vlmparse 0.1.3__py3-none-any.whl → 0.1.5__py3-none-any.whl

vlmparse/benchpdf2md/st_visu_benchmark/test_form.py

@@ -1,95 +0,0 @@
-import streamlit as st
-
-
-def edit_test_form(test_obj, test_type):
-    st.markdown("### Edit Test Fields")
-    with st.form("edit_test_fields"):
-        type_fields = {}
-        type_fields["max_diffs"] = st.number_input(
-            "Max Diffs", value=test_obj.max_diffs, min_value=0, step=1
-        )
-        type_fields["unidecode"] = st.checkbox("Unidecode", value=test_obj.unidecode)
-        type_fields["alphanum"] = st.checkbox("Alphanum", value=test_obj.alphanum)
-        type_fields["ignore_str"] = st.text_input(
-            "Ignore strings (seperarated by spaces)",
-            value=" ".join(test_obj.ignore_str),
-        )
-        type_fields["ignore_space"] = st.checkbox(
-            "Ignore space", value=test_obj.ignore_space
-        )
-
-        type_fields["ignore_str"] = (
-            type_fields["ignore_str"].split(" ") if type_fields["ignore_str"] else []
-        )
-
-        if test_type == "present" or test_type == "absent":
-            type_fields["text"] = st.text_area(
-                "Text", value=test_obj.text, height="content"
-            )
-            layout_cat_options = [
-                "text",
-                "footer",
-                "header",
-                "footnote",
-                "image",
-                "image_caption",
-            ]
-
-            type_fields["layout_cat"] = st.selectbox(
-                "Layout Category",
-                layout_cat_options,
-                index=layout_cat_options.index(test_obj.layout_cat),
-            )
-            type_fields["case_sensitive"] = st.checkbox(
-                "Case Sensitive", value=test_obj.case_sensitive
-            )
-            type_fields["first_n"] = st.number_input(
-                "First N",
-                value=test_obj.first_n if test_obj.first_n else 0,
-                min_value=0,
-                step=100,
-            )
-            type_fields["last_n"] = st.number_input(
-                "Last N",
-                value=test_obj.last_n if test_obj.last_n else 0,
-                min_value=0,
-                step=100,
-            )
-            if type_fields["first_n"] == 0:
-                type_fields["first_n"] = None
-            if type_fields["last_n"] == 0:
-                type_fields["last_n"] = None
-        elif test_type == "order":
-            type_fields["before"] = st.text_area(
-                "Before", value=test_obj.before, height="content"
-            )
-            type_fields["after"] = st.text_area(
-                "After", value=test_obj.after, height="content"
-            )
-        elif test_type == "table":
-            type_fields["cell"] = st.text_input("Cell", value=test_obj.cell)
-            type_fields["up"] = st.text_input(
-                "Up", value=test_obj.up if test_obj.up else ""
-            )
-            type_fields["down"] = st.text_input(
-                "Down", value=test_obj.down if test_obj.down else ""
-            )
-            type_fields["left"] = st.text_input(
-                "Left", value=test_obj.left if test_obj.left else ""
-            )
-            type_fields["right"] = st.text_input(
-                "Right", value=test_obj.right if test_obj.right else ""
-            )
-            type_fields["top_heading"] = st.text_input(
-                "Top Heading",
-                value=test_obj.top_heading if test_obj.top_heading else "",
-            )
-            type_fields["left_heading"] = st.text_input(
-                "Left Heading",
-                value=test_obj.left_heading if test_obj.left_heading else "",
-            )
-        if st.form_submit_button("Save Changes"):
-            for field, value in type_fields.items():
-                setattr(test_obj, field, value)
-
-            return test_obj

vlmparse/benchpdf2md/st_visu_benchmark/ui_elements.py

@@ -1,20 +0,0 @@
-from pathlib import Path
-from typing import Optional
-
-import streamlit as st
-
-from vlmparse.benchpdf2md.st_visu_benchmark.utils import get_pdf_bytes
-
-
-def download_pdf_page(
-    pdf_path: Path, page_no: int = 0, file_name: Optional[str] = None
-):
-    pdf_bytes = get_pdf_bytes(pdf_path, page_no)
-    if pdf_bytes:
-        st.download_button(
-            label="📄 Download PDF Page",
-            data=pdf_bytes,
-            file_name=file_name if file_name else f"{pdf_path.stem}.pdf",
-            mime="application/pdf",
-            use_container_width=True,
-        )

vlmparse/benchpdf2md/st_visu_benchmark/utils.py

@@ -1,50 +0,0 @@
-import io
-from pathlib import Path
-
-import pypdfium2 as pdfium
-import streamlit as st
-
-from vlmparse.data_model.document import Document
-
-
-@st.cache_data
-def get_pdf_bytes(pdf_path, page_no=0):
-    pdf_reader = pdfium.PdfDocument(pdf_path)
-    if page_no >= len(pdf_reader):
-        pdf_reader.close()
-        return None
-
-    # Create a new PDF
-    new_pdf = pdfium.PdfDocument.new()
-
-    # Import the chosen page into the new PDF
-    new_pdf.import_pages(pdf_reader, pages=[page_no])
-
-    bytes_io = io.BytesIO()
-    # Get bytes
-    new_pdf.save(bytes_io)
-
-    pdf_bytes = bytes_io.getvalue()
-
-    # Clean up
-    new_pdf.close()
-    pdf_reader.close()
-
-    return pdf_bytes
-
-
-@st.cache_data
-def get_doc(doc_path: Path):
-    return Document.from_zip(doc_path)
-
-
-def save_new_test(tests, test_obj_edited, test_path):
-    from vlmparse.benchpdf2md.bench_tests.benchmark_tsts import save_tests
-
-    for test in tests:
-        if test.id == test_obj_edited.id:
-            test = test_obj_edited
-        else:
-            test = test
-    save_tests(tests, test_path)
-    st.success("Test updated successfully!")

vlmparse/benchpdf2md/utils.py

@@ -1,56 +0,0 @@
-import base64
-from io import BytesIO
-
-import numpy as np
-import pandas as pd
-from PIL import Image
-
-
-def vectorized_bootstrap_grouped_std(df, group_col, value_col, n_bootstrap=1000):
-    group_col = [group_col] if isinstance(group_col, str) else group_col
-    grouped = df.groupby(group_col)[value_col]
-
-    def bootstrap_group(group):
-        values = group.values
-        n = len(values)
-        bootstrap_samples = np.random.choice(
-            values, size=(n_bootstrap, n), replace=True
-        )
-        bootstrap_means = np.mean(bootstrap_samples, axis=1)
-        return pd.Series(
-            {"mean": np.mean(values), "bootstrap_std": np.std(bootstrap_means)}
-        )
-
-    result = grouped.apply(bootstrap_group)
-    return result.unstack(-1)
-
-
-def format_results_vectorized(result_df, precision=2):
-    means = result_df["mean"].values
-    margins = 2 * result_df["bootstrap_std"].values
-
-    formatted = np.char.add(
-        np.char.add(np.round(means, precision).astype(str), " ± "),
-        np.round(margins, precision).astype(str),
-    )
-
-    return pd.DataFrame({"formatted_result": formatted}, index=result_df.index)
-
-
-def bootstrap_and_format_results(
-    df, group_col, value_col, n_bootstrap=1000, precision=2
-):
-    result_df = vectorized_bootstrap_grouped_std(df, group_col, value_col, n_bootstrap)
-    return format_results_vectorized(result_df, precision)
-
-
-def to_base64(image: Image, extension="PNG"):
-    img_byte_arr = BytesIO()
-    image.save(img_byte_arr, format=extension)
-    img_byte_arr = img_byte_arr.getvalue()
-    return base64.b64encode(img_byte_arr).decode("utf-8")
-
-
-def from_base64(base64_str: str):
-    image_data = base64.b64decode(base64_str)
-    return Image.open(BytesIO(image_data))

vlmparse-0.1.3.dist-info/RECORD

@@ -1,50 +0,0 @@
-vlmparse/base_model.py,sha256=4U4UPe8SNArliKnUf8pp8zQugWYsnhg9okylt7mrW1U,381
-vlmparse/build_doc.py,sha256=9evdU6GwVAZU15dZ1Qght6hNo_QxBQN8X3gmYdU2ltg,1965
-vlmparse/cli.py,sha256=tQma1IkOsFnqPKqqHVO1PJh18n1w82gp4ewA7oraJkE,15855
-vlmparse/converter.py,sha256=5wTA_cFyDMDSY8YgLzZV9SVBKmHjEbJCW8KPoJjmVVA,6880
-vlmparse/converter_with_server.py,sha256=9yoqfv8akB0xZZ7Snjq3aHW5NPNam2AgbK7_rfFqNkk,3909
-vlmparse/registries.py,sha256=TdSR1fx1Tz3roGk4Tk5ckIK6Iz-e4UD4erWUk96fFpQ,5846
-vlmparse/utils.py,sha256=jZWbNMwpZSZL--ZzvL8wPG_7mwpw9Pi36qTO9TjvHZU,1239
-vlmparse/benchpdf2md/create_dataset.py,sha256=0o4I0O3pHm1W7NYOTnW1JvPmgxJM8KLElKFvAbPAIic,1855
-vlmparse/benchpdf2md/run_benchmark.py,sha256=LMHElWyWIgB4ppBL0s-qjfMz5FZQnZOEm5mXxd0p0C8,9800
-vlmparse/benchpdf2md/utils.py,sha256=Q62vtvLIzxOEzSi-w210d7qnaRz-q_5ykmLNTkmbs-8,1732
-vlmparse/benchpdf2md/bench_tests/benchmark_tsts.py,sha256=XISZKlN1CiPBGDHgiDkjFkhGWxPxMFRu1GfStiOprmo,69527
-vlmparse/benchpdf2md/bench_tests/utils.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-vlmparse/benchpdf2md/olmocrbench/repeatdetect.py,sha256=SWCfYgAVmtyNypCsEZnuj6HoNGjFjTOqPBn2fCnEsiQ,5509
-vlmparse/benchpdf2md/olmocrbench/run_olmocr_bench.py,sha256=XnVlN5Sy7mXkr7nqWbcCgROsoavFBXqn4haSoxVsyLM,8495
-vlmparse/benchpdf2md/olmocrbench/tests.py,sha256=bszavrGQNCB3l60gPrsMgxtcMJvDjwZBBL_0B_K1oFA,53625
-vlmparse/benchpdf2md/olmocrbench/katex/__init__.py,sha256=DD9Knd52Ur9WiatymQGI9B1-yJ7OkL2w5E0quITDkPY,72
-vlmparse/benchpdf2md/olmocrbench/katex/render.py,sha256=l8mYRnz2I-10RE3JoxYrMfZ2UhxECTiRFazuf-dHnqU,18219
-vlmparse/benchpdf2md/st_visu_benchmark/app.py,sha256=ng8o_q20rY88BFQSEi62dBylklV6yyGyrrMiTrl7xtM,8921
-vlmparse/benchpdf2md/st_visu_benchmark/highligh_text.py,sha256=vsGtd9nPrtL5uDqALMumVlsllLvVZr2E6_d9TOLQoN4,3915
-vlmparse/benchpdf2md/st_visu_benchmark/test_form.py,sha256=qNmFZoSdbWcw1EJKesgO7rYV_LpslYPzij_gyJEuDP8,3701
-vlmparse/benchpdf2md/st_visu_benchmark/ui_elements.py,sha256=WkKncexShO3SU-DO7dPT4DOe-8UNjsCaHlj9L1B2mkI,572
-vlmparse/benchpdf2md/st_visu_benchmark/utils.py,sha256=JSmOJQY1DDETtWmjWv07SlQlORE6yBewiMcE5qRZI_Q,1109
-vlmparse/clients/chandra.py,sha256=zfu-A6Slh-fIAyrtrlVoCb6QHLBimnimefap_K9YwYw,9775
-vlmparse/clients/deepseekocr.py,sha256=iCG5wI5yPv98hIPgVJX4gkkkH1OekblZjFhh5ORVWAk,1813
-vlmparse/clients/docling.py,sha256=K-Grl_nZiSdooEdEaflevprE56l3Keby9xSMBtFwdis,5355
-vlmparse/clients/dotsocr.py,sha256=9ygvIVVOi9UhTUJwmrI-h6AjMV9vL9J2vMaBfUyTorY,9895
-vlmparse/clients/granite_docling.py,sha256=EQpsv5qSJG0HtMSacmJStER2sq4TGf1EMU5_NmJsl4g,4634
-vlmparse/clients/hunyuanocr.py,sha256=Xw0Q1l-3pQzaEgFngnfM8vrSWpnT3I99QvDaGZ8XooM,1712
-vlmparse/clients/lightonocr.py,sha256=wx1Im8Z3wlRWwYbPqnSd3LqTtdAU8CnX5mzu1BuCUY8,1314
-vlmparse/clients/mineru.py,sha256=OL56O6cMGksoVMVDyMYCQvwIvZEjKiAZXjEMEshqXaY,3549
-vlmparse/clients/nanonetocr.py,sha256=BT5vaeerCsK5agvOaHK3NvLUqWd1FfDmrMmDYbp646I,1543
-vlmparse/clients/olmocr.py,sha256=mQEDpfyLY8a80Zlps5mG0QaWytIgnNQZVEVWKWjPIjk,1849
-vlmparse/clients/openai_converter.py,sha256=QBHomw3_K_BpMVboaMd4rlhrntsqEFT46kxpg9KqBIs,5636
-vlmparse/clients/paddleocrvl.py,sha256=tmaqg3boV4edywiiiNiNiI3dBHi111wz4dFb52OISXw,1376
-vlmparse/clients/prompts.py,sha256=-J60lqxgRzlkQ9VsQLxmWsIMaDt-gNqWqWoqHIw9CLc,4228
-vlmparse/clients/pipe_utils/cleaner.py,sha256=oxBkBTOkluN1lmeNbzajRIe0_D__ZGwUOBaI_Ph0uxE,2396
-vlmparse/clients/pipe_utils/html_to_md_conversion.py,sha256=cFFqzD2jCNw_968_eu3Wt--Ox7iJj2Rn5UoP_DZWosU,4112
-vlmparse/clients/pipe_utils/utils.py,sha256=935ecIO446I0pstszE_1nrIPHn1Ffrxunq7fVd0dsd8,315
-vlmparse/data_model/box.py,sha256=lJsh4qhjgYXZF5vTSJ1qMXD5GVlBi2_SBedBMlfJikU,16868
-vlmparse/data_model/document.py,sha256=pdCZvWzRFkez53ZJpNaB4ezUW-OVUlbR3_SBmmgVzGQ,4217
-vlmparse/servers/docker_server.py,sha256=nI7K8CEzJwSZxLY7Jg9IuYHHLR5YQpOSgY8Ln71POJA,6608
-vlmparse/servers/utils.py,sha256=gMk5Y8FA1nlSxi7JzKxZu7XyljkYUZ5AnsTb3YFqu28,8821
-vlmparse/st_viewer/fs_nav.py,sha256=7GNH68h2Loh5pQ64Pe72-D2cs2BLhqRXevEmKdFmPX0,1616
-vlmparse/st_viewer/st_viewer.py,sha256=m2rQTtk5rlwErNmivNAg-4rkHkvNkvLhoJZxFQi7Dwk,2105
-vlmparse-0.1.3.dist-info/licenses/LICENSE,sha256=3TKJHk8hPBR5dbLWZ3IpfCftl-_m-iyBwpYQGZYxj14,1080
-vlmparse-0.1.3.dist-info/METADATA,sha256=JkSI4uFnnF59WReyhfRFZZVoe6KLk0ZJrjG0FQkUIPI,5112
-vlmparse-0.1.3.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-vlmparse-0.1.3.dist-info/entry_points.txt,sha256=gD5berP6HwE2wNIkls-Lw5goiceA8uMgPEd7ifnFJXs,47
-vlmparse-0.1.3.dist-info/top_level.txt,sha256=k4ni-GNH_iAX7liQEsk_KY_c3xgZgt8k9fsSs9IXLXs,9
-vlmparse-0.1.3.dist-info/RECORD,,