Initial commit
Browse files- .gitignore +1 -0
- README.md +6 -5
- app.py +171 -0
- leaderboard.csv +17 -0
- packages.txt +12 -0
- requirements.txt +2 -0
.gitignore
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
.venv
|
README.md
CHANGED
|
@@ -1,13 +1,14 @@
|
|
| 1 |
---
|
| 2 |
title: Lexoid
|
| 3 |
-
emoji:
|
| 4 |
-
colorFrom:
|
| 5 |
-
colorTo:
|
| 6 |
sdk: gradio
|
| 7 |
-
sdk_version: 5.25.
|
| 8 |
app_file: app.py
|
| 9 |
pinned: false
|
| 10 |
-
|
|
|
|
| 11 |
---
|
| 12 |
|
| 13 |
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|
|
|
|
| 1 |
---
|
| 2 |
title: Lexoid
|
| 3 |
+
emoji: ๐
|
| 4 |
+
colorFrom: yellow
|
| 5 |
+
colorTo: red
|
| 6 |
sdk: gradio
|
| 7 |
+
sdk_version: 5.25.1
|
| 8 |
app_file: app.py
|
| 9 |
pinned: false
|
| 10 |
+
license: apache-2.0
|
| 11 |
+
short_description: Try out Lexoid, an efficient document parsing library.
|
| 12 |
---
|
| 13 |
|
| 14 |
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|
app.py
ADDED
|
@@ -0,0 +1,171 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import gradio as gr
|
| 3 |
+
import pandas as pd
|
| 4 |
+
from lexoid.api import parse
|
| 5 |
+
|
| 6 |
+
parser_options = ["LLM_PARSE", "STATIC_PARSE", "AUTO"]
|
| 7 |
+
|
| 8 |
+
|
| 9 |
+
# Function to set the environment variables and parse the document
|
| 10 |
+
def run_parser(
|
| 11 |
+
file,
|
| 12 |
+
parser_type,
|
| 13 |
+
model,
|
| 14 |
+
pages_per_split,
|
| 15 |
+
max_processes,
|
| 16 |
+
as_pdf,
|
| 17 |
+
x_tolerance,
|
| 18 |
+
y_tolerance,
|
| 19 |
+
save_dir,
|
| 20 |
+
page_nums,
|
| 21 |
+
router_priority,
|
| 22 |
+
framework,
|
| 23 |
+
temperature,
|
| 24 |
+
depth,
|
| 25 |
+
google_api_key,
|
| 26 |
+
openai_api_key,
|
| 27 |
+
huggingfacehub_api_token,
|
| 28 |
+
together_api_key,
|
| 29 |
+
openrouter_api_key,
|
| 30 |
+
):
|
| 31 |
+
# Set environment variables
|
| 32 |
+
os.environ["GOOGLE_API_KEY"] = google_api_key
|
| 33 |
+
os.environ["OPENAI_API_KEY"] = openai_api_key
|
| 34 |
+
os.environ["HUGGINGFACEHUB_API_TOKEN"] = huggingfacehub_api_token
|
| 35 |
+
os.environ["TOGETHER_API_KEY"] = together_api_key
|
| 36 |
+
os.environ["OPENROUTER_API_KEY"] = openrouter_api_key
|
| 37 |
+
|
| 38 |
+
if file is None:
|
| 39 |
+
return "Please upload a file to parse."
|
| 40 |
+
|
| 41 |
+
kwargs = {
|
| 42 |
+
"model": model,
|
| 43 |
+
"pages_per_split": pages_per_split,
|
| 44 |
+
"max_processes": max_processes,
|
| 45 |
+
"as_pdf": as_pdf,
|
| 46 |
+
"x_tolerance": x_tolerance,
|
| 47 |
+
"y_tolerance": y_tolerance,
|
| 48 |
+
"save_dir": save_dir,
|
| 49 |
+
"page_nums": (
|
| 50 |
+
[int(num.strip()) for num in page_nums.split(",")] if page_nums else None
|
| 51 |
+
),
|
| 52 |
+
"router_priority": router_priority,
|
| 53 |
+
"framework": framework,
|
| 54 |
+
"temperature": temperature,
|
| 55 |
+
"depth": depth,
|
| 56 |
+
}
|
| 57 |
+
|
| 58 |
+
# Clean None values
|
| 59 |
+
kwargs = {k: v for k, v in kwargs.items() if v is not None}
|
| 60 |
+
|
| 61 |
+
result = parse(path=file.name, parser_type=parser_type, **kwargs)
|
| 62 |
+
|
| 63 |
+
if "raw" in result:
|
| 64 |
+
return result["raw"]
|
| 65 |
+
elif "segments" in result:
|
| 66 |
+
return "\n\n".join([seg.get("content", "") for seg in result["segments"]])
|
| 67 |
+
else:
|
| 68 |
+
return str(result)
|
| 69 |
+
|
| 70 |
+
|
| 71 |
+
with gr.Blocks(title="Lexoid Document Parser") as app:
|
| 72 |
+
gr.Markdown(
|
| 73 |
+
"## ๐ Lexoid Document Parser\nUpload a document and customize how you'd like to parse it."
|
| 74 |
+
)
|
| 75 |
+
|
| 76 |
+
with gr.Row():
|
| 77 |
+
file_input = gr.File(
|
| 78 |
+
label="Upload Document",
|
| 79 |
+
file_types=[".pdf", ".docx", ".html", ".txt"],
|
| 80 |
+
type="filepath",
|
| 81 |
+
)
|
| 82 |
+
parser_type = gr.Dropdown(
|
| 83 |
+
choices=parser_options, value="AUTO", label="Parser Type"
|
| 84 |
+
)
|
| 85 |
+
model_input = gr.Textbox(value="gemini-2.0-flash", label="LLM Model")
|
| 86 |
+
framework = gr.Textbox(
|
| 87 |
+
value="pdfplumber",
|
| 88 |
+
label="Static Framework",
|
| 89 |
+
placeholder="e.g., pdfplumber, slate",
|
| 90 |
+
)
|
| 91 |
+
|
| 92 |
+
with gr.Accordion("Advanced Options", open=False):
|
| 93 |
+
pages_per_split = gr.Slider(
|
| 94 |
+
minimum=1, maximum=20, value=4, step=1, label="Pages per Split"
|
| 95 |
+
)
|
| 96 |
+
max_processes = gr.Slider(
|
| 97 |
+
minimum=1, maximum=16, value=4, step=1, label="Max Parallel Processes"
|
| 98 |
+
)
|
| 99 |
+
as_pdf = gr.Checkbox(label="Convert to PDF before parsing")
|
| 100 |
+
x_tolerance = gr.Number(label="X-axis Tolerance", value=None)
|
| 101 |
+
y_tolerance = gr.Number(label="Y-axis Tolerance", value=None)
|
| 102 |
+
save_dir = gr.Textbox(
|
| 103 |
+
label="Save Directory",
|
| 104 |
+
placeholder="Path to save intermediate files (optional)",
|
| 105 |
+
)
|
| 106 |
+
page_nums = gr.Textbox(
|
| 107 |
+
label="Page Numbers",
|
| 108 |
+
placeholder="Comma-separated page numbers (e.g., 1,3,5)",
|
| 109 |
+
)
|
| 110 |
+
router_priority = gr.Dropdown(
|
| 111 |
+
choices=["speed", "accuracy"], value="accuracy", label="Router Priority"
|
| 112 |
+
)
|
| 113 |
+
temperature = gr.Number(label="LLM Temperature", value=None)
|
| 114 |
+
depth = gr.Number(label="Recursive Depth", value=None)
|
| 115 |
+
|
| 116 |
+
# Adding the text boxes for the environment variables
|
| 117 |
+
with gr.Row():
|
| 118 |
+
google_api_key = gr.Textbox(
|
| 119 |
+
label="Google API Key", placeholder="Enter Google API Key"
|
| 120 |
+
)
|
| 121 |
+
openai_api_key = gr.Textbox(
|
| 122 |
+
label="OpenAI API Key", placeholder="Enter OpenAI API Key"
|
| 123 |
+
)
|
| 124 |
+
huggingfacehub_api_token = gr.Textbox(
|
| 125 |
+
label="HuggingFaceHub API Token",
|
| 126 |
+
placeholder="Enter HuggingFaceHub API Token",
|
| 127 |
+
)
|
| 128 |
+
together_api_key = gr.Textbox(
|
| 129 |
+
label="Together API Key", placeholder="Enter Together API Key"
|
| 130 |
+
)
|
| 131 |
+
openrouter_api_key = gr.Textbox(
|
| 132 |
+
label="OpenRouter API Key", placeholder="Enter OpenRouter API Key"
|
| 133 |
+
)
|
| 134 |
+
|
| 135 |
+
output = gr.Markdown(label="Parsed Output")
|
| 136 |
+
|
| 137 |
+
parse_button = gr.Button("Parse Document")
|
| 138 |
+
parse_button.click(
|
| 139 |
+
fn=run_parser,
|
| 140 |
+
inputs=[
|
| 141 |
+
file_input,
|
| 142 |
+
parser_type,
|
| 143 |
+
model_input,
|
| 144 |
+
pages_per_split,
|
| 145 |
+
max_processes,
|
| 146 |
+
as_pdf,
|
| 147 |
+
x_tolerance,
|
| 148 |
+
y_tolerance,
|
| 149 |
+
save_dir,
|
| 150 |
+
page_nums,
|
| 151 |
+
router_priority,
|
| 152 |
+
framework,
|
| 153 |
+
temperature,
|
| 154 |
+
depth,
|
| 155 |
+
google_api_key,
|
| 156 |
+
openai_api_key,
|
| 157 |
+
huggingfacehub_api_token,
|
| 158 |
+
together_api_key,
|
| 159 |
+
openrouter_api_key,
|
| 160 |
+
],
|
| 161 |
+
outputs=output,
|
| 162 |
+
)
|
| 163 |
+
|
| 164 |
+
# Leaderboard loaded from leaderboard.csv
|
| 165 |
+
df = pd.read_csv("leaderboard.csv")
|
| 166 |
+
leaderboard = gr.Dataframe(
|
| 167 |
+
value=df,
|
| 168 |
+
label="Leaderboard",
|
| 169 |
+
)
|
| 170 |
+
|
| 171 |
+
app.launch()
|
leaderboard.csv
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Rank,Model,Mean Similarity,Std. Dev.,Time (s),Cost($)
|
| 2 |
+
1,gemini-2.0-flash,0.829,0.102,7.41,0.000480
|
| 3 |
+
2,gemini-2.0-flash-001,0.814,0.176,6.85,0.000421
|
| 4 |
+
3,gemini-1.5-flash,0.797,0.143,9.54,0.000238
|
| 5 |
+
4,gemini-2.0-pro-exp,0.764,0.227,11.95,TBA
|
| 6 |
+
5,gemini-2.0-flash-thinking-exp,0.746,0.266,10.46,TBA
|
| 7 |
+
6,gemini-1.5-pro,0.732,0.265,11.44,0.003332
|
| 8 |
+
7,gpt-4o,0.687,0.247,10.16,0.004736
|
| 9 |
+
8,gpt-4o-mini,0.642,0.213,9.71,0.000275
|
| 10 |
+
9,gemma-3-27b-it (via OpenRouter),0.628,0.299,18.79,0.000096
|
| 11 |
+
10,gemini-1.5-flash-8b,0.551,0.223,3.91,0.000055
|
| 12 |
+
11,Llama-Vision-Free (via Together AI),0.531,0.198,6.93,0
|
| 13 |
+
12,Llama-3.2-11B-Vision-Instruct-Turbo (via Together AI),0.524,0.192,3.68,0.000060
|
| 14 |
+
13,qwen/qwen-2.5-vl-7b-instruct (via OpenRouter),0.482,0.209,11.53,0.000052
|
| 15 |
+
14,Llama-3.2-90B-Vision-Instruct-Turbo (via Together AI),0.461,0.306,19.26,0.000426
|
| 16 |
+
15,Llama-3.2-11B-Vision-Instruct (via Hugging Face),0.451,0.257,4.54,0
|
| 17 |
+
16,microsoft/phi-4-multimodal-instruct (via OpenRouter),0.366,0.287,10.80,0.000019
|
packages.txt
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
libnss3-dev
|
| 2 |
+
libxcomposite1
|
| 3 |
+
libxcursor1
|
| 4 |
+
libxdamage1
|
| 5 |
+
libxi6
|
| 6 |
+
libxtst6
|
| 7 |
+
libnss3
|
| 8 |
+
libxrandr2
|
| 9 |
+
libasound2
|
| 10 |
+
libpangocairo-1.0-0
|
| 11 |
+
libatk1.0-0
|
| 12 |
+
libgtk-3-0
|
requirements.txt
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
lexoid==0.1.12
|
| 2 |
+
matplotlib==3.10.1
|