Chinese PDF to Excel: Characters, Fonts and Dates

A Chinese PDF converts to Excel well when its fonts map to real characters, its full-width digits become ordinary numbers, its dates are rebuilt from the 年, 月 and 日 markers, and any CSV is saved as UTF-8. Test a copy and paste first. If characters come out as boxes, the file needs OCR instead.

Why Chinese PDF characters copy as boxes or question marks

Chinese PDF characters copy as boxes or question marks for one of two reasons. Either the file has no usable map from its character codes to Unicode, so nothing meaningful reaches the clipboard, or the characters arrive correctly but the place you paste them has no font that can show them. The first needs OCR; the second, a different font.

Tell the two apart with one test. Paste the copied text into a plain text editor that you know displays Chinese, or into a web search box. If the characters appear there, the PDF is fine and your spreadsheet font is the problem. Change the cell font to one that covers Chinese and the boxes turn into characters.

If the text is still boxes, symbols or question marks everywhere, the file is the cause. The PDF 32000-1:2008 specification, section 9.10.2, explains why Chinese files are prone to this. A reader can map Chinese characters to Unicode in two main ways: through a ToUnicode table stored in the font, or through standard character collections such as Adobe-GB1 for simplified Chinese and Adobe-CNS1 for traditional Chinese. A font that uses the generic Identity encoding and carries no ToUnicode table falls outside both. The specification says that when these methods fail "there is no way to determine what the character code represents".

Two other causes look similar:

• Outlined text. Some files draw every character as a shape, with no text behind it at all. Nothing selects, and nothing copies. How to spot a PDF with outlined text explains why that file needs OCR like a scan.
• A scanned page. The whole page is an image. You cannot select a single character.

Full-width digits and punctuation in a Chinese PDF

Full-width digits are versions of 0 to 9 that take the same width as a Chinese character, such as ０１２３ instead of 0123. They look like numbers, but Excel treats them as text, so it will not sum or sort them. Full-width commas, periods and minus signs cause the same problem. Replace them with ordinary characters before you calculate anything.

Chinese documents mix the two widths freely. A statement may print amounts in ordinary digits and reference numbers in full-width ones, or use a full-width comma as the thousands separator. Each mix gives a different failure.

The table shows example values and what Excel does with each. The values are made up.

To fix full-width digits, select the amount column and use Find and Replace ten times, once for each digit: replace ０ with 0, １ with 1, and so on to ９. Then replace the full-width comma ， with nothing and the full-width period ． with an ordinary period. It is dull but safe, because it touches only those exact characters.

After the replacements, check the result. Numbers should now sit on the right of their cells. Select the column and look at the sum in the status bar; it should match the total printed on the statement.

How do you convert Chinese dates like 2026年3月5日 in Excel?

Convert a Chinese date by reading the year, month and day out from between the 年, 月 and 日 markers, then passing them to Excel's DATE function. Do not just strip the markers and hope, because "202635" is ambiguous. Splitting on the markers keeps each part separate, whatever its length.

In Excel 365, with the text date in A2, this formula returns a real date:

=DATE(TEXTBEFORE(A2,"年"),TEXTBEFORE(TEXTAFTER(A2,"年"),"月"),TEXTBEFORE(TEXTAFTER(A2,"月"),"日"))

For 2026年3月5日, the three parts are 2026, 3 and 5, and the result is 5 March 2026. Format the cell as a date to see it that way. The formula works whether the month and day have one digit or two.

Other Chinese date formats

• Slashes or dots. 2026/03/05 and 2026.03.05 put the year first. Excel usually reads the slash form directly. For the dot form, replace the dots with hyphens first.
• Full-width digits in dates. Fix the digits first, using the replacements in the previous section, then run the formula.
• Years counted from 1912. Some documents from Taiwan count years from 1912, so 115年 is 2026. Add 1911 to the year part: wrap the first TEXTBEFORE in 1911+VALUE(...).
• Month and day only. A transaction list may print 03-05 and leave the year to the statement header. Take the year from the header, and watch for a period that crosses New Year.

Once the dates are real, sort the sheet by date and check the first and last rows against the statement period. Fixing converted dates that came out wrong covers swapped day and month and dates stored as text in more detail.

Saving a Chinese spreadsheet as CSV in the right encoding

Save a Chinese spreadsheet as CSV UTF-8, not plain CSV. A CSV file is plain text, and the encoding decides how each character is written as bytes. The older CSV option uses your computer's regional code page, which may not cover Chinese, so characters are replaced or garbled. UTF-8 covers every Chinese character.

Microsoft's guide, Save a worksheet as a CSV file, names the option: under File Format, choose "CSV UTF-8 (Comma delimited)(*.csv)". Pick it in the Save As dialog whenever the file holds Chinese text.

Garbled text after a round trip has a name, mojibake, and a typical look: Chinese characters replaced by strings of accented Latin letters such as "工资" for 工资. When you see that, the file is usually fine and the program opening it guessed the wrong encoding. Open it again through Excel's Data tab, choose From Text/CSV, and set File Origin to 65001: Unicode (UTF-8). The preview shows the characters correctly before you load them.

Check what will read the file next. A budgeting app, a database import or a script may expect a particular encoding. If it shows mojibake after import, the importer is reading UTF-8 as something else, and its own settings need changing, not your file.

Is OCR the answer for a Chinese PDF scan?

OCR is the answer for a Chinese PDF scan and for any Chinese file whose text will not copy. Pick a tool with the right Chinese language selected, because simplified and traditional Chinese are usually separate options, and expect to check digits and look-alike characters afterwards. Numbers are often read more reliably than names.

Set the language before you run it. An OCR tool set to English will read Chinese characters as nonsense Latin letters. Many tools let you choose more than one language, which helps on statements that mix Chinese descriptions with English bank codes.

Things to check after OCR on Chinese pages:

• Look-alike characters. Characters that differ by one stroke, such as 日 and 曰 or 己 and 已, are easy to confuse. They matter in names more than in amounts.
• Full-width output. Some OCR tools produce full-width digits from full-width originals. Run the replacements above.
• Vertical text. Older or formal documents may run text top to bottom. Tell the OCR tool the text direction, or rotate the page image.
• Fixed-box tools. Zonal OCR reads fixed regions of a page, which works on forms where every copy has the same layout. Why zonal OCR breaks on statements whose rows move explains why a statement, where rows shift month to month, needs a different approach.

Some Chinese statements mask parts of an ID number or account number with black boxes or asterisks. If the masking was drawn on top of the text rather than removing it, the hidden digits may still come out in your sheet. What the black boxes hide when a redacted PDF goes to Excel shows how to check.

Chinese PDF to Excel checks before you trust the sheet

A short set of checks catches most problems in a converted Chinese sheet: characters display, numbers are numbers, dates are dates, totals match, and the saved file reopens cleanly. Run them in that order, because each one depends on the one before.

1. Characters display. Scroll through the description column. No boxes, no question marks, no strings of accented Latin letters.
2. Amounts are numbers. Every amount sits on the right of its cell. No 元, ￥, full-width digits or full-width minus signs remain.
3. Dates are dates. Change the date column's format to a long date. Every cell should change; any cell that does not is still text.
4. Totals match. The sum of the amount column, or the closing balance, matches the figure printed on the statement.
5. Row count matches. Count the transactions on one page of the PDF and compare with the same page in the sheet.
6. The CSV reopens clean. If you saved a CSV, close it and open it again through From Text/CSV with UTF-8. Characters should be unchanged.

For bank statements, step 4 is the one that matters most. FastBankConvert is getting ready to open as a converter that warns you before download if a statement's balances do not add up. Get started to join.