Shareware Grab Bag/Shareware Grab Bag.iso/007/pcfilerd.arc/PCFRD.DOC

Text File | 1988-09-20 | 5.6 KB | 94 lines

   The text below is from The C Gazette, Summer 1988, Volume 3
No. 1, and is the magazine text for PCFILERD.C.

----------------------------------------------------------------------------
   PC-FILE and it's successor PC-FILE+ are the most widely distributed data-
base managers for the PC. Their success is due to their ease of use and their
availability as shareware. This article discusses the file formats and access
to the data and index records using C.

   PC-FILE+ provides programmers with an inexpensive front-end to database
applications. The programs that handle data-entry and record management are
commonly available as shareware. They are ideal for the management of small
databases. Few packages are as easy to install and use, and none offers so
much functionality at such a low price. The complete system with a bound
manual and technical support is available for $69.95 from Buttonware (ad-
dress at end of article).
   PC-FILE manages data as a flat file that is a collection of fixed-length
sequential records. A separate index file permits random access to individual
data records.
   The index file contains one record for each data record. Every field in
the data record has a corresponding two-byte index in the index record, such
that every field of every data record is accessible directly. The index field
contains the first two bytes of the data field, such that a name field cont-
aining "ERIC", for example, would have an index entry of "ER". The index
record contains two bytes for each data field plus a two-byte unsigned integer
at the end which holds the record number of the data. This number is limited
to the range of an unsigned int, 65536, and predictably, this is the upper
limit on the number of records supported in one database. (Early versions used
a signed integer and were limited thereby to half the current capacity.)
   To find a data record based on the record number, the following formula
is to calculate the record's offset into the data file:
      (record number - 1) * length of data
   The record number should be decremented because the first record (number
1) starts at an offset of zero bytes into the file. The length of the data
record is the sum of all the data fields plus one byte. This byte is a
carriage return (0x0d) which PC-FILE appends to each data record.
   To access a database, we need to know the number of fields in each record
(to know the length of the index record) and how many bytes of data each
record holds (to know the length of the data records). This information is
available in a separate header file maintained by PC-FILE.
   Every database contains at least three files: filename.inx, filename.dta,
and filename.hdr. These are the index, data, and header files respectively.
The header file is a straight ASCII file which lists each data field: it's
name, size, and placement in the record. The keys in the index file follow the
same order as the data record layout.
   Access to a data file is done as follows. Suppose a mailing list file
were being queried for all persons living in Dubuque. The index file would be
read one record at a time and the city key would be checked against the first
two bytes of the search key, namely "DU". When a match occurs, the data record
is read (based on calculation the offset into the data file) and the actual
city name is verified. This check avoids denizens of DUluth from being
accepted by the query.
   A sequential read of the file requires reading each index record and look-
ing up the data record. The accompanying program, PCFILERD.C, demonstrates the
technique of a sequential read through a file.
   Two special cases need to be accounted for. Deleted records are flagged
by the placement of a forward slash in the first byte of the index record.
Such records should simply be skipped. End-of-file is indicated by a backslash
in the first byte. This should always be checked for and considered a true EOF
condition. PC-FILE maintains one record at file's end with this flag; hence a
sequential read will never attempt to go beyond the end-of-file. The function
get_record() illustrates the handling of these special cases.
   ( Typist's note: The accompanying program displays my database. The func-
tion print_line() as it appears in the magazine article follows below. If you
have trouble printing your database, try replacing the print_line() in the
accompanying program PCFILERD.C with this version: )


void print_line()       /* Just a sample application: print the first */
{                       /*     forty characters of the data record */

   char line [40];

   strncpy (line, Data_buffer, 40);
   if (strlen (line) > 39) /* Remember, strncpy() may not end */
      line[39] = '\0';     /*    the string with a null.      */
   printf ("%s\n", line);
}

   This application is simply for demonstration, since all forms of data
manipulation are of course possible.
   The information in this article relies in no small part on material pre-
sented in the manual to PC-FILE+. (Typist's note: I used this program with
PC-FILE III Ver.4 and it works fine.)  This book should be bought by anyone
who wishes to use PC-FILE+ to it's full effect. An appendix contains some
sample C source code which demonstrates file access. However, it would be kind
to say that the author's native language is not C. The coding techniques
should not be imitated at all, and the routines presented here should be used
instead.
   PC-FILE+ is sold by Buttonware, P.O. Box 5786, Bellevue, WA 98006.
(800) JBUTTON. Not surprisingly, PC-FILE and PC-FILE+ are trademarks of
Buttonware.