speed up incremental update with deleted files and usegdbm

Author: Peter McCluskey

Changelog

@@ -1,6 +1,10 @@
 Version Changes for Hypermail
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
+ Peter McCluskey (Mar 23, 2004)
+   Changes to speed up incremental update when there are deleted files if
+  using usegdbm (it's still slow if usegdbm = 0).
+
 ============================
 HYPERMAIL VERSION 2.1.9:
 ============================

src/hypermail.h

@@ -274,6 +274,8 @@ struct emailinfo {
     long exp_time;
     int is_deleted;	/* 1=deleted, 2=expired, 4=filtered out, */
 			/* 8=filtered (required line missing) */
+    int deletion_completed; /* -1 or delete_level that reflects last time */
+                            /* that file was rewritten to reflect is_deleted */
 };
 
 struct header {

src/parse.c

@@ -3204,6 +3204,7 @@ int loadoldheadersfromGDBMindex(char *dir, int get_count_only)
       GDBM_FILE gp;
       int num;
       int num_added = 0;
+      int old_delete_level = -1;
 
       if (!get_count_only)
 	authorlist = subjectlist = datelist = NULL;
@@ -3235,6 +3236,12 @@ int loadoldheadersfromGDBMindex(char *dir, int get_count_only)
 	datum key;
 	int max_num;
 
+	key.dptr = "delete_level";
+	key.dsize = strlen(key.dptr);
+	content = gdbm_fetch(gp, key);
+	if (content.dptr)
+	    old_delete_level = atoi(content.dptr);
+
 	key.dptr = (char *) #
 	key.dsize = sizeof(num);
 
@@ -3309,6 +3316,7 @@ int loadoldheadersfromGDBMindex(char *dir, int get_count_only)
 			   fromdate, charset, isodate, isofromdate, bp))) {
 	      emp->exp_time = exp_time;
 	      emp->is_deleted = is_deleted;
+	      emp->deletion_completed = old_delete_level;
 	      check_expiry(emp);
 	      if (insert_in_lists(emp, NULL, 0))
 		  ++num_added;

src/print.c

@@ -1775,6 +1775,17 @@ static GDBM_FILE gdbm_init()
 
 	    unlink(indexname);
 	}
+	else {
+	    datum key;
+	    datum content;
+	    char buf[512];
+	    key.dptr = "delete_level";
+	    key.dsize = strlen(key.dptr);
+	    sprintf(buf, "%d", set_delete_level);
+	    content.dsize = strlen(buf) + 1;
+	    content.dptr = buf; /* the value is in this string */
+	    gdbm_store(gp, key, content, GDBM_REPLACE);
+	}
     }
     return gp;
 }
@@ -1796,17 +1807,22 @@ void update_deletions(int num_old)
 	if (num >= num_old)
 	    continue;		/* new message - already done */
 	if (hashnumlookup(num, &ep)) {
-	    char *filename = articlehtmlfilename(ep);
-	    if (set_delete_level != DELETE_REMOVES_FILES) {
-		struct body *bp = ep->bodylist;
-		if (bp == NULL)
-		    parse_old_html(num, ep, 1, 0, NULL, 0);
-		writearticles(num, num + 1);
-	    }
-	    else if (isfile(filename)) {
-		unlink(filename);
+	    char *filename;
+	    if (ep->deletion_completed == set_delete_level) /* done already? */
+	        continue;
+	    if (set_delete_level != DELETE_LEAVES_TEXT) {
+	        filename = articlehtmlfilename(ep);
+		if (set_delete_level != DELETE_REMOVES_FILES) {
+		    struct body *bp = ep->bodylist;
+		    if (bp == NULL)
+		        parse_old_html(num, ep, 1, 0, NULL, 0);
+		    writearticles(num, num + 1);
+		}
+		else if (isfile(filename)) {
+		    unlink(filename);
+		}
+		free(filename);
 	    }
-	    free(filename);
 #ifdef FASTREPLYCODE
 	    for (rp = ep->replylist; rp != NULL; rp = rp->next) {
 #else

src/struct.c

@@ -266,6 +266,7 @@ struct emailinfo *addhash(int num, char *date, char *name, char *email, char *ms
     e->charset = strsav(charset);
     e->flags = 0;
     e->is_deleted = 0;
+    e->deletion_completed = -1;
     e->exp_time = -1;
     e->bodylist = sp;
     e->initial_next_in_thread = -1;