CSV Viewer
Open the file, sort the column, find the row. Without waiting for a spreadsheet to launch.
Drop a CSV here
Read in this page. Nothing is uploaded, so a confidential export stays put.
Filters every column at once.
Detected automatically.
Or paste rows from a spreadsheet
- Rows
- —
- Columns
- —
- Showing
- —
This is a reader, not a spreadsheet
No editing, no formulas. Sort, search and export what you are looking at — the file itself is never modified and never leaves your device.
Everything here runs on your device. Nothing you enter is uploaded or stored.
Drop a CSV in and it becomes a table you can sort and search. The delimiter is detected, quoted fields survive intact, and the column panel tells you what each column contains before you start reading rows.
How to use it
Drop the file on the panel or paste rows directly from a spreadsheet. Click any column heading to sort by it and click again to reverse. The search box filters every column at once as you type. When you have the view you want, export it as CSV or JSON — the filter and sort come with it.
The delimiter is detected, not assumed
A large share of the CSV files in circulation are not comma-separated. Excel writes semicolons whenever the system’s decimal separator is a comma, so files from most of Europe arrive semicolon-delimited with a .csv extension. Tab-separated exports are common from databases, and pipe-delimited files turn up in older systems.
Rather than trust the extension, the first dozen lines are scored for each candidate. The winner is the one that appears the same number of times in every row — consistency, not frequency, because a semicolon that shows up five times in a single prose cell is punctuation rather than structure. The dropdown lets you override the guess when a genuinely ambiguous file defeats it.
Column types earn their place
Each column is typed from the majority of its non-empty cells: number, date, boolean or text. That is not decoration. Numeric columns right-align with tabular figures so the digits line up; they sort numerically rather than lexically, which is the difference between 2, 10, 100 and 10, 100, 2; and they get a minimum, maximum and mean in the column panel. Date columns sort chronologically instead of alphabetically.
Majority rather than unanimity matters here. A single N/A in a column of four thousand prices would otherwise reclassify the whole column as text, and you would lose the alignment, the sorting and the statistics for the sake of one cell.
Large files page rather than render
The table only ever holds 200 rows. The rest of the file stays as data and is paged through. This is the difference between a viewer that opens a 400,000-row file instantly and one that hangs the tab trying to build 400,000 table rows — that is several seconds of layout work and a great deal of memory for rows nobody is looking at.
Rows that do not match are flagged
If any row has more or fewer fields than the header, the count is reported along with the first few row numbers. Ragged rows usually mean an unescaped quote somewhere upstream, and knowing which rows are affected is the fastest way to find the bad record. Nothing is dropped or silently repaired — the row is shown as it exists, with any missing cells blank, because a viewer that quietly fixes your data is a viewer you cannot use to diagnose it.
A byte-order mark will not break the first column
Files exported from Windows tools frequently begin with an invisible three-byte marker. Left in place, it becomes part of the first column’s name, so id is actually id and every lookup against that column fails for reasons nothing on screen explains. It is stripped before parsing. This is a small thing that costs people an hour surprisingly often.
What it does not do
No cell editing, no formulas, no charts. This is a reader, and the reading case is worth doing properly on its own. Nothing is uploaded — the file is read in the page, so a confidential export never leaves your device.
Which tool for which job
| You want to | Use |
|---|---|
| Read, sort and search a CSV | CSV Viewer |
| Reformat or validate JSON | JSON Formatter |
| Cut a file too big to open | File Splitter |
| Pull a CSV out of an archive | ZIP Extractor |
Questions
Why does my file open as one wide column in other tools?
Because it is almost certainly semicolon-separated and the tool assumed a comma. Excel writes semicolons rather than commas in any locale where the comma is the decimal separator, which covers most of continental Europe and a good part of South America — so a colleague's export can look completely broken in a viewer that hardcodes the comma. This tool sniffs the delimiter from the first dozen lines, scoring candidates by how consistently they appear per row rather than by raw frequency. That distinction matters, because a semicolon appearing five times in one prose cell is not structure. You can override the guess from the dropdown if it gets it wrong.
How large a file can it open?
Up to 120 MB, which is roughly a million rows of typical data. Beyond that, a browser cannot hold the file as a string without risking the tab, so the limit is stated rather than discovered by crashing. The table itself only ever puts 200 rows in the page at a time and pages through the rest — this is why sorting a 400,000-row file feels instant while a spreadsheet is still opening. If your file is larger, File Splitter will cut it into pieces this can read.
Does it handle commas inside quoted fields?
Yes, and it is the reason this is a real parser rather than a line-splitter. Splitting on commas fails on the first address field, and splitting on newlines fails on the first multi-line note. The parser walks the file character by character, tracking whether it is inside a quoted region, so "Smith, John" stays one field and a quoted field containing a line break stays one cell. Doubled quotes are unescaped correctly, and a stray quote in the middle of an unquoted field is treated as a literal character rather than an error, because that pattern is common in real exports.
What does the column type actually do?
Three things. Numeric columns are right-aligned with tabular figures so digits line up. Sorting is type-aware, so a numeric column sorts 2 before 10 rather than lexically, and a date column sorts chronologically rather than alphabetically. And numeric columns get a min, max and mean in the column panel, which is often the entire question you opened the file to answer. Types are decided by majority of non-empty cells rather than unanimity, so one "N/A" in four thousand numbers does not turn the column into text.
Can I edit cells?
No, and that is deliberate. Editing means an undo history, validation, and a save path — a spreadsheet, in other words, and a worse one than the free spreadsheet already on your device. This is built for the far more common task of looking — open the file someone sent, find out whether the column you need is populated, sort by the value you care about, and get on with your day. You can export the filtered and sorted view if you want to carry that state somewhere else.
Last updated