[ Languages: English | 日本語 (Japanese) ]

ble.sh (/blɛʃ/) ―Bash Line Editor―

[ README | Manual | Q&A | contrib | Recipes ]

Bash Line Editor (ble.sh^†1) is a command line editor written in pure Bash^†2 which replaces the default GNU Readline.

The current devel version is 0.4. This script supports Bash 3.0 or higher although we recommend using ble.sh with release versions of Bash 4.0 or higher. The POSIX standard utilities are also required. Currently, only UTF-8 encoding is supported for non-ASCII characters. This script is provided under the BSD License (3-clause BSD license).

Quick instructions

There are two ways to get ble.sh: to get the source using git and build ble.sh, or to download the nightly build using curl or wget. See Sec 1.1 and Sec 1.2 for the details of trial and installation. See Sec 1.3 for the details of the setup of your ~/.bashrc.

[!NOTE] If you want to use fzf with ble.sh, you need to check Sec 2.8.

Features

• Syntax highlighting: Highlight command lines input by users as in fish and zsh-syntax-highlighting. Unlike the simple highlighting in zsh-syntax-highlighting, ble.sh performs syntactic analysis to enable the correct highlighting of complex structures such as nested command substitutions, multiple here documents, etc. Highlighting colors and styles are fully configurable.
• Enhanced completion: Extend completion by syntax-aware completion, completion with quotes and parameter expansions in prefix texts, ambiguous candidate generation, etc. Also, menu-complete supports the selection of candidates in the menu (candidate list) by cursor keys, TAB, and S-TAB. The feature auto-complete supports the automatic suggestion of completed texts as in fish and zsh-autosuggestions (with Bash 4.0+). The feature menu-filter integrates automatic filtering of candidates into menu completion (with Bash 4.0+). There are other functionalities such as dabbrev and sabbrev like zsh abbreviations or zsh-abbr.
• Vim editing mode: Enhance readline's vi editing mode available with set -o vi. Vim editing mode supports various vim modes such as char/line/block visual/select mode, replace mode, command mode, operator pending mode as well as insert mode and normal mode. Vim editing mode supports various operators, text objects, registers, keyboard macros, marks, etc. It also provides vim-surround as an option.
• Other interesting features include status line, history share, right prompt, transient prompt, xterm title, etc.

Note: ble.sh does not provide specific settings of the prompt, aliases, functions, etc. ble.sh provides a more fundamental infrastructure so that users can set up their own prompt, aliases, functions, etc. Of course ble.sh can be used in combination with other Bash configurations such as bash-it and oh-my-bash.

Demo (version 0.2)

History and roadmap

My little experiment took place in one corner of my bashrc at the end of May 2013 after I enjoyed an article on zsh-syntax-highlighting. I initially thought something could be achieved by writing a few hundred lines of code but soon realized that everything needs to be re-implemented for the authentic support of syntax highlighting in Bash. I decided to make it as an independent script ble.sh. The name stemmed from that of Zsh's line editor, ZLE (Zsh Line Editor), but suffixed with .sh for the implication of being written in a shell script. I'm occasionally asked about the pronunciation of ble.sh, but you can pronounce it as you like. After the two-week experiment, I was satisfied with my conclusion that it is possible to implement a full-featured line editor in Bash that satisfies the actual daily uses. The real efforts to improve the prototype implementation for real uses started in February 2015. I released the initial version in the next December. Until then, the basic part of the line editor was completed. The implementation of vim mode was started in September 2017 and completed in the next March. I started working on the enhancement of the completion in August 2018 and released it in the next February.

• 2013-06 v0.0 -- prototype
• 2015-12 v0.1 -- Syntax highlighting [v0.1.15]
• 2018-03 v0.2 -- Vim mode [v0.2.7]
• 2019-02 v0.3 -- Enhanced completion [v0.3.4]
• 20xx-xx v0.4 (plan) -- programmable highlighting [nightly build]
• 20xx-xx v0.5 (plan) -- TUI configuration
• 20xx-xx v0.6 (plan) -- error diagnostics?

Limitations and assumptions

There are some limitations due to the way ble.sh is implemented. Also, some user configurations or other Bash frameworks may conflict with ble.sh. For example,

• By default, ble.sh does not set PIPESTATUS for the previous command line because it adds extra execution costs. Instead, the array BLE_PIPESTATUS contains the values of PIPESTATUS of the previous command line. If you need to access the values directly through the variable PIPESTATUS, please set the option bleopt exec_restore_pipestatus=1.
• By default, ble.sh assumes that PROMPT_COMMAND and PRECMD hooks do not change the cursor position and the layout in the terminal display to offer smooth rendering. If you have settings that output texts or changes the cursor position in PROMPT_COMMAND and PRECMD, please set the option bleopt prompt_command_changes_layout=1.
• ble.sh assumes that common variable names and environment variables (such as LC_*) are not used for the global readonly variables. In Bash, global readonly variables take effect in any scope including the local scope of the function, which means that we cannot even define a local variable that has the same name as a global readonly variable. This is not the problem specific to ble.sh, but any Bash framework may suffer from the global readonly variables. It is generally not recommended to define global readonly variables in Bash except for the security reasoning (Refs. [1], [2], [3]). Also, ble.sh overrides the builtin readonly with a shell function to prevent it from making global variables readonly. It allows only uppercase global variables and _* to become readonly except _ble_*, __ble_*, and some special uppercase variables.
• ble.sh overrides Bash's built-in commands (trap, readonly, bind, history, read, and exit) with shell functions to adjust the behavior of each built-in command and prevent them from interfering with ble.sh. If the user or another framework directly calls the original builtins through builtin BUILTIN, or if the user or another framework replaces the shell functions, the behavior is undefined.
• The shell and terminal settings for the line editor and the command execution are different. ble.sh adjusts them for the line editor and try to restore the settings for the command execution. However, there are settings that cannot be restored or are intentionally not restored for various reasons. Some of them are summarized on a wiki page.

Criticism

• ^†1Q. It is hard to pronounce "ble-sh". How should I pronounce it? --- A. The easiest pronunciation of ble.sh that users use is /blɛʃ/, but you can pronounce it as you like. I do not specify the canonical way of pronouncing ble.sh. In fact, I personally call it simply /biːɛliː/ or verbosely read it as /biːɛliː dɑt ɛseɪtʃ/ in my head.
• ^†2Q. It cannot be pure Bash because the user should be able to input and run external commands. What does the pure Bash mean? --- A. It means that the core part of the line editor is written in pure Bash. Of course, the external commands will be run when the user inputs them and requests the execution of them. In addition, before and after the execution of user commands, ble.sh relies on POSIX stty to set up the correct TTY states for user commands. It also uses other POSIX utilities for acceleration in some parts of initialization and cleanup code, processing of large data in completions, pasting large data, etc. The primary goal of the ble.sh implementation is not being pure Bash, but the performance in the Bash implementation with the POSIX environment. Being pure Bash is usually useful for reducing the fork/exec cost, but if implementation by external commands is more efficient in specific parts, ble.sh will use the external commands there.
• ^†3Q. Why does ble.sh use make to generate the script files? You should not use make for a script framework. --- A. Because it is not a good idea to directly edit a large script file of tens of thousands of lines. I split the codebase of ble.sh into source files of reasonable sizes and edit the source files. In the build process, some source files are combined to form the main script ble.sh, and some other files are arranged in appropriate places. The reason for combining the files into one file instead of sourcing the related files in runtime is to minimize the shell startup time, which has a large impact on the shell experience. Opening and reading many files can take a long time. Some people seem to be angry about ble.sh using make to build and arrange script files. They seem to believe that one always needs to use make with C/C++/Fortran compilers to generate binaries. They complain about ble.sh's make process, but it comes from the lack of knowledge on the general principle of make. Some people seem to be angry about ble.sh having C/C++ source codes in the repository, but they are used to update the Unicode character table from the Unicode database when a new Unicode standard appears. The generated table is included in the repository: canvas.GraphemeClusterBreak.sh, canvas.c2w.musl.sh, canvas.c2w.sh, and canvas.emoji.sh, so there is no need to run these C/C++ programs in the build process. Another C file is used as an adapter in an old system MSYS1, which is used with an old compiler toolchain in Windows, but it will never be used in Unix-like systems. Each file used in the build process is explained in make/README.md.

1 Usage

1.1 Build from source^†

Generate

To generate ble.sh, gawk (GNU awk) and gmake (GNU make) (in addition to Bash and POSIX standard utilities) are required. The file ble.sh can be generated using the following commands. If you have GNU make installed on gmake, please use gmake instead of make.

git clone --recursive https://github.com/akinomyoga/ble.sh.git
cd ble.sh
make

A script file ble.sh will be generated in the directory ble.sh/out.

Try

Then, you can load ble.sh in the Bash session using the source command:

source out/ble.sh

Install

To install ble.sh in a specified directory, use make install.

# INSTALL to ~/.local/share/blesh and ~/.local/share/doc/blesh
make install

# INSTALL into a specified directory
make install INSDIR=/path/to/blesh

The install locations of ble.sh and related script files can be specified by make variable INSDIR. The locations of the license files and the documentation files can be specified by make variables INSDIR_LICENSE and INSDIR_DOC, respectively. When INSDIR is specified, the default values of INSDIR_LICENSE and INSDIR_DOC are $INSDIR/licenses and $INSDIR/doc. When INSDIR and the below-mentioned DESTDIR/PREFIX are not specified, the default values of INSDIR, INSDIR_LICENSES, and INSDIR_DOC are $data/blesh, $data/blesh/licenses, and $data/doc/blesh, respectively, where $data represents ${XDG_DATA_HOME:-$HOME/.local/share}/blesh.

When USE_DOC=no is specified, the documentation files are disabled.

By default, the comment lines and blank lines in the script files are stripped in the installation process. If you would like to keep these lines in the script files, please specify the argument strip_comment=no to make.

To set up .bashrc see Sec. 1.3.

Package

Package maintainers may use make variables DESTDIR and PREFIX to quickly set up the default values for INSDIR, INSDIR_LICENSE, and INSDIR_DOC.

# PACKAGE - Example 1
make install DESTDIR=/tmp/blesh-package PREFIX=/usr/local

# PACKAGE - Example 2
make install DESTDIR="$build" PREFIX="$prefix" \
  INSDIR_LICENSE="$build/$prefix/share/licenses/blesh"

# PACKAGE - Example 3
make install DESTDIR="$build" PREFIX="$prefix" \
  INSDIR_LICENSE="$build/$prefix/share/blesh/doc" \
  INSDIR_DOC="$build/$prefix/share/blesh/doc"

# PACKAGE - Example 4
make install USE_DOC=no DESTDIR="$build" PREFIX="$prefix"

If make variable DESTDIR or PREFIX is specified instead of INSDIR, the value of INSDIR is set to $DESTDIR/$PREFIX/share/blesh, and the default install locations of the license and documentation files, INSDIR_LICENSE and INSDIR_DOC, will be $DESTDIR/$PREFIX/share/doc/blesh.

_package.bash

When you want to tell ble.sh the way to update the package for ble-update, you can place _package.bash at ${prefix}/share/blesh/lib/_package.bash. The file _package.bash is supposed to define a shell variable and a shell function as illustrated in the following example (please replace XXX with a name representing the package management system):

# ${prefix}/share/blesh/lib/_package.bash

_ble_base_package_type=XXX

function ble/base/package:XXX/update {
  update-the-package-in-a-proper-way
  return 0
}

When the shell function returns exit status 0, it means that the update has been successfully completed, and ble.sh will be reloaded automatically. When the shell function returns exit status 6, the timestamp of ble.sh will be checked so ble.sh is reloaded only when ble.sh is actually updated. When the shell function returns exit status 125, the default ble.sh update procedure is attempted. Otherwise, the updating procedure is canceled, where any message explaining situation should be output by the shell function. An example _package.bash for AUR can be found here.

1.2 Download a tarball^†

You can also download a tarball of ble.sh from GitHub releases. See each release page for the description of downloading, trial and installation. Many features are unavailable in the stable versions since they are significantly old compared to the devel version.

• Devel v0.4.0-devel3 (2023-04), nightly build
• Stable v0.3.4 (2019-02 fork) Enhanced completions
• Stable v0.2.7 (2018-03 fork) Vim mode
• Stable v0.1.15 (2015-12 fork) Syntax highlighting

1.3 Set up .bashrc^†

If you want to load ble.sh in interactive sessions of bash by default, usually one can just source ble.sh in ~/.bashrc, but a more reliable way is to add the following codes to your .bashrc file:

# bashrc

# Add this lines at the top of .bashrc:
[[ $- == *i* ]] && source /path/to/blesh/ble.sh --noattach

# your bashrc settings come here...

# Add this line at the end of .bashrc:
[[ ! ${BLE_VERSION-} ]] || ble-attach

Basically, when source /path/to/ble.sh and ble-attach are performed, standard streams (stdin, stdout, and stderr) should not be redirected but should be connected to the controlling TTY of the current session. Also, please avoid calling source /path/to/ble.sh in shell functions. The detailed conditions where the above more reliable setup is needed are explained in an answer in Discussion #254.

1.4 User settings ~/.blerc

User settings can be placed in the init script ~/.blerc (or ${XDG_CONFIG_HOME:-$HOME/.config}/blesh/init.sh if ~/.blerc is not available) whose template is available as the file blerc.template in the repository. The init script is a Bash script that is sourced during the load of ble.sh, so any shell commands can be used in ~/.blerc. If you want to change the default path of the init script, you can add the option --rcfile INITFILE to source ble.sh as the following example:

# in bashrc

# Example 1: ~/.blerc will be used by default
[[ $- == *i* ]] && source /path/to/blesh/ble.sh --noattach

# Example 2: /path/to/your/blerc will be used
[[ $- == *i* ]] && source /path/to/blesh/ble.sh --noattach --rcfile /path/to/your/blerc

1.5 Update

You need Git (git), GNU awk (gawk) and GNU make (make). For ble-0.3+, you can run ble-update in the session with ble.sh loaded:

$ ble-update

For ble.0.4+, you can also update it outside the ble.sh session using

$ bash /path/to/ble.sh --update

You can instead download the latest version by git pull and install it:

cd ble.sh   # <-- enter the git repository you already have
git pull
git submodule update --recursive --remote
make
make INSDIR="$HOME/.local/share/blesh" install

1.6 Uninstall

Basically you can simply delete the installed directory and the settings that the user added.

• Close all the ble.sh sessions (the Bash interactive sessions with ble.sh)
• Remove related user data. If you would like to keep them, you can skip these steps.
  • Remove the added lines in .bashrc.
  • Remove blerc files (~/.blerc or ~/.config/blesh/init.sh) if any.
  • Remove the state directory ~/.local/state/blesh if any.
• Remove the directory where ble.sh is installed. When you use out/ble.sh inside the working tree of the git repository, the installed directory is the directory of the repository. When you use ble.sh installed by make install, the installed directory is <PREFIX>/share/blesh where <PREFIX> (default: ~/.local) is the prefix specified to make install in the installation stage. When you use the version extracted from a tarball, the directory created by extracting the tarball is the installed directory.
• Remove the cache directory ~/.cache/blesh if any.
• Remove the temporary directory /tmp/blesh if any [ Only needed when your system does not automatically clear /tmp ].

2 Basic settings

Here, some of the settings for ~/.blerc are picked up. You can find useful settings also in Q\&A, Recipes, and contrib repository. The complete list of setting items can be found in the file blerc.template. For detailed explanations please refer to Manual.

2.1 Vim mode

For the vi/vim mode, check the wiki page.

2.2 Disable features

One of frequently asked questions is the way to disable a specific feature that ble.sh adds. Here the settings for disabling features are summarized.

# Disable syntax highlighting
bleopt highlight_syntax=

# Disable highlighting based on filenames
bleopt highlight_filename=

# Disable highlighting based on variable types
bleopt highlight_variable=

# Disable auto-complete (Note: auto-complete is enabled by default in bash-4.0+)
bleopt complete_auto_complete=
# Tip: you may instead specify the delay of auto-complete in millisecond
bleopt complete_auto_delay=300

# Disable auto-complete based on the command history
bleopt complete_auto_history=

# Disable ambiguous completion
bleopt complete_ambiguous=

# Disable menu-complete by TAB
bleopt complete_menu_complete=

# Disable menu filtering (Note: auto-complete is enabled by default in bash-4.0+)
bleopt complete_menu_filter=

# Disable EOF marker like "[ble: EOF]"
bleopt prompt_eol_mark=''
# Tip: you may instead specify another string:
bleopt prompt_eol_mark='⏎'

# Disable error exit marker like "[ble: exit %d]"
bleopt exec_errexit_mark=
# Tip: you may instead specify another string:
bleopt exec_errexit_mark=$'\e[91m[error %d]\e[m'

# Disable elapsed-time marker like "[ble: elapsed 1.203s (CPU 0.4%)]"
bleopt exec_elapsed_mark=
# Tip: you may instead specify another string
bleopt exec_elapsed_mark=$'\e[94m[%ss (%s %%)]\e[m'
# Tip: you may instead change the threshold of showing the mark
bleopt exec_elapsed_enabled='sys+usr>=10*60*1000' # e.g. ten minutes for total CPU usage

# Disable exit marker like "[ble: exit]"
bleopt exec_exit_mark=

# Disable some other markers like "[ble: ...]"
bleopt edit_marker=
bleopt edit_marker_error=

2.3 CJK Width

The option char_width_mode controls the width of the Unicode characters with East_Asian_Width=A (Ambiguous characters). Currently, four values emacs, west, east, and auto are supported. With the value emacs, the default width in emacs is used. With west, all the ambiguous characters have width 1 (Hankaku). With east, all the ambiguous characters have width 2 (Zenkaku). With auto, the width mode west or east is automatically chosen based on the terminal behavior. The default value is auto. The appropriate value should be chosen in accordance with your terminal behavior. For example, the value can be changed to west as:

bleopt char_width_mode='west'

2.4 Input Encoding

The option input_encoding controls the encoding scheme used in the decode of input. Currently UTF-8 and C are available. With the value C, byte values are directly interpreted as character codes. The default value is UTF-8. For example, the value can be changed to C as:

bleopt input_encoding='C'

2.5 Bell

The option edit_bell controls the behavior of the edit function (widget) called bell. It is a colon-separated list of the values vbell, abell, and visual. When a value is contained, the corresponding type of the bell is enabled. The value abell corresponds to the audible bell, which prints ASCII Control Character BEL (0x07) will be written to stderr. The value vbell corresponds to the visible bell, which shows the message in the terminal display. The value visual corresponds to the visual bell, which flashes the terminal screen by turning on the DECSCNM mode for a short moment. By default, only the audible bell is enabled.

The option vbell_default_message specifies the default message shown by the visual bell. The default value of this setting is ' Wuff, -- Wuff!! '. The option vbell_duration specifies the display duration of the visual-bell message. The unit is a millisecond. The default value is 2000. The option vbell_align specifies the position of vbell by left, center, or right.

For example, the audible bell can be disabled, and the visual bell can be set up as:

bleopt edit_bell=vbell vbell_{default_message=' BEL ',duration=3000,align=right}

2.6 Highlight Colors

The colors and attributes used in the syntax highlighting are controlled by the function ble-face. The following code reproduces the default configuration:

# highlighting related to editing
ble-face -s region                    bg=60,fg=231
ble-face -s region_target             bg=153,fg=black
ble-face -s region_match              bg=55,fg=231
ble-face -s region_insert             fg=27,bg=254
ble-face -s disabled                  fg=242
ble-face -s overwrite_mode            fg=black,bg=51
ble-face -s vbell                     reverse
ble-face -s vbell_erase               bg=252
ble-face -s vbell_flash               fg=green,reverse
ble-face -s prompt_status_line        fg=231,bg=240

# syntax highlighting
ble-face -s syntax_default            none
ble-face -s syntax_command            fg=brown
ble-face -s syntax_quoted             fg=green
ble-face -s syntax_quotation          fg=green,bold
ble-face -s syntax_escape             fg=magenta
ble-face -s syntax_expr               fg=33
ble-face -s syntax_error              bg=203,fg=231
ble-face -s syntax_varname            fg=202
ble-face -s syntax_delimiter          bold
ble-face -s syntax_param_expansion    fg=133
ble-face -s syntax_history_expansion  bg=94,fg=231
ble-face -s syntax_function_name      fg=99,bold
ble-face -s syntax_comment            fg=242
ble-face -s syntax_glob               fg=198,bold
ble-face -s syntax_brace              fg=37,bold
ble-face -s syntax_tilde              fg=63,bold
ble-face -s syntax_document           fg=100
ble-face -s syntax_document_begin     fg=100,bold
ble-face -s command_builtin_dot       fg=red,bold
ble-face -s command_builtin           fg=red
ble-face -s command_alias             fg=teal
ble-face -s command_function          fg=99
ble-face -s command_file              fg=green
ble-face -s command_keyword           fg=blue
ble-face -s command_jobs              fg=red
ble-face -s command_directory         fg=33,underline
ble-face -s command_suffix            fg=231,bg=28
ble-face -s command_suffix_new        fg=231,bg=124
ble-face -s filename_directory        underline,fg=33
ble-face -s filename_directory_sticky underline,fg=231,bg=26
ble-face -s filename_link             underline,fg=teal
ble-face -s filename_orphan           underline,fg=16,bg=224
ble-face -s filename_executable       underline,fg=green
ble-face -s filename_setuid           underline,fg=black,bg=220
ble-face -s filename_setgid           underline,fg=black,bg=191
ble-face -s filename_other            underline
ble-face -s filename_socket           underline,fg=cyan,bg=black
ble-face -s filename_pipe             underline,fg=lime,bg=black
ble-face -s filename_character        underline,fg=231,bg=black
ble-face -s filename_block            underline,fg=yellow,bg=black
ble-face -s filename_warning          underline,fg=red
ble-face -s filename_url              underline,fg=blue
ble-face -s filename_ls_colors        underline
ble-face -s varname_array             fg=orange,bold
ble-face -s varname_empty             fg=31
ble-face -s varname_export            fg=200,bold
ble-face -s varname_expr              fg=99,bold
ble-face -s varname_hash              fg=70,bold
ble-face -s varname_new               fg=34
ble-face -s varname_number            fg=64
ble-face -s varname_readonly          fg=200
ble-face -s varname_transform         fg=29,bold
ble-face -s varname_unset             fg=245
ble-face -s argument_option           fg=teal
ble-face -s argument_error            fg=black,bg=225

# highlighting for completions
ble-face -s auto_complete             fg=238,bg=254
ble-face -s menu_desc_default         none
ble-face -s menu_desc_type            ref:syntax_delimiter
ble-face -s menu_desc_quote           ref:syntax_quoted
ble-face -s menu_filter_fixed         bold
ble-face -s menu_filter_input         fg=16,bg=229

The current list of faces can be obtained by the following command (ble-face without arguments):

$ ble-face

The color codes can be checked in output of the function ble-color-show (defined in ble.sh):

$ ble-color-show

2.7 Key Bindings

Key bindings can be controlled with the shell function, ble-bind. For example, with the following setting, "Hello, world!" will be inserted on typing C-x h

ble-bind -f 'C-x h' 'insert-string "Hello, world!"'

The details on the key representation, such as C-x h in the above example, are described in Manual §3.1. The representations of Space, Tab, Enter, Backspace, Escape, etc. are described in Manual §3.1.1: The space is represented as SP, the tab key is represented as C-i or TAB depending on the terminal, the enter/return key is represented as C-m or RET depending on the terminal, and the backspace key is represented as C-?, DEL, C-h, or BS depending on the terminal. The representations of modified special keys such as Ctrl+Return and Shift+Return are described in Manual §3.6.4: If your terminal does not support modifyOtherKeys, you need to manually configure the escape sequences of modified special keys.

For another example, if you want to invoke a command on typing M-c, you can write it as follows:

ble-bind -c 'M-c' 'my-command'

Or, if you want to invoke a edit function (designed for Bash bind -x) on typing C-r, you can write it as follows:

ble-bind -x 'C-r' 'my-edit-function'

The existing key bindings are shown by the following command:

$ ble-bind -P

The list of widgets is shown by the following command:

$ ble-bind -L

If you want to run multiple widgets with a key, you can define your own widget by creating a function of the name ble/widget/YOUR_WIDGET_NAME as illustrated in the following example. It is highly recommended to prefix the widget name with YOUR_NAME/, my/, blerc/, dotfiles/, etc. in order not to conflict with the names of the existing standard widgets.

# Example of calling multiple widgets with the key C-t
function ble/widget/my/example1 {
  ble/widget/beginning-of-logical-line
  ble/widget/insert-string 'echo $('
  ble/widget/end-of-logical-line
  ble/widget/insert-string ')'
}
ble-bind -f C-t my/example1

2.8 fzf integration^†

If you would like to use fzf in combination with ble.sh, you need to configure fzf using the contrib/fzf integration. Please follow the instructions in the link for the detailed description.

# blerc

# Note: If you want to combine fzf-completion with bash_completion, you need to
# load bash_completion earlier than fzf-completion.  This is required
# regardless of whether to use ble.sh or not.
source /etc/profile.d/bash_completion.sh

ble-import -d integration/fzf-completion
ble-import -d integration/fzf-key-bindings

The option -d of ble-import delays the initialization. In this way, the fzf settings are loaded in background after the prompt is shown. See ble-import - Manual §8 for details.

When you have additional configuration for fzf

When you want to run codes of the additional configuration after the fzf settings are loaded, you cannot simply write them after the above settings because of the delayed loading of the fzf settings. In this case, there are four options. The easiest way is to drop the -d option (Option 1 below) to disable the delayed loading:

# [1] Drop -d
ble-import integration/fzf-completion
ble-import integration/fzf-key-bindings
<settings>

However, the above setting may make the initialization time longer. As another option, you may also delay the additional settings with ble-import -d [2] or ble/util/idle.push [3]. Or, you can hook into the loading of the fzf settings by ble-import -C [4].

# [2] Use ble-import -d for additional settings
ble-import -d integration/fzf-completion
ble-import -d integration/fzf-key-bindings
ble-import -d '<filename containing the settings>'

# [3] Use "ble/util/idle.push" for additional settings
ble-import -d integration/fzf-completion
ble-import -d integration/fzf-key-bindings
ble/util/idle.push '<settings>'

# [4] Use "ble-import -C" for additional settings
ble-import -d integration/fzf-completion
ble-import -d integration/fzf-key-bindings
ble-import -C '<settings>' integration/fzf-key-bindings

3 Tips

3.1 Use multiline mode

When the command line string contains a newline character, ble.sh enters the MULTILINE mode.

By typing C-v C-j or C-q C-j, you can insert a newline character in the command line string. In the MULTILINE mode, RET (C-m) causes the insertion of a new newline character. In the MULTILINE mode, the command can be executed by typing C-j.

When the shell option shopt -s cmdhist is set (which is the default), RET (C-m) inserts a newline if the current command line string is syntactically incomplete.

3.2 Use vim editing mode

If set -o vi is specified in .bashrc or set editing-mode vi is specified in .inputrc, the vim mode is enabled. For details, please check the wiki page.

3.3 Use auto-complete

The feature auto-complete is available in Bash 4.0 or later. auto-complete automatically suggests a possible completion on user input. The suggested contents can be inserted by typing S-RET (when the cursor is at the end of the command line, you can also use right, C-f, or end to insert the suggestion). If you want to insert only the first word of the suggested contents, you can use M-right or M-f. If you want to accept the suggestion and immediately run the command, you can use C-RET (if your terminal does not support special key combinations like C-RET, please check Manual §3.6.4).

3.4 Use sabbrev (static abbrev expansions)

By registering words to sabbrev, the words can be expanded to predefined strings. When the cursor is just after a registered word, typing SP causes the sabbrev expansion. For example, with the following settings, when you type SP after the string command L, the command line will be expanded to command | less.

# blerc
ble-sabbrev L='| less'

The sabbrev names that start with \ plus alphabetical letters are also recommended since it is unlikely to conflict with real words that are a part of the executed command.

# blerc
ble-sabbrev '\L'='| less'

The sabbrevs starting with ~ can be expanded also by /. This can be used to approximate Zsh's named directories. For example, with the following settings, typing ~mybin/ expands it to e.g. /home/user/bin/ (where we assumed HOME=/home/user).

# blerc

ble-sabbrev "~mybin=$HOME/bin"

4 Contributors

I received much feedback from many people in GitHub Issues/PRs. I thank all such people for supporting the project. Among them, the following people have made particularly significant contributions.

• @cmplstofB helped me implement vim-mode by testing it and giving me a lot of suggestions.
• @dylankb reported many issues with the fzf integration, initialization, etc.
• @rux616 reported several issues and created a PR for fixing the default path of .blerc
• @timjrd suggested and contributed to performance improvements in completion.
• @3ximus reported many issues for a wide variety of problems.
• @SuperSandro2000 reported many issues related to NixOS and others