Text::BibTeX::Entry - read and parse BibTeX files


Text::BibTeX::Entry - read and parse BibTeX files


   use Text::BibTeX;            # do not use Text::BibTeX::Entry alone!
   # ...assuming that $bibfile and $newbib are both objects of class
   # Text::BibTeX::File, opened for reading and writing (respectively):
   # Entry creation/parsing methods:
   $entry = new Text::BibTeX::Entry;
   $entry->read ($bibfile);
   $entry->parse ($filename, $filehandle);
   $entry->parse_s ($entry_text);
   # or:
   $entry = new Text::BibTeX::Entry $bibfile;
   $entry = new Text::BibTeX::Entry $filename, $filehandle;
   $entry = new Text::BibTeX::Entry $entry_text;

   # Entry query methods
   warn "error in input" unless $entry->parse_ok;
   $metatype = $entry->metatype;
   $type = $entry->type;
   # if metatype is BTE_REGULAR or BTE_MACRODEF:
   $key = $entry->key;                  # only for BTE_REGULAR metatype
   $num_fields = $entry->num_fields;
   @fieldlist = $entry->fieldlist;
   $has_title = $entry->exists ('title');
   $title = $entry->get ('title');
   # or:
   ($val1,$val2,...$valn) = $entry->get ($field1, $field2, ..., $fieldn);
   # if metatype is BTE_COMMENT or BTE_PREAMBLE:
   $value = $entry->value;
   # Author name methods 
   @authors = $entry->split ('author');
   ($first_author) = $entry->names ('author');
   # Entry modification methods
   $entry->set_type ($new_type);
   $entry->set_key ($new_key);
   $entry->set ('title', $new_title);
   # or:
   $entry->set ($field1, $val1, $field2, $val2, ..., $fieldn, $valn);
   $entry->delete (@fields);
   $entry->set_fieldlist (\@fieldlist);
   # Entry output methods
   $entry->write ($newbib);
   $entry->print ($filehandle);
   $entry_text = $entry->print_s;
   # Miscellaneous methods
   $entry->warn ($entry_warning);
   # or:
   $entry->warn ($field_warning, $field);


Text::BibTeX::Entry does all the real work of reading and parsing BibTeX files. (Well, actually it just provides an object-oriented Perl front-end to a C library that does all that. But that's not important right now.)

BibTeX entries can be read either from Text::BibTeX::File objects (using the read method), or directly from a filehandle (using the parse method), or from a string (using parse_s). The first is preferable, since you don't have to worry about supplying the filename, and because of the extra functionality provided by the Text::BibTeX::File class. Currently, this means that you may specify the database structure to which entries are expected to conform via the File class. This lets you ensure that entries follow the rules for required fields and mutually constrained fields for a particular type of database, and also gives you access to all the methods of the structured entry class for this database structure. See the Text::BibTeX::Structure manpage for details on database structures.

Once you have the entry, you can query it or change it in a variety of ways. The query methods are parse_ok, type, key, num_fields, fieldlist, exists, and get. Methods for changing the entry are set_type, set_key, set_fieldlist, delete, and set.

Finally, you can output BibTeX entries, again either to an open Text::BibTeX::File object, a filehandle or a string. (A filehandle or File object must, of course, have been opened in write mode.) Output to a File object is done with the write method, to a filehandle via print, and to a string with print_s. Using the File class is recommended for future extensibility, although it currently doesn't offer anything extra.


Entry creation/parsing methods

new ([SOURCE])
Creates a new Text::BibTeX::Entry object. If the SOURCE parameter is supplied, it must be one of the following: a Text::BibTeX::File (or descendant class) object, a filename/filehandle pair, or a string. Calls read to read from a Text::BibTeX::File object, parse to read from a filehandle, and parse_s to read from a string.

A filehandle can be specified as a GLOB reference, or as an IO::Handle (or descendants) object, or as a FileHandle (or descendants) object. (But there's really no point in using FileHandle objects, since Text::BibTeX requires Perl 5.004, which always includes the IO modules.) You can not pass in the name of a filehandle as a string, though, because Text::BibTeX::Entry conforms to the use strict pragma (which disallows such symbolic references).

The corresponding filename should be supplied in order to allow for accurate error messages; if you simply don't have the filename, you can pass undef and you'll get error messages without a filename. (It's probably better to rearrange your code so that the filename is available, though.)

Thus, the following are equivalent to read from a file named by $filename (error handling ignored):

   # good ol' fashioned filehandle and GLOB ref
   open (BIBFILE, $filename);
   $entry = new Text::BibTeX::Entry ($filename, \*BIBFILE);
   # newfangled IO::File thingy
   $file = new IO::File $filename;
   $entry = new Text::BibTeX::Entry ($filename, $file);

But using a Text::BibTeX::File object is simpler and preferred:

   $file = new Text::BibTeX::File $filename;
   $entry = new Text::BibTeX::Entry $file;

Returns the new object, unless SOURCE is supplied and reading/parsing the entry fails (e.g., due to end of file) -- then it returns false.

read (BIBFILE)
Reads and parses an entry from BIBFILE, which must be a Text::BibTeX::File object (or descendant). The next entry will be read from the file associated with that object.

Returns the same as parse (or parse_s): false if no entry found (e.g., at end-of-file), true otherwise. To see if the parse itself failed (due to errors in the input), call the parse_ok method.

Reads and parses the next entry from FILEHANDLE. (That is, it scans the input until an '@' sign is seen, and then slurps up to the next '@' sign. Everything between the two '@' signs [including the first one, but not the second one -- it's pushed back onto the input stream for the next entry] is parsed as a BibTeX entry, with the simultaneous construction of an abstract syntax tree [AST]. The AST is traversed to ferret out the most interesting information, and this is stuffed into a Perl hash, which coincidentally is the Text::BibTeX::Entry object you've been tossing around. But you don't need to know any of that -- I just figured if you've read this far, you might want to know something about the inner workings of this module.)

The success of the parse is stored internally so that you can later query it with the parse_ok method. Even in the presence of syntax errors, you'll usually get something resembling your input, but it's usually not wise to try to do anything with it. Just call parse_ok, and if it returns false then silently skip to the next entry. (The error messages printed out by the parser should be quite adequate for the user to figure out what's wrong. And no, there's currently no way for you to capture or redirect those error messages -- they're always printed to stderr by the underlying C code. That should change in future releases.)

If no '@' signs are seen on the input before reaching end-of-file, then we've exhausted all the entries in the file, and parse returns a false value. Otherwise, it returns a true value -- even if there were syntax errors. Hence, it's important to check parse_ok.

The FILENAME parameter is only used for generating error messages, but anybody using your program will certainly appreciate your setting it correctly!

parse_s (TEXT)
Parses a BibTeX entry (using the above rules) from the string TEXT. The string is not modified; repeatedly calling parse_s with the same string will give you the same results each time. Thus, there's no point in putting multiple entries in one string.

Entry query methods

parse_ok ()
Returns false if there were any serious errors encountered while parsing the entry. (A ``serious'' error is a lexical or syntax error; currently, warnings such as ``undefined macro'' result in an error message being printed to stderr for the user's edification, but no notice is available to the calling code.)

type ()
Returns the type of the entry. (The `type' is the word that follows the '@' sign; e.g. `article', `book', `inproceedings', etc. for the standard BibTeX styles.)

metatype ()
Returns the metatype of the entry. (The `metatype' is a numeric value used to classify entry types into four groups: comment, preamble, macro definition (@string entries), and regular (all other entry types). Text::BibTeX exports four constants for these metatypes: BTE_COMMENT, BTE_PREAMBLE, BTE_MACRODEF, and BTE_REGULAR.)

key ()
Returns the key of the entry. (The key is the token immediately following the opening `{' or `(' in ``regular'' entries. Returns undef for entries that don't have a key, such as macro definition (@string) entries.)

num_fields ()
Returns the number of fields in the entry. (Note that, currently, this is not equivalent to putting scalar in front of a call to fieldlist. See below for the consequences of calling fieldlist in a scalar context.)

fieldlist ()
Returns the list of fields in the entry. In a scalar context, returns a reference to the object's own list of fields. That way, you can change or reorder the field list with minimal interference from the class. I'm not entirely sure if this is a good idea, so don't rely on it existing in the future; feel free to play around with it and let me know if you get bitten in dangerous ways or find this enormously useful.

exists (FIELD)
Returns true if a field named FIELD is present in the entry, false otherwise.

get (FIELD, ...)
Returns the value of one or more FIELDs, as a list of values. For example:
   $author = $entry->get ('author');
   ($author, $editor) = $entry->get ('author', 'editor');

If a FIELD is not present in the entry, undef will be returned at its place in the return list. However, you can't completely trust this as a test for presence or absence of a field; it is possible for a field to be present but undefined. Currently this can only happen due to certain syntax errors in the input, or if you pass an undefined value to set, or if you create a new field with set_fieldlist (the new field's value is implicitly set to undef).

Normally, the field value is what the input looks like after ``maximal processing''--quote characters are removed, whitespace is collapsed (the same way that BibTeX itself does it), macros are expanded, and multiple tokens are pasted together. (See bt_postprocess for details on the post-processing performed by btparse.)

For example, if your input file has the following:

   @string{of = "of"}
   @string{foobars = "Foobars"}
     title = {   The Mating Habits      } # of # " Adult   " # foobars

then using get to query the value of the title field from the foobar entry would give the string ``The Mating Habits of Adult Foobars''.

However, in certain circumstances you may wish to preserve the values as they appear in the input. This is done by setting a preserve_values flag at some point; then, get will return not strings but Text::BibTeX::Value objects. Each Value object is a list of Text::BibTeX::SimpleValue objects, which in turn consists of a simple value type (string, macro, or number) and the text of the simple value. Various ways to set the preserve_values flag and the interface to both Value and SimpleValue objects are described in the Text::BibTeX::Value manpage.

value ()
Retuns the single string associated with @comment and @preamble entries. For instance, the entry
   @preamble{" This is   a preamble" # 
             {---the concatenation of several strings}}

would return a value of ``This is a preamble---the concatenation of several strings''.

If this entry was parsed in ``value preservation'' mode, then value acts like get, and returns a Value object rather than a simple string.

Author name methods

This is the only part of the module that makes any assumption about the nature of the data, namely that certain fields are lists delimited by a simple word such as ``and'', and that the delimited sub-strings are human names of the ``First von Last'' or ``von Last, Jr., First'' style used by BibTeX. If you are using this module for anything other than bibliographic data, you can most likely forget about these two methods. However, if you are in fact hacking on BibTeX-style bibliographic data, these could come in very handy -- the name-parsing done by BibTeX is not trivial, and the list-splitting would also be a pain to implement in Perl because you have to pay attention to brace-depth. (Not that it wasn't a pain to implement in C -- it's just a lot more efficient than a Perl implementation would be.)

Incidentally, both of these methods assume that the strings being split have already been ``collapsed'' in the BibTeX way, i.e. all leading and trailing whitespace removed and internal whitespace reduced to single spaces. This should always be the case when using these two methods on a Text::BibTeX::Entry object, but these are actually just front ends to more general functions in Text::BibTeX. (More general in that you supply the string to be parsed, rather than supplying the name of an entry field.) Should you ever use those more general functions directly, you might have to worry about collapsing whitespace; see the Text::BibTeX manpage (the split_list and split_name functions in particular) for more information.

Please note that the interface to author name parsing is experimental, subject to change, and open to discussion. Please let me know if you have problems with it, think it's just perfect, or whatever.

split (FIELD [, DELIM [, DESC]])
Splits the value of FIELD on DELIM (default: `and'). Don't assume that this works the same as Perl's builtin split just because the names are the same: in particular, DELIM must be a simple string (no regexps), and delimiters that are at the beginning or end of the string, or at non-zero brace depth, or not surrounded by whitespace, are ignored. Some examples might illuminate matters:
   if field F is...                then split (F) returns...
   'Name1 and Name2'               ('Name1', 'Name2')
   'Name1 and and Name2'           ('Name1', undef, 'Name2')
   'Name1 and'                     ('Name1 and')
   'and Name2'                     ('and Name2')
   'Name1 {and} Name2 and Name3'   ('Name1 {and} Name2', 'Name3')
   '{Name1 and Name2} and Name3'   ('{Name1 and Name2}', 'Name3')

Note that a warning will be issued for empty names (as in the second example above). A warning ought to be issued for delimiters at the beginning or end of a string, but currently this isn't done. (Hmmm.)

DESC is a one-word description of the substrings; it defaults to 'name'. It is only used for generating warning messages.

names (FIELD)
Splits FIELD as described above, and further splits each name into four components: first, von, last, and jr.

Returns a list of Text::BibTeX::Name objects, each of which represents one name. Use the part method to query these objects; see the Text::BibTeX::Name manpage for details on the interface to name objects (and on name-parsing as well).

For example if this entry:

            author = {John Smith and 
                      Hacker, J. Random and
                      Ludwig van Beethoven and
                      {Foo, Bar and Company}}}

has been parsed into a Text::BibTeX::Entry object $entry, then

   @names = $entry->names ('author');

will put a list of Text::BibTeX::Name objects in @names. These can be queried individually as described in the Text::BibTeX::Name manpage; for instance,

   @last = $names[0]->part ('last');

would put the list of tokens comprising the last name of the first author into the @last array: ('Smith').

Entry modification methods

set_type (TYPE)
Sets the entry's type.

set_metatype (METATYPE)
Sets the entry's metatype (must be one of the four constants BTE_COMMENT, BTE_PREAMBLE, BTE_MACRODEF, and BTE_REGULAR, which are all optionally exported from Text::BibTeX).

set_key (KEY)
Sets the entry's key.

set (FIELD, VALUE, ...)
Sets the value of field FIELD. (VALUE might be undef or unsupplied, in which case FIELD will simply be set to undef -- this is where the difference between the exists method and testing the definedness of field values becomes clear.)

Multiple (FIELD, VALUE) pairs may be supplied; they will be processed in order (i.e. the input is treated like a list, not a hash). For example:

   $entry->set ('author', $author);
   $entry->set ('author', $author, 'editor', $editor);

VALUE can be either a simple string or a Text::BibTeX::Value object; it doesn't matter if the entry was parsed in ``full post-processing'' or ``preserve input values'' mode.

delete (FIELD)
Deletes field FIELD from an entry.

set_fieldlist (FIELDLIST)
Sets the entry's list of fields to FIELDLIST, which must be a list reference. If any of the field names supplied in FIELDLIST are not currently present in the entry, they are created with the value undef and a warning is printed. Conversely, if any of the fields currently present in the entry are not named in the list of fields supplied to set_fields, they are deleted from the entry and another warning is printed.

Entry output methods

write (BIBFILE)
Prints a BibTeX entry on the filehandle associated with BIBFILE (which should be a Text::BibTeX::File object, opened for output). Currently the printout is not particularly human-friendly; a highly configurable pretty-printer will be developed eventually.

Prints a BibTeX entry on FILEHANDLE.

print_s ()
Prints a BibTeX entry to a string, which is the return value.

Miscellaneous methods

warn (WARNING [, FIELD])
Prepends a bit of location information (filename and line number(s)) to WARNING, appends a newline, and passes it to Perl's warn. If FIELD is supplied, the line number given is just that of the field; otherwise, the range of lines for the whole entry is given. (Well, almost -- currently, the line number of the last field is used as the last line of the whole entry. This is a bug.)

For example, if lines 10-15 of file foo.bib look like this:

     author = {Homer Simpson and Ned Flanders},
     title = {Territorial Imperatives in Modern Suburbia},
     journal = {Journal of Suburban Studies},
     year = 1997

then, after parsing this entry to $entry, the calls

   $entry->warn ('what a silly entry');
   $entry->warn ('what a silly journal', 'journal');

would result in the following warnings being issued:

   foo.bib, lines 10-14: what a silly entry
   foo.bib, line 13: what a silly journal

line ([FIELD])
Returns the line number of FIELD. If the entry was parsed from a string, this still works--it's just the line number relative to the start of the string. If the entry was parsed from a file, this works just as you'd expect it to: it returns the absolute line number with respect to the whole file. Line numbers are one-based.

If FIELD is not supplied, returns a two-element list containing the line numbers of the beginning and end of the whole entry. (Actually, the ``end'' line number is currently inaccurate: it's really the the line number of the last field in the entry. But it's better than nothing.)

filename ()
Returns the name of the file from which the entry was parsed. Only works if the file is represented by a Text::BibTeX::File object---if you just passed a filename/filehandle pair to parse, you can't get the filename back. (Sorry.)


the Text::BibTeX manpage, the Text::BibTeX::File manpage, the Text::BibTeX::Structure manpage


Greg Ward <gward@python.net>


Copyright (c) 1997-2000 by Gregory P. Ward. All rights reserved. This file is part of the Text::BibTeX library. This library is free software; you may redistribute it and/or modify it under the same terms as Perl itself.

 Text::BibTeX::Entry - read and parse BibTeX files