# Arabic PDF to Excel: Right-to-Left Text and Digits

- Author: FastBankConvert (https://www.fastbankconvert.com)
- Published: 2026-06-05
- Updated: 2026-06-05
- Category: Difficult and locked PDFs
- Canonical: https://www.fastbankconvert.com/blog/arabic-pdf-to-excel

> Convert an Arabic PDF to Excel: right-to-left text that comes out reversed, Eastern Arabic digits, joined letters, and column order.
An Arabic PDF goes to Excel cleanly when three things are handled: the right-to-left text is put back in reading order, Eastern Arabic digits such as ١٢٣ become numbers Excel can add, and joined letter shapes are turned back into plain letters. Get those right and the column order is a matter of display, not data.

## Why does Arabic PDF text come out reversed in Excel?

Arabic text comes out reversed when the PDF stores letters in the order they are drawn on the page instead of the order they are read. Many PDF producers place Arabic glyphs, the drawn letter shapes, from left to right, so a tool that reads them in drawing order gets every word backwards.

Software that handles Arabic properly keeps text in what the Unicode Standard calls logical order. The [Unicode Bidirectional Algorithm, UAX #9](https://www.unicode.org/reports/tr9/), says text is stored in logical order and the display order is derived from it, line by line. In logical order, the first letter you read is the first character stored, even though it appears at the right edge. Excel expects logical order. Give it visual order and each cell shows the letters in reverse.

The same standard explains why numbers behave differently. Digits inside right-to-left text keep their own left-to-right sequence, so 2026 is still read as two thousand and twenty-six. A conversion that reverses whole lines therefore damages numbers too. A date stored as ١٥/٠٣/٢٠٢٦ can arrive as ٦٢٠٢/٣٠/٥١, which looks like a date but is not one.

How to tell which problem you have: copy one payee name from the PDF and paste it into a cell. If it reads correctly, the file stores logical order and the rest of the work is digits and letter shapes. If the letters are backwards, the PDF stores visual order, and you need a converter that reorders right-to-left runs, or OCR on the page image.

## Eastern Arabic digits in Arabic PDF amounts

Eastern Arabic digits are the digit shapes ٠١٢٣٤٥٦٧٨٩ used in many Arabic-speaking countries. They are separate characters from the digits 0 to 9, so in most setups Excel treats a cell such as ١٢٥٠ as text. It will not add up, sort as a number or format as currency until each digit is replaced.

Persian and Urdu statements use a related set, sometimes called extended Arabic-Indic digits, with slightly different shapes for four, five and six. They are different characters again, so a fix for one set misses the other. The Excel formula in this section handles both.

Statements often also use the Arabic decimal separator ٫ and the Arabic thousands separator ٬, which look like a comma and an apostrophe but are neither. They need replacing too.

### A formula that turns Arabic digits into numbers

In Excel 365, with the amount text in A2, this formula swaps each digit for its Western form, replaces the two separators, and returns a number:

```
=LET(t, A2,
  d, REDUCE(t, SEQUENCE(10,,0), LAMBDA(s, n,
      SUBSTITUTE(SUBSTITUTE(s, UNICHAR(1632+n), n), UNICHAR(1776+n), n))),
  VALUE(SUBSTITUTE(SUBSTITUTE(d, UNICHAR(1643), "."), UNICHAR(1644), "")))
```

UNICHAR(1632) is the Eastern Arabic zero and UNICHAR(1776) is the Persian zero; the rest of each set follows in order. UNICHAR(1643) is the Arabic decimal separator and UNICHAR(1644) the thousands separator. The formula assumes your Excel uses a full stop as its decimal point. If your regional settings use a comma, replace "." with "," in the last line.

Run it on dates with the VALUE call removed, then turn the cleaned text into a date with a separate step, because a date also depends on day and month order.

## Joined Arabic letters that break search after conversion

Arabic letters change shape by position in a word: alone, at the start, in the middle or at the end. Some PDFs store each of those shapes as its own character, called a presentation form, instead of the plain letter. The text looks right in a cell but does not match the same word typed on a keyboard.

That mismatch causes quiet failures. A search for a payee finds nothing. A lookup against a list of vendor names returns errors. A pivot table splits one payee into several rows because one statement stored plain letters and another stored presentation forms. Nothing looks wrong on screen, which is what makes it costly.

Ligatures add a second layer. The pair lam and alef is usually drawn as a single combined shape, and a PDF may store that pair as one presentation-form character rather than two letters. A word containing it is one character shorter than the typed version, so length checks and partial matches go wrong too.

The fix is Unicode normalization in the compatibility form called NFKC, which maps presentation forms and ligatures back to plain letters. Excel has no built-in function for it. A converter that normalizes text on the way out avoids the problem. Without one, a short script does it; in Python, `unicodedata.normalize("NFKC", text)` cleans a column in one pass. Chinese statements run into a similar issue with full-width digits and punctuation, covered in [converting a Chinese PDF to Excel](/blog/chinese-pdf-to-excel).

## Right-to-left column order in an Arabic statement sheet

An Arabic statement prints its table right to left, with the date column at the right edge and the balance at the left. A converter may keep that visual order, putting the balance in column A, or put the date first. Neither is wrong as data, but your formulas and imports need to know which you have.

Decide on one order and stick to it. For a sheet you will sum, filter or import into accounting software, the usual choice is date, description, debit, credit, balance from left to right, because most import templates expect it. You can still read it right to left: Excel can show a whole sheet from right to left, with column A at the right edge. The setting appears when a right-to-left language is enabled in Office. This changes only the display, so formulas and imports see the same columns either way.

Do not reverse column order with copy and paste after conversion if there is any doubt about which column is which. Instead, check one row against the PDF. The balance column is easy to identify: it changes on every row by exactly that row's amount.

### A worked example, one statement line

The table shows one example line from an imaginary Arabic statement. The values are made up for illustration.

| Field | What the PDF shows | What a naive copy can give | Cleaned value |
| --- | --- | --- | --- |
| Date | ١٥/٠٣/٢٠٢٦ | ٦٢٠٢/٣٠/٥١ | 15/03/2026 |
| Description | تحويل إلى أحمد | دمحأ ىلإ ليوحت | تحويل إلى أحمد |
| Debit | ١٬٢٥٠٫٠٠ | ٠٠٫٠٥٢٬١ | 1250.00 |
| Balance | ٨٬٤٠٠٫٥٠ | ٠٥٫٠٠٤٬٨ | 8400.50 |

The naive column shows the worst case, a tool that reverses every character in the line, so the Arabic words and the digits both come out backwards. A tool that only reads drawing order reverses the Arabic words but leaves the digits alone. Either way every value has the right characters in the wrong order, which is why the order must be fixed before the digits are converted.

## Do Arabic statement amounts keep their sign and decimals?

Arabic statement amounts keep their decimals only if the Arabic decimal separator is replaced, and their sign only if the minus is found where the bank put it. In right-to-left text, a minus sign or a "Dr" marker may sit on the far side of the number from where Excel looks, so a debit can arrive as a positive value.

Common patterns to look for:

- **A trailing minus,** such as 1250.00- after conversion. Excel's Text to Columns wizard has an advanced option for trailing minus signs that turns these into negatives.
- **A separate debit and credit column,** where no sign is printed at all. The column the amount sits in is the sign, so keep the two columns apart until you combine them deliberately.
- **An Arabic word for debit or credit** beside the amount. Search for it, set the sign, then delete the word.

After cleaning, test the balance arithmetic on one page. Each row's balance should equal the previous balance plus credits minus debits. A wrong sign shows up immediately as a balance that drifts by twice the amount.

## Arabic PDF files that need OCR instead of text extraction

Some Arabic PDFs need OCR, reading the page as an image, because their text layer is beyond repair. This happens when the fonts carry no map from shapes to letters, when every word is reversed and broken into fragments, or when the statement is a scan. OCR with an Arabic language model reads the page as a person would.

Signs that extraction will not work:

- copied text is a run of boxes or Latin symbols rather than Arabic letters
- each letter copies as a separate word, with spaces between them
- the file is a scan or a phone photo, so there is no text to copy at all

OCR has its own limits with Arabic. Joined script with dots above and below letters is harder to read than separated Latin print, so check payee names closely. Account numbers covered with black boxes are a separate matter again; if your statement has them, [what happens to redacted text in a PDF to Excel conversion](/blog/redacted-pdf-to-excel) explains what may still sit under the box.

A converter that promises a single click has to make all of these decisions for you: reading order, digits, letter shapes and columns. [What happens inside a one click PDF to Excel conversion](/blog/one-click-pdf-to-excel) walks through those steps so you know what to check afterwards.

FastBankConvert reads a statement's layout rather than relying on a fixed list of banks, and it tells you before download when the balances do not add up, which is often the first sign that a sign or digit went wrong. It is getting ready to open; [join through Get started](/get-started) to hear when trial conversions begin.