Ebib is a program with which you can manage BibTeX database files without having to edit the raw .bib files. It runs in GNU/Emacs, version 24.3 or higher. (Users of GNU/Emacs 23.1 and up can install Ebib 1.x, which is available in the pasture branch on Github or as a zip or tar archive on Github's releases page.)

It should be noted that Ebib is not a minor or major mode for editing BibTeX files. It is a program in itself, which just happens to make use of Emacs as a working environment, in the same way that for example Gnus is.

The advantage of having a BibTeX database manager inside Emacs is that X is no longer required, as Emacs can run on the console, and also that some integration with Emacs' TeX and LaTeX modes becomes possible. For example, you can push a BibTeX key from Ebib to a LaTeX buffer, or, vice versa, when you're in a LaTeX buffer, you can consult your BibTeX database and insert a key from it into the document. Another advantage of Ebib is that it is completely controlled by key commands: no stressful mouse movements are required, as with most other (usually X-based) BibTeX database managers.


Package manager

The easiest way to install Ebib is to use Emacs' package manager. Ebib is available as a package from the Melpa package archive. If you add the Melpa archive to your package-archive list, you can install Ebib from the package manager. This will also install the Info file so you can access the Ebib manual within Emacs.

Manual installation

You can also install Ebib manually by copying the file ebib.el to somewhere in your load path and add the following line to Emacs' init file ~/.emacs:

(autoload 'ebib "ebib" "Ebib, a BibTeX database manager." t)

On a default installation, the load path probably only contains system directories. If you want to run ebib.el from somewhere else (e.g., directly from the source directory), you can add this directory to your load path:

(add-to-list 'load-path "~/src/ebib"))

Note: if you do not know what your load path is set to, go to the *scratch* buffer, type load-path on an empty line, put the cursor right after it and type C-j. The value of load-path will then appear in the buffer.

Alternatively, you can load Ebib directly by putting the following in .emacs:

(load "/path/to/ebib")

The difference between load and autoload is that the former loads Ebib immediately and thus slows down Emacs' startup (a bit), while the latter loads Ebib only when it is called (and thus slows down Ebib's first time start up a bit).

It is recommended to byte-compile the source, Ebib runs quite a lot faster when it is byte-compiled. You can do this either within Emacs with M-x byte-compile-file (recommended on OS X), or from your shell by going into the directory where you put ebib.el and typing:

emacs -batch -f batch-byte-compile ebib.el

This will create a file ebib.elc, which Emacs will load instead of ebib.el. Do this for every .el file in the source.

Installing the info file

If you install Ebib through the package manager, the Info file is installed automatically as well. However, if you install Ebib manually, or run it from the source directory, this is not the case. If you want the Info file to be accessible from within Emacs, you need to put it in a directory where Emacs can find it. One way to do this is to put ebib.info in one of the standard info directories. Make sure to run install-info ebib.info dir, so that Emacs knows about the new info file.

Alternatively, you can leave ebib.info in Ebib's source directory and add that dir to Info-additional-directory-list. It contains a suitable dir file, so there's no need to run install-info.

Starting Ebib

Once Ebib has been installed and is loaded, you can start it with M-x ebib. This command is also used to return to Ebib when you have put the program in the background. You can bind this command to a key sequence by putting something like the following in Emacs' init file:

(global-set-key "\C-ce" 'ebib)

You can of course choose any key combination you like. (In Emacs, key combinations of C-c <letter> are reserved for the user, which means that no package may set them.)

You can also call Ebib from an Eshell command line. This in itself is entirely unspectacular (in Eshell, you can invoke any Emacs function), but the nice thing is that you can then provide a filename to load. So, provided a file references.bib exists in ~/Work/Papers/, the following command:

~/Work/Papers $ ebib references.bib

starts Ebib and loads the file references.bib.

Upgrading from Ebib 1.x

If you upgrade from Ebib 1.x to the current version 2, there are a few things you need to be aware of. First, Ebib 2 does not support Emacs versions below 24.3. If you're on an older version of Emacs, either upgrade Emacs, or do not upgrade Ebib. Ebib 1.x will continue to be available in the pasture branch on Github or as a package on the Github releases page, so if for some reason you do not want to upgrade, you can continue to use this branch. Ebib 1.x will continue to be supported with bug fixes and possibly the occasional (small) feature addition, though no major changes will take place.

If you do upgrade, it is important to note that a few customisation options have changed. If you never made any customisations to Ebib, the upgrade should not pose any problems. If you have, however, read on.

The most important change is that the variable ebib-entry-types (option "Entry Types") has been removed. Ebib now uses the entry type definitions of the built-in bibtex-mode, which means that if you wish to customise the entry types, you should customise bibtex-BibTeX-entry-alist or bibtex-biblatex-entry-alist.

Secondly, a few user options have been renamed because their types have changed. (Mainly from symbol to string). The name change means that existing customisations are lost, but it also prevents Ebib from throwing cryptic errors and refusing to run. If you customised any of the following options, you should redo them after upgrading Ebib (and preferably remove the old ones from your init file):

Note that even if you customised ebib-biblatex-inheritance, there is most likely no need to customise its replacement ebib-biblatex-inheritances, because the latter is set up with the default inheritance rules that Biblatex defines.

Another thing to note is that four user options have a new default value. First, the default location of the rc file has moved to ~/.emacs.d/ebibrc, in line with the recommendation to keep all Emacs-related files in ~/.emacs.d. If you kept an .ebibrc file in your home directory, move it to ~/.emacs.d. Second, a \documentclass definition has been added to the user options "Latex Preamble" (ebib-latex-preamble) and "Print Preamble" (ebib-print-preamble). If you customised one of these, you should add a document class specification, otherwise the LaTeX file won't compile. Third, the option ebib-save-xrefs-first now defaults to t.

Lastly, if you use Biblatex, make sure to read the section on using Ebib with Biblatex files.

Getting Started

A BibTeX database is somewhat of a free-form database. A BibTeX entry consists of a set of field-value pairs. Furthermore, each entry is known by a unique key. The way that Ebib navigates this database is by having two windows, one that contains a list of all the entry keys in the database, and one that contains the fields and values of the currently highlighted entry.

When Ebib is started, the current windows in Emacs are hidden and the Emacs frame is divided into two windows. The top one contains a buffer that is called the index buffer, while the lower window contains the entry buffer. When a database is loaded, the index buffer holds a list of all the keys in the database. You can move through these keys with the cursor keys. In the entry buffer, the fields of the currently highlighted entry are shown, with their values.

This manual first describes Ebib's basic functionality, so that you can get startet with it. At times, reference will be made to later sections, where more specific functions are described.

Ebib has a menu through which most functions can be accessed. (Especially some of the lesser used functions can only be accessed through the menu, unless you assign key shortcuts to them, of course.)

You can start Ebib with the command M-x ebib. Entering this command hides all the windows in the current Emacs frame and replaces them with two windows: the top one contains the index buffer, the bottom one, taking up the larger part of the screen, contains the entry buffer. The index buffer is named none, to indicate that no database has been loaded. If you open a database, or start a new one, the index buffer will carry its name.

You can quit Ebib by typing q. You will be asked for confirmation, and you will receive a warning if you happen to have an unsaved database. You can also leave Ebib with the command z. However, unlike q, which completely quits Ebib, z only lowers it, so that it remains active in the background. The .bib files that you have opened remain loaded, and you can return to them by typing M-x ebib again.

Opening a .bib File

Loading a .bib file into Ebib is done with the command o. Ebib reads the file that you specify, and reports how many entries it found, how many @string definitions it found, and whether a @preamble was found. Note that when Ebib reads a .bib file, it only reads entry types (e.g. book, article, phdthesis etc.) that it knows about. Therefore, you should make sure that all the entry types and fields that your databases use are defined. Ebib uses the entry type definitions of bibtex.el, which is fairly complete, but if you use non-standard entry types, make sure they have been defined or customise bibtex-bibtex-entry-alist or bibtex-biblatex-entry-alist, depending on which of the two you use.

Every time Ebib reads a .bib file, it produces a few log messages. These are written into a special buffer *Ebib-log*. If Ebib encounters entry types in the .bib file that it doesn't know, a warning will be logged. If Ebib finds something that it cannot parse, it will log an error. If warnings and/or errors occurred during loading, Ebib will issue a message when it finishes loading the .bib and direct you to the log buffer.

If Ebib finds entry types in a .bib file that are not defined, those entries will still be loaded, but their entry type is displayed using Emacs' error face. The most likely case in which this will happen is when you load a Biblatex file without letting Ebib know the file is Biblatex-specific. By default, Ebib assumes that a .bib file it loads is a BibTeX file. If you intend to use Biblatex files, make sure to read the section on Biblatex (Using Biblatex).

Preloading .bib Files

Chances are that you will be doing most of your work with one or a few .bib files, and you may find yourself opening the same file or files every time you start Ebib. If so, you can tell Ebib to always load specific .bib files on startup. To do this, specify the files in Ebib's customisation buffer, under the option "Preload Bib Files".

By default, .bib files are searched for in your home directory. Since this is most likely not where you keep the files, you need to specify either the file's full path or a relative path starting from your home directory. Alternatively, you can customise the option "Bib Search Dirs" to specify one or more directories in which Ebib should search the .bib files.

Once you've opened a .bib file, the keys of all the entries in the file are shown in alphabetical order in the index buffer in the top Ebib window. (In fact, it is possible to show more than just the entry key in this buffer. See the option "Index Display Fields" on how to accomplish this.) The first entry is highlighted, meaning it is the current entry. The fields it holds and their values are shown in the entry buffer in the bottom Ebib window. The first field is the type field, which tells you what kind of entry you're dealing with (i.e. book, article, etc.).

Below the type field, Ebib displays (up to) four sets of fields. The first set are the so-called required fields, the fields that Bib(La)TeX requires to be filled. The second group are the optional fields, which do not have to be filled but which Bib(La)TeX will normally add to the bibliography if they do have a value. The third group are the so-called extra fields. These fields are usually ignored by Bib(La)TeX (note that Bib(La)TeX normally ignores all fields it does not know), although there are bibliography styles that treat some of these fields as optional rather than as extra. Extra fields are defined in the user option "Extra Fields". Lastly, the fourth set of fields shown in the entry buffer are fields that exist in the entry but are not defined as part of the entry type nor as extra fields.

The first two groups of fields are different for each entry type, while the third group are common to all entry types. You can use the extra fields, for example, to add personal comments to the works in your database. Ebib by default defines the following extra fields: crossref, url (BibTeX only), annote (annotation for Biblatex), abstract, keywords, file, timestamp, and doi (BibTeX only). url and doi are defined only for BibTeX, since Biblatex defines them as optional fields for most entry types. If these are not sufficient for you, you can customise the option "Extra Fields".

To move around in the index buffer, you can use the up and down cursor keys, C-p and C-n, or for those more used to mutt's key bindings, k and j. Furthermore, Space and PgDn move a screenful of entries down, while b and PgUp move in the other direction. Lastly, g and Home move to the first entry, while G and End move to the last one.

Ebib is not restricted to opening just one .bib file at a time. You can open more files by just typing o again and entering the filename. Ebib numbers the databases: the number of each database is shown in the mode line of the index buffer, directly before the database name. The keys 1--9 provide a quick way of jumping from one database to another. Note that the numbering is dynamic: if you have three databases opened and then close the second, database 3 becomes database 2.

With the left and right cursor keys, you can move to the previous or next database. These keys wrap, so if you hit the left cursor key while the first database is active, you move to the last database. If you are done with a database and want to close it, type c. This closes the current database. It does not leave Ebib, and all other databases you have open will remain so.

Starting a New .bib File

If you want to start a new .bib file from scratch, you cannot just go and enter entries. You first have to give the database a name. So, to start a new database, type o first, and give the new file a name. Once you have done this, you can start adding entries to the database.

Editing the Database

Of course, being able to open and view .bib files is only half the fun. One needs to be able to edit the files as well. Ebib's essential editing facilities are discussed here.

Adding and Deleting Entries

To add an entry to a database, you type a. When you do this, Ebib first asks you for an entry key, as every entry must be identified by a unique key. Just type a name for the new entry (say Jones1998). Since the entry key must be unique, Ebib will complain if you enter a key that already exists.

You can also let Ebib automatically generate entry keys: if you set the customisation option "Autogenerate Keys", Ebib does not ask you for a key when you add a new entry. Instead, it creates a temporary key (of the form <new-entry>). When you have finished entering the field values for the new entry, Ebib automatically replaces the temporary key with a key that is based on the contents of the author (or editor), year and title fields. (It uses the function bibtex-generate-autokey for this; see that function's documentation string for customisation options.)

Note that if you should later decide that you want to change the key of an entry, you can do so with the command E. So if you have an entry with the key Jones1998 and you want to add another entry by Jones from 1998, you can call the new one Jones1998b and rename the existing one to Jones1998a. Similarly, it is possible to let Ebib recreate an autogenerated key by pressing !.

Deleting an entry is done with d. Be careful with this: you will be asked for confirmation, but once you've confirmed, the entry is gone, and it is not possible to bring it back. There is no undo in Ebib. (If you haven't saved the database yet, it is still possible to retrieve the deleted entry from the .bib file, and otherwise it may still be in the backup file that Ebib creates; Saving a Database.)

Editing Fields Values

Editing the field values for an entry is done in the lower of the two Ebib buffers, the so-called entry buffer. You can move focus to the entry buffer and start editing field values by typing the command e in the index buffer.

You can move between fields with the same keys that you use to move between entries in the index buffer: the cursor keys up and down, C-p and C-n, or j and k. Space and PgDn move to the next set of fields, while PgUp and b move to the previous set of fields. g and G, and Home and End also work as expected.

Editing a field value can be done with e. (In fact, in the entry buffer, RET is equivalent to e.) For most fields, Ebib simply asks you for a string value in the minibuffer. (Here, RET confirms the edit, while C-g cancels it.) Although BibTeX requires that field values be surrounded by braces {} (or double quotes "", but Ebib does not use those, even though it can of course handle them when they are used in an existing .bib file) you do not need to type these. Ebib adds them when it saves the .bib file.

Some fields, however, are handled in a special way. The first of these is the type field: if you edit this field, you must enter one of the predefined entry types. Ebib won't allow you to enter anything else. You can use TAB completion in this case. Similarly, if you edit the crossref field, Ebib requires that you fill in a key from the database. Here, too, you can use TAB completion. The fields keywords and file are also treated differently, see Managing keywords and Viewing Files, respectively.

Note that if you're adding a new entry, Ebib automatically puts you in the entry buffer after you've typed the entry key: you don't have to type e to move to the entry buffer. When creating a new entry, it is best to set the type field first, because the type field determines which other fields are available for an entry. After editing a field, Ebib (usually) puts you on the next field. This is convenient if you're creating a new entry and need to fill out several fields in a row.

If you're done editing the fields of the entry, type q to move focus back to the index buffer. (Note: keys may have different functions in the index buffer and the entry buffer. q is a typical example: in the entry buffer, it quits editing the entry and moves focus back to the index buffer. In the index buffer, however, q quits Ebib.)

Editing Multiline Values

Apart from the type, keywords, file and crossref fields, there is another field that Ebib handles in a special way when you edit its value. This is the annote field (annotation in Biblatex). Most field values normally consist of a single line of text. However, because the annote field is meant for creating annotated bibliographies, it would not be very useful if you could only write one line of text in this field. Therefore, when you edit the annote field, Ebib puts you in the so-called multiline edit buffer. This is essentially a text mode buffer that allows you to enter as much text as you like. To store the text and leave the multiline edit buffer, type C-c | q.

If you want to leave the multiline edit buffer without saving the text you have just typed, type C-c | c. This command cancels the edit and leaves the multiline edit buffer. The text that is stored in the field you were editing is not altered.

Multiline values are not restricted to the annote field. Any field (except the type and crossref fields) can in fact hold a multiline value. To give a field a multiline value, use m instead of e. You will again be put in the multiline edit buffer, where you can edit the value. Note that you can use m even if a field already has a single line value. Ebib will just make that the first line in the multiline edit buffer.

When a field has a multiline value, only the first line is shown in the entry buffer, for space reasons. To indicate that the value is multiline, a plus sign + is placed in front of the value. If you want to see the whole contents of a multiline field, you can use v: this will display the contents of the current field in a *Help* buffer. This buffer can be dismissed again with q. This is quicker than entering the multiline edit buffer just to read the contents of a field and it has the advantage that the other fields stay visible.

By the way, the e key is smart about the way an entry must be edited. If you press e on a field that already has a multiline value, regardless of the fact whether it is the annote field or not, Ebib puts you in the multiline edit buffer. Therefore, you need m only if you want to give a field a multiline value when it doesn't have one yet.

For more details on working with the multiline edit buffer, see The Multiline Edit Buffer.

Undefined Fields

BibTeX ignores fields that it does not know, which is a property that can be exploited to add any kind of information to a BibTeX entry. Ebib accommodates this by allowing fields with any name, not just the ones that are predefined. Such undefined fields are displayed last in the entry buffer, following the extra fields.

It is even possible to add such fields to an entry by pressing a in the entry buffer. This asks for a field name and then a value. If you make heavy use of this option, it may be better to define the relevant fields through the user option "Extra Fields", but undefined fields may be useful when reading or importing BibTeX entries from a different source that contain non-standard fields.

Note that if you delete the contents of an undefined field, the field itself remains in the database until you quit Ebib (or close the database). It will not be saved, however, so the next time you load the .bib file, the field is gone.

Hidden Fields

Biblatex defines a large number of fields, many of which are optional for most entry types. Displaying all these fields in the entry buffer would not be very practical, because you are most likely interested in only a few of them. For this reason, Ebib defines a (fairly large) number of fields as 'hidden', meaning that they are not shown in the entry buffer. You can make these fields visible with the key H in the index buffer. Which fields are treated as hidden is controlled by the option "Hidden Fields", which can be customised.

Most of the fields defined as hidden are Biblatex-specific, because BibTeX recognises a much smaller number of fields and there isn't much of a need to hide the lesser used ones. However, the functionality is available: if you wish to use it, just add the relevant fields to the option "Hidden Fields".


Ebib provides the possibility to add a timestamp to every new entry, recording the time it was added to the database. The timestamp is recorded in the (extra) field timestamp, which is hidden by default.

You can tell Ebib to create timestamps by setting the option Use Timestamp in Ebib's customisation buffer. With this option set, a timestamp is included in entries added to the database with a. Ebib will also add a timestamp to entries imported from a buffer or merged from a file, and to entries exported to another database or to a file. When importing or exporting entries, existing timestamps will be overwritten. The logic behind this is that the timestamp records the date and time when the entry was added to the database, not when it was first created.

Note that if this option is unset, the timestamp of an entry is retained when it's imported or exported. Therefore, if you record timestamps and want to im-/export entries without changing their timestamps, temporarily unset this option, which can be done in the menu under “Options”.

Ebib uses the function format-time-string to create the timestamp. The format string that Ebib uses can be customised. The default string is "%a %b %e %T %Y", which produces a timestamp of the form "Mon Mar 12 01:03:26 2007". This string is not directly suitable for sorting, so if you want to be able to sort on timestamps, you'll need to customise the format string. See the documentation for format-time-string on the options that are available. (Alternatively, the default time stamp format can be converted into a sortable time format using date-to-time, but currently Ebib is not able to do this automatically.)

Adding timestamps in a format that date-to-time can parse makes it possible to list the most recent additions to the database. Ebib provides a function to do this: ebib-list-recent, which asks for a number of days and lists the entries that were added since then.

Copy, Cut, Paste (Yank), and Delete

A few more commands are available when you're in the entry buffer editing field values. The commands c, x and y implement copy, kill and yank: c copies the contents of the current field to the kill ring, x kills the contents of the current field to the kill ring, and y yanks (pastes) the most recently killed text in the kill ring. You can type y repeatedly to get the same effect you get in Emacs when you type M-y after an initial C-y.

Lastly, there is the command d, which deletes the contents of the current field without storing the text in the kill ring. (It asks for confirmation, though, just to make sure.)

Note that y only works when the current field does not have a value yet. This is to prevent you from accidentally overwriting a field value. If you do want to yank text into a field that already has a value, simply hit d first to delete the text.

Saving a Database

When you have undertaken any kind of editing action on a database, it is marked as modified, which is indicated in the mode line for the index buffer. A modified database can be saved by typing s. This saves the database to the file it was loaded from without asking for confirmation. (It is similar to C-x C-s in Emacs.) If you're saving a file for the first time after loading it, Ebib creates a backup file. (Ebib honours backup-directory-alist when saving backups. Note that you can also disable backups altogether with the option "Create Backups".)

If you want to force-save a database that has not been modified, you can use a prefix argument: C-u s. In either case, however, Ebib checks whether the underlying file was modified and warns you if it was. (Ebib does this by storing the .bib file’s modification time when reading the file and comparing this time with the modification time when requested to save the file.) If you also want to forego this check, use a double prefix argument: C-u C-u s. This saves the file unconditionally.

If you have multiple databases open, have made changes in more than one of them, and want to save all of them without going through each yourself, you can save all databases at once through the menu. You can also save the database to another name, similar to C-x C-w in Emacs: the new .bib file becomes associated with the database. The command for this is w. This command can also be prefixed with C-u C-u in order to overwrite any existing file without asking for confirmation. (A single prefix argument C-u does nothing with w.)

Using Biblatex

Compared to BibTeX, Biblatex has a greatly expanded number of fields, a somewhat different set of entry types and a much more complex system of field value inheritances. Ebib can handle both kinds of files, provided it is set up for the right “dialect”.

Setting the BibTeX Dialect

Biblatex files use the same .bib suffix that BibTeX files use. Whether a file is interpreted as a BibTeX or a Biblatex is determined by the user option “Bibtex Dialect” (i.e., the variable ebib-bibtex-dialect). Possible values for this option are BibTeX and biblatex. (These values are taken from the variable bibtex-dialect-list.) The default value of this option is BibTeX. If you wish to read or create Biblatex files in Ebib, you’ll need to change the value of ebib-bibtex-dialect.

The dialect specified determines which entry types Ebib recognises and which fields it expects. Reading a file with the wrong dialect setting will most likely result in a series of “Illegal entry type” errors. Note, however, that these entries will still be loaded and displayed, but they will be highlighted with Emacs’ error face. Fields that are not defined for the current dialect are displayed as undefined fields (i.e., below all other fields in the entry buffer).

The option “Bibtex Dialect” sets the default dialect. It is possible to set the dialect for individual files as well. To do this, load the file and then set the dialect through the menu option «Ebib | BibTeX Dialect» or with the command M-x ebib-set-dialect. You only need to do this once for a file, because the setting is saved in the .bib file in the local variable block. (If no local variable block exists, one is created.) The setting is actually saved as a file-local value for the variable bibtex-dialect, which means that if you should open the file directly in bibtex-mode, Emacs will apply the dialect setting as well. (Note, by the way, that Ebib’s option to set the BibTeX Dialect is different from the one provided by bibtex-mode. Specifically, the former uses the variable ebib-bibtex-dialect, while the latter uses bibtex-dialect.)

The mode line of the index buffer shows the dialect that Ebib assumes for the current database. Note that this does not necessarily mean that the dialect is set in the .bib file: if the file does not have a dialect setting, the mode line shows the default setting.

Alias Types and Fields

The set of entry types defined by Biblatex differs from the set used by BibTeX. Mostly, Biblatex adds new entry types, but there are a few BibTeX entry types that have been dropped. For legacy reasons, Biblatex still recognises these entry types, but it treats them as aliases for some of its own types. The relevant entry types are @conference (treated as an alias for @inproceedings), @electronic (alias for @online), @mastersthesis (alias for @thesis with the type field set to 'Master's thesis'), @phdthesis (alias for @thesis with the type field set to 'PhD thesis'), @techreport (alias for @report with the type field set to 'technical report') and @www (alias for @online). If an entry has such an alias as entry type, Ebib displays the entry type that Biblatex treats it as in the entry buffer. (For example, the entry type alias phdthesis is shown as phdthesis [==> Thesis].)

Similarly, a number of fields are deprecated but still accepted as aliases. These are address (alias for location), annote (alias for annotation), archiveprefix (for eprinttype), journal (for journaltitle), key (for sortkey), pdf (for file), primaryclass (for eprintclass), and school (for institution). These aliases are also indicated in the entry buffer, albeit in a slightly different way. For example, if an entry has a journal field, its value is shown as the value of the journaltitle field; a tag [<== journal] is placed after the field value, indicating that the value is actually contained in the journal field. The journal field itself is shown as an undefined field (i.e., after all other fields). That is, in such cases, the value is shown twice: once as the journaltitle field (among the obligatory fields), and once as the journal field. This is done so that you can easily copy the value of the journal field to the journaltitle field, if you wish to bring your entries into line with Biblatex's conventions.


Simple Searches

Ebib provides several search methods. The simplest method is to type /, which searches for a string (more precisely, a regular expression) starting from the current entry (i.e., not from the first entry) and will display the entry with the first occurrence of the search string that it finds. All the occurrences of the search string in that entry are highlighted.

Ebib searches all the fields of each entry. It is not possible with / to specify the fields to search. Note that if the search term is found in a field with a multiline value, Ebib will highlight the + sign that it displays in front of the field value. Keep an eye out for this when doing a search, because Ebib only shows the first line of multiline values, and if the search term appears in another line, the highlighted + is the only indication that the search term was found. (Well, that and the fact that Ebib does not say "Search string not found", of course...)

A search term may of course appear more than once in the database. To search for the next occurrence, type n. This will continue searching for the search string in the rest of the database. Again, the first entry found to contain the search string is displayed. Note that n does not wrap: if the end of the database is reached, Ebib stops searching. To continue searching from the top, hit g and then n.

Note, by the way, that if you are in the index buffer, the normal Emacs incremental searches, C-s and C-r, can be used to search entry keys. Once you've found the key you're searching, you must hit RET to quit the search and again RET to make the entry you found active: Ebib does not update the entry buffer during incremental search. If you make frequent use of this option, you may want to make the cursor visible in the Ebib buffers. Unset the customisation option "Hide Cursor" to do so.


Ebib also has a much more sophisticated search mechanism that makes use of filters. A filter is basically a search expression that selects entries from the current database. When you apply a filter to a database, only the entries that match are shown. With filters, you can, for example, select all entries from a database that contain the string "Jones" in their author field. A filter can be as complex as you want: you can select all entries that do not contain "Jones" in the author field, or all entries that contain "Jones" in either the author or the editor field, or all entries that contain "Jones" in the author field, and "symbiotic hibernation" in the keyword field, etc. Basically, the filter can consist of an arbitary number of search criteria combined with the logical operators and, or and not.

Simple Selection

Creating a filter is simple: press &, and Ebib will ask you for a field to select on, and for a regular expression to select with. So if you want to select all entries that contain "Jones" in the author field, you press & and type author as the field and Jones as the regexp to filter on. Ebib then runs this filter on the database, and only shows those entries that match the filter. To indicate that a filter is active, the active filter is displayed in the mode line of index buffer. (The filter can be displayed in Lisp form, if you prefer: customise "Filters Display As Lisp" to do so.)

If you don't want to filter on one specific field but rather want to select all entries that match a certain regexp in any field, you can type any as the field to filter on. So specifying any as the field and Jones as the regexp will give you all entries that have a field that contains "Jones" in them.

Note that you can also select items based on their entry type. In order to do that, you need to specify =type= as the field to search, which is the field name in which Ebib stores the entry type internally. (There is also a "normal" field called type, hence the equal signs.) If you search the =type= field, only exact matches are returned, so if you search for book, only the entries that are of type book are returned, not those of type inbook. You can use TAB completion in this case, by the way.

If you specify the keywords field, the keywords associated with your database are available for TAB completion as well. Though you can enter any search term, of course.

Complex Filters

Once you have a filter on your database, you can refine or extend it. For example, suppose you have a filter selecting all entries with "Jones" in the author field and want to add all entries that have "Jones" in the editor field to your selection. In this case you need to do a logical or operation: you want to select an entry if it contains "Jones" in the author field (which you already did) or if it contains "Jones" in the editor field.

A short sidenote: the first impulse in a case like this might be to use and instead of or: after all, you want to select all entries that contain "Jones" in the author field and all entries that contain "Jones" in the editor field. However, the filter that you build up is used to test each entry individually whether it meets the selection criterion. An entry meets the criterion if it contains "Jones" in the author field or if it contains "Jones" in the editor field. Therefore, or is the required operator in this case. If you would use and, you would only get those entries that contain "Jones" in both the author and editor fields.

To perform a logical or operation, press the key |. As before, you will be asked which field you want to filter on, and which regexp you want to filter with. Ebib will then update the index buffer.

It is also possible to perform a logical and on the filter. Use this if you want to select those entries that contain "Jones" in the author field and e.g. "symbiotic hibernation" in the keyword field. A logical and operation is done with the key &. (Note: this is the same key that is used to create the filter. In fact, you can create a filter with | as well: when used in an unfiltered database, & and | are equivalent. They are only different when a filter is already active.)

Both the & and | commands can be used with the negative prefix argument M-- (or C-u -, which is identical). In this case, the search criterion is negated. That is, the negative prefix argument performs a logical not operation on the search criterion. For example, if you want to select all entries from a database that do not contain "Jones" in the author field, you can do this by typing M-- & and then filling out the relevant field and regexp.

There is another way of performing a logical not operation, which is only available when a filter is active: by pressing the key ~, you invert the current filter. That is, if you have a filtered database with all the entries containing "Jones" in the author or in the editor field, and you press ~, the selection is inverted, and now contains all entries that do not have "Jones" in the author or editor field.

Although ~ and the negative prefix argument to & or | both perform logical not operations, they are not equivalent: ~ negates the entire filter built up so far, while the negative prefix argument only negates the single selection criterion you enter with it.

When a filter is active, the filter itself is displayed at the top of the index buffer. If the index window is too small to display the entire filter (which can easily happen if Ebib is set to split the frame vertically rather than horizontally), you can press F v (uppercase F, small v), which will display the filter in the minibuffer.

To cancel the filter and return to the normal view of the database, press F c. For convenience, this action is also available with c, which normally closes a database. If a filter is active, however, it simply cancels the filter. (If you find this behaviour confusing, you can rebind the c key to the function ebib-close-database. See Modifying Key Bindings for details.)

Storing and Saving Filters

When you cancel a filter, it is automatically stored so that it can be reapplied later. To reapply a filter, type F L. This will reapply the last used filter regardless of which database you're in. That is, you can use this to search more than one database without having to type the filter over and over.

However, Ebib only stores one filter this way. If you want to store more filters, you have to name them. You can store the currently active filter or the last used filter with F s. Ebib will ask you for a name for the filter in order to identify it later. By default, filter names are case-insensitive, but if you prefer to use case-sensitive filter names, you can unset the option "Filters Ignore Case". When Ebib is closed, all stored filters are saved to a file and they're automatically reloaded when you open Ebib again. Stored filters are not associated with a particular database: once a filter is stored, it is available to all databases.

You can apply a stored filter with F a. This will ask for the name of a filter and apply it to the current database. You can extend the filter in the normal way, though the changes will not be stored automatically. To store it, type F s again. You can store the extended filter under the old name, in which case Ebib will ask you for confirmation, or under a new name, which will store it as a new filter, keeping the old one.

The file that Ebib uses to store filters is ~/.emacs.d/ebib-filters, although that can of course be customised (see the option "Filters Default File"). As mentioned, stored filters are saved automatically when Ebib closes, but you can also save them manually with F S. Note that if there are no stored filters when Ebib is closed (or when you press F S), the file is deleted.

You can also save your filters to a different file with F w. Such a filter file can be reloaded later with F l. If you load filters from a file while you still have stored filters, you are asked if you want to replace them completely or if you want to add the new filters to the existing ones. In the latter case, however, filters whose name conflict with existing filters are not loaded. (Ebib will log a message about this when this happens.)

To see what filters are currently stored, use F V. If you want to rename a filter, you can do so with F R.

Note that cancelling a filter with F c does not delete it from the list of stored filters, it will remain available for later application. If you want to delete a filter from the list of stored filters, use F d. You can also delete all stored filters with F D. These deletion commands do not ask for confirmation, but if you delete any filters by accident, you can reload them from ~/.emacs.d/ebib-filters with F l.

Special Filters

Filters are essentially Lisp expressions that consist of the functions and, or, and not, together with a special macro contains. However, filters are not limited to these forms. They can essentially contain any Lisp expression. It is not possible to create such special filters interactively, but it is possible to write such filters and put them in a filter file, or to write a function that creates a special filter.

A filter is a Lisp expression that should return either t or nil, indicating whether the entry being tested matches the filter or not. The contents of the entry is available in a variable entry. This variable is given a value by the function that runs the filter, but it is not passed as an argument. Rather, it is a local dynamic variable, which means that the file that defines the filter function should have the variable lexical-binding set to nil.

The value of entry is an alist of fields and their values. These include the fields =key= and =type= for the entry key and type:

(("author" . "{Noam Chomsky}")
 ("title" . "{Syntactic Structures}")
 ("publisher" . "{The Hague: Mouton}")
 ("year" . "{1957}")
 ("remark" . "{copy}")
 ("timestamp" . "{2007-12-30}")
 ("file" . "{c/Chomsky1957.pdf}")
 ("=type=" . "book")
 ("=key=" . "Chomsky1957"))

An Example: Listing Recent Additions

One special filter is included with Ebib. It filters recent additions to the database. The command that creates the filter is ebib-list-recent:

(defun ebib-list-recent (days)
  "List entries created in the last DAYS days."
  (interactive "nNumber of days: ")
  (let ((filter (ebib-db-get-filter ebib--cur-db)))
    (when filter (setq ebib--filters-last-filter filter)))
  (let* ((date (time-subtract (current-time) (days-to-time days)))
         (filter `(ebib--newer-than (quote ,date))))
    (ebib-db-set-filter filter ebib--cur-db)

First, this function saves the current filter if there is one. It then creates a date in Emacs’ internal time format by subtracting the number of days provided by the user from the current date and uses this date to create a Lisp expression which is then installed as the filter for the current database. A call to ebib--redisplay then updates the display, taking the filter into account.

The function ebib--newer-than is defined as follows:

(defun ebib--newer-than (date)
  "Function for use in filters.
Return t if the entry being tested is newer than DATE.  DATE must
be a list of the format returned by `current-time' and is
compared to the timestamp of the entry being tested.  If the
entry has no timestamp, or a timestamp that cannot be converted
into a date representation, return nil."
  (let ((timestamp (cdr (assoc-string "timestamp" entry))))
    (when (and timestamp
               (setq timestamp (ignore-errors (date-to-time timestamp))))
      (time-less-p date timestamp))))

This function obtains the time stamp of the entry being tested from the variable entry and then tries to convert it to Emacs’ time format. If successful, it compares this time to the date passed as an argument and returns t if the latter precedes the former.

Properties of Filtered Databases

When a filter is active, there are a few things that are not possible or function differently. First, it is not possible to add or delete entries, either interactively or by merging or exporting. Exporting from a filtered database or saving a filtered database is also disabled. (The reasons for these limitations are mainly historic and they may be lifted in future version). Editing existing entries is possible, however. Note that if the entry doesn't match the filter anymore after the edit, it doesn't disappear from view. In order to reapply the filter, press F r.

It is also possible to mark entries. Marked entries stay marked when you cancel the filter, so in order to do something with all the entries matching a filter, you can mark them all in the filter view with M, then cancel the filter and perform an action on them.

Although a filtered database cannot be saved, it is possible to write the filtered entries to a file. To do so, use F W. This command also writes all the @string definitions and the @preamble, so you end up with a complete .bib file. (Note that this is in fact the same command as the menu option "Save Database As". This command behaves differently depending on whether a filter is active. If there is, only the filtered entries are written to the new file and the original database remains open. If no filter is active, all the entries are written and the database in Ebib is renamed.)

One final note: of all the filter-related commands, ~, F c, F r, F s and F v and of course the special function of F w are only available when a filter is active. The other commands operate on the stored filters and can be used when no filter is active.

Consulting BibTeX Databases outside Ebib

Having a BibTeX database manager running inside Emacs has an additional advantage: it makes it easy to consult your BibTeX databases from a text buffer. By default, this functionality is set up for LaTeX, for Orgmode and for Pandoc Markdown.

In any (text mode) buffer, you can call the command ebib-insert-bibtex-key to insert a citation. When you invoke this command, Emacs prompts you for a key from the database(s) associated with the current buffer, a citation command and any optional argument(s) the command allows. You can use TAB completion to enter a citation key, and after hitting RET, Emacs puts a citation at the cursor position in the current buffer with the key you selected. This citation includes the additional markup for the text mode in question. That is, if you call ebib-insert-bibtex-key from a LaTeX buffer, Ebib inserts \<cite>{<key>} into your buffer, substituting the citation command you select for <cite> and the key for <key>.

You can also do it the other way around: if you're in the index buffer in Ebib, you can push an entry to a text buffer. To do this, use the command p. Ebib will ask you for a buffer to push the entry to (which defaults to the buffer you started Ebib from, or the buffer you previously pushed an entry to), a citation command and also any optional arguments, and then inserts a citation at the current cursor position in the buffer you've supplied.

The citation command that ebib-insert-bibtex-key and the command key p ask for have to be defined before you can use them. How to do this is explained below, Defining Citation Commands. For LaTeX, only the \cite command is set up by default, so you may want to add some commands. In Orgmode files, Ebib inserts a link of the form [[ebib:Jones1992][Jones (1992)]]. (Actually, you'll need to provide the Jones (1992) yourself, when Ebib asks for a description.) If you use this, you should add the following line to your init file:

(org-add-link-type "ebib" 'ebib-open-org-link)

This will allow Orgmode to recognise Ebib links and allow you to jump to the BibTeX entry in Ebib using C-c C-o (org-open-at-point, the normal Orgmode command for following links). Note that such org links aren't properly translated to LaTeX citation commands when you export your org file. For that, you'll need to set up a different link type for each citation command and provide functions for exporting them to LaTeX. The orgmode documentation explains how this is done. (You could, in fact, create similar functions for exporting to other formats.)

If you use ebib-insert-bibtex-key in a markdown-mode buffer, you can choose one of the citation commands defined for Pandoc Markdown: text, which inserts a citation of the form @Jones1992, paren, which inserts a citation of the form [@Jones1992] and year, which inserts [-@Jones1992]. Since these are the only types of citations that Pandoc Markdown knows, you shouldn't need to change anything.

Calling Ebib from a text mode buffer has another small advantage. If point is on a BibTeX key when Ebib is called, it jumps to that entry in your database. This allows you to quickly check a reference in your text. Ebib will search the entry in the current database, or, if you're calling Ebib from a LaTeX file and there is a \bibliography or \addbibresource command in the file, in the databases in these commands, if they are opened in Ebib, Associating a Database with a LaTeX File below for details.

There is one more command that can be useful: ebib-create-bib-from-bbl creates a .bib file from the .bbl file associated with the LaTeX document in the current buffer. This makes it easy to create a .bib file containing just the BibTeX entries that are used in the document. Note, however, that this command does not add cross-referenced entries to the newly created .bib file, nor does it add @string definitions and the @preamble. So make sure to add those after using ebib-create-bib-from-bbl.

Key Bindings

Of course, the easiest way to use the commands discussed here is to bind them to a key sequence. For example, the following binds C-c b to ebib-insert-bibtex-key in AUCTeX's LaTeX mode:

(define-key 'LaTeX-mode-map "\C-cb" 'ebib-insert-bibtex-key)

Note that commands of the form C-c <letter> are reserved for the user, and should therefore not be set by any package. For this reasons, Ebib does not set this command itself.

ebib-insert-bibtex-key recognises the major mode of the buffer it is called from and uses this information to determine which kinds of citations to insert. So you can bind the ebib-insert-bibtex-key to the same key sequence in every text mode in which you use citations and Ebib will do the right thing.

Defining Citation Commands

As mentioned, you should probably set up some citation commands if you use Ebib with LaTeX documents. The customisation option that allows you to do this is "Citation Commands".

Citation commands are defined for specific major modes. You can also define commands for all modes under the heading any. (By default, this option is used for LaTeX commands, since there is more than one Emacs mode for (La)TeX). Each command consists of an identifier, which you type when Ebib prompts you for a citation command, plus a format string, which is used to create the actual citation command.

The identifier should be a simple string which you can type easily when Ebib asks you for a citation command (TAB completion is available, though). The format string can contain a few directives, which are used to add the citation key and any optional arguments. The following directives are recognised:

the entry key to be inserted.
an argument, for which the user is prompted.
optional material surrounding a %A directive.
a so-called repeater, which contains material that can be repeated. If present, the repeater must contain the entry key directive %K.
a description, for which the user is prompted. Mainly for use in Orgmode.

In the simplest case, the format string contains just a %K directive: \cite{%K}. In this case, %K is replaced with the citation key and the result inserted. Usually, however, citation commands allow for optional arguments that are formatted as pre- or postnotes to the citation. For example, using the natbib package, you have citation commands available of the form:

\citet[cf.][p. 50]{Jones1992}

In order to be able to insert such citations, the format string must contain %A directives:


With such a format string, Ebib asks the user to provide text for the two arguments and inserts it at the locations specified by the directives. Of course, it is possible to leave the arguments empty (by just hitting RET). With the format string above, this would yield the following citation in the LaTeX buffer:


The empty brackets are completely harmless, because LaTeX will simply ignore the empty arguments. However, you may prefer for the brackets not to appear if the arguments are empty. In that case, you can wrap the brackets and the %A directives in a %<...%> pair:


Now, if you leave the arguments empty, Ebib produces the following citation:


Note however, that in this format string is problematic. If you fill out the first argument but not the second, Ebib produces the wrong format string:


If only one optional argument is provided, natbib assumes that it is a postnote, while what you intended is actually a prenote. Therefore, it is best not to make the second argument optional:


This way, the second pair of brackets is always inserted, regardless of whether you provide a second argument or not.

Natbib commands also accept multiple citation keys. When you call ebib-insert-bibtex-key from within a LaTeX buffer, you can only provide one key, but when you're in Ebib, you can mark multiple entry keys and then use p to push them to a buffer. In this case, Ebib asks you for a separator and then inserts all keys into the position of %K:


It is, however, also possible to specify in the format string that a certain sequence can be repeated and how the different elements should be separated. This is done by wrapping that portion of the format string that can be repeated in a %(...%) pair. Normally, you'll want to provide a separator, which is done by placing it between the % and the closing parenthesis:


This format string says that the directive %K can be repeated and that multiple keys must be separated with a comma. The advantage of this is that you are no longer asked to provide a separator.

It is also possible to put %A directives in the repeating part. This is useful for biblatex package, which has so-called multicite commands that take the following form:

\footcites[cf.][p. 50]{Jones1992}[][p. 201]{Haddock2004}

Multicite commands can take more than one citation key in braces {} and each of those citation keys can take two optional arguments in brackets []. In order to get such citations, you can provide the following format string:


Here, the entire sequence of two optional arguments and the obligatory citation key is wrapped in %(...%), so that Ebib knows it can be repeated. If you now mark multiple entries in Ebib, press p and select the footcites command, Ebib will put all the keys in the citation, asking you for two arguments for each citation key.

Of course it is also possible to combine parts that are repeated with parts that are not repeated. In fact, that already happens in the previous example, because the part \footcites is not repeated. But the part that is not repeated may contain %A directives as well:


Multicite commands in biblatex take two additional arguments surrounded with parentheses. These are pre- and postnotes for the entire sequence of citations. They can be accommodated as shown.

Lastly, a citation command can also contain a %D directive. This is mainly for use in Orgmode citations, which take the form [[ebib:<key>][<description>]]. The description is not an argument to the citation command but the string that will be displayed in the Orgmode buffer.

Associating a Database with a LaTeX File

The commands ebib-insert-bibtex-key and ebib-entry-summary must consult the database or databases loaded in Ebib, and Ebib tries to be smart about which database(s) to consult. Usually, a LaTeX file has a \bibliography command somewhere toward the end, or an \addbibresource command in the preamble, which names the .bib file or files that contain the bibliography entries. If you consult a BibTeX database from within a LaTeX file, Ebib first looks for one of these commands, reads the names of the .bib file(s) from it, and then sees if those files happen to be open. If they are, Ebib uses them to let you pick an entry key (in the case of ebib-insert-entry-key) or to search for the entry (in the case of ebib-entry-summary).

Of course, it may be the case that the LaTeX file is actually part of a bigger project, and that only the master file contains a \bibliography or \addbibresource command. To accommodate for this, Ebib checks whether the (buffer-local) variable TeX-master is set to a filename. If it is, it reads that file and tries to find the relevant command there. (Note: TeX-master is an AUCTeX variable, which is used to keep track of multi-file projects. If you don't use AUCTeX, this functionality doesn't work, and Ebib will only check the current file for a \bibliography or \addbibresource command.)

Note that if one of the .bib files in the \bibliography or \addbibresource command isn't loaded, Ebib issues a warning message about this, and continues to check for the next .bib file. These warning messages appear in the minibuffer, but are probably directly overwritten again by further messages or prompts Ebib produces, so check the *Messages* buffer if Ebib doesn't seem to be able to find an entry that you're sure is in one of your databases.

Another thing to keep in mind is that Ebib only looks for a \bibliography or \addbibresource command once: the first time either ebib-insert-bibtex-entry or ebib-entry-summary is called. It stores the result of this search and uses it the next time either of these commands is used. Therefore, if you make a change to the bibliography command, you must reload the file (use M-x revert-buffer or C-x C-v RET) to make sure Ebib rereads the bibliography command.

If no \bibliography or \addbibresource command is found at all, either in the LaTeX file itself, or in the master file, Ebib simply consults the current database, i.e. the database that was active when Ebib was lowered with z. This is also what Ebib does for Orgmode or Pandoc Markdown files, since they do not have the equivalent of a bibliography command. (Note, however, that you can set ebib-local-bibtex-filenames to a list of .bib files in the file-local variable section of a file.)


Both BibTeX and Biblatex have a cross-referencing facility. Suppose you have an entry Jones1998, which appeared in a book that is also in your database, say under Miller1998. You can tell Bib(La)TeX that Jones1998 is contained in Miller1998 by putting Miller1998 in the crossref field. When Bib(La)TeX finds such a cross-reference, all the fields of Jones1998 that don't have a value inherit their values from Miller1998. At the very least, this saves you some typing, but more importantly, if two or more entries cross-reference the same entry, Bib(La)TeX automatically includes the cross-referenced entry in the bibliography (and puts a reduced reference in the cross-referencing entries).

When you fill in the crossref field in Ebib, Ebib displays the values of the cross-referenced entry in the entry buffer. To indicate that they are just inherited values, they are marked with ebib-crossref-face, which by default inherits from font-lock-comment-face. (You can customise it, of course. See the customisation option "Crossref Face".) These values are merely displayed for convenience: they cannot be edited. (They can be copied, however).

BibTeX inheritance mechanism is rather crude: cross-referencing entries simply inherit the fields of the same name from the cross-referenced entry. Biblatex's mechanism is more sophisticated. Biblatex inheritance rules depend on the field and on the types of the cross-referencing and the cross-referenced entry. Thus it is possible to specify that the inbook entry type can inherit a maintitle field from the title field if the cross-referenced entry is of type mvbook, and a booktitle field if the cross-referenced entry is of type book. The inheritance scheme for Biblatex is defined by the option "Biblatex Inheritances", which is set up with the default inheritance relations defined by Biblatex, but which can be customised if needed.

If you're viewing an entry that has a cross-reference and you want to go to the cross-referenced entry you can type C. This command reads the value of the crossref field and then displays that entry. If you want to do the reverse, i.e., see if the current entry is cross-referenced by any other entries, you can use the same key C: if you type C on an entry that does not have a cross-reference, Ebib makes the key of the current entry the current search string and searches for its first occurrence after the current entry. Note that after Ebib has jumped to the first cross-referencing entry, you cannot type C again to find the next one. (Instead, it would take you back to the cross-referenced entry.) In order to find the next cross-referencing entry, you have to type n, as with a normal search. (Also, if the cross-referenced entry appears alphabetically before the cross-referencing entry, you need to type g and then n.)

Note that if you want to use Bib(La)TeX's cross-referencing options, the option "Save Xrefs first" needs to be set (which it is by default). This tells Ebib to save all entries with a crossref field first in the .bib file. Without this, cross-referencing will not work reliably.

Marking Entries

Commands in the index buffer generally operate on one single entry, or on all entries. For some commands, however, it may sometimes be useful to perform them on more than one entry, but not necessarily all of them. This can be achieved by selecting entries. You can select the entries you want to perform a command on with the key m. This selects (or unselects) the current entry. Selected entries are highlighted. (Note that the face properties of selected entries can be customised through the customisation option "Selected Face".)

Commands for which it makes sense automatically operate on all marked entries if there are any. Of the commands discussed so far, these are d to delete entries and p to push entries to a LaTeX buffer. (Note that Ebib creates a single citation command with commas separating the entry keys.)

The command M unmarks all entries or, if there are no marked entries, marks all entries in the current database.

Printing the Database

Sometimes it may be useful to have a .pdf file or print-out of your database. Although Ebib does not actually do the printing itself, it can create a LaTeX file for you that you can compile and print. In fact, there are two ways of doing this.

The print options are available in the Ebib menu when the index buffer is active. You can print the entries as index cards or as a bibliography.

If you print your entries as a bibliography, Ebib creates a simple LaTeX document that essentially contains a \nocite{*} command followed by a \bibliography command referring to the .bib file belonging to the current database. You can then run the usual sequence of LaTeX, BibTeX, LaTeX, LaTeX on this file, creating a document containing a list of all the references in your database.

If you choose to print as index cards, Ebib also creates a LaTeX file. However, instead of simply providing a \nocite{*} command, this file contains a tabular environment for each entry in the database listing all the fields of that entry and their values.

The entries are separated by a \bigskip, but if you set the option Print Newpage in the customisation buffer (or in the Print menu), the entries are separated by a \newpage, so that every entry is on a separate page. The latter option is useful when printing actual index cards (though you'd probably have to change the page size with the geometry package as well).

By default, the index cards only show single-line field values. That is, multiline values are normally excluded. If you want to include multiline values in the print-out, you have to set the option Print Multiline in the Options menu or in Ebib's customisation buffer. With this option set, Ebib includes all multiline values in the LaTeX file that it creates. Note however that Ebib does not change anything about the formatting of the text in a multiline value. So if you plan to make (heavy) use of this option, make sure that the way you type your text conforms to LaTeX's conventions (e.g. empty lines to mark paragraphs, etc.) and doesn't contain any characters such as & that are illegal in LaTeX. (Or, alternatively, use LaTeX code in your multiline fields.)

As mentioned, when you "print" the database, Ebib really just creates a LaTeX file. More precisely, it creates a temporary buffer and writes the LaTeX code into it, and then saves the contents of that buffer to a file. After it has done that, Ebib lowers itself and instruct Emacs to open the file in a buffer, which will then be properly set up as a LaTeX buffer. From there you can run LaTeX and view the result.

Before doing all this, Ebib asks you which file to write to. Be careful with this: since this is supposed to be a temporary file, Ebib simply assumes that if you provide a filename of an existing file, it can overwrite that file without warning!

A better way to tell Ebib which file to use is to set the option "Print Tempfile" in Ebib's customisation buffer to some temporary file. When this option is set, Ebib will always use this file to write to, and will not ask you for a filename anymore.

Note that both print options operate on all entries of the database or on the selected entries.

There are two more customisation options for printing the database. These are "Print Preamble" and "LaTeX Preamble". With these options, you can specify what Ebib should put in the preamble of the LaTeX files it creates. Use this if you want to use specific packages. This is especially useful for printing a bibliography, since by default, Ebib uses BibTeX's standard bibliography style. With the option "LaTeX Preamble" you can set your preferred bibliography style.

Calling a Browser

With most scientific literature nowadays being available on-line, it is common to store URLs and DOIs in a BibTeX database. (Biblatex of course has standardised fields for this information.) Sometimes you may want to load such a URL or a DOI in your browser. Ebib provides a convenient way for doing so.

If you type u in the index buffer, Ebib takes the URL stored in the url field of the current entry and passes it to your browser. Furthermore, in the entry buffer, you can use u on any field. If you happen to have more than one URL stored in the relevant field, Ebib will ask you which one you want to open. Alternatively, you can use a prefix argument: typing M-2 u sends the second URL to your browser.

It is not even necessary that the relevant field contains only URLs. It may contain other text mixed with the URLs: Ebib simply searches the URLs in the field and ignores the rest of the text. Ebib considers every string of characters that starts with http:// or https:// and that does not contain whitespace or any of the characters " ' ; < or > as a URL. The semicolon is included here even though it is actually a valid character in URLs. This is done for consistency, because the semicolon (actually, semicolon+space) is the standard separator for files in the file field and in this way, you can use the same separator to distinguish multiple URLs in the url field. By default Ebib also regards everything that is enclosed in a LaTeX \url{...} command as a URL. So if you use ; to separate URLs and then happen to run into a URL that contains a semicolon, you can enclose it in \url{...} and it will be recognised properly. You can, of course, customise the regular expression that controls this behaviour. See the option "Url Regexp" for details.

Similarly, with the key i in the index buffer you can send a DOI to a browser. The DOI must be stored in the doi field. Unlike URLs, there can only be one DOI in this field. The whole contents of the field is assumed to be the DOI and is sent to the browser. Ebib adds the URL http://dx.doi.org/ before sending the DOI to the browser, so the doi field should contain just the DOI itself.

Ebib uses the Emacs function browse-url to call the default browser on the system. If you prefer to use another browser, however, you can specify this with the option "Browser Command". You can also customise the field that Ebib takes URLs from, with the option "Standard Url Field". The same can be done for the DOI field.

Viewing Files

If you have electronic versions of the papers in your database stored on your computer, or any other file associated with your entries (e.g., notes, if you store those in separate files) you can use Ebib to call external viewers for these files or have them opened in Emacs. The interface for this is similar to that for calling a browser: if you press f in the index buffer, Ebib searches the file field for a filename and when it finds one, calls an appropriate viewer. And just as with u, you can use f in the entry buffer as well, in which case it can be used on any field, not just the file field. It is also possible to have more than one filename in a field. In that case, Ebib asks you which one you want to open. If you want to open the first one you can just hit ENTER. Alternatively, you can provide a prefix argument: M-2 f for the second file, etc.

The file names in the file field do not have to have full paths. You can set the option "File Search Dirs" to a list of directories that Ebib should search when opening a file from the file field. Note that Ebib searches only the directories in this list, not their subdirectories. However, you can specify a relative path in the file field: if you put something like a/Abney1987.pdf in the file field, Ebib searches for the relevant file in a subdirectory a/ of the directories listed in the option "File Search Dirs". As an extra service, Ebib also searches for the base filename, i.e., Abney1987 in this particular case.

Ebib can call different external programs depending on the file extension of the relevant file. The option "File Associations" allows you to specify which programs to call for which types of files. By default, .pdf and .ps files are handled, by xpdf and gv, respectively. You can specify further file types by their extensions (do not include the dot). The program is searched for in PATH, but you can of course specify the full path to the program. There is also the option to open files in Emacs. Use this if you have separate text files for notes, or if you want to read pdf files in Emacs with doc-view-mode.

If the file field is empty, Ebib will search for a pdf file with a name based on the entry key. By default, Ebib just appends .pdf to the entry key and tries to find a file by the name thus created. If you want, you can modify the file name that Ebib searches for by setting the option ebib-name-transform-function to a function that performs the transformation. This function takes the key of the current entry as its argument (as a string), and should return the file name to use (without .pdf, which is added automatically).

For example, if your entry keys separate authors with colons (i.e., you have keys of the form Hauser:Chomsky:Fitch2002), but your pdf files have underscores (i.e., Hauser_Chomsky_Fitch2002.pdf), then you can set ebib-name-transform-function to:

(lambda (key) (replace-regexp-in-string ":" "_" key))

Note that you can also use ebib-name-transform-function to add a subdirectory to the file name. For example, if you store the pdf files in subdirectories A/ to Z/ corresponding to the first letter of the file name, you can use the following function:

(lambda (key) (format "%c/%s" (aref key 0) key))

For the entry key Chomsky1995, this produces C/Chomsky1995, which is then appended with the extension .pdf.

Of course, the function in ebib-name-transform-function does not have to be a lambda. It can also be the name of a function that you’ve defined for the purpose.

Editing the file field

As mentioned above, editing the file field is a bit different from editing other fields. Instead of typing the full contents of the file field, you are asked to specify a single file name. When you hit ENTER, Ebib adds the filename to the file field, appending it no any existing contents (adding a separator if necessary), and then asks you for the next file. If you don't want to add another, just hit ENTER. The default separator is "; " (semicolon-space), but this can be customised (see the option "Filename Separator" for details). The advantage of this method is that you can use TAB completion to complete file names.

The first directory in the option "File Search Dirs" is used as the starting directory for filename completion when editing the file field. Note that when completing file names, Ebib does not take the directories in "File Search Dirs" into account: completion is done using the standard Emacs file name completion mechanism. However, when you enter a file name, Ebib checks if it is in a (subdirectory of) one of the directories in "File Search Dirs", and if so, cuts off the relevant part of the file name to turn it into a relative path.

Notes files

Ebib supports the annote (or annotation) field, but if you prefer to keep notes outside the .bib file, there is an easy way to do that as well. When you hit N on an entry in the index buffer, Ebib opens the notes file for the current entry or, if no such file exists yet, creates one. The file name of the notes file is derived from the key of the entry by adding the extension .org to it (which means that it is an orgmode file). The file is saved in the directory in ebib-notes-directory, or, if that is nil (the default value), to the first directory listed in ebib-file-search-dirs.

The extension (and with that, the major mode) of the file can be changed with the user option ebib-notes-file-extension. Furthermore, the function in ebib-notes-name-transform-function can be used to customise the file name itself in the same way that ebib-name-transform-function does for pdf files. See Viewing Files for some examples. Note that if ebib-notes-name-transform-function is nil, which is the default, the function in ebib-name-transform-function is applied to the key to create the filename for the notes file. If you do no want this to happen but also do not want to change the key for the notes filename, you can set to ebib-notes-name-transform-function to identity.

Window Management

By default, Ebib takes over the entire Emacs frame it is started in, displaying the index window at the top and the entry window below it. There are a few options to change this behaviour, however. They are all part of the customisation group ebib-windows, and allow you to specify two alternative ways to deal with Ebib windows. The main layout option is simply called "Layout" and has four options: use the full frame (the default), use the current window, use the right part of the frame, or display only the index window.

If you set the layout to use only the right part of the frame, the Ebib buffers are displayed on the right of the frame, with the (usually larger) left part of the frame displaying some other buffer, normally the buffer from which you called Ebib. The width of the Ebib windows can be set with the option "Width", which defaults to 80, and which can be specified as an absolute value (the number of columns), but also as a value relative to the current window. In that case, you must specify a value between 0 and 1. Note that when this option is used, the key z does not hide the Ebib buffers, it simply switches to a non-Ebib window in the same frame. You can use (uppercase) Z to hide the Ebib buffers. Furthermore, with this option, the multiline edit buffer is not displayed in the same window as the entry buffer. Rather, Ebib uses another, non-Ebib window to display it.

The fourth option that Ebib provides is to only show the index buffer on start-up. In this case, Ebib does not display the entry buffer when it is started. Instead, only the index buffer is displayed, which can be navigated in the usual manner. The entry buffer is only displayed when you add or edit an entry. When you've finished editing and move back to the index buffer, the entry buffer is hidden again.

The entry buffer is also displayed if you press RET. When you do this, the index buffer remains selected, so you can use this to display the fields of an entry without moving focus to the entry window. If you navigate the index buffer, the entry buffer remains visible, updating its contents as you move around.

In this case, too, the key z does not hide the index window. Rather, it just selects another, non-Ebib window. In order to hide the index window, you can use (uppercase) Z.

If you set Ebib's layout to display only the index buffer on startup, you can additionally set the option "Popup Entry Window". Normally, Ebib will reuse an existing window to display the entry buffer (and restore its original buffer when you leave the entry buffer). With this option set, however, Ebib uses the Emacs function display-buffer-popup-window to create a new window (which is destroyed again when you leave the entry buffer).

Further relevant options are "Window Vertical Split", which displays the index buffer to the left of the frame rather than at the top, and "Index Window Size", which determines the size of the index window (either its height or its width, depending on whether the index window is displayed at the top or on the left of the frame.)

Creating Entry Stubs

If you have a directory full of (pdf) files of articles that you want to add to your database, Ebib can make the task a little bit easier by creating entry stubs for all the files. You can do this with the command M-x ebib-add-file-entry. This command asks you for a file or a directory and creates an entry in the current database for that file or each file in the directory. The entries only contain a file field pointing to the file, all the other information still has to be filled out by hand, but this way you can at least keep track of which files are already in your database.

Note that the entry keys for the stubs are temporary keys. They will be replaced by more permanent keys automatically when you edit the entries. This behaviour is controlled by the function bibtex-generate-autokey, which has a number of customisation options. Check out its doc string for details. If you prefer to edit the keys by hand, you can do so by pressing E in the index buffer.

@Preamble Definition

Apart from database entries, BibTeX allows three more types of elements to appear in a .bib file. These are @comment, @preamble and @string definitions. Ebib provides facilities to handle these, which are discussed here and in the following sections.

Ebib allows you to add one @preamble definition to the database. In principle, BibTeX allows more than one such definition, but really one suffices, because you can use the concatenation character # to include multiple TeX or LaTeX commands. So, rather than having two @preamble definitions such as:

@preamble{ "\newcommand{\noopsort}[1]{} " }
@preamble{ "\newcommand{\singleletter}[1]{#1} " }

you can write this in your .bib file:

@preamble{ "\newcommand{\noopsort}[1]{} "
         # "\newcommand{\singleletter}[1]{#1} " }

Creating or editing a @preamble definition in Ebib is done by hitting (uppercase) P in the index buffer. Ebib uses the multiline edit buffer for editing the text of the @preamble definition, which means that C-c | q stores the @preamble text and returns focus to the index buffer, while C-c | c returns focus to the index buffer while abandoning any changes you may have made. (For details on using the multiline edit buffer, see The Multiline Edit Buffer.)

In order to create a @preamble as shown above in Ebib, you only have to type the text between the braces. Ebib takes care of including the braces of the @preamble command, but otherwise it saves the text exactly as you enter it. So in order to get the preamble above, you'd have to type the following in Ebib:

"\newcommand{\noopsort}[1]{} " # "\newcommand{\singleletter}[1]{#1} "

Note that when Ebib loads a .bib file that contains more than one @preamble definition, it concatenates all the strings in them in the manner just described and saves them in one @preamble definition.

@String Definitions

If you press (uppercase) S in the index buffer, Ebib hides the entry buffer in the lower window and replaces it with the strings buffer. In this buffer, you can add, delete and edit @string definitions.

Adding a @string definition is done with the command a. This will first ask you for an abbreviation and then for the value to be associated with that abbreviation. Once you've entered these, Ebib will sort the new abbreviation into the buffer.

Moving between the @string definitions can be done in the usual way: the cursor keys up and down, C-p and C-n and k and j move up and down. Space and PgDn move ten strings down, while b and PgUp move in the other direction. The keys g, G, Home and End also function as expected.

To delete a @string definition, use d. To edit the value of a definition, use e. There is also a command c, which copies the value of the current @string definition to the kill ring. Unlike in the entry buffer, there are no corresponing commands y and x. (In fact, x does exist, but has another function.) Yanking from the kill ring can be done with C-y/M-y in the minibuffer when you edit a @string's value. Cutting a @string's value is pointless, because a @string definition must have a value.

Having defined @string definitions, there must of course be a way to use them. Just giving a field a string abbreviation as value will not do, because Ebib puts braces around the value that you enter when it writes the .bib file, so that BibTeX will not recognise the abbreviation, and will not expand it. BibTeX will only recognise an abbreviation if it appears in the .bib file outside of any braces.

To accomplish this, you must mark a field's value as special. A special field is a field whose value is not surrounded by braces when the database is saved, so that BibTeX recognises it as an abbreviation. To mark a field special, press r. An asterisk will appear before the field, indicating that has no braces. Pressing r again will change the field back to normal. If you press r on a field that does not have a value yet, Ebib will ask you for one.

Note that this also makes it possible to enter field values that are composed of concatenations of strings and abbreviations. The BibTeX documentation for example explains that if you have defined:

@string{WGA = "World Gnus Almanac"}

you can create a BibTeX field like this:

title = 1966 # WGA

which will produce "1966 World Gnus Almanac". Or you can do:

month = "1~" # jan

which will produce someting like "1 January", assuming your bibliography style has defined the abbreviation jan. All this is possible with Ebib, simply by entering the exact text including quotes or braces around the strings, and marking the relevant field as special.

An easy way to enter a @string abbreviation as a field value is to use the key s instead of e. If you type s, Ebib asks you for a @string abbreviation to put in the current field, and automatically marks the field as special. With this command, Ebib only accepts @string definitions that are in the database, so that by using s you can make sure you don't make any typos. Note that you can use TAB completion to complete a partial string.


If Ebib finds a @comment in a .bib file, it will read it and store it in the database. When the database is saved, all the @comments will be saved with it, at the top of the file (right after the @preamble.) There is no way to edit comments, nor can you specify where in the .bib file a comment is placed, but they won't be lost.

Managing Keywords

Ebib provides some special functionality for handling keywords. By default, there is a keywords field in the list of extra fields. Editing this field is a bit different from other fields. Instead of just entering a string and hitting ENTER to store it and return to the entry buffer, you should enter a single keyword and hit enter. The keyword will then be added to the keywords already present and you are asked to enter the next keyword. If you've added all keywords you want, just hit ENTER to finish.

The advantage of doing it this way is that you can reuse keywords: once you've added a keyword to one entry, Ebib remembers it. The next time you want to use the same keyword for a different entry, you just need to type the first (few) letters, hit TAB and the keyword will be completed. That makes it easier to ensure you use the exact same keywords in different entries. Note that Ebib's keyword functionality is not used to check the contents of keyword fields. It is simply a way to make it easier to stick to specific keywords, which should make it easier to categorise and search your entries.

It is still possible to edit the keywords field directly. To do so, use a prefix argument: C-u e (or any other prefix argument) on the keywords field will allow you to edit the entire contents in the normal way. Use this method if you want to remove single keywords. (Blanking the entire keywords field is quicker with x or d.)

It is also possible to add keywords to an entry from the index buffer, using the command K a (ebib-keywords-add). This works essentially the same way as when adding keywords from the entry buffer. There is one advantage, however: this command can also operate on marked entries, so that you can add keywords to multiple entries in one go.

Remembering keywords is practical, but it is even more useful if remembered keywords can be saved, so that they are available the next time you start Ebib. There are two ways of doing this: first, there is an option "Keywords List" that you can use to store keywords. Keywords stored in this option will be available for TAB completion to all databases in Ebib. New keywords, however, will not automatically be stored. If you find you need a keyword not on the list and want to make it permanent, you'll have to add it to "Keywords List" yourself.

The other way of making keywords permanent is by storing them in a file. Ebib offers two ways of doing this (which are mutually exclusive, so you have to choose one). You can either configure a single keyword file, with keywords that are available to all databases, or you can configure per-directory keyword files, with keywords that are available for all .bib files in the same directory. You can set up keyword files by configuring the option "Keywords File". You can either set it to use a single keyword file, in which case you need to specify a file with its full path, or you can use per-directory keyword files, in which case you must provide a filename without a path. That is, if you use per-directory keyword files, the files have the same name in each directory. The default name is ebib-keywords.txt, but you can change that if you like, of course.

Keywords that have not been made permanent are marked as such in the entry buffer in ebib-warning-face, (usually a red foreground colour, but this can be customised). If you run into such keywords and want to make them permanent, use the command K s. This will take the keywords in the keywords field of the current entry and store all keywords that are not permanent yet.

Keyword files have a very simple format: they are text files with one keyword per line. So you can easily create or edit keyword files by hand, or have them created by some other programme. Keep in mind, though, that Ebib does not check for changes to keyword files. If you have a single keyword file, it is loaded when Ebib starts up; per-directory keyword files are loaded when the first .bib file in that directory is opened. If you open a second .bib file from the same directory, Ebib won't reload the keywords file.

When you close a database, Ebib checks if you have added new keywords to it and asks you if you want to save them. You can tell Ebib to save new keywords automatically by setting the option "Keywords File Save On Exit" to always. Note that this doesn't save the keywords when you enter them, only when you close the database or quit Ebib. You can also set this option to never, which means Ebib will discard new keywords when the database is closed. Note that if you want to save the keywords file without having to close the database, you can do so through the menu.

The option "Keywords Use Only File" controls whether Ebib uses only the keyword file, or both the keyword file and the configured keyword list. This option is only useful when you have configured a keyword file. In that case, setting this option to use both the keyword list and the keyword file tells Ebib to offer keywords from both sources when you edit the keyword field. Otherwise, only the keyword file is used.

It is also possible to tell Ebib to sort the keywords in the keywords field in alphabetical order. Set the option "Keywords Field Keep Sorted" if you want to do this. Note that setting this option also automatically removes duplicates.

Lastly, you can configure the separator used between keywords in the keyword field. By default, it is set to ", ", i.e., comma-space. (The separator is a comma because that is what Biblatex expects in the keywords field.) If you change it, keep in mind that Ebib does not add a space between keywords, so if you want a space, make sure to add it to the separator.

Sorting the .bib File

By default, the entries in the database are saved to the .bib file in alphabetical order according to entry key. If you only deal with the .bib file through Ebib, you may not care in which order the entries are saved. However, it may sometimes be desirable to be able to specify the sort order of entries in more detail. (Apparently, this can be useful with ConTeXt, for example.)

You can specify a sort order in Ebib's customisation buffer. To sort the entries, you must set at least one sort level (that is, a field to sort the entries on). You can also specify more than one sort level: if two entries have identical values for the first sort level, they will be sorted on the second sort level. E.g., if the first sort level is author and the second is year, then the entries are sorted by author, and those entries that have identical values for the author field are sorted by year.

A sort level is not restricted to a single field. You can specify more fields for a single sort level. Within a single sort level, a second sort field is used if the first sort field does not have a value. For example, books that have an editor instead of an author will have an empty author field. If you sort the database on the author field, such entries will all appear at the beginning of the .bib file, which is most likely not what you want.

To remedy this, you can specify both the author and the editor fields for the first sort level. Ebib will then sort an entry on its author field if it has a value, and will otherwise use the value of the editor field.

The difference between two sort fields within one sort level and two sort levels is that a second sort field is an alternative for the first field when it has no value, while a second sort level is an additional sort criterion when two or more entries cannot be sorted on the first level, because they have identical values.

By default, the option Sort Order has no value, which means that the entries in the .bib file are sorted according to entry key. Those that wish to customise the sort order will usually want to set the first sort level to author editor, and the second to year. In that way, the entries in the .bib file are sorted according to author/editor, and entries with the same author/editor are sorted by year.

Entries that cannot be sorted on some sort level, because the sort fields are empty, are sorted on entry key. (Keep in mind that if the first sort level yields no value for a specific entry, Ebib does not use the second sort level to sort that entry. It uses the entry key. The second sort level is only used if the first yields identical values for two or more entries.)

Note that if you wish to make use of this option, you need to unset the option "Save Xrefs First" (Cross-referencing). It is pointless to set a sort order if cross-referenced entries are saved first. Since "Save Xrefs First" is set by default, you need to unset if you set "Sort Order".

Merging and Importing

As described in the previous chapter, adding entries to a database can be done manually with the key a. There are other ways of adding entries to a database, however.

In the index buffer, the Ebib menu has an option to merge a second .bib file into the current database. Ebib reads the entries in this file and adds them to the database. Duplicate entries (that is, entries with an entry key that already exists in the database) will normally not be loaded. Ebib logs a warning about each duplicate entry to its log buffer and displays a warning after loading the .bib file when this happens. However, if you've customised Ebib to automatically generate keys, duplicate entries will be merged into the current database under a unique key.

Another way to add entries to a database is to import them from an Emacs buffer. If, for example, you find ready-formatted BibTeX entries in a text file or on the internet, you can copy & paste them to any Emacs buffer (e.g. the *scratch* buffer), and then execute the command M-x ebib-import. Ebib then goes through the buffer and loads all BibTeX entries it finds into the current database (i.e. the database that was active when you lowered Ebib). If you call ebib-import while the region is active, Ebib only reads the BibTeX entries in the region.

If a BibTeX entry in the buffer lack an entry key (which sometimes happens with BibTeX entries found on the internet), Ebib will generate a temporary key for it of the form <new-keyXX>, where XX is a number. You can change such keys by hitting E in the index buffer. They will also automatically be replaced with a more sensible key when you edit them. See the option "Autogenerate Keys" for details.

Exporting Entries

Sometimes it can be useful to copy entries from one database to another, or to create a new .bib file with several entries from an existing database. For this purpose, Ebib provides exporting facilities. To export an entry to a .bib file, use the command x. This command operates on a single entry or on all marked entries.

When you export one or more entries, Ebib will ask you for a filename to export them to. If you have already exported an entry before, Ebib will present the filename you used as default, but you can of course change it. For obvious reasons, Ebib appends the entry to the file that you enter if it already exists, it does not overwrite the file. If this is not what you want, delete the file first, as Ebib provides no way to do this.

If you have more than one database open in Ebib, it is also possible to copy entries from one database to another. To do this, use the x command with a numeric prefix argument. E.g., if the database you want to export an entry to is the second database, type M-2 x to export the current entry to it. The number of the database is given in the modeline of the index buffer. If the database you're copying an entry to already contains an entry with the same entry key, Ebib won't copy the entry, and issues an appropriate warning message.

Apart from entries, it is also possible to export the @preamble and @string definitions. The @preamble definition is exported with the command X in the index buffer. @string definitions can be exported in the strings buffer: x in this buffer exports the current string, while X exports all @string definitions in one go. All these commands function in the same way: when used without a prefix argument, they ask for a filename, and then append the relevent data to that file. With a numeric prefix argument, they copy the relevant data to the corresponding open database.

Multiple Identical Fields

Under normal circumstances, a BibTeX entry only contains one occurrence of each field. If BibTeX notices that an entry contains more than one occurrence of an required or optional field, it issues a warning. Ebib is somewhat less gracious, it simply takes the value of the last occurrence without giving any warning. (Note, by the way, that BibTeX will use the value of the first occurrence, not the last.) When extra fields appear more than once in an entry, BibTeX does not warn you, since it ignores those fields anyway. Here, too, Ebib's standard behaviour is to ignore all but the last value.

However, some online reference management services "use" this feature of BibTeX in that they put multiple keywords fields in the BibTeX entries that they produce. If you were to import such an entry into Ebib, you would lose all your keywords except the last one. To remedy this, you can tell Ebib that it should allow multiple occurrences of a single field in a BibTeX entry. You can do this by setting the customisation option "Allow Identical Fields".

With this option set, Ebib collapses the multiple occurrences into a single occurrence. All the values of the different occurrences are collected and stored in the single occurrence, separated by colons (actually, the value of the option "Keywords Separator"). That is, Ebib does not retain the multiple occurrences, but it does retain the values. So suppose you have an entry that contains the following keywords fields:

    author = {Jones, Joan},
    year = {1998},
    keywords = {sleep},
    keywords = {winter},
    keywords = {hibernation}

If you load this entry into Ebib with the option "Allow Identical Fields" set, you will get the following:

    author = {Jones, Joan},
    year = {1998},
    keywords = {sleep, winter, hibernation}

The Multiline Edit Buffer

As mentioned several times before, Ebib has a special multiline edit buffer, which is used to edit field values that contain newlines (so-called multiline fields), and also to edit the contents of the @preamble command. This section discusses the details of this buffer.

Ebib enters multiline edit mode in one of three cases: when you edit the @preamble definition, when you hit m in the entry buffer to edit the current field as multiline, or when you hit e on the annote field, or on a field whose value already is multiline.

The mode that is used in the multiline edit buffer is user-configurable. The default value is text-mode, but if you prefer to use some other mode, you can specify this through the customisation options. (Personally, I use markdown-mode in the multiline edit buffer, so that I can use Markdown to write annotations, which provides an easy way to create headers, use bold and italic, etc., in plain text.)

Three commands are relevant for interacting with Ebib when you're in the multiline edit buffer, which are bound to key sequences in the minor mode ebib-multiline-edit-mode, which is activated automatically in the multiline edit buffer.

ebib-quit-multiline-buffer-and-save, bound to C-c | q, leaves the multiline edit buffer and stores the text in the database. If you invoke this command when you've deleted all contents of the buffer (including the final newline!) and you were editing a field value or the @preamble, the field value or preamble is deleted. (This is in fact the only way to delete the @preamble definition. Field values on the other hand can also be deleted by hitting x or d on them in the entry buffer.) If you were editing a @string value, Ebib will just complain, because string definitions cannot be empty.

ebib-cancel-multiline-buffer, bound to C-c | c, also leaves the multiline edit buffer, but it does so without storing the text. The original value of the field, string or preamble will be retained. If the text was modified, Ebib will ask for a confirmation before leaving the buffer.

ebib-save-from-multiline-buffer, bound to C-c | s, can be used in the multiline edit buffer to save the database. This command first stores the text in the database and then saves it. Because Ebib does not do an autosave of the current database, it is advisable to save the database manually every now and then to prevent data loss in case of crashes. It would be annoying to have to leave the multiline edit buffer every time you want to do this, so this command has been provided to allow you to do this from within the buffer.

Admittedly, the key combinations of the multiline edit buffer are somewhat awkward. The reason for this is that these commands are part of a minor mode, which restricts the available keys to combinations of C-c plus a non-alphanumeric character. However, it is possible to change the key commands, if you wish. Ebib itself provides a method to change the second key of these commands, Modifying Key Bindings. You could change the | to e.g., c or C-c, if these do not conflict with any key commands in the major mode used for the multiline edit buffer.

Even more drastically, you could put something like the following in your ~/.emacs:

(eval-after-load 'ebib
     (define-key ebib-multiline-mode-map
                 "\C-c\C-c" 'ebib-quit-multiline-buffer-and-save)
     (define-key ebib-multiline-mode-map
                 "\C-c\C-q" 'ebib-cancel-multiline-buffer)
     (define-key ebib-multiline-mode-map
                 "\C-c\C-s" 'ebib-save-from-multiline-buffer)))

This sets up C-c C-c, C-c C-q and C-c C-s for use in the multiline edit buffer. Since such key combinations are restricted for use with major modes, however, Ebib cannot set these up automatically, but as an Emacs user, you are free do do as you like, of course.

The Options Menu

In the index buffer, Ebib's menu has an Options submenu. This menu gives you quick access to Ebib's customisation buffer, and it also provides checkboxes for several settings that can be toggled on and off. All of these settings have defaults that can be defined in the customisation buffer. Setting or unsetting them in the Options menu only changes them for the duration of your Emacs session, it doesn't affect the default setting.

The same is true for the printing options that are in the Print menu. When set or unset in the menu, the default values specified in the customisation buffer do not change.

The Ebib Buffers

This chapter lists all the key commands that exist in Ebib, with a short description and the actual command that they call. The latter information is needed if you want to customise Ebib's key bindings. (See Modifying Key Bindings.)

The Index Buffer

go to previous entry. (ebib-prev-entry)
go to next entry. (ebib-next-entry)
move to the next database. (ebib-next-database)
move to the previous database. (ebib-prev-database)
scroll the index buffer down. (ebib-index-scroll-down)
scroll the index buffer up. (ebib-index-scroll-up)
go to first entry. (ebib-goto-first-entry)
go to last entry. (ebib-goto-last-entry)
make the entry under the cursor current. Use after e.g. C-s. (ebib-select-entry)
equivalent to PgDn.
jump to the corresponding database.
search the database. (ebib-search)
create a filter, or perform a logical and on the current filter. With negative prefix argument: apply a logical not to the selectional criterion. (ebib-filter-and)
create a filter, or perform a logical or on the current filter. With negative prefix argument: apply a logical not to the selectional criterion. (ebib-filter-or)
perform a logical not on the current filter. (ebib-filter-not)
display Ebib info. (ebib-info)
recreate an autogenerated key. (ebib-generate-autokey)
add an entry. (ebib-add-entry)
equivalent to Pgup.
close the database or cancel the current filter. (ebib-index-c)
follow crossref field. (ebib-follow-crossref)
delete the current entry or marked entries. (ebib-delete-entry)
edit the current entry. (ebib-edit-entry)
edit the key of the current entry. (ebib-edit-keyname)
extract a filename from the file field and send it to an appropriate viewer. With numeric prefix argument, extract the n-th filename.
prefix key for filter actions.
F &
equivalent to &.
F |
equivalent to |.
F ~
equivalent to ~
F a
apply a stored filter. (ebib-filters-apply-filter)
F c
cancel the current filter. (ebib-filters-cancel-filter)
F d
delete a stored filter. (ebib-filters-delete-filter)
delete all stored filters. (ebib-filters-delete-all-filters)
F l
load filters from a file. (ebib-filters-load-from-file)
reapply the last used filter. (ebib-filters-reapply-last-filter)
F r
reapply the current filter. (ebib-filters-reapply-filter)
rename a stored filter. (ebib-filters-rename-filter)
F s
store the current filter. (ebib-filters-store-filter)
save stored filters to ebib-filters-default-file. (ebib-filters-save-filters)
F v
display the filter in the minibuffer. (ebib-filters-view-filter)
show a list of filters. (ebib-filters-view-all-filters)
F w
write the stored filters to a file. (ebib-filters-write-to-file)
equivalent to Home.
equivalent to End.
show the info node on the index buffer. (ebib-index-help)
show/hide the hidden fields. (ebib-toggle-hidden)
extract a DOI from the doi field, append it to a URL and open that in a browser (ebib-browse-doi)
equivalent to Down.
jump to another database. This accepts a numeric prefix argument, but will ask you for a database number if there is none. (ebib-switch-to-database)
equivalent to Up.
prefix for keyword commands
K a
add keywords to the current entry or to the marked entries. (ebib-keywords-add)
K s
store non-permanent keywords present in the keywords field. (ebib-keywords-save-from-entry)
show the log buffer. (ebib-show-log)
mark or unmark the current entry. (ebib-mark-entry)
find next occurrence of the search string. (ebib-search-next)
create or open a notes file for the curret entry. (ebib-open-notes-file)
equivalent to Down.
equivalent to PgDn.
open a .bib file. (ebib-load-bibtex-file)
push a single entry or the marked entries to a LaTeX buffer (ebib-push-bibtex-key)
equivalent to Up.
equivalent to PgUp.
show and edit the @preamble definition in the database. (ebib-edit-preamble)
quit Ebib. This sets all variables to nil, unloads the database(s) and quits Ebib. (ebib-quit)
save the database. (ebib-save-current-database)
show and edit the @string definitions in the database. (ebib-edit-strings)
extract a URL from the url field and send it to a browser. With numeric prefix argument, extract the n-th url. (ebib-browse-url)
save the database under a new name. If a filter is active, save the filtered entries to a new .bib file. (ebib-write-database)
export the current entry or the marked entries to a file, or, when used with numeric prefix argument, to another database. (ebib-export-entry)
C-x b
equivalent to z.
C-x k
equivalent to q.
export the @preamble definition to a file or, when used with a numeric prefix argument, to another database. (ebib-export-preamble)
move focus away from the Ebib windows, or, if using the full layout, put Ebib in the background. (ebib-leave-ebib-windows)
put Ebib in the background. (ebib-lower)

Functions not bound to any key:

The Entry Buffer

go to the previous field. (ebib-prev-field)
go to the next field. (ebib-next-field)
go to the previous set of fields. (ebib-goto-prev-set)
go to the next set of fields. (ebib-goto-next-set)
go to the first field. (ebib-goto-first-field)
go to the last field. (ebib-goto-last-field)
equivalent to PgDn.
equivalent to PgUp.
copy the contents of the current field to the kill ring. (ebib-copy-field-contents)
delete the value of the current field. The deleted contents will not be put in the kill ring, and is therefore irretrievably lost. (ebib-delete-field-contents)
edit the current field. (ebib-edit-fields)
extract a filename from the current field and send it to an appropriate viewer. With numeric prefix argument, extract the n-th filename.
equivalent to Home.
equivalent to End.
show the info node on the entry buffer. (ebib-entry-help)
go to the next field. (ebib-next-field)
go to the previous field. (ebib-prev-field)
edit the current field as multiline. (ebib-edit-multiline-field)
equivalent to Down.
equivalent to PgDn.
equivalent to Up.
equivalent to PgUp.
quit editing the current entry and return focus to the index buffer. (ebib-quit-entry-buffer)
toggle a field's "special" status. (ebib-toggle-raw)
insert an abbreviation from the @string definitions in the database. (ebib-insert-abbreviation)
extract a URL from the current field and send it to a browser. With numeric prefix argument, extract the n-th url.
cut the contents of the current field. Like c, x puts the contents of the current field in the kill ring. (ebib-cut-field-contents)
C-x b
equivalent to q.
C-x k
equivalent to q.
yank the last element in the kill ring to the current field. Repeated use of y functions like C-y/M-y. Note that no text will be yanked if the field already has a value. (ebib-yank-field-contents)

The Strings Buffer

Up (ebib-prev-string)
go to the previous string.
go to the next string. (ebib-next-string)
go ten strings up. (ebib-strings-page-up)
go ten strings down. (ebib-strings-page-down)
go to the first string. (ebib-goto-first-string)
go to the last string. (ebib-goto-last-string)
equivalent to PgDn.
add a new @string definition. (ebib-add-string)
equivalent to PgUp.
copy the text of the current string to the kill ring. (ebib-copy-string-contents)
delete the current @string definition from the database. You will be asked for confirmation. (ebib-delete-string)
edit the value of the current string. (ebib-edit-string)
equivalent to Home.
equivalent to End.
show the info node on the strings buffer. (ebib-strings-help)
equivalent to Down.
equivalent to Up.
equivalent to Down.
equivalent to PgDn.
equivalent to Up.
equivalent to PgUp.
quit the strings buffer and return focus to the index buffer. (ebib-quit-strings-buffer)
export the current @string definition to a file or, when used with a prefix argument, to another database. (ebib-export-string)
export all the @string definitions to a file or, when used with a prefix argument, to another database. (ebib-export-all-strings)
C-x b
equivalent to q.
C-x k
equivalent to q.


Ebib can be customised through Emacs' standard customisation interface. The relevant customisation group is (obviously) called ebib, which has four subgroups: ebib-faces, ebib-filters, and ebib-keywords, whose functions should be obvious, and ebib-windows, where options for Ebib's window management can be set. All options are documented in the customisation buffers. You can go to Ebib's customisation buffer with M-x customize-group RET ebib RET, or by using the menu «Ebib | Options | Customize Ebib».

Another way to customise Ebib is to use the initialisation file ~/.emacs.d/ebibrc. (Actually, the location and name of this file can be customised as well.) This file contains normal Lisp code and is the best way to customise key bindings (as discussed in the next section), but it can in fact be used to run any Lisp code.

Modifying Key Bindings

If you would like to change Ebib's standard key bindings, or if you would like to bind a command that is only available through the menu to a key, you can do so by creating a file ~/.emacs.d/ebibrc and writing your preferred key bindings in it. A key binding can be defined as follows:

(ebib-key <buffer> <key> <command>)

<buffer> is either index, entry or strings, for the corresponding buffer. <key> is a standard Emacs key description, and <command> is the Ebib command to be associated with the key (which does not have to be quoted, unlike define-key and friends). The commands that can be used here are listed in The Ebib Buffers. Note that it is possible to bind more than one key to a single function: just add as many ebib-key statements as necessary.

As an example, the following binds C-s to ebib-search in the index buffer, so that the database can be searched with C-s as well as with /:

(ebib-key index "\C-s" ebib-search)

If you want to unbind a key, you can simply leave out <command>. So if you want to bind the command ebib-delete-entry to D rather than d, you need to put the following in ~/.emacs.d/ebibrc:

(ebib-key index "D" ebib-delete-entry)
(ebib-key index "d")

The first line binds D to the command ebib-delete-entry. The second line unbinds d.

ebib-key allows you to redefine the second key in the key bindings of the multiline edit buffer:

(ebib-key multiline "&")

This piece of code changes the commands C-c | q, C-c | c and C-c | s to C-c & q, C-c & c and C-c & s, respectively.

Alternatively, if you find these key bindings awkward, you could put something like the following code in ~/.emacs.d/ebibrc:

(define-key ebib-multiline-mode-map
            "\C-c\C-c" 'ebib-quit-multiline-buffer-and-save)
(define-key ebib-multiline-mode-map
            "\C-c\C-q" 'ebib-cancel-multiline-buffer)
(define-key ebib-multiline-mode-map
            "\C-c\C-s" 'ebib-save-from-multiline-buffer)

This will bind the relevant commands to C-c C-c, C-c C-q and C-c C-s`, which are of course more convenient than the standard key bindings.