We currently have a problem with our meta-data directory which it maybe
that Elib members have already solved. Our directory is presented as
dynamic HTML from a database, but we would also like to offer the
records as static pages for indexing by Internet search-engines,
Eldis records are held in an Inmagic DbTextworks database. There are
currently some 7,000 records. (the number will grow, certainly triple in
the next 12 months). We need to write each of these records out of the
database as a single file. These files will be in HTML format and will
become separate HTML pages on our WWW service. Whilst the software
allows great flexibility on creating output formats, it does not allow
you to select all 7000 records and write them out as 7000 separate
files: it only allows you to write out one file of 7000 records. The
objective is to take this large output file and, using a programme,
break it into its component records, writing each record to a separate
file and naming that file in a consistent way.
Does any0ne know of a programme for this, or of someone who might be interested in writing one?
Method
Database output file will be
(a) in ascii format. The file can be produced with either DOS or
Microsoft standard characters
(b) for the main application of the programme, this file will contain
material with HTML codes
© the file will include diacritics, which should be retained on
viewing in HTML browser. These diacritics
(d) The beginning and end of each record will be tagged (eg beginning
XXX, end YYY). The precise tagging required can be decided by the
programmer
(e) Each record will also be identified with a unique number (this
number can itself be tagged, eg ***1076$$$). The format of the numbers
is currently either (1) straight number eg 2457 (these are the
organisation description) or (2) number with preceding DOC eg DOC3275.
Naming of separated records
(a) each record to be given a unique name based on the record number and with an .htm extension. (eg 1076.htm, doc1555.htm (b) as this process will be repeated regularly (fortnightly/monthly), the same record must be given the same name on each repetition of the process. This is important as the resulting files will be indexed by WWW search engines which will store the file URL (which is based on the file name given): if the file contains a different record, this search service will not work. © these separated records will then need to be copied into directories on the WWW server. Can all 7000 go into one directory? If not, then the programme will also need to copy each record into a predictable directory, overwriting any record of the same name in that directory.
A second maintenance programme is needed which will change the contents of all files in a specified directory/directories so that their content is replaced with a message such "this page has moved. Use the Eldis Search to locate iit" with an appropriate URL The Programme can be designed to run on either our UNIX/Solaris server or under Windows/WindowsNT. We would prefer a programme that we can alter if parameters change in the future (eg record numbers, directory names)
Peter Ferguson
Eldis Project
Institute of Development Studies, University of Sussex
[log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|