Upload app.py with huggingface_hub
Browse files
app.py
CHANGED
|
@@ -97,7 +97,7 @@ def _prune_downloads(max_age_seconds: int = 3600) -> None:
|
|
| 97 |
pass
|
| 98 |
|
| 99 |
|
| 100 |
-
def
|
| 101 |
if not markdown or not markdown.strip():
|
| 102 |
return None
|
| 103 |
try:
|
|
@@ -177,7 +177,7 @@ def _run_harness(text: str | None, num_candidates: int | None, structured_fallba
|
|
| 177 |
|
| 178 |
|
| 179 |
@spaces.GPU(duration=_estimate_duration)
|
| 180 |
-
def convert(text: str | None = None, num_candidates: int | None = None, structured_fallback: bool | None = None, max_output_tokens: int | None = None) -> tuple[str, str, str, str]:
|
| 181 |
"""Convert a plain-text document into faithful GitHub-Flavored Markdown.
|
| 182 |
|
| 183 |
PASITA samples the document several times, every candidate is post-processed and
|
|
@@ -191,11 +191,12 @@ def convert(text: str | None = None, num_candidates: int | None = None, structur
|
|
| 191 |
structured_fallback: Use the deterministic converter for HTML and delimited data.
|
| 192 |
max_output_tokens: Optional cap on generated tokens. When omitted, the length is chosen automatically.
|
| 193 |
Returns:
|
| 194 |
-
A tuple of the rendered Markdown, the raw Markdown, a status line
|
|
|
|
| 195 |
"""
|
| 196 |
text = (text or "").strip()
|
| 197 |
if not text:
|
| 198 |
-
return EMPTY_MD, "", _status("READY / PASTE A DOCUMENT TO CONVERT"), DETAILS_EMPTY
|
| 199 |
if len(text) > MAX_INPUT_CHARS:
|
| 200 |
message = f"Input too long ({len(text):,} chars). The limit is {MAX_INPUT_CHARS:,} characters."
|
| 201 |
return (
|
|
@@ -203,6 +204,7 @@ def convert(text: str | None = None, num_candidates: int | None = None, structur
|
|
| 203 |
"",
|
| 204 |
_status("ERROR / INPUT TOO LONG", "error"),
|
| 205 |
_details_md(None, 0.0, error=message),
|
|
|
|
| 206 |
)
|
| 207 |
|
| 208 |
try:
|
|
@@ -219,6 +221,7 @@ def convert(text: str | None = None, num_candidates: int | None = None, structur
|
|
| 219 |
"",
|
| 220 |
_status(f"ERROR / {err_type}", "error"),
|
| 221 |
_details_md(None, 0.0, error=f"{plain} Technical detail: {err_type}: {exc}"),
|
|
|
|
| 222 |
)
|
| 223 |
|
| 224 |
metrics = result["metrics"]
|
|
@@ -234,7 +237,7 @@ def convert(text: str | None = None, num_candidates: int | None = None, structur
|
|
| 234 |
detail = f"{source_label} / {numbers} / {overlap} / {taken:.1f}S"
|
| 235 |
if result.get("strategy", {}).get("truncated_input"):
|
| 236 |
detail += " / INPUT TRUNCATED TO 2048 TOKENS"
|
| 237 |
-
return result["markdown"], result["markdown"], _status("DONE", "ready", detail), _details_md(result, taken)
|
| 238 |
|
| 239 |
|
| 240 |
@spaces.GPU(duration=_estimate_duration)
|
|
@@ -458,8 +461,8 @@ def _extract_user_text(messages: list[ChatMessage]) -> str | None:
|
|
| 458 |
return None
|
| 459 |
|
| 460 |
|
| 461 |
-
def reset() -> tuple[str, str, str, str, str]:
|
| 462 |
-
return EMPTY_MD, "", _status("READY"), DETAILS_EMPTY, _counter("")
|
| 463 |
|
| 464 |
|
| 465 |
@server.post("/v1/chat/completions", response_model=None)
|
|
@@ -528,12 +531,12 @@ def chat_completions(body: ChatCompletionRequest, request: fastapi.Request) -> d
|
|
| 528 |
}
|
| 529 |
|
| 530 |
|
| 531 |
-
BUSY_JS = """(text, num_candidates, structured_fallback, md, source, status, details) => {
|
| 532 |
const n = Number(num_candidates) || 1;
|
| 533 |
const label = n > 1 ? 'PASITA ' + n + ' SAMPLES / RERANK' : 'PASITA SAMPLE / RERANK';
|
| 534 |
return ['_Converting…_', '',
|
| 535 |
'<div class="p-status busy"><span class="p-dot"></span><span class="p-text">CONVERTING</span><span class="p-detail">' + label + '</span></div>',
|
| 536 |
-
'_Checking drafts…_'];
|
| 537 |
}"""
|
| 538 |
|
| 539 |
COUNTER_JS = """(text) => {
|
|
@@ -893,13 +896,13 @@ with gr.Blocks(title="PASITA: plain text to Markdown") as demo:
|
|
| 893 |
details = gr.Markdown(DETAILS_EMPTY, elem_id="p-details")
|
| 894 |
with gr.Row(elem_classes=["p-out-head"]):
|
| 895 |
out_label = gr.HTML('<span class="p-label">MARKDOWN</span>')
|
| 896 |
-
dl = gr.DownloadButton("DOWNLOAD .MD",
|
| 897 |
|
| 898 |
gr.Examples(
|
| 899 |
examples=EXAMPLES,
|
| 900 |
example_labels=["ES REPORT", "EN MEETING", "HTML", "CSV"],
|
| 901 |
inputs=[text_in],
|
| 902 |
-
outputs=[md_out, source, status, details],
|
| 903 |
fn=convert,
|
| 904 |
cache_examples=True,
|
| 905 |
cache_mode="lazy",
|
|
@@ -912,7 +915,7 @@ with gr.Blocks(title="PASITA: plain text to Markdown") as demo:
|
|
| 912 |
run.click(
|
| 913 |
convert,
|
| 914 |
inputs=[text_in, candidates, structured],
|
| 915 |
-
outputs=[md_out, source, status, details],
|
| 916 |
api_name="convert",
|
| 917 |
js=BUSY_JS,
|
| 918 |
show_progress="minimal",
|
|
@@ -920,13 +923,13 @@ with gr.Blocks(title="PASITA: plain text to Markdown") as demo:
|
|
| 920 |
text_in.submit(
|
| 921 |
convert,
|
| 922 |
inputs=[text_in, candidates, structured],
|
| 923 |
-
outputs=[md_out, source, status, details],
|
| 924 |
api_visibility="undocumented",
|
| 925 |
js=BUSY_JS,
|
| 926 |
show_progress="minimal",
|
| 927 |
)
|
| 928 |
text_in.change(None, inputs=[text_in], outputs=[label_in], js=COUNTER_JS, api_visibility="undocumented")
|
| 929 |
-
clear.click(reset, inputs=None, outputs=[md_out, source, status, details, label_in], api_visibility="undocumented")
|
| 930 |
|
| 931 |
demo.queue()
|
| 932 |
demo.launch(_app=server, theme=theme, css=CSS, mcp_server=True)
|
|
|
|
| 97 |
pass
|
| 98 |
|
| 99 |
|
| 100 |
+
def _write_download(markdown: str | None) -> str | None:
|
| 101 |
if not markdown or not markdown.strip():
|
| 102 |
return None
|
| 103 |
try:
|
|
|
|
| 177 |
|
| 178 |
|
| 179 |
@spaces.GPU(duration=_estimate_duration)
|
| 180 |
+
def convert(text: str | None = None, num_candidates: int | None = None, structured_fallback: bool | None = None, max_output_tokens: int | None = None) -> tuple[str, str, str, str, str | None]:
|
| 181 |
"""Convert a plain-text document into faithful GitHub-Flavored Markdown.
|
| 182 |
|
| 183 |
PASITA samples the document several times, every candidate is post-processed and
|
|
|
|
| 191 |
structured_fallback: Use the deterministic converter for HTML and delimited data.
|
| 192 |
max_output_tokens: Optional cap on generated tokens. When omitted, the length is chosen automatically.
|
| 193 |
Returns:
|
| 194 |
+
A tuple of the rendered Markdown, the raw Markdown, a status line, run details
|
| 195 |
+
and a download path for the Markdown file (or None).
|
| 196 |
"""
|
| 197 |
text = (text or "").strip()
|
| 198 |
if not text:
|
| 199 |
+
return EMPTY_MD, "", _status("READY / PASTE A DOCUMENT TO CONVERT"), DETAILS_EMPTY, None
|
| 200 |
if len(text) > MAX_INPUT_CHARS:
|
| 201 |
message = f"Input too long ({len(text):,} chars). The limit is {MAX_INPUT_CHARS:,} characters."
|
| 202 |
return (
|
|
|
|
| 204 |
"",
|
| 205 |
_status("ERROR / INPUT TOO LONG", "error"),
|
| 206 |
_details_md(None, 0.0, error=message),
|
| 207 |
+
None,
|
| 208 |
)
|
| 209 |
|
| 210 |
try:
|
|
|
|
| 221 |
"",
|
| 222 |
_status(f"ERROR / {err_type}", "error"),
|
| 223 |
_details_md(None, 0.0, error=f"{plain} Technical detail: {err_type}: {exc}"),
|
| 224 |
+
None,
|
| 225 |
)
|
| 226 |
|
| 227 |
metrics = result["metrics"]
|
|
|
|
| 237 |
detail = f"{source_label} / {numbers} / {overlap} / {taken:.1f}S"
|
| 238 |
if result.get("strategy", {}).get("truncated_input"):
|
| 239 |
detail += " / INPUT TRUNCATED TO 2048 TOKENS"
|
| 240 |
+
return result["markdown"], result["markdown"], _status("DONE", "ready", detail), _details_md(result, taken), _write_download(result["markdown"])
|
| 241 |
|
| 242 |
|
| 243 |
@spaces.GPU(duration=_estimate_duration)
|
|
|
|
| 461 |
return None
|
| 462 |
|
| 463 |
|
| 464 |
+
def reset() -> tuple[str, str, str, str, str | None, str]:
|
| 465 |
+
return EMPTY_MD, "", _status("READY"), DETAILS_EMPTY, None, _counter("")
|
| 466 |
|
| 467 |
|
| 468 |
@server.post("/v1/chat/completions", response_model=None)
|
|
|
|
| 531 |
}
|
| 532 |
|
| 533 |
|
| 534 |
+
BUSY_JS = """(text, num_candidates, structured_fallback, md, source, status, details, dl) => {
|
| 535 |
const n = Number(num_candidates) || 1;
|
| 536 |
const label = n > 1 ? 'PASITA ' + n + ' SAMPLES / RERANK' : 'PASITA SAMPLE / RERANK';
|
| 537 |
return ['_Converting…_', '',
|
| 538 |
'<div class="p-status busy"><span class="p-dot"></span><span class="p-text">CONVERTING</span><span class="p-detail">' + label + '</span></div>',
|
| 539 |
+
'_Checking drafts…_', null];
|
| 540 |
}"""
|
| 541 |
|
| 542 |
COUNTER_JS = """(text) => {
|
|
|
|
| 896 |
details = gr.Markdown(DETAILS_EMPTY, elem_id="p-details")
|
| 897 |
with gr.Row(elem_classes=["p-out-head"]):
|
| 898 |
out_label = gr.HTML('<span class="p-label">MARKDOWN</span>')
|
| 899 |
+
dl = gr.DownloadButton("DOWNLOAD .MD", size="sm", elem_id="p-dl")
|
| 900 |
|
| 901 |
gr.Examples(
|
| 902 |
examples=EXAMPLES,
|
| 903 |
example_labels=["ES REPORT", "EN MEETING", "HTML", "CSV"],
|
| 904 |
inputs=[text_in],
|
| 905 |
+
outputs=[md_out, source, status, details, dl],
|
| 906 |
fn=convert,
|
| 907 |
cache_examples=True,
|
| 908 |
cache_mode="lazy",
|
|
|
|
| 915 |
run.click(
|
| 916 |
convert,
|
| 917 |
inputs=[text_in, candidates, structured],
|
| 918 |
+
outputs=[md_out, source, status, details, dl],
|
| 919 |
api_name="convert",
|
| 920 |
js=BUSY_JS,
|
| 921 |
show_progress="minimal",
|
|
|
|
| 923 |
text_in.submit(
|
| 924 |
convert,
|
| 925 |
inputs=[text_in, candidates, structured],
|
| 926 |
+
outputs=[md_out, source, status, details, dl],
|
| 927 |
api_visibility="undocumented",
|
| 928 |
js=BUSY_JS,
|
| 929 |
show_progress="minimal",
|
| 930 |
)
|
| 931 |
text_in.change(None, inputs=[text_in], outputs=[label_in], js=COUNTER_JS, api_visibility="undocumented")
|
| 932 |
+
clear.click(reset, inputs=None, outputs=[md_out, source, status, details, dl, label_in], api_visibility="undocumented")
|
| 933 |
|
| 934 |
demo.queue()
|
| 935 |
demo.launch(_app=server, theme=theme, css=CSS, mcp_server=True)
|