HLEDGER-WEB(1) hledger User Manuals HLEDGER-WEB(1)

hledger-web - robust, friendly plain text accounting (Web version)

hledger-web [--serve|--serve-api] [OPTS] [ARGS]

hledger web -- [--serve|--serve-api] [OPTS] [ARGS]

This manual is for hledger's web interface, version 1.31. See also the hledger manual for common concepts and file formats.

hledger is a robust, user-friendly, cross-platform set of programs for tracking money, time, or any other commodity, using double-entry accounting and a simple, editable file format. hledger is inspired by and largely compatible with ledger(1), and largely interconvertible with beancount(1).

hledger-web is a simple web application for browsing and adding transactions. It provides a more user-friendly UI than the hledger CLI or hledger-ui TUI, showing more at once (accounts, the current account register, balance charts) and allowing history-aware data entry, interactive searching, and bookmarking.

hledger-web also lets you share a journal with multiple users, or even the public web. There is no access control, so if you need that you should put it behind a suitable web proxy. As a small protection against data loss when running an unprotected instance, it writes a numbered backup of the main journal file (only) on every edit.

Like hledger, it reads from (and appends to) a journal file specified by the LEDGER_FILE environment variable (defaulting to $HOME/.hledger.journal); or you can specify files with -f options. It can also read timeclock files, timedot files, or any CSV/SSV/TSV file with a date field. (See hledger(1) -> Input for details.)

hledger-web can be run in three modes:

Transient mode (the default): your default web browser will be opened to show the app if possible, and the app exits automatically after two minutes of inactivity (no requests received and no open browser windows viewing it).
With --serve: the app runs without stopping, and without opening a browser.
With --serve-api: only the JSON API is served.

In all cases hledger-web runs as a foreground process, logging requests to stdout.

Command-line options and arguments may be used to set an initial filter on the data. These filter options are not shown in the web UI, but it will be applied in addition to any search query entered there.

hledger-web provides the following options:

serve and log requests, don't browse or auto-exit after timeout
like --serve, but serve only the JSON web API, without the server-side web UI
listen on this IP address (default:
listen on this TCP port (default: 5000)
use a unix domain socket file to listen for requests instead of a TCP socket. Implies --serve. It can only be used if the operating system can provide this type of socket.
set the base url (default: http://IPADDR:PORT). Note: affects url generation but not route parsing. Can be useful if running behind a reverse web proxy that does path rewriting.
set the static files url (default: BASEURL/static). hledger-web normally serves static files itself, but if you wanted to serve them from another server for efficiency, you would set the url with this.
enable the view, add, and/or manage capabilities (default: view,add)
read capabilities to enable from a HTTP header, like X-Sandstorm-Permissions (default: disabled)
run hledger-web's tests and exit. hspec test runner args may follow a --, eg: hledger-web --test -- --help

By default the server listens on IP address, accessible only to local requests. You can use --host to change this, eg --host to listen on all configured addresses.

Similarly, use --port to set a TCP port other than 5000, eg if you are running multiple hledger-web instances.

Both of these options are ignored when --socket is used. In this case, it creates an AF_UNIX socket file at the supplied path and uses that for communication. This is an alternative way of running multiple hledger-web instances behind a reverse proxy that handles authentication for different users. The path can be derived in a predictable way, eg by using the username within the path. As an example, nginx as reverse proxy can use the variable $remote_user to derive a path from the username used in a HTTP basic authentication. The following proxy_pass directive allows access to all hledger-web instances that created a socket in /tmp/hledger/:

proxy_pass http://unix:/tmp/hledger/${remote_user}.socket;

You can use --base-url to change the protocol, hostname, port and path that appear in hyperlinks, useful eg for integrating hledger-web within a larger website. The default is http://HOST:PORT/ using the server's configured host address and TCP port (or http://HOST if PORT is 80).

With --file-url you can set a different base url for static files, eg for better caching or cookie-less serving on high performance websites.

hledger-web also supports many of hledger's general options (and the hledger manual's command line tips also apply here):

-h --help
show general or COMMAND help
show general or COMMAND user manual with man
show general or COMMAND user manual with info
show general or ADDONCMD version
show debug output (levels 1-9, default: 1)

-f FILE --file=FILE
use a different input file. For stdin, use - (default: $LEDGER_FILE or $HOME/.hledger.journal)
Conversion rules file to use when reading CSV (default: FILE.rules)
Field separator to expect when reading CSV (default: ',')
rename accounts named OLD to NEW
anonymize accounts and payees
use some other field or tag for the account name
-I --ignore-assertions
disable balance assertion checks (note: does not disable balance assignments)
-s --strict
do extra error checking (check that all posted accounts are declared)

-b --begin=DATE
include postings/txns on or after this date (will be adjusted to preceding subperiod start when using a report interval)
-e --end=DATE
include postings/txns before this date (will be adjusted to following subperiod end when using a report interval)
-D --daily
multiperiod/multicolumn report by day
-W --weekly
multiperiod/multicolumn report by week
-M --monthly
multiperiod/multicolumn report by month
-Q --quarterly
multiperiod/multicolumn report by quarter
-Y --yearly
multiperiod/multicolumn report by year
-p --period=PERIODEXP
set start date, end date, and/or reporting interval all at once using period expressions syntax
match the secondary date instead (see command help for other effects)
override today's date (affects relative smart dates, for tests/examples)
-U --unmarked
include only unmarked postings/txns (can combine with -P or -C)
-P --pending
include only pending postings/txns
-C --cleared
include only cleared postings/txns
-R --real
include only non-virtual postings
-NUM --depth=NUM
hide/aggregate accounts or postings more than NUM levels deep
-E --empty
show items with zero amount, normally hidden (and vice-versa in hledger-ui/hledger-web)
-B --cost
convert amounts to their cost/selling amount at transaction time
-V --market
convert amounts to their market value in default valuation commodities
-X --exchange=COMM
convert amounts to their market value in commodity COMM
convert amounts to cost or market value, more flexibly than -B/-V/-X
infer conversion equity postings from costs
infer costs from conversion equity postings
use costs as additional market prices, as if they were P directives
generate transactions from periodic rules, between the latest recorded txn and 6 months from today, or during the specified PERIOD (= is required). Auto posting rules will be applied to these transactions as well. Also, in hledger-ui make future-dated transactions visible.
generate extra postings by applying auto posting rules to all txns (not just forecast txns)
add visible tags indicating transactions or postings which have been generated/modified
Override the commodity style in the output for the specified commodity. For example 'EUR1.000,00'.
--color=WHEN (or --colour=WHEN)
Should color-supporting commands use ANSI color codes in text output. 'auto' (default): whenever stdout seems to be a color-supporting terminal. 'always' or 'yes': always, useful eg when piping output into 'less -R'. 'never' or 'no': never. A NO_COLOR environment variable overrides this.
Show prettier output, e.g. using unicode box-drawing characters. Accepts 'yes' (the default) or 'no' ('y', 'n', 'always', 'never' also work). If you provide an argument you must use '=', e.g. '--pretty=yes'.

When a reporting option appears more than once in the command line, the last one takes precedence.

Some reporting options can also be written as query arguments.

By default, hledger-web allows anyone who can reach it to view the journal and to add new transactions, but not to change existing data.

You can restrict who can reach it by

setting the IP address it listens on (see --host above). By default it listens on, accessible to all users on the local machine.
putting it behind an authenticating proxy, using eg apache or nginx
custom firewall rules

You can restrict what the users who reach it can do, by

using the --capabilities=CAP[,CAP..] flag when you start it, enabling one or more of the following capabilities. The default value is view,add:
view - allows viewing the journal file and all included files
add - allows adding new transactions to the main journal file
manage - allows editing, uploading or downloading the main or included files
using the --capabilities-header=HTTPHEADER flag to specify a HTTP header from which it will read capabilities to enable. hledger-web on Sandstorm uses the X-Sandstorm-Permissions header to integrate with Sandstorm's permissions. This is disabled by default.

If you enable the manage capability mentioned above, you'll see a new "spanner" button to the right of the search form. Clicking this will let you edit, upload, or download the journal file or any files it includes.

Note, unlike any other hledger command, in this mode you (or any visitor) can alter or wipe the data files.

Normally whenever a file is changed in this way, hledger-web saves a numbered backup (assuming file permissions allow it, the disk is not full, etc.) hledger-web is not aware of version control systems, currently; if you use one, you'll have to arrange to commit the changes yourself (eg with a cron job or a file watcher like entr).

Changes which would leave the journal file(s) unparseable or non-valid (eg with failing balance assertions) are prevented. (Probably. This needs re-testing.)

hledger-web detects changes made to the files by other means (eg if you edit it directly, outside of hledger-web), and it will show the new data when you reload the page or navigate to a new page. If a change makes a file unparseable, hledger-web will display an error message until the file has been fixed.

(Note: if you are viewing files mounted from another machine, make sure that both machine clocks are roughly in step.)

In addition to the web UI, hledger-web also serves a JSON API that can be used to get data or add new transactions. If you want the JSON API only, you can use the --serve-api flag. Eg:

$ hledger-web -f examples/sample.journal --serve-api

You can get JSON data from these routes:


Eg, all account names in the journal (similar to the accounts command). (hledger-web's JSON does not include newlines, here we use python to prettify it):

$ curl -s | python -m json.tool

"liabilities:debts" ]

Or all transactions:

$ curl -s | python -m json.tool

"tcode": "",
"tcomment": "",
"tdate": "2008-01-01",
"tdate2": null,
"tdescription": "income",
"tindex": 1,
"tpostings": [
"paccount": "assets:bank:checking",
"pamount": [
"acommodity": "$",
"aismultiplier": false,
"aprice": null, ...

Most of the JSON corresponds to hledger's data types; for details of what the fields mean, see the Hledger.Data.Json haddock docs and click on the various data types, eg Transaction. And for a higher level understanding, see the journal docs.

In some cases there is outer JSON corresponding to a "Report" type. To understand that, go to the Hledger.Web.Handler.MiscR haddock and look at the source for the appropriate handler to see what it returns. Eg for /accounttransactions it's getAccounttransactionsR, returning a "accountTransactionsReport ...". Looking up the haddock for that we can see that /accounttransactions returns an AccountTransactionsReport, which consists of a report title and a list of AccountTransactionsReportItem (etc).

You can add a new transaction to the journal with a PUT request to /add, if hledger-web was started with the add capability (enabled by default). The payload must be the full, exact JSON representation of a hledger transaction (partial data won't do). You can get sample JSON from hledger-web's /transactions or /accounttransactions, or you can export it with hledger-lib, eg like so:

.../hledger$ stack ghci hledger-lib
>>> writeJsonFile "txn.json" (head $ jtxns samplejournal)
>>> :q

Here's how it looks as of hledger-1.17 (remember, this JSON corresponds to hledger's Transaction and related data types):


"tcomment": "",
"tpostings": [
"pbalanceassertion": null,
"pstatus": "Unmarked",
"pamount": [
"aprice": null,
"acommodity": "$",
"aquantity": {
"floatingPoint": 1,
"decimalPlaces": 10,
"decimalMantissa": 10000000000
"aismultiplier": false,
"astyle": {
"ascommodityside": "L",
"asdigitgroups": null,
"ascommodityspaced": false,
"asprecision": 2,
"asdecimalpoint": "."
"ptransaction_": "1",
"paccount": "assets:bank:checking",
"pdate": null,
"ptype": "RegularPosting",
"pcomment": "",
"pdate2": null,
"ptags": [],
"poriginal": null
"pbalanceassertion": null,
"pstatus": "Unmarked",
"pamount": [
"aprice": null,
"acommodity": "$",
"aquantity": {
"floatingPoint": -1,
"decimalPlaces": 10,
"decimalMantissa": -10000000000
"aismultiplier": false,
"astyle": {
"ascommodityside": "L",
"asdigitgroups": null,
"ascommodityspaced": false,
"asprecision": 2,
"asdecimalpoint": "."
"ptransaction_": "1",
"paccount": "income:salary",
"pdate": null,
"ptype": "RegularPosting",
"pcomment": "",
"pdate2": null,
"ptags": [],
"poriginal": null
"ttags": [],
"tsourcepos": {
"tag": "JournalSourcePos",
"contents": [
"tdate": "2008-01-01",
"tcode": "",
"tindex": 1,
"tprecedingcomment": "",
"tdate2": null,
"tdescription": "income",
"tstatus": "Unmarked" }

And here's how to test adding it with curl. This should add a new entry to your journal:

$ curl -X PUT -H 'Content-Type: application/json' --data-binary @txn.json

You can add --debug[=N] to the command line to log debug output. N ranges from 1 (least output, the default) to 9 (maximum output). Typically you would start with 1 and increase until you are seeing enough. Debug output goes to stderr, interleaved with the requests logged on stdout. To capture debug output in a log file instead, you can usually redirect stderr, eg:

hledger-web --debug=3 2>hledger-web.log.

LEDGER_FILE The main journal file to use when not specified with -f/--file. Default: $HOME/.hledger.journal.

We welcome bug reports in the hledger issue tracker (shortcut: http://bugs.hledger.org), or on the #hledger chat or hledger mail list (https://hledger.org/support).

Some known issues:

Does not work well on small screens, or in text-mode browsers.

Simon Michael <simon@joyful.com> and contributors.
See http://hledger.org/CREDITS.html

Copyright 2007-2023 Simon Michael and contributors.

Released under GNU GPL v3 or later.

hledger(1), hledger-ui(1), hledger-web(1), ledger(1)

September 2023 hledger-web-1.31