From 84cf2dcc12c5f978d0563bce2e9c85e9ff3c5956 Mon Sep 17 00:00:00 2001 From: Sebastian Hammer Date: Mon, 15 Jan 2007 19:01:15 +0000 Subject: [PATCH] Dealing with empty subject headings and punctionation at the end of SHs --- src/termlists.c | 17 +++++++++++++---- 1 file changed, 13 insertions(+), 4 deletions(-) diff --git a/src/termlists.c b/src/termlists.c index 027d057..19517d7 100644 --- a/src/termlists.c +++ b/src/termlists.c @@ -1,9 +1,10 @@ /* - * $Id: termlists.c,v 1.4 2007-01-10 10:04:23 adam Exp $ + * $Id: termlists.c,v 1.5 2007-01-15 19:01:15 quinn Exp $ */ #include #include +#include #include #if HAVE_CONFIG_H @@ -124,11 +125,19 @@ void termlist_insert(struct termlist *tl, const char *term) { unsigned int bucket; struct termlist_bucket **p; + char buf[256], *cp; - bucket = hash((unsigned char *)term) & tl->hashmask; + if (strlen(term) > 255) + return; + strcpy(buf, term); + for (cp = buf + strlen(buf) - 1; cp > buf && + (*cp == ',' || *cp == '.' || *cp == ' '); cp--) + *cp = '\0'; + + bucket = hash((unsigned char *)buf) & tl->hashmask; for (p = &tl->hashtable[bucket]; *p; p = &(*p)->next) { - if (!strcmp(term, (*p)->term.term)) + if (!strcmp(buf, (*p)->term.term)) { (*p)->term.frequency++; update_highscore(tl, &((*p)->term)); @@ -139,7 +148,7 @@ void termlist_insert(struct termlist *tl, const char *term) { struct termlist_bucket *new = nmem_malloc(tl->nmem, sizeof(struct termlist_bucket)); - new->term.term = nmem_strdup(tl->nmem, term); + new->term.term = nmem_strdup(tl->nmem, buf); new->term.frequency = 1; new->next = 0; *p = new; -- 1.7.10.4