Clean Up Messy Markdown After PDF Conversion

You converted a PDF to Markdown, opened the result, and it is a mess: sentences chopped into fragments, words split with stray hyphens, page numbers floating mid-paragraph, and lists that no longer look like lists. This is normal. A PDF stores how a document looks on the page, not what it means, so even a good converter has to reconstruct structure from visual cues — and some of those cues are misleading.

The good news: almost all of the mess falls into a handful of predictable categories, and each one has a quick fix. This guide walks through the most common problems and how to clean them up fast, whether you are doing it by hand, with find-and-replace, or with a quick script.

Why PDF-to-Markdown output looks broken

PDFs position text as glyphs at x/y coordinates. There is often no real concept of a paragraph, a heading, or a list item baked into the file — those are things your eye infers from font size, spacing, and indentation. When a tool reads a PDF, it has to guess where one paragraph ends and the next begins, whether a large bold line is a heading or just emphasis, and what to do with the repeating text at the top and bottom of every page.

Most artifacts come from three sources:

• Hard line breaks baked into each visual line of the page.
• Repeating page furniture — running headers, footers, and page numbers.
• Typographic substitutions — hyphenation at line ends, ligatures, and "smart" quotes.

Once you know the categories, cleanup becomes mechanical. Here is the checklist.

The cleanup checklist

1. Rejoin broken line breaks

The single most common problem: every visual line in the PDF becomes its own line in the Markdown, so a paragraph that wrapped over six lines on the page arrives as six short fragments. In Markdown, a single line break inside a paragraph is usually fine when rendered, but it makes the source ugly and breaks downstream tools that treat each line as a unit.

The fix is to join lines that belong to the same paragraph while preserving the blank lines between paragraphs. A reliable find-and-replace approach in an editor that supports regex:

• Find (\S)\n(\S) and replace with $1 $2 to merge mid-paragraph breaks into spaces.
• Leave double newlines (\n\n) untouched so paragraph boundaries survive.

Run the paragraph-boundary-safe version first; never do a blanket "remove all newlines," or you will weld separate paragraphs together.

2. Fix hyphenated words split across lines

When a word was broken at the right margin — conver- at the end of one line and sion at the start of the next — naive extraction keeps the hyphen: conver- sion or conver-\nsion. Rejoin these before you merge line breaks, so you do not turn the hyphen into a permanent scar.

• Find (\w)-\n(\w) and replace with $1$2 to stitch the word back together.
• Watch out for genuinely hyphenated terms like state-of-the-art — those should keep their hyphen. A line-end hyphen is almost always a soft hyphen; a mid-line one is real.

3. Remove headers, footers, and page numbers

Running heads ("CHAPTER 3 — METHODS"), footers, and bare page numbers repeat on every page and end up sprinkled through your text, often interrupting a sentence that spanned a page break. Because they repeat, they are easy to spot:

• Scan for a short line that appears many times (the document title, author, or section name) and delete the duplicates.
• Delete lines that are just a number, or a number surrounded by whitespace.
• Re-join the sentence that the page break split — this is where most "why is this paragraph cut in half" confusion comes from.

If you have dozens of pages, a script that strips lines matching the repeating header/footer text is far faster than manual deletion.

4. Repair headings

Converters often miss headings (a bold 16pt line becomes plain bold text) or over-promote them (a figure caption becomes an ##). Skim the document outline and fix the hierarchy by hand:

• Add #, ##, ### where the original used larger or bolder type for section titles.
• Keep levels consistent — do not jump from # straight to ####.
• Demote anything that became a heading but is really a caption or a label.

Good headings matter beyond looks: they are what note apps, static-site generators, and AI tools use to chunk and navigate a document. If you are feeding the file to an LLM, clean headings noticeably improve results — see why LLMs work better with Markdown.

5. Rebuild lists

Bulleted and numbered lists frequently lose their markers or get the indentation wrong, so they render as ordinary paragraphs. Restore them:

• Replace stray bullet glyphs (•, ◦, ‣) with - followed by a space.
• Make sure each list item starts on its own line with the marker at the correct indent.
• For numbered lists, 1., 2., 3. works, but Markdown also renders a list where every item is 1. — handy when reordering.

6. Normalize typographic characters

PDFs love "smart" typography that can confuse plain-text tools and search:

• Ligatures: fi, fl, ff may come through as single glyphs or vanish entirely, turning "file" into "le". Search for missing letters near fi/fl combinations.
• Curly quotes and dashes: " " ' — are valid but inconsistent; normalize to straight quotes and -/-- if your pipeline expects ASCII.
• Non-breaking spaces ( ) hide between words and break alignment — replace them with regular spaces.

7. Check tables and code last

Tables are the hardest thing to recover from a PDF because the visual grid carries the structure. Expect to rebuild column alignment by hand or re-extract them carefully. We cover this in depth in extract tables from PDF to Markdown. Code blocks similarly lose their fences and indentation — wrap them in triple backticks and restore leading whitespace.

Quick reference: problem → fix

Symptom	Cause	Fix
Sentences chopped into short lines	Hard line breaks per visual line	Merge \n between non-blank lines into spaces
conver- sion, stray hyphens	Line-end hyphenation	Remove hyphen + newline between word parts
Repeating title/number in text	Page headers and footers	Delete repeated lines; rejoin split sentences
Bold line that should be a title	Heading not detected	Add #/##/### by hand
List renders as a paragraph	Lost bullet/number markers	Restore - or 1. markers and indentation
"le" instead of "file"	Dropped fi ligature	Search and restore missing letters
Garbled table	Visual grid lost	Rebuild columns or re-extract

Start with a cleaner conversion

The fastest cleanup is the one you do not have to do. A converter that reconstructs paragraphs, headings, and lists well leaves you far less to fix. Our free PDF to Markdown converter runs entirely in your browser — the file is never uploaded to a server, which matters when the document is a contract, a draft, or anything you would not paste into a random website — and it rebuilds structure from the PDF's layout so the raw output already needs less work. For scanned documents it runs OCR locally too, though OCR output always deserves a closer proofread.

Whatever tool you use, do the cleanup in this order: rejoin hyphenated words, merge line breaks, strip page furniture, fix headings and lists, then normalize typography and tables. Working top-down means each step does not undo the last. If you want the broader picture of how conversion works and why these artifacts appear, the how to convert PDF to Markdown guide is a good companion read.

FAQ

Why does my converted Markdown have a line break after every line?

Because the PDF stored the text as separate visual lines and the converter preserved them. Merge single newlines between non-blank lines into spaces while keeping the blank lines that separate paragraphs. Never strip all newlines, or paragraphs will run together.

How do I remove page numbers and headers automatically?

They repeat on every page, so they are easy to target: delete lines that are just a number, and delete the repeated title or section text that appears many times. For long documents, a small find-and-replace pattern or script that matches the exact header/footer text is much faster than deleting by hand.

Why are some words missing letters like "fi" or "fl"?

Those are ligatures — single glyphs the typesetter used for letter pairs. Some PDFs drop them during extraction, turning "file" into "le" or "flow" into "ow". Search the text for these gaps near fi, fl, and ff combinations and restore the missing letters.

Can I avoid cleanup entirely?

Not entirely — PDFs simply do not carry the structure Markdown needs, so some reconstruction is unavoidable. But a converter that does a good job rebuilding headings, paragraphs, and lists dramatically reduces the work. Digital-native PDFs clean up in minutes; scanned documents that went through OCR always need a more careful pass.