderWidth=1, borderColor=colors.HexColor("#444444"), padding=8, leading=12, fontSize=9)
style_table_header = ParagraphStyle('TableHeader', parent=style_normal, fontName=f"{font_name}-Bold" if font_name != 'Helvetica' else 'Helvetica-Bold')
emoji_pattern = re.compile(f"([{re.escape(''.join(map(chr, range(0x1f600, 0x1f650))))}"
f"{re.escape(''.join(map(chr, range(0x1f300, 0x1f5ff))))}"
f"{re.escape(''.join(map(chr, range(0x1f900, 0x1f9ff))))}"
f"{re.escape(''.join(map(chr, range(0x2600, 0x26ff))))}"
f"{re.escape(''.join(map(chr, range(0x2700, 0x27bf))))}]+)")
def create_flowables_for_line(text, style):
"""Splits a line of text into text and emoji flowables, returning a list."""
parts = emoji_pattern.split(text)
flowables = []
for part in parts:
if not part: continue
if emoji_pattern.match(part):
for emoji_char in part:
emoji_flowable = None
if use_svg_engine:
emoji_flowable = render_emoji_as_svg(emoji_char, style.fontSize)
else:
img_buffer = render_emoji_as_image(emoji_char, style.fontSize)
if img_buffer:
emoji_flowable = ReportLabImage(img_buffer, height=style.fontSize * 1.2, width=style.fontSize * 1.2)
if emoji_flowable:
flowables.append(emoji_flowable)
else:
formatted_part = re.sub(r'_(.*?)_', r'\1', re.sub(r'\*\*(.*?)\*\*', r'\1', part))
flowables.append(Paragraph(formatted_part, style))
if not flowables: return [Spacer(0, 0)]
table = Table([flowables], colWidths=[None]*len(flowables))
table.setStyle(TableStyle([('VALIGN', (0,0), (-1,-1), 'MIDDLE'), ('LEFTPADDING', (0,0), (-1,-1), 0), ('RIGHTPADDING', (0,0), (-1,-1), 0)]))
return [table]
story = []
lines = markdown_text.split('\n')
in_code_block, in_table = False, False
code_block_text, table_data = "", []
first_heading = True
document_title = "Untitled Document"
for line in lines:
stripped_line = line.strip()
if stripped_line.startswith("```"):
if in_code_block:
escaped_code = code_block_text.replace('&', '&').replace('<', '<').replace('>', '>')
story.append(Paragraph(escaped_code.replace('\n', '
'), style_code))
story.append(Spacer(1, 0.1 * inch))
in_code_block, code_block_text = False, ""
else: in_code_block = True
continue
if in_code_block:
code_block_text += line + '\n'; continue
if stripped_line.startswith('|'):
if not in_table: in_table = True
if all(c in '-|: ' for c in stripped_line): continue
cells = [cell.strip() for cell in stripped_line.strip('|').split('|')]
table_data.append(cells)
continue
if in_table:
in_table = False
if table_data:
processed_table_data = [[create_flowables_for_line(cell, style_normal)[0] for cell in row] for row in table_data]
header_row = [create_flowables_for_line(cell, style_table_header)[0] for cell in table_data[0]]
final_table_data = [header_row] + processed_table_data[1:]
table = Table(final_table_data, hAlign='LEFT', repeatRows=1)
table.setStyle(TableStyle([
('BACKGROUND', (0, 0), (-1, 0), colors.HexColor("#4a044e")),
('GRID', (0, 0), (-1, -1), 1, colors.HexColor("#6b21a8")),
('VALIGN', (0,0), (-1,-1), 'MIDDLE'),
('TOPPADDING', (0,0), (-1,-1), 6), ('BOTTOMPADDING', (0,0), (-1,-1), 6),
]))
story.append(table); story.append(Spacer(1, 0.2 * inch))
table_data = []
continue
if not stripped_line: continue
content, style, bullet_text = stripped_line, style_normal, None
if stripped_line.startswith("# "):
if not first_heading: story.append(PageBreak())
content, style = stripped_line.lstrip('# '), style_h1
if first_heading: document_title = content
first_heading = False
elif stripped_line.startswith("## "): content, style = stripped_line.lstrip('## '), style_h2
elif stripped_line.startswith("### "): content, style = stripped_line.lstrip('### '), style_h3
elif stripped_line.startswith(("- ", "* ")):
content = stripped_line[2:]
bullet_text = '• '
line_flowables = create_flowables_for_line(content, style)
if bullet_text:
bullet_p = Paragraph(bullet_text, style)
# Adjust colWidths to give bullet a fixed space
list_item_table = Table([[bullet_p] + line_flowables], colWidths=[style.fontSize*1.5] + [None]*len(line_flowables))
list_item_table.setStyle(TableStyle([('VALIGN', (0,0), (-1,-1), 'TOP'), ('LEFTPADDING', (0,0), (-1,-1), 0), ('RIGHTPADDING', (0,0), (-1,-1), 0)]))
story.append(list_item_table)
else:
story.extend(line_flowables)
return story, document_title
def create_pdf_preview(pdf_path: Path):
"""Generates a PNG preview of the first page of a PDF."""
preview_path = PREVIEW_DIR / f"{pdf_path.stem}.png"
try:
doc = fitz.open(pdf_path)
page = doc.load_page(0)
pix = page.get_pixmap(dpi=150)
pix.save(str(preview_path))
doc.close()
return str(preview_path)
except Exception as e:
print(f"Could not create preview for {pdf_path.name}: {e}")
try:
img = Image.new('RGB', (400, 500), color = '#111827')
img.save(str(preview_path))
return str(preview_path)
except: return None
# --- Main API Function ---
def generate_pdfs_api(files, ai_content, layouts, fonts, num_columns, header_text, footer_text, font_size_body, font_size_h1, font_size_h2, font_size_h3, margin_top, margin_bottom, margin_left, margin_right, use_svg_engine, progress=gr.Progress(track_tqdm=True)):
if not files and (not ai_content or "Golem awaits" in ai_content): raise gr.Error("Please conjure some content or upload a file before alchemizing!")
if not layouts: raise gr.Error("You must select a scroll (page layout)!")
if not fonts: raise gr.Error("A scribe needs a font! Please choose one.")
if not EMOJI_FONT_PATH: raise gr.Error("CRITICAL: Cannot generate PDFs. 'NotoColorEmoji-Regular.ttf' not found. Please add it to the app directory.")
shutil.rmtree(OUTPUT_DIR, ignore_errors=True); shutil.rmtree(PREVIEW_DIR, ignore_errors=True)
OUTPUT_DIR.mkdir(); PREVIEW_DIR.mkdir()
image_files, pdf_files, txt_files = [], [], []
if files:
for f in files:
file_path = Path(f.name)
ext = file_path.suffix.lower()
if ext in ['.png', '.jpg', '.jpeg', '.gif', '.bmp', '.tiff']: image_files.append(file_path)
elif ext == '.pdf': pdf_files.append(file_path)
elif ext == '.txt': txt_files.append(file_path)
log_updates = ""
all_text_content = []
if ai_content and "Golem awaits" not in ai_content: all_text_content.append(ai_content)
for txt_path in txt_files:
try:
all_text_content.append(txt_path.read_text(encoding='utf-8'))
except Exception as e:
log_updates += f"⚠️ Failed to read text file {txt_path.name}: {e}\n"
md_content = "\n\n---\n\n".join(all_text_content)
generated_pdf_paths = []
EMOJI_IMAGE_CACHE.clear(); EMOJI_SVG_CACHE.clear()
for layout_name in progress.tqdm(layouts, desc=" brewing potions..."):
for font_name in progress.tqdm(fonts, desc=f" enchanting scrolls with {layout_name}..."):
merger = PdfWriter()
if md_content:
md_buffer = io.BytesIO()
story, title = markdown_to_story(md_content, font_name, EMOJI_FONT_NAME, font_size_body, font_size_h1, font_size_h2, font_size_h3, use_svg_engine)
pagesize = LAYOUTS[layout_name]["size"]
doc = BaseDocTemplate(md_buffer, pagesize=pagesize, leftMargin=margin_left*inch, rightMargin=margin_right*inch, topMargin=margin_top*inch, bottomMargin=margin_bottom*inch)
frame_width = (doc.width / num_columns) - (num_columns - 1) * 0.1*inch
frames = [Frame(doc.leftMargin + i * (frame_width + 0.2*inch), doc.bottomMargin, frame_width, doc.height, id=f'col_{i}') for i in range(num_columns)]
header_footer_callback = partial(_draw_header_footer, header_text=header_text, footer_text=footer_text, title=title)
page_template = PageTemplate(id='main_template', frames=frames, onPage=header_footer_callback)
doc.addPageTemplates([page_template])
doc.build(story)
md_buffer.seek(0)
merger.append(fileobj=md_buffer)
for img_path in image_files:
try:
with Image.open(img_path) as img: img_width_px, img_height_px = img.size
img_width_pt, img_height_pt = img_width_px * (inch / 72), img_height_px * (inch / 72)
img_buffer = io.BytesIO()
img_doc = SimpleDocTemplate(img_buffer, pagesize=(img_width_pt, img_height_pt), leftMargin=0, rightMargin=0, topMargin=0, bottomMargin=0)
img_doc.build([ReportLabImage(img_path, width=img_width_pt, height=img_height_pt)])
img_buffer.seek(0)
merger.append(fileobj=img_buffer)
except Exception as e: log_updates += f"⚠️ Failed to process image {img_path.name}: {e}\n"
for pdf_path in pdf_files:
try:
merger.append(str(pdf_path))
except Exception as e: log_updates += f"⚠️ Failed to merge PDF {pdf_path.name}: {e}\n"
if len(merger.pages) > 0:
time_str = datetime.datetime.now().strftime('%H%M%S')
clean_layout = layout_name.replace(' ', '')
filename = f"Scroll_{clean_layout}_{font_name}_x{num_columns}_{time_str}.pdf"
output_path = OUTPUT_DIR / filename
with open(output_path, "wb") as f: merger.write(f)
generated_pdf_paths.append(output_path)
log_updates += f"✅ Successfully alchemized: {filename}\n"
gallery_previews = [create_pdf_preview(p) for p in generated_pdf_paths]
final_gallery = [g for g in gallery_previews if g is not None]
return final_gallery, log_updates if log_updates else "✨ All scrolls alchemized successfully! ✨", [str(p) for p in generated_pdf_paths]
# --- Gradio UI Definition ---
AVAILABLE_FONTS, EMOJI_FONT_NAME = register_local_fonts()
def get_theme():
"""Dynamically selects a font for the theme to avoid warnings for missing fonts."""
desired_font = "MedievalSharp"
font_family = None
if any(desired_font in s for s in AVAILABLE_FONTS):
font_family = (gr.themes.GoogleFont(desired_font), "ui-sans-serif", "system-ui", "sans-serif")
elif AVAILABLE_FONTS:
first_font = AVAILABLE_FONTS[0]
print(f"WARNING: '{desired_font}' font not found. Using '{first_font}' for UI theme instead.")
font_family = (gr.themes.GoogleFont(first_font), "ui-sans-serif", "system-ui", "sans-serif")
else:
print(f"WARNING: '{desired_font}' font not found and no other fonts available. Using system default.")
font_family = ("ui-sans-serif", "system-ui", "sans-serif")
return gr.themes.Base(primary_hue=gr.themes.colors.purple, secondary_hue=gr.themes.colors.indigo, neutral_hue=gr.themes.colors.gray, font=font_family).set(
body_background_fill="#111827", body_text_color="#d1d5db", button_primary_background_fill="#a855f7", button_primary_text_color="#ffffff",
button_secondary_background_fill="#6366f1", button_secondary_text_color="#ffffff", block_background_fill="#1f2937",
block_label_background_fill="#1f2937", block_title_text_color="#a855f7", input_background_fill="#374151"
)
with gr.Blocks(theme=get_theme(), title="The PDF Alchemist") as demo:
gr.Markdown("# ✨ The PDF Alchemist ✨")
gr.Markdown("A single-page grimoire to turn your ideas into beautifully crafted PDF scrolls. Use the power of AI or upload your own treasures.")
with gr.Row(equal_height=False):
with gr.Column(scale=2):
# --- Accordion Groups for Inputs ---
with gr.Accordion("📜 Content Crucible (Your Ingredients)", open=True):
gr.Markdown("### 🤖 Command Your Idea Golem")
ai_prompt = gr.Textbox(label="Incantation (Prompt)", placeholder="e.g., 'A recipe for a dragon's breath chili...'")
generate_ai_btn = gr.Button("🧠 Animate Golem!")
ai_content_output = gr.Textbox(label="Golem's Manuscript (Editable)", lines=10, interactive=True, value="# The Golem awaits your command!\n\n")
gr.Markdown("
Toggle for higher quality, resolution-independent emojis. May be slower.
*") gr.Markdown("