# HG changeset patch
# User ali <ali@juiblex.co.uk>
# Date 1381960289 -3600
# Node ID 719d0682a16a4590e61baf1c1468b206a76e0e4c
# Parent  37da646396b98688519d402293acf70d6a6aa300
Fix bug #28: Don't report ., as double punctuation after "etc" or "&c"

diff -r 37da646396b9 -r 719d0682a16a bookloupe/bookloupe.c
--- a/bookloupe/bookloupe.c	Tue Oct 15 07:57:45 2013 +0100
+++ b/bookloupe/bookloupe.c	Wed Oct 16 22:51:29 2013 +0100
@@ -2130,6 +2130,27 @@
 }
 
 /*
+ * str_follows_word:
+ *
+ * Given a position p within a string str, determine whether it follows the
+ * given word. This is roughly equivalent to the regular expression (?<=\bword)
+ * but has different boundary conditions.
+ */
+static gboolean str_follows_word(const char *str,const char *p,const char *word)
+{
+    int len=strlen(word);
+    if (p-len<str)
+	return FALSE;
+    else if (!g_str_has_prefix(p-len,word))
+	return FALSE;
+    else if (p-len==str)
+	return TRUE;
+    else
+	/* Using non-alpha as a word boundary. See UAX #29 for a better way. */
+	return !g_unichar_isalpha(g_utf8_get_char(g_utf8_prev_char(p-len)));
+}
+
+/*
  * check_for_double_punctuation:
  *
  * Look for double punctuation like ,. or ,,
@@ -2140,11 +2161,14 @@
  * common errors. What to do? Make these cases paranoid?
  * ".," is the most common, so warnings->dotcomma is used
  * to suppress detailed reporting if it occurs often.
+ * Indeed, ".," is so common after "etc" or "&c" that
+ * we don't warn on these cases at all.
  */
 void check_for_double_punctuation(const char *aline,struct warnings *warnings)
 {
     const char *s;
     gunichar c,nc;
+    gboolean is_query;
     nc=g_utf8_get_char(aline);
     for (s=aline;*s;s=g_utf8_next_char(s))
     {
@@ -2155,36 +2179,30 @@
 	  g_utf8_strchr(".?!,;:",-1,nc))
 	{
 	    /* followed by punctuation, it's a query, unless . . . */
-	    if (c==nc && (c=='.' || c=='?' || c=='!') ||
-	      !warnings->dotcomma && c=='.' && nc==',' ||
-	      warnings->isFrench && g_str_has_prefix(s,",...") ||
-	      warnings->isFrench && g_str_has_prefix(s,"...,") ||
-	      warnings->isFrench && g_str_has_prefix(s,";...") ||
-	      warnings->isFrench && g_str_has_prefix(s,"...;") ||
-	      warnings->isFrench && g_str_has_prefix(s,":...") ||
-	      warnings->isFrench && g_str_has_prefix(s,"...:") ||
-	      warnings->isFrench && g_str_has_prefix(s,"!...") ||
-	      warnings->isFrench && g_str_has_prefix(s,"...!") ||
-	      warnings->isFrench && g_str_has_prefix(s,"?...") ||
-	      warnings->isFrench && g_str_has_prefix(s,"...?"))
+	    is_query=TRUE;
+	    if (warnings->isFrench &&
+	      (g_str_has_prefix(s,",...") || g_str_has_prefix(s,"...,") ||
+	       g_str_has_prefix(s,";...") || g_str_has_prefix(s,"...;") ||
+	       g_str_has_prefix(s,":...") || g_str_has_prefix(s,"...:") ||
+	       g_str_has_prefix(s,"!...") || g_str_has_prefix(s,"...!") ||
+	       g_str_has_prefix(s,"?...") || g_str_has_prefix(s,"...?")))
 	    {
-		if (warnings->isFrench && g_str_has_prefix(s,",...") ||
-		  warnings->isFrench && g_str_has_prefix(s,"...,") ||
-		  warnings->isFrench && g_str_has_prefix(s,";...") ||
-		  warnings->isFrench && g_str_has_prefix(s,"...;") ||
-		  warnings->isFrench && g_str_has_prefix(s,":...") ||
-		  warnings->isFrench && g_str_has_prefix(s,"...:") ||
-		  warnings->isFrench && g_str_has_prefix(s,"!...") ||
-		  warnings->isFrench && g_str_has_prefix(s,"...!") ||
-		  warnings->isFrench && g_str_has_prefix(s,"?...") ||
-		  warnings->isFrench && g_str_has_prefix(s,"...?"))
-		{
-		    s+=4;
-		    nc=g_utf8_get_char(g_utf8_next_char(s));
-		}
-		; /* do nothing for .. !! and ?? which can be legit */
+		s+=4;
+		nc=g_utf8_get_char(g_utf8_next_char(s));
+		is_query=FALSE;
 	    }
-	    else
+	    else if (c==nc && (c=='.' || c=='?' || c=='!'))
+	    {
+		/* do nothing for .. !! and ?? which can be legit */
+		is_query=FALSE;
+	    }
+	    else if (c=='.' && nc==',')
+	    {
+		if (!warnings->dotcomma || str_follows_word(aline,s,"etc") || 
+		  str_follows_word(aline,s,"&c"))
+		    is_query=FALSE;
+	    }
+	    if (is_query)
 	    {
 		if (pswit[ECHO_SWITCH])
 		    g_print("\n%s\n",aline);
diff -r 37da646396b9 -r 719d0682a16a test/bookloupe/Makefile.am
--- a/test/bookloupe/Makefile.am	Tue Oct 15 07:57:45 2013 +0100
+++ b/test/bookloupe/Makefile.am	Wed Oct 16 22:51:29 2013 +0100
@@ -1,6 +1,6 @@
 TESTS_ENVIRONMENT=BOOKLOUPE=../../bookloupe/bookloupe ../harness/loupe-test
 TESTS=non-ascii.tst long-line.tst curved-single-quotes.tst curved-quotes.tst \
 	runfox-quotes.tst curved-genitives.tst multi-line-illustration.tst \
-	emdash.tst
+	emdash.tst dot-comma.tst
 
 dist_pkgdata_DATA=$(TESTS)
diff -r 37da646396b9 -r 719d0682a16a test/bookloupe/dot-comma.tst
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test/bookloupe/dot-comma.tst	Wed Oct 16 22:51:29 2013 +0100
@@ -0,0 +1,6 @@
+**************** INPUT ****************
+All men should read the Gospels, &c., in their vulgar tongue.
+
+I would have redeemed them from the grip of Sheol, etc., if they had been
+wise, but being foolish I will bring on them the plagues of death.
+**************** EXPECTED ****************