developer-tools

liteparse

Name: liteparse
Availability: InStock
Author: AI Demos

Open-source PDF-to-markdown parsing that works well on native-digital reports, but degrades on scans, tables, and charts.

Visit liteparse

Native PDFsScanned OCR degradesTables fragileTOC preserved

TL;DR — our verdictUpdated July 2026 · 31 test artifacts

Strong on clean reports, weak on scans and complex visuals

Where it wins

You need a self-hosted, open-source PDF-to-markdown parser for native-digital reports.
You care most about preserving prose, headings, reading order, and table-of-contents structure on clean PDFs.
You can tolerate image placeholders and some cleanup in tables.

Main limitation

Your PDFs are mostly scanned or OCR-heavy.

Pricing (verified plans)

Free $0

Strongest test artifacts

Text Extraction →Table Extraction and Reconstruction →Document Hierarchy Preservation →

Our take

LiteParse is a strong self-hosted parser for native-digital reports: it keeps prose, headings, reading order, and table-of-contents structure in good shape, and it at least preserves image placeholders instead of silently dropping assets on clean PDFs. The tradeoff is sharp: table structure is brittle, charts and figures are not faithfully reconstructed, and scanned layouts push OCR and image retention into clearly degraded territory.

General walkthrough demo from the financial PDF test.

In-Depth Review

Our detailed analysis of liteparse — features, performance, and real-world testing.

AI Demos Team

Expert Reviewer

Verified Review

Feature-by-Feature Breakdown

Text Extraction

Strong on native-digital prose; scanned text becomes noisy on complex layouts.

▾

Test Summary

Feature tested: Text Extraction

Result: Partial — Strong on native-digital prose; scanned text becomes noisy on complex layouts.

Feature tested: Text Extraction

Result: Partial

Verdict: Strong on native-digital prose; scanned text becomes noisy on complex layouts.

Expected behavior: LiteParse extracts readable prose from native-digital report pages, including long paragraphs, headings, and disclaimer copy, and can also OCR scanned research-paper pages with reduced reliability. The same capability is exercised on both the native reports and the scanned paper snippets, where OCR introduces substitutions, word splitting, and clipped lines.

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input1_hybridearnings_ceo_letter.png

Observed output: Output artifact (Image): The CEO letter prose is extracted cleanly with the opening narrative and bullet points intact, while the original page styling is dropped. — liteparse_input1_hybridearnings_ceo_letter_parsed_success.png

Input artifact: Input artifact (Image): Input — liteparse_input1_hybridearnings_ceo_letter.png

Output artifact: Output artifact (Image): The CEO letter prose is extracted cleanly with the opening narrative and bullet points intact, while the original page styling is dropped. — liteparse_input1_hybridearnings_ceo_letter_parsed_success.png

What changed: Image transformed into Image

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input2_financialpdf_disclaimer_text.png

Observed output: Output artifact (Image): The Japanese GAAP disclaimer is extracted cleanly as a single readable paragraph with the translation and rounding note preserved. — liteparse_input2_financialpdf_disclaimer_text_parsed_success.png

Input artifact: Input artifact (Image): Input — liteparse_input2_financialpdf_disclaimer_text.png

Output artifact: Output artifact (Image): The Japanese GAAP disclaimer is extracted cleanly as a single readable paragraph with the translation and rounding note preserved. — liteparse_input2_financialpdf_disclaimer_text_parsed_success.png

What changed: Image transformed into Image

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input3_scannedpaper_superscript.png

Observed output: Output artifact (Image): A narrow scanned-paper snippet loses superscript formatting and substitutes characters in the unit expression, showing OCR degradation on scanned text. — liteparse_input3_scannedpaper_superscript_parsed_failure.png

Input artifact: Input artifact (Image): Input — liteparse_input3_scannedpaper_superscript.png

Output artifact: Output artifact (Image): A narrow scanned-paper snippet loses superscript formatting and substitutes characters in the unit expression, showing OCR degradation on scanned text. — liteparse_input3_scannedpaper_superscript_parsed_failure.png

What changed: Image transformed into Image

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input3_scannedpaper_text_fragment.png

Observed output: Output artifact (Image): A cropped scanned-forest-paper fragment is clipped and OCR-garbled, with word splits and missing phrase boundaries. — liteparse_input3_scannedpaper_text_fragment_parsed_failure.png

Input artifact: Input artifact (Image): Input — liteparse_input3_scannedpaper_text_fragment.png

Output artifact: Output artifact (Image): A cropped scanned-forest-paper fragment is clipped and OCR-garbled, with word splits and missing phrase boundaries. — liteparse_input3_scannedpaper_text_fragment_parsed_failure.png

What changed: Image transformed into Image

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input1_hybridearnings_liquidity_section.png

Observed output: Output artifact (Image): A dense later-section page is preserved as readable prose with financial figures intact, showing no quality drop later in the 84-page report. — liteparse_input1_hybridearnings_liquidity_section_parsed_success.png

Input artifact: Input artifact (Image): Input — liteparse_input1_hybridearnings_liquidity_section.png

Output artifact: Output artifact (Image): A dense later-section page is preserved as readable prose with financial figures intact, showing no quality drop later in the 84-page report. — liteparse_input1_hybridearnings_liquidity_section_parsed_success.png

What changed: Image transformed into Image

Why it matters / Conclusion: Clean on native PDFs; OCR degrades on complex scans, especially when layout and typography get harder.

LiteParse extracts readable prose from native-digital report pages, including long paragraphs, headings, and disclaimer copy, and can also OCR scanned research-paper pages with reduced reliability. The same capability is exercised on both the native reports and the scanned paper snippets, where OCR introduces substitutions, word splitting, and clipped lines.

image

Input artifact for "Text Extraction" test: Input, liteparse_input1_hybridearnings_ceo_letter.png

↓→

image

Output artifact for "Text Extraction" test: The CEO letter prose is extracted cleanly with the opening narrative and bullet points intact, while the original page styling is dropped., liteparse_input1_hybridearnings_ceo_letter_parsed_success.png

The CEO letter prose is extracted cleanly with the opening narrative and bullet points intact, while the original page styling is dropped.

INPUT

↓→

OUTPUT

The Japanese GAAP disclaimer is extracted cleanly as a single readable paragraph with the translation and rounding note preserved.

INPUT

↓→

OUTPUT

A narrow scanned-paper snippet loses superscript formatting and substitutes characters in the unit expression, showing OCR degradation on scanned text.

INPUT

↓→

OUTPUT

A cropped scanned-forest-paper fragment is clipped and OCR-garbled, with word splits and missing phrase boundaries.

image

Input artifact for "Text Extraction" test: Input, liteparse_input1_hybridearnings_liquidity_section.png

↓→

image

Output artifact for "Text Extraction" test: A dense later-section page is preserved as readable prose with financial figures intact, showing no quality drop later in the 84-page report., liteparse_input1_hybridearnings_liquidity_section_parsed_success.png

A dense later-section page is preserved as readable prose with financial figures intact, showing no quality drop later in the 84-page report.

Bottom Line

Clean on native PDFs; OCR degrades on complex scans, especially when layout and typography get harder.

From our researchPDF to markdown using open source libraries

Table Extraction and Reconstruction

Usable on some clean tables, but headers, merged rows, and scan-heavy tables are fragile.

▾

Test Summary

Feature tested: Table Extraction and Reconstruction

Result: Partial — Usable on some clean tables, but headers, merged rows, and scan-heavy tables are fragile.

Feature tested: Table Extraction and Reconstruction

Result: Partial

Verdict: Usable on some clean tables, but headers, merged rows, and scan-heavy tables are fragile.

Expected behavior: LiteParse can turn some financial tables into usable markdown, but denser or scanned tables lose header structure, split multi-line labels, and drift out of row/column alignment. The evidence comes from the tested financial tables and scan-based table cases.

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input1_hybridearnings_financial_table.png

Observed output: Output artifact (Image): The Target financial summary table keeps some values, but the header row is flattened and the row structure is degraded. — liteparse_input1_hybridearnings_financial_table_parsed_failure.png

Input artifact: Input artifact (Image): Input — liteparse_input1_hybridearnings_financial_table.png

Output artifact: Output artifact (Image): The Target financial summary table keeps some values, but the header row is flattened and the row structure is degraded. — liteparse_input1_hybridearnings_financial_table_parsed_failure.png

What changed: Image transformed into Image

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input2_financialpdf_fy2025_table.png

Observed output: Output artifact (Image): The FY2025 forecasts table is flattened into awkward lines of text, with row and column relationships lost. — liteparse_input2_financialpdf_fy2025_table_parsed_failed.png

Input artifact: Input artifact (Image): Input — liteparse_input2_financialpdf_fy2025_table.png

Output artifact: Output artifact (Image): The FY2025 forecasts table is flattened into awkward lines of text, with row and column relationships lost. — liteparse_input2_financialpdf_fy2025_table_parsed_failed.png

What changed: Image transformed into Image

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input2_financialpdf_liabilities_table.png

Observed output: Output artifact (Image): A liabilities row is split across multiple lines, showing broken row segmentation in an otherwise numeric table. — liteparse_input2_financialpdf_liabilities_table_parsed_failure.png

Input artifact: Input artifact (Image): Input — liteparse_input2_financialpdf_liabilities_table.png

Output artifact: Output artifact (Image): A liabilities row is split across multiple lines, showing broken row segmentation in an otherwise numeric table. — liteparse_input2_financialpdf_liabilities_table_parsed_failure.png

What changed: Image transformed into Image

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input3_scannedpaper_table_1.png

Observed output: Output artifact (Image): The scanned Table 1 loses clean column alignment and becomes a fragmented markdown table with misassigned values. — liteparse_input3_scannedpaper_table_1_parsed_failure.png

Input artifact: Input artifact (Image): Input — liteparse_input3_scannedpaper_table_1.png

Output artifact: Output artifact (Image): The scanned Table 1 loses clean column alignment and becomes a fragmented markdown table with misassigned values. — liteparse_input3_scannedpaper_table_1_parsed_failure.png

What changed: Image transformed into Image

Test case: Image → Text/code file

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input2_financialpdf_segment_table.png

Observed output: Output artifact (Text/code file): The segment orders-received table is readable in markdown and keeps its year-over-year columns, though the report still shows table-fidelity issues elsewhere. — liteparse_input2_financialpdf_output.md

Input artifact: Input artifact (Image): Input — liteparse_input2_financialpdf_segment_table.png

Output artifact: Output artifact (Text/code file): The segment orders-received table is readable in markdown and keeps its year-over-year columns, though the report still shows table-fidelity issues elsewhere. — liteparse_input2_financialpdf_output.md

What changed: Image transformed into Text/code file

Test case: Image → Text/code file

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input3_scannedpaper_table_5.png

Observed output: Output artifact (Text/code file): This scanned table is the weakest case: row alignment breaks down severely and part of the table is left as plain text. — liteparse_input3_scannedpaper_output.md

Input artifact: Input artifact (Image): Input — liteparse_input3_scannedpaper_table_5.png

Output artifact: Output artifact (Text/code file): This scanned table is the weakest case: row alignment breaks down severely and part of the table is left as plain text. — liteparse_input3_scannedpaper_output.md

What changed: Image transformed into Text/code file

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input3_scannedpaper_table_6.png

Observed output: Output artifact (Image): The scanned Table 6 drops row structure and leaves the last row outside the table, making the data unreliable. — liteparse_input3_scannedpaper_table_6_parsed_failure.png

Input artifact: Input artifact (Image): Input — liteparse_input3_scannedpaper_table_6.png

Output artifact: Output artifact (Image): The scanned Table 6 drops row structure and leaves the last row outside the table, making the data unreliable. — liteparse_input3_scannedpaper_table_6_parsed_failure.png

What changed: Image transformed into Image

Why it matters / Conclusion: Some cleaner tables are usable, but structural fidelity is inconsistent and scan-based tables are not dependable.

LiteParse can turn some financial tables into usable markdown, but denser or scanned tables lose header structure, split multi-line labels, and drift out of row/column alignment. The evidence comes from the tested financial tables and scan-based table cases.

INPUT

↓→

OUTPUT

The Target financial summary table keeps some values, but the header row is flattened and the row structure is degraded.

INPUT

↓→

OUTPUT

The FY2025 forecasts table is flattened into awkward lines of text, with row and column relationships lost.

INPUT

↓→

OUTPUT

A liabilities row is split across multiple lines, showing broken row segmentation in an otherwise numeric table.

INPUT

↓→

OUTPUT

The scanned Table 1 loses clean column alignment and becomes a fragmented markdown table with misassigned values.

image

Input artifact for "Table Extraction and Reconstruction" test: Input, liteparse_input2_financialpdf_segment_table.png

↓→

text

liteparse_input2_financialpdf_output.md

Loading file...

The segment orders-received table is readable in markdown and keeps its year-over-year columns, though the report still shows table-fidelity issues elsewhere.

image

Input artifact for "Table Extraction and Reconstruction" test: Input, liteparse_input3_scannedpaper_table_5.png

↓→

text

liteparse_input3_scannedpaper_output.md

Loading file...

This scanned table is the weakest case: row alignment breaks down severely and part of the table is left as plain text.

INPUT

↓→

OUTPUT

The scanned Table 6 drops row structure and leaves the last row outside the table, making the data unreliable.

Bottom Line

Some cleaner tables are usable, but structural fidelity is inconsistent and scan-based tables are not dependable.

From our researchPDF to markdown using open source libraries

Document Hierarchy Preservation

Strong on native report headings and TOCs; scanned titles and section headers fragment or flatten.

▾

Test Summary

Feature tested: Document Hierarchy Preservation

Result: Partial — Strong on native report headings and TOCs; scanned titles and section headers fragment or flatten.

Feature tested: Document Hierarchy Preservation

Result: Partial

Verdict: Strong on native report headings and TOCs; scanned titles and section headers fragment or flatten.

Expected behavior: LiteParse preserves markdown hierarchy on native-digital reports, keeping section headers and table-of-contents structure intact. On scanned paper, those same headings can collapse into plain text and lose their structural roles.

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input1_hybridearnings_item_header.png

Observed output: Output artifact (Image): The Item 1 page is converted into proper Markdown heading levels, with the section structure preserved in readable form. — liteparse_input1_hybridearnings_item_header_parsed_success.png

Input artifact: Input artifact (Image): Input — liteparse_input1_hybridearnings_item_header.png

Output artifact: Output artifact (Image): The Item 1 page is converted into proper Markdown heading levels, with the section structure preserved in readable form. — liteparse_input1_hybridearnings_item_header_parsed_success.png

What changed: Image transformed into Image

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input2_financialpdf_summary_section.png

Observed output: Output artifact (Image): The summary section keeps genuine markdown heading levels, even though the extracted body text is dense and long. — liteparse_input2_financialpdf_summary_section_parsed_failed.png

Input artifact: Input artifact (Image): Input — liteparse_input2_financialpdf_summary_section.png

Output artifact: Output artifact (Image): The summary section keeps genuine markdown heading levels, even though the extracted body text is dense and long. — liteparse_input2_financialpdf_summary_section_parsed_failed.png

What changed: Image transformed into Image

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Table of contents page — liteparse_input2_financialpdf_toc.png

Observed output: Output artifact (Image): The table of contents keeps its hierarchy and dot leaders, even though the output file label is inconsistent. — liteparse_input2_financialpdf_toc_parsed_failed.png

Input artifact: Input artifact (Image): Table of contents page — liteparse_input2_financialpdf_toc.png

Output artifact: Output artifact (Image): The table of contents keeps its hierarchy and dot leaders, even though the output file label is inconsistent. — liteparse_input2_financialpdf_toc_parsed_failed.png

What changed: Image transformed into Image

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Scanned paper title page — liteparse_input3_scannedpaper_title.png

Observed output: Output artifact (Image): The scanned title page is fragmented across several heading levels instead of being kept as a single clean title. — liteparse_input3_scannedpaper_title_parsed_failure.png

Input artifact: Input artifact (Image): Scanned paper title page — liteparse_input3_scannedpaper_title.png

Output artifact: Output artifact (Image): The scanned title page is fragmented across several heading levels instead of being kept as a single clean title. — liteparse_input3_scannedpaper_title_parsed_failure.png

What changed: Image transformed into Image

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input3_scannedpaper_study_area_header.png

Observed output: Output artifact (Image): The study-area heading is left as plain text instead of a true markdown section header. — liteparse_input3_scannedpaper_study_area_header_parsed_failure.png

Input artifact: Input artifact (Image): Input — liteparse_input3_scannedpaper_study_area_header.png

Output artifact: Output artifact (Image): The study-area heading is left as plain text instead of a true markdown section header. — liteparse_input3_scannedpaper_study_area_header_parsed_failure.png

What changed: Image transformed into Image

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input1_hybridearnings_ceo_letter_reading_order.png

Observed output: Output artifact (Image): The two-column CEO letter is read in the correct logical order, with the left column completed before the right column begins. — liteparse_input1_hybridearnings_ceo_letter_reading_order_parsed_success.png

Input artifact: Input artifact (Image): Input — liteparse_input1_hybridearnings_ceo_letter_reading_order.png

Output artifact: Output artifact (Image): The two-column CEO letter is read in the correct logical order, with the left column completed before the right column begins. — liteparse_input1_hybridearnings_ceo_letter_reading_order_parsed_success.png

What changed: Image transformed into Image

Why it matters / Conclusion: Reliable for headings and TOCs on clean PDFs, but scanned titles and section headers lose their structure.

LiteParse preserves markdown hierarchy on native-digital reports, keeping section headers and table-of-contents structure intact. On scanned paper, those same headings can collapse into plain text and lose their structural roles.

image

Input artifact for "Document Hierarchy Preservation" test: Input, liteparse_input1_hybridearnings_item_header.png

↓→

image

Output artifact for "Document Hierarchy Preservation" test: The Item 1 page is converted into proper Markdown heading levels, with the section structure preserved in readable form., liteparse_input1_hybridearnings_item_header_parsed_success.png

The Item 1 page is converted into proper Markdown heading levels, with the section structure preserved in readable form.

INPUT

↓→

OUTPUT

The summary section keeps genuine markdown heading levels, even though the extracted body text is dense and long.

file

Table of contents page

↓→

image

Output artifact for "Document Hierarchy Preservation" test: The table of contents keeps its hierarchy and dot leaders, even though the output file label is inconsistent., liteparse_input2_financialpdf_toc_parsed_failed.png

The table of contents keeps its hierarchy and dot leaders, even though the output file label is inconsistent.

file

Scanned paper title page

↓→

image

Output artifact for "Document Hierarchy Preservation" test: The scanned title page is fragmented across several heading levels instead of being kept as a single clean title., liteparse_input3_scannedpaper_title_parsed_failure.png

The scanned title page is fragmented across several heading levels instead of being kept as a single clean title.

INPUT

↓→

OUTPUT

The study-area heading is left as plain text instead of a true markdown section header.

image

Input artifact for "Document Hierarchy Preservation" test: Input, liteparse_input1_hybridearnings_ceo_letter_reading_order.png

↓→

image

Output artifact for "Document Hierarchy Preservation" test: The two-column CEO letter is read in the correct logical order, with the left column completed before the right column begins., liteparse_input1_hybridearnings_ceo_letter_reading_order_parsed_success.png

The two-column CEO letter is read in the correct logical order, with the left column completed before the right column begins.

Bottom Line

Reliable for headings and TOCs on clean PDFs, but scanned titles and section headers lose their structure.

From our researchPDF to markdown using open source libraries

Reading Order Preservation

Correct on clean multi-column and ordered layouts; scan-heavy pages can fragment flow.

▾

Test Summary

Feature tested: Reading Order Preservation

Result: Partial — Correct on clean multi-column and ordered layouts; scan-heavy pages can fragment flow.

Feature tested: Reading Order Preservation

Result: Partial

Verdict: Correct on clean multi-column and ordered layouts; scan-heavy pages can fragment flow.

Expected behavior: LiteParse usually keeps logical sequence intact in the CEO letter’s multi-column layout and the report’s ordered table of contents. The scanned paper test shows only partial breakdown, mainly through a fragmented list item that interrupts flow.

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input1_hybridearnings_ceo_letter_reading_order.png

Observed output: Output artifact (Image): The CEO letter is read column by column in the correct sequence instead of interleaving lines across columns. — liteparse_input1_hybridearnings_ceo_letter_reading_order_parsed_success.png

Input artifact: Input artifact (Image): Input — liteparse_input1_hybridearnings_ceo_letter_reading_order.png

Output artifact: Output artifact (Image): The CEO letter is read column by column in the correct sequence instead of interleaving lines across columns. — liteparse_input1_hybridearnings_ceo_letter_reading_order_parsed_success.png

What changed: Image transformed into Image

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input2_financialpdf_toc.png

Observed output: Output artifact (Image): The table of contents keeps the right section order and page sequence, even though the formatting is flattened. — liteparse_input2_financialpdf_toc_parsed_failed.png

Input artifact: Input artifact (Image): Input — liteparse_input2_financialpdf_toc.png

Output artifact: Output artifact (Image): The table of contents keeps the right section order and page sequence, even though the formatting is flattened. — liteparse_input2_financialpdf_toc_parsed_failed.png

What changed: Image transformed into Image

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input3_scannedpaper_pointer.png

Observed output: Output artifact (Image): A scanned list item is broken into fragments, disrupting the natural reading flow of the paragraph. — liteparse_input3_scannedpaper_pointer_parsed_failure.png

Input artifact: Input artifact (Image): Input — liteparse_input3_scannedpaper_pointer.png

Output artifact: Output artifact (Image): A scanned list item is broken into fragments, disrupting the natural reading flow of the paragraph. — liteparse_input3_scannedpaper_pointer_parsed_failure.png

What changed: Image transformed into Image

Why it matters / Conclusion: Good at preserving logical flow in native PDFs, with only partial breakdown on the scan test.

LiteParse usually keeps logical sequence intact in the CEO letter’s multi-column layout and the report’s ordered table of contents. The scanned paper test shows only partial breakdown, mainly through a fragmented list item that interrupts flow.

INPUT

↓→

OUTPUT

The CEO letter is read column by column in the correct sequence instead of interleaving lines across columns.

INPUT

↓→

OUTPUT

The table of contents keeps the right section order and page sequence, even though the formatting is flattened.

INPUT

↓→

OUTPUT

A scanned list item is broken into fragments, disrupting the natural reading flow of the paragraph.

Bottom Line

Good at preserving logical flow in native PDFs, with only partial breakdown on the scan test.

From our researchPDF to markdown using open source libraries

Embedded Image Preservation

Native-page images are kept as placeholders, but scanned visuals can disappear entirely.

▾

Test Summary

Feature tested: Embedded Image Preservation

Result: Partial — Native-page images are kept as placeholders, but scanned visuals can disappear entirely.

Feature tested: Embedded Image Preservation

Result: Partial

Verdict: Native-page images are kept as placeholders, but scanned visuals can disappear entirely.

Expected behavior: On native PDFs, LiteParse keeps portraits, signatures, and logos as image placeholders or references. On scanned pages, the visuals can be dropped entirely, showing the same image-retention behavior under weaker source conditions.

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input1_hybridearnings_ceo_image.png

Observed output: Output artifact (Image): The CEO portrait is retained only as a linked image reference, so the asset is present in the markdown but not actually viewable. — liteparse_input1_hybridearnings_ceo_image_parsed_partial_success.png

Input artifact: Input artifact (Image): Input — liteparse_input1_hybridearnings_ceo_image.png

Output artifact: Output artifact (Image): The CEO portrait is retained only as a linked image reference, so the asset is present in the markdown but not actually viewable. — liteparse_input1_hybridearnings_ceo_image_parsed_partial_success.png

What changed: Image transformed into Image

Test case: Image → Image

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input1_hybridearnings_scanned_signs.png

Observed output: Output artifact (Image): The signature images are not recovered as usable visuals, and the parse only leaves broken image references and fragments. — liteparse_input1_hybridearnings_scanned_signs_parsed_failure.png

Input artifact: Input artifact (Image): Input — liteparse_input1_hybridearnings_scanned_signs.png

Output artifact: Output artifact (Image): The signature images are not recovered as usable visuals, and the parse only leaves broken image references and fragments. — liteparse_input1_hybridearnings_scanned_signs_parsed_failure.png

What changed: Image transformed into Image

Test case: Image → Text prompt

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input3_scannedpaper_usda_logo.png

Observed output: Output artifact (Text prompt): Output

Input artifact: Input artifact (Image): Input — liteparse_input3_scannedpaper_usda_logo.png

Output artifact: Output artifact (Text prompt): Output

What changed: Image transformed into Text prompt

Test case: Image → Text/code file

Input type: Image

Input used: Input artifact (Image): Input — liteparse_input2_financialpdf_logo.png

Observed output: Output artifact (Text/code file): The report logo is preserved as a markdown image placeholder at the top of the file, but there is no accessible image preview. — liteparse_input2_financialpdf_output.md

Input artifact: Input artifact (Image): Input — liteparse_input2_financialpdf_logo.png

Output artifact: Output artifact (Text/code file): The report logo is preserved as a markdown image placeholder at the top of the file, but there is no accessible image preview. — liteparse_input2_financialpdf_output.md

What changed: Image transformed into Text/code file

Why it matters / Conclusion: Native images are technically retained, but scanned visuals are dropped outright.

On native PDFs, LiteParse keeps portraits, signatures, and logos as image placeholders or references. On scanned pages, the visuals can be dropped entirely, showing the same image-retention behavior under weaker source conditions.

image

↓→

image

The CEO portrait is retained only as a linked image reference, so the asset is present in the markdown but not actually viewable.

image

↓→

image

The signature images are not recovered as usable visuals, and the parse only leaves broken image references and fragments.

INPUT

↓→

OUTPUT

The USDA logo is omitted entirely from the extracted output, with no placeholder or markdown image syntax generated.

image

↓→

text

liteparse_input2_financialpdf_output.md

Loading file...

The report logo is preserved as a markdown image placeholder at the top of the file, but there is no accessible image preview.

Bottom Line

Native images are technically retained, but scanned visuals are dropped outright.

From our researchPDF to markdown using open source libraries

Chart and Figure Reconstruction

Charts and figures are not faithfully reconstructed; visual meaning is lost.

▾

Test Summary

Feature tested: Chart and Figure Reconstruction

Result: Failed — Charts and figures are not faithfully reconstructed; visual meaning is lost.

LiteParse attempts to preserve or reconstruct charts and figures from the tested report pages, but the observed outputs often fall back to placeholders or garbled text. The behavior is shown on both the Target report chart and the scanned research-paper figure.

file

Five-year total return comparison chart

↓→

image

Output artifact for "Chart and Figure Reconstruction" test: The five-year return chart is replaced by a placeholder, with most chart structure missing., liteparse_input1_hybridearnings_comparison_chart_parsed_failure.png

The five-year return chart is replaced by a placeholder, with most chart structure missing.

file

Scanned mountain-beetle figure

↓→

image

Output artifact for "Chart and Figure Reconstruction" test: The scanned bar chart is reduced to fragmented text and symbols rather than a readable figure., liteparse_input3_scannedpaper_figure_1_parsed_failure.png

The scanned bar chart is reduced to fragmented text and symbols rather than a readable figure.

Bottom Line

Chart/figure reconstruction is one of LiteParse’s weakest areas.

From our researchPDF to markdown using open source libraries

Caption/Figure Association

Captions stay attached when the source layout is intact, but become orphaned when visuals vanish on scans.

▾

Test Summary

Feature tested: Caption/Figure Association

Result: Partial — Captions stay attached when the source layout is intact, but become orphaned when visuals vanish on scans.

LiteParse keeps captions and nearby explanatory text attached to the right table or figure on native PDFs, and also keeps caption-like text linked to associated visuals when the page structure is intact. When scanned visuals disappear, those associations break and the text becomes detached.

image

Input artifact for "Caption/Figure Association" test: Input, liteparse_input1_hybridearnings_financial_table_caption.png

↓→

image

Output artifact for "Caption/Figure Association" test: The financial summary header stays directly attached to the table, so the caption-to-table association is preserved on the native PDF., liteparse_input1_hybridearnings_financial_table_caption_parsed_success.png

The financial summary header stays directly attached to the table, so the caption-to-table association is preserved on the native PDF.

file

Figure 1 caption on the scanned paper

↓→

image

Output artifact for "Caption/Figure Association" test: The figure caption survives as text, but the figure itself is missing, so the caption is orphaned., liteparse_input3_scannedpaper_figure_1_parsed_partial_success.png

The figure caption survives as text, but the figure itself is missing, so the caption is orphaned.

Bottom Line

Works on clean pages, but scan-heavy documents leave captions detached from their visuals.

From our researchPDF to markdown using open source libraries

Long-Document Robustness

Quality stays consistent across document length; source type is the real separator.

▾

Test Summary

Feature tested: Long-Document Robustness

Result: Passed — Quality stays consistent across document length; source type is the real separator.

Feature tested: Long-Document Robustness

Result: Passed

Verdict: Quality stays consistent across document length; source type is the real separator.

Expected behavior: LiteParse maintains a consistent baseline across the 84-page hybrid earnings report, the 18-page financial report, and the 12-page scanned paper. The test results suggest document length itself is not the limiting factor; source quality is.

Test case: PDF document → Text/code file

Input type: PDF document

Input used: Input artifact (PDF document): Input — input1_hybridearnings.pdf

Observed output: Output artifact (Text/code file): The 84-page hybrid earnings report was processed in about 1 minute 21 seconds, with text, headings, and reading order staying strong throughout. — liteparse_input1_hybridearnings_output.md

Input artifact: Input artifact (PDF document): Input — input1_hybridearnings.pdf

Output artifact: Output artifact (Text/code file): The 84-page hybrid earnings report was processed in about 1 minute 21 seconds, with text, headings, and reading order staying strong throughout. — liteparse_input1_hybridearnings_output.md

What changed: PDF document transformed into Text/code file

Test case: PDF document → Text/code file

Input type: PDF document

Input used: Input artifact (PDF document): Input — input2_financialpdf.pdf

Observed output: Output artifact (Text/code file): The 18-page financial report was processed in about 1 minute 2 seconds, with consistent extraction quality across the document. — liteparse_input2_financialpdf_output.md

Input artifact: Input artifact (PDF document): Input — input2_financialpdf.pdf

Output artifact: Output artifact (Text/code file): The 18-page financial report was processed in about 1 minute 2 seconds, with consistent extraction quality across the document. — liteparse_input2_financialpdf_output.md

What changed: PDF document transformed into Text/code file

Test case: PDF document → Text/code file

Input type: PDF document

Input used: Input artifact (PDF document): Input — input3_scannedpaper.pdf

Observed output: Output artifact (Text/code file): The 12-page scanned paper was processed in about 49 seconds, but the same structural weaknesses persisted from start to finish. — liteparse_input3_scannedpaper_output.md

Input artifact: Input artifact (PDF document): Input — input3_scannedpaper.pdf

Output artifact: Output artifact (Text/code file): The 12-page scanned paper was processed in about 49 seconds, but the same structural weaknesses persisted from start to finish. — liteparse_input3_scannedpaper_output.md

What changed: PDF document transformed into Text/code file

Why it matters / Conclusion: Document length itself was not a problem; the source type determined whether output quality stayed strong or fell apart.

LiteParse maintains a consistent baseline across the 84-page hybrid earnings report, the 18-page financial report, and the 12-page scanned paper. The test results suggest document length itself is not the limiting factor; source quality is.

INPUT

input1_hybridearnings.pdf

↓→

OUTPUT

liteparse_input1_hybridearnings_output.md

Loading file...

The 84-page hybrid earnings report was processed in about 1 minute 21 seconds, with text, headings, and reading order staying strong throughout.

INPUT

input2_financialpdf.pdf

↓→

OUTPUT

liteparse_input2_financialpdf_output.md

Loading file...

The 18-page financial report was processed in about 1 minute 2 seconds, with consistent extraction quality across the document.

INPUT

input3_scannedpaper.pdf

↓→

OUTPUT

liteparse_input3_scannedpaper_output.md

Loading file...

The 12-page scanned paper was processed in about 49 seconds, but the same structural weaknesses persisted from start to finish.

Bottom Line

Document length itself was not a problem; the source type determined whether output quality stayed strong or fell apart.

From our researchPDF to markdown using open source libraries

Open source

Free

Open source

✓ Use This If

●You need a self-hosted, open-source PDF-to-markdown parser for native-digital reports.

●You care most about preserving prose, headings, reading order, and table-of-contents structure on clean PDFs.

●You can tolerate image placeholders and some cleanup in tables.

✕ Skip This If

●Your PDFs are mostly scanned or OCR-heavy.

●You need faithful chart and figure reconstruction.

●You need dependable complex table reconstruction with merged headers and split rows.

developer-toolsdocumentation-toolstext

It worked much better on the two native-digital reports. Text, headings, reading order, and table-of-contents structure stayed strong there, while the scanned paper showed OCR garbling, broken tables, omitted images, and fragmented headings.

Mixed. Some cleaner tables were usable, but headers flattened and rows split on the hybrid earnings and financial report, and the scanned paper tables were severely degraded or unusable.

Yes on native-digital reports. Section headings and TOC ordering were preserved with real markdown hierarchy, but scanned titles and section headers fragmented or flattened.

On native PDFs, LiteParse keeps images and signatures as markdown image placeholders, but they are not actually viewable. On the scanned paper, the USDA logo and other visuals were dropped entirely.

Yes in the CEO letter test, where it read the columns in the correct sequence. The scanned paper was only partially reliable, with at least one fragmented list item disrupting flow.

The report identifies LiteParse as open source and lists it as free at $0.