Merge branch 'experimental'

I'm not entirely happy with the new paragraph heuristics, but I've been
using them for a while and have grown used to them and have grown to
rely on the behavior.  Since there probably aren't all that many other
people who use uemacs, let's just merge that behavior and see if anybody
else even notices.

The "NBSP to SP on keyboard input" is similarly convenient to me, and
might be offensive to others.  Let's see.

* experimental:
  Turn NBSP into regular SP on input
  Try updated rule for "is new paragraph"
  Make some minor code legibility changes

2 files changed, 31 insertions, 23 deletions

diff --git a/basic.c b/basic.c
index 5071047..010288c 100644
--- a/basic.c
+++ b/basic.c
@@ -270,6 +270,28 @@ int backline(int f, int n)
 }
 
 #if	WORDPRO
+static int is_new_para(void)
+{
+	int i, len;
+
+	len = llength(curwp->w_dotp);
+
+	for (i = 0; i < len; i++) {
+		int c = lgetc(curwp->w_dotp, i);
+		if (c == ' ' || c == TAB) {
+#if PKCODE
+			if (justflag)
+				continue;
+#endif
+			return 1;
+		}
+		if (!isletter(c))
+			return 1;
+		return 0;
+	}
+	return 1;
+}
+
 /*
  * go back to the beginning of the current paragraph
  * here we look for a <NL><NL> or <NL><TAB> or <NL><SPACE>
@@ -294,19 +316,11 @@ int gotobop(int f, int n)
 
 		/* and scan back until we hit a <NL><NL> or <NL><TAB>
 		   or a <NL><SPACE>                                     */
-		while (lback(curwp->w_dotp) != curbp->b_linep)
-			if (llength(curwp->w_dotp) != 0 &&
-#if	PKCODE
-			    ((justflag == TRUE) ||
-#endif
-			     (lgetc(curwp->w_dotp, curwp->w_doto) != TAB &&
-			      lgetc(curwp->w_dotp, curwp->w_doto) != ' '))
-#if	PKCODE
-			    )
-#endif
-				curwp->w_dotp = lback(curwp->w_dotp);
-			else
+		while (lback(curwp->w_dotp) != curbp->b_linep) {
+			if (is_new_para())
 				break;
+			curwp->w_dotp = lback(curwp->w_dotp);
+		}
 
 		/* and then forward until we are in a word */
 		suc = forwchar(FALSE, 1);
@@ -343,18 +357,9 @@ int gotoeop(int f, int n)
 		/* and scan forword until we hit a <NL><NL> or <NL><TAB>
 		   or a <NL><SPACE>                                     */
 		while (curwp->w_dotp != curbp->b_linep) {
-			if (llength(curwp->w_dotp) != 0 &&
-#if	PKCODE
-			    ((justflag == TRUE) ||
-#endif
-			     (lgetc(curwp->w_dotp, curwp->w_doto) != TAB &&
-			      lgetc(curwp->w_dotp, curwp->w_doto) != ' '))
-#if	PKCODE
-			    )
-#endif
-				curwp->w_dotp = lforw(curwp->w_dotp);
-			else
+			if (is_new_para())
 				break;
+			curwp->w_dotp = lforw(curwp->w_dotp);
 		}
 
 		/* and then backward until we are in a word */
diff --git a/posix.c b/posix.c
index 97edd9f..74ba7b9 100644
--- a/posix.c
+++ b/posix.c
@@ -213,6 +213,9 @@ int ttgetc(void)
 	}
 	bytes = utf8_to_unicode(buffer, 0, pending, &c);
 
+	/* Hackety hack! Turn no-break space into regular space */
+	if (c == 0xa0)
+		c = ' ';
 done:
 	pending -= bytes;
 	memmove(buffer, buffer+bytes, pending);