Vik Paruchuri
commited on
Commit
·
2b17801
1
Parent(s):
ed75e8a
Better support for chunk format
Browse files
marker/scripts/server.py
CHANGED
|
@@ -84,7 +84,9 @@ class CommonParams(BaseModel):
|
|
| 84 |
|
| 85 |
|
| 86 |
async def _convert_pdf(params: CommonParams):
|
| 87 |
-
assert params.output_format in ["markdown", "json", "html"
|
|
|
|
|
|
|
| 88 |
try:
|
| 89 |
options = params.model_dump()
|
| 90 |
config_parser = ConfigParser(options)
|
|
|
|
| 84 |
|
| 85 |
|
| 86 |
async def _convert_pdf(params: CommonParams):
|
| 87 |
+
assert params.output_format in ["markdown", "json", "html", "chunks"], (
|
| 88 |
+
"Invalid output format"
|
| 89 |
+
)
|
| 90 |
try:
|
| 91 |
options = params.model_dump()
|
| 92 |
config_parser = ConfigParser(options)
|
marker/scripts/streamlit_app.py
CHANGED
|
@@ -94,7 +94,7 @@ page_range = st.sidebar.text_input(
|
|
| 94 |
value=f"{page_number}-{page_number}",
|
| 95 |
)
|
| 96 |
output_format = st.sidebar.selectbox(
|
| 97 |
-
"Output format", ["markdown", "json", "html"], index=0
|
| 98 |
)
|
| 99 |
run_marker = st.sidebar.button("Run Marker")
|
| 100 |
|
|
@@ -155,6 +155,8 @@ with col2:
|
|
| 155 |
st.json(text)
|
| 156 |
elif output_format == "html":
|
| 157 |
st.html(text)
|
|
|
|
|
|
|
| 158 |
|
| 159 |
if debug:
|
| 160 |
with col1:
|
|
|
|
| 94 |
value=f"{page_number}-{page_number}",
|
| 95 |
)
|
| 96 |
output_format = st.sidebar.selectbox(
|
| 97 |
+
"Output format", ["markdown", "json", "html", "chunks"], index=0
|
| 98 |
)
|
| 99 |
run_marker = st.sidebar.button("Run Marker")
|
| 100 |
|
|
|
|
| 155 |
st.json(text)
|
| 156 |
elif output_format == "html":
|
| 157 |
st.html(text)
|
| 158 |
+
elif output_format == "chunks":
|
| 159 |
+
st.json(text)
|
| 160 |
|
| 161 |
if debug:
|
| 162 |
with col1:
|