A GEDCOM to HTML Translator
Gene Stark
(stark@cs.sunysb.edu)
Version 2.0 (6 April, 1995)
This program reads a GEDCOM file as input and produces a set of HTML files,
one for each individual record in the GEDCOM file. Each individual file
contains hypertext links (relative URL's) to the files for the parents,
spouses, and children for that individual. Birth, marriage, and death events
are included in the file, as are notes records associated with the individual,
and also certain source information. A sorted index file can also be created,
with a hypertext link to each individual. The program can also automatically
insert additional HTML text into each individual file. This feature can be
used add additional text, images, audio, or whatever you want to each
individual. Moreover, you only have to create the included files for the
individuals on which you have extra information, and you can maintain all your
other data on whatever system you are using to produce the GEDCOM output.
If you get an updated GEDCOM file, simply go to the directory where the
individual files are stored and run "ged2html" again. It will update each
individual file with the new GEDCOM data and will automatically insert the
additional HTML text again.
I wrote this program because I tried the GEDCOM to HTML translator posted
by Frode Kvam (frode@ifi.unit.no), and found it insufficiently flexible.
Since it only parses a limited portion of the GEDCOM file, not including
notes records, there wasn't an easy way to modify his program to get
all my notes into the output files. So, I decided to write a YACC-based
parser for the GEDCOM standard, and to base the translator on that.
The YACC parser was used in version 1 of this program, however as I got
more experience with the GEDCOM standard and how it is actually used in
practice, I decided that it was too difficult to make the YACC-based parser
accept the variety of GEDCOM's that actually exist. So, for the current
version I have rewritten the parser so that it will accept essentially
"any" GEDCOM file, and will only complain about grossly malformed input.
I have tested this program on a large number of GEDCOM's, including the
following:
* The "royal92.ged" file from ROOTS-L, which contains 30,682
lines of GEDCOM and 3,010 individuals.
* The "Richard Austin database" (produced by PAF 2.3), which
contains 82,874 lines of GEDCOM and 6,482 individuals.
This database was provided to me by Bill Minnick
(svpafug@rahul.net).
* A database (produced by Family Scrapbook), which contains
30521 lines and 2121 individuals. This database was
provided to me by Bill Spurlock (shadow@mindspring.com).
* All the databases on Yvon Cyr's Acadian/French Canadian CD-ROM.
The program parses GEDCOM's fairly rapidly, at about 3700 GEDCOM lines
per second on my Intel 486/33MHz PC with 16MB of RAM running the FreeBSD
operating system. Parsing of the 82,874 line "Richard Austin database"
takes 22 seconds of wall clock time, and it takes another 10 seconds or
so to produce the 500KB sorted index file. Under FreeBSD, complete
processing of Austin database, including the production of the 6482 output
files, takes just over 5 minutes of wall clock time. Under DOS or Windows,
creation of a large number of output files can take substantially longer,
due to the less efficient filesystems used in those operating systems.
I have used this program to prepare my own data for presentation on the
World-Wide Web. You can view this data by starting from URL
http://www.cs.sunysb.edu/~stark
and following the links. I preprocessed my GEDCOM file to produce
approximately 700 individual files, which are linked together between
themselves and to my hypertext family history document. Birger Wathne
(Birger.Wathne@vest.sdata.no) and others are using version 1 of this program
in various demonstrations of genealogy over the World-Wide Web. Many of these
demonstrations do not preprocess the data into HTML files, but rather use
LifeLines to manage the database in GEDCOM format, and ged2html to process the
output of queries for presentation over the Web.
I have developed and run this program on an Intel 486DX/33 under the
FreeBSD operating system. If you using another flavor of *ix, you
shouldn't have too much trouble getting it to run. You do need an ANSI C
compiler (like GCC), as I am no longer interested in writing old-style C.
I have also compiled the program for MS-DOS and MS-WINDOWS using Microsoft
Visual C. See the file MSDOS for details.
The GEDCOM parser in the program is built around the GEDCOM 5.3 standard.
Whereas version 1 of this program checked the GEDCOM input fairly stringently
for conformance to the standard, the current version attempts to make sense
out of anything that looks like a GEDCOM file. It will complain about
grossly malformed GEDCOM files, but it still tries to get through to the
end and produce whatever output it can.
The output processor is template-driven. That is, it consists of an
interpreter for a simple macro language, which produces output files by
processing template strings and filling in information from the GEDCOM
database. The template-driven output scheme was used to obtain flexibility
and language-independence. The default templates use the cross-reference ID's
in the GEDCOM file to name the HTML files, and will insert one "image" file
(if it exists) near the beginning of each individual file and one "additional
information" file (if it exists) at the end of each individual file.
For example, an individual with cross-reference ID "I101" would receive an
HTML file "I101.html". As this file is created, the file "I101.img" would be
inserted near the beginning, and the file "I101.inc" would be inserted at the
end. Default templates are compiled into the program, and they will be used
unless you specify an alternative template using the appropriate command-line
argument.
If you like the default output format, and you are happy with English as
the output language, then there is no need for you to understand anything
about the template macro language. This is a good thing, because the
macro language is not that pleasant to program in. If you do want to
write your own templates, have a look at the files "template_individual"
and "template_index". These use most, if not all, the available constructs
in the output language. A template file consists of text interspersed
with variable references and control commands. Variable references start
with "$", and are used to insert in-line information from the GEDCOM database.
Constructs that can appear in variables are as follows:
@ denotes the "current individual"
[i] is a subscripting operation that selects the i-th
family, event, note, etc. in a list. The identifier
i is an "index variable", which takes on values
1, 2, 3, etc.
.fieldname is a selection operation that follows associations
in the database. For example ${@.FATHER} denotes
the individual record corresponding to the father
of the current individual. You have to look at
the sample template files and the code in output.c
to see what selectors are understood.
.& is a selection operation that turns a string into
a URL to be output in an HTML anchor.
i refers to the index variable i.
{} appearing in a variable name act as delimiters.
They must be properly matched.
Control constructs are signalled by a "!" appearing at the beginning of
a template line. The control constructs are:
!IF condition
!THEN
!ELSE
!ENDIF
The above constructs provide for conditional output based on whether
particular fields in the database have non-null values.
!WHILE condition
!END
These constructs provide for repetitive output of a particular section of
the template, based on what is in the database. For example, it can be
used to iterate over all the marriages of a particular individual.
In both the !IF and !WHILE constructs, the "condition" is a variable,
only without the $ that would normally precede it if the variable were
to appear in normal text.
!RESET index_variable
Resets the value of an index variable to 1.
!INCREMENT index_variable
Increments the value of an index variable.
!NEXT
Advances "@" to refer to the "next" individual in the database.
!INCLUDE filename_template
Any occurrences of "@" in the "filename_template" are replaced by the
cross-reference ID of the current individual. The substituted template
is then used as the name of a file to be included in the output stream.
If the file does not exist, this construct is ignored. The included
file is inserted verbatim into the output stream; no macroprocessing is
performed on it.
I have reorganized the program so that it is language-independent,
except for the tables in "tags.c". All strings in the output come either
from the templates or from those tables. If you want to make the program
produce output in another language, have a look at "tags.h" and "tags.c"
to see what to do. You should also change the compilation flags in
"Makefile". At the moment, only English is supported. If you create
tables for another language, I'd appreciate receiving them so that I
can integrate them back into the source. Thanks!
DISCLAIMER: I'm not real proud of this program, in the sense that I want
to promote it as an example of great (or even good) coding. However, it
does the job for me, and I spent a reasonable amount of time on it, so I
figured it might be useful for others as well. If you make substantial
additions or changes to this program, please let me know. I will probably
make some changes myself, and I will make the new versions available by
anonymous FTP.
THANKS: go to Birger Wathne for contributing useful ideas and code.
Gene Stark
stark@cs.sunysb.edu