Docjure makes reading and writing Office documents in Clojure easy.
(use 'dk.ative.docjure.spreadsheet)
;; Load a spreadsheet and read the first two columns from the
;; price list sheet:
(->> (load-workbook "spreadsheet.xlsx")
(select-sheet "Price List")
(select-columns {:A :name, :B :price}))
;=> [{:name "Foo Widget", :price 100}, {:name "Bar Widget", :price 200}]
If you want to read a single cell value, you can use the select-cell
function which
takes an Excel-style cell reference (A2) and returns the cell. In order to get the
actual value, use read-cell
(use 'dk.ative.docjure.spreadsheet)
(read-cell
(->> (load-workbook "spreadsheet.xslx")
(select-sheet "Price List")
(select-cell "A1")))
This example loads a workbook from a named file. In the case of running in the application server, the file typically resides in the resources directory, and it's not on the caller's path. To cover this scenario, we provide the function 'load-workbook-from-resource' that takes a named resource as the parameter. After a minor modification, the same example will look like:
(->> (load-workbook-from-resource "spreadsheet.xlsx")
(select-sheet "Price List")
(select-columns {:A :name, :B :price}))
The function 'load-workbook' is a multimethod, and the first example takes a file name as a parameter. The overloaded version of 'load-workbook' takes an InputStream. This may be useful when uploading a workbook to the server over HTTP connection as multipart form data. In this case, the web framework passes a byte buffer, and the example should be modified as (note that you have to use 'with-open' to ensure that the stream will be closed):
(with-open [stream (clojure.java.io/input-stream bytes)]
(->> (load-workbook stream)
(select-sheet "Price List")
(select-columns {:A :name, :B :price})))
This example creates a spreadsheet with a single sheet named "Price List". It has three rows. We apply a style of yellow background colour and bold font to the top header row, then save the spreadsheet.
(use 'dk.ative.docjure.spreadsheet)
;; Create a spreadsheet and save it
(let [wb (create-workbook "Price List"
[["Name" "Price"]
["Foo Widget" 100]
["Bar Widget" 200]])
sheet (select-sheet "Price List" wb)
header-row (first (row-seq sheet))]
(do
(set-row-style! header-row (create-cell-style! wb {:background :yellow,
:font {:bold true}}))
(save-workbook! "spreadsheet.xlsx" wb)))
This example creates a spreadsheet with multiple sheets. Simply add more
sheet-name and data pairs. To create a sheet with no data, pass nil
as
the data argument.
(use 'dk.ative.docjure.spreadsheet)
;; Create a spreadsheet and save it
(let [wb (create-workbook "Squares"
[["N" "N^2"]
[1 1]
[2 4]
[3 9]]
"Cubes"
[["N" "N^3"]
[1 1]
[2 8]
[3 27]])]
(save-workbook! "exponents.xlsx" wb)))
Docjure allows you not only to evaluate a formula cell in a speadsheet, it also
provides a way of exposing a formla in a cell as a Clojure function using the
cell-fn
function.
(use 'dk.ative.docjure.spreadsheet)
;; Load a speadsheet and take the first sheet, construct a function from cell A2, taking
;; A1 as input.
(def formula-from-a2 (cell-fn "A2"
(first (sheet-seq (load-workbook "spreadsheet.xslx")))
"A1"))
;; Returns value of cell A2, as if value in cell A1 were 1.0
(formula-from-a2 1.0)
If the spreadsheet being read contains cells with errors the default behaviour of the library is to return a keyword representing the error as the cell value.
For example, given a spreadsheet with errors:
(use 'dk.ative.docjure.spreadsheet)
(def sample-cells (->> (load-workbook "spreadsheet.xlsx")
(sheet-seq)
(mapcat cell-seq)))
sample-cells
;=> (#<XSSFCell 15.0> #<XSSFCell NA()> #<XSSFCell 35.0> #<XSSFCell 13/0> #<XSSFCell 33.0> #<XSSFCell 96.0>)
Reading error cells, or cells that evaluate to an error (e.g. divide by
zero) returns a keyword representing the type of error from
read-cell
.
(->> sample-cells
(map read-cell))
;=> (15.0 :NA 35.0 :DIV0 33.0 96.0)
How you handle errors will depend on your application. You may want to replace specific errors with a default value and remove others for example:
(->> sample-cells
(map read-cell)
(map #(get {:DIV0 0.0} % %))
(remove keyword?))
;=> (15.0 35.0 0.0 33.0 96.0)
The following is a list of all possible error values:
#{:VALUE :DIV0 :CIRCULAR_REF :REF :NUM :NULL :FUNCTION_NOT_IMPLEMENTED :NAME :NA}
It's worth understanding a bit about the underlying structure of a spreadsheet before you start iterating over the contents.
Spreadsheets are designed to be sparse - not all rows in the spreadsheet must physically exist, and not all cells in a row must physically exist. This is how you can create data at ZZ:65535 without using huge amounts of storage.
Thus each cell can be in 3 states - with data, blank, or nonexistent (null). There's a special type CellType.BLANK for blank cells, but missing cells are just returned as nil.
Similarly rows can exist with cells, or exist but be empty, or they can not exist at all.
Prior to Docjure 1.11 the iteration functions wrapped the underlying Apache POI iterators, which skipped over missing data - this could cause surprising behaviour, especially when there were missing cells inside tabular data.
Since Docjure 1.11 iteration now returns nil
values for missing rows and cells - this is a breaking change - any code that calls row-seq
or cell-seq
now needs to deal with possible nil values.
You can iterate over all the rows in a worksheet with row-seq
:
(->> (load-workbook "test.xls")
(select-sheet "first")
row-seq)
This will return a sequence of org.apache.poi.usermodel.Row
objects, or nil
for any missing rows. You can use (remove nil? (row-seq ...) )
if you are happy to ignore missing rows, but then be aware the nth result in the sequence might not match the nth row in the spreadsheet.
You can iterate over all the cells in a row with cell-seq
- this returns a sequence of org.apache.poi.usermodel.Cell
objects, or nil
for missing cells. Note that (read-cell nil)
returns nil
so it's safe to apply read-cell
to the results of cell-seq
(->> (load-workbook "test.xls")
(select-sheet "first")
row-seq
(remove nil?)
(map cell-seq)
(map #(map read-cell %)))
For example, if you run the above snippet on a sparse spreadsheet like:
| First Name | Middle Name | Last Name |
|---
| Edger | Allen | Poe |
| (missing row)
|
| John | (missing)
| Smith |
Then it will return:
(("First Name" "Middle Name" "Last Name")
("Edger" "Allen" "Poe")
("John" nil "Smith"))
The Docjure jar is distributed on Clojars. Here you can find both release builds and snapshot builds of pre-release versions.
If you are using the Leiningen build tool just add this line to the :dependencies list in project.clj to use it:
[dk.ative/docjure "1.11.0"]
Remember to issue the 'lein deps' command to download it.
(defproject some.cool/project "1.0.0-SNAPSHOT"
:description "Spreadsheet magic using Docjure"
:dependencies [[org.clojure/clojure "1.8.0"]
[dk.ative/docjure "1.11.0"]])
You need to install the Leiningen build tool to build the library. You can get it here: Leiningen
The library uses the Apache POI library which will be downloaded by the "lein deps" command.
Then build the library:
lein deps
lein compile
lein test
To run the tests on all supported Clojure versions use:
lein all test
Copyright (c) 2009-2016 Martin Jul
Docjure is licensed under the MIT License. See the LICENSE file for the license terms.
Docjure uses the Apache POI library, which is licensed under the Apache License v2.0.
For more information on Apache POI refer to the Apache POI web site.
Martin Jul
- Email: [email protected]
- Twitter: mjul
- GitHub: mjul
This library includes great contributions from
- Carl Baatz (cbaatz)
- Michael van Acken (mva)
- Ragnar Dahlén (ragnard)
- Vijay Kiran (vijaykiran)
- Jon Neale (jonneale)
- "Naipmoro" (naipmoro)
- Nikolay Durygin (nidu)
- Oliver Holworthy (oholworthy)
- "rakhra" (rakhra)
- Igor Tovstopyat-Nelip (igortn)
- Dino Kovač (reisub)
- Lars Trieloff (trieloff)
- Jens Bendisposto (bendisposto)
- Stuart Hinson (stuarth)
- Dan Petranek (dpetranek)
- Aleksander Madland Stapnes (madstap)
- Korny Sietsma (kornysietsma)
- Antti Virtanen (lokori)
Thank you very much!