jaw@ames.UUCP (01/27/87)
# "i put a spell on you" -- screamin' jay hawkins
though the "one-off" correction capability of 'ispell' is useful,
a more general interactive lookup capability using wildcards can
easily be added via an 'L' command. for example, to look up the word
silhouette
the mods below let one input
sil*et*
for quick search using my fast 'e?grep' (less than a second on the
venerable 11/750). shell syntax is chosen over 'grep' syntax
for ease of use by novices, as well as minimum typing.
if you want to get fancy (or move up to /usr/dict/web2), consider
pre-processing the grep command below with 'look' where possible, though
not all systems have this.
the only other things i'd want to see in future 'ispell' versions would
be the user dictionary 'fix'er mentioned in this forum ('awk' is natural
for this) run in conjunction with hashing the smaller local lists into
the large memory table for uniform affix handling. it's odd that
this is not done currently.
ames!jaw
-----
the diffs to file ispell.c:
add to givehelp(), after ... printf ("0-9 Replace ..."); ...
< printf ("L Look up words in system dictionary.\r\n");
while you're at it, practice what 'ispell' preaches by correcting
the spelling error (!)
< printf ("X Exit immediately. Asks for confirmation. ");
---
> printf ("X Exit immediately. Asks for conformation. ");
then fold in this command switch:
< case 'l': case 'L':
< {
< char buf[100];
< move (18, 0);
< printf ("Lookup string ('*' is wildcard): ");
< if (getline (buf) == NULL) {
< putchar (7);
< erase ();
< goto checkagain;
< }
< printf ("\r\n\r\n");
< lookharder (buf);
< erase ();
< goto checkagain;
< }
finally, append routine lookharder() to the end of ispell.c:
< lookharder(string)
< char *string;
< {
< char cmd[150];
< char *g, *s, grepstr[100];
<
< g = grepstr;
< for (s = string; *s != '\0'; s++)
< if (*s == '*')
< *g++ = '.', *g++ = '*';
< else
< *g++ = *s;
< *g = '\0';
< if (grepstr[0]) {
< sprintf (cmd, "/usr/local/egrep -i '^%s$' /usr/dict/words", grepstr);
< shellescape (cmd);
< }
< }
be sure to use the 5-10x faster 'egrep' in the mod.sources archive.
the kanji-ready, grep-unified, even swifter in cases 'e?grep' version
is slated to appear in mod.sources "soon".