Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -8,6 +8,68 @@ import shutil
|
|
| 8 |
|
| 9 |
app = FastAPI()
|
| 10 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
@app.post("/extract")
|
| 12 |
async def extract(file: UploadFile = File(...)):
|
| 13 |
pdf_bytes = await file.read()
|
|
@@ -64,6 +126,57 @@ async def convert_to_markdown(file: UploadFile = File(...)):
|
|
| 64 |
if os.path.exists(temp_path):
|
| 65 |
os.remove(temp_path)
|
| 66 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 67 |
@app.get("/health")
|
| 68 |
async def health():
|
| 69 |
return {
|
|
|
|
| 8 |
|
| 9 |
app = FastAPI()
|
| 10 |
|
| 11 |
+
|
| 12 |
+
EPUB_CSS = """
|
| 13 |
+
body {
|
| 14 |
+
font-family: serif;
|
| 15 |
+
line-height: 1.5;
|
| 16 |
+
margin: 5%;
|
| 17 |
+
color: #e0e0e0;
|
| 18 |
+
background-color: #1a1a1a;
|
| 19 |
+
}
|
| 20 |
+
h1 {
|
| 21 |
+
text-align: center;
|
| 22 |
+
color: #f4a261;
|
| 23 |
+
text-transform: uppercase;
|
| 24 |
+
margin-bottom: 0.2em;
|
| 25 |
+
}
|
| 26 |
+
h3 {
|
| 27 |
+
border-bottom: 1px solid #333;
|
| 28 |
+
padding-bottom: 5px;
|
| 29 |
+
margin-top: 30px;
|
| 30 |
+
color: #e76f51;
|
| 31 |
+
}
|
| 32 |
+
blockquote {
|
| 33 |
+
font-style: italic;
|
| 34 |
+
border-left: 3px solid #e76f51;
|
| 35 |
+
padding-left: 15px;
|
| 36 |
+
color: #b0b0b0;
|
| 37 |
+
margin: 1.5em 10px;
|
| 38 |
+
}
|
| 39 |
+
li {
|
| 40 |
+
margin-bottom: 8px;
|
| 41 |
+
}
|
| 42 |
+
table {
|
| 43 |
+
width: 100%;
|
| 44 |
+
border-collapse: separate;
|
| 45 |
+
border-spacing: 0;
|
| 46 |
+
margin: 20px 0;
|
| 47 |
+
border: 1px solid #333;
|
| 48 |
+
border-radius: 8px;
|
| 49 |
+
overflow: hidden;
|
| 50 |
+
}
|
| 51 |
+
th {
|
| 52 |
+
background-color: #2d2d2d;
|
| 53 |
+
color: #f4a261;
|
| 54 |
+
font-weight: bold;
|
| 55 |
+
text-align: left;
|
| 56 |
+
padding: 12px;
|
| 57 |
+
border-bottom: 2px solid #3d3d3d;
|
| 58 |
+
}
|
| 59 |
+
td {
|
| 60 |
+
padding: 10px 12px;
|
| 61 |
+
border-bottom: 1px solid #2d2d2d;
|
| 62 |
+
vertical-align: top;
|
| 63 |
+
font-size: 0.95em;
|
| 64 |
+
}
|
| 65 |
+
tr:last-child td {
|
| 66 |
+
border-bottom: none;
|
| 67 |
+
}
|
| 68 |
+
tr:nth-child(even) {
|
| 69 |
+
background-color: #222222;
|
| 70 |
+
}
|
| 71 |
+
"""
|
| 72 |
+
|
| 73 |
@app.post("/extract")
|
| 74 |
async def extract(file: UploadFile = File(...)):
|
| 75 |
pdf_bytes = await file.read()
|
|
|
|
| 126 |
if os.path.exists(temp_path):
|
| 127 |
os.remove(temp_path)
|
| 128 |
|
| 129 |
+
@app.post("/export-epub")
|
| 130 |
+
async def export_epub(file: UploadFile = File(...)):
|
| 131 |
+
base_name = os.path.splitext(file.filename)[0]
|
| 132 |
+
temp_input = f"temp_{file.filename}"
|
| 133 |
+
temp_css = "style.css"
|
| 134 |
+
output_epub = f"{base_name}.epub"
|
| 135 |
+
|
| 136 |
+
try:
|
| 137 |
+
# 1. Save the upload
|
| 138 |
+
with open(temp_input, "wb") as buffer:
|
| 139 |
+
content = await file.read()
|
| 140 |
+
buffer.write(content)
|
| 141 |
+
|
| 142 |
+
# 2. Use MarkItDown to get the Markdown content
|
| 143 |
+
result = md.convert(temp_input)
|
| 144 |
+
markdown_content = result.text_content
|
| 145 |
+
|
| 146 |
+
# 3. Create a temporary CSS file for Pandoc
|
| 147 |
+
with open(temp_css, "w") as f:
|
| 148 |
+
f.write(EPUB_CSS)
|
| 149 |
+
|
| 150 |
+
# 4. Call Pandoc to convert Markdown string to EPUB
|
| 151 |
+
# We pipe the markdown_content directly into pandoc's stdin
|
| 152 |
+
process = subprocess.Popen(
|
| 153 |
+
['pandoc', '--from=markdown', '--to=epub', '--css', temp_css,
|
| 154 |
+
'--metadata', f'title={base_name}', '-o', output_epub],
|
| 155 |
+
stdin=subprocess.PIPE,
|
| 156 |
+
stderr=subprocess.PIPE,
|
| 157 |
+
text=True
|
| 158 |
+
)
|
| 159 |
+
stdout, stderr = process.communicate(input=markdown_content)
|
| 160 |
+
|
| 161 |
+
if process.returncode != 0:
|
| 162 |
+
raise Exception(f"Pandoc Error: {stderr}")
|
| 163 |
+
|
| 164 |
+
# 5. Return the generated EPUB file
|
| 165 |
+
return FileResponse(
|
| 166 |
+
path=output_epub,
|
| 167 |
+
filename=output_epub,
|
| 168 |
+
media_type='application/epub+zip'
|
| 169 |
+
)
|
| 170 |
+
|
| 171 |
+
except Exception as e:
|
| 172 |
+
raise HTTPException(status_code=500, detail=str(e))
|
| 173 |
+
|
| 174 |
+
finally:
|
| 175 |
+
# Cleanup
|
| 176 |
+
for path in [temp_input, temp_css]:
|
| 177 |
+
if os.path.exists(path):
|
| 178 |
+
os.remove(path)
|
| 179 |
+
|
| 180 |
@app.get("/health")
|
| 181 |
async def health():
|
| 182 |
return {
|