docbook2mdoc: optional parent-child histogram for one relation

docbook2mdoc: optional parent-child histogram for one relation

[schwarze]

Log Message:
-----------
optional parent-child histogram for one relation

Modified Files:
--------------
    docbook2mdoc:
        statistics.c

Revision Data
-------------
Index: statistics.c
===================================================================
RCS file: /home/cvs/mdocml/docbook2mdoc/statistics.c,v
retrieving revision 1.3
retrieving revision 1.4
diff -Lstatistics.c -Lstatistics.c -u -p -r1.3 -r1.4
--- statistics.c
+++ statistics.c
@@ -36,6 +36,16 @@
  * Typical usage:
  * statistics < filenames.txt | sort -n
  * statistics < filenames.txt | grep '\<listitem\>' | sort -n
+ *
+ * Relations already fully implemented are excluded by default.
+ * The option -a shows all relations.
+ *
+ * If two arguments (parent and child) are given, a histogram
+ * of the number of children of the kind in each parent is given
+ * in addition to the normal output.
+ *
+ * Example usage:
+ * statistics tgroup colspec < filenames.txt | grep colspec
  */
 
 struct entry {
@@ -52,6 +62,17 @@ static char		**stack;
 static size_t		  stacksz;
 static size_t		  stacki;
 
+static const int	  nchildsz = 8;
+struct nchild {
+	char	*parent;
+	char	*child;
+	int	 freq[nchildsz];
+	int	 count;
+};
+
+static struct nchild	  nchild;
+static char		 *fname;
+
 
 /*
  * Count one instance of a parent-child relation.
@@ -70,6 +91,21 @@ table_add(const char *parent, const char
 		return;
 	}
 
+	/* Optional parent-child histogram. */
+
+	if (init_done && parent != NULL && child != NULL &&
+	    nchild.parent != NULL && nchild.child != NULL &&
+	    strcmp(parent, nchild.parent) == 0 &&
+	    strcmp(child, nchild.child) == 0) {
+		if (nchild.count < nchildsz) {
+			nchild.freq[nchild.count]++;
+			if (nchild.count > 0)
+				nchild.freq[nchild.count - 1]--;
+		} else if (nchild.count == nchildsz)
+			puts(fname);
+		nchild.count++;
+	}
+
 	/* If the table entry already exists, increment its count. */
 
 	for (i = 0; i < tablei; i++) {
@@ -109,6 +145,9 @@ table_add(const char *parent, const char
 static void
 stack_push(const char *name)
 {
+	if (nchild.parent != NULL && strcmp(name, nchild.parent) == 0)
+		nchild.count = 0;
+
 	if (stacki == stacksz) {
 		stacksz += 8;
 		stack = reallocarray(stack, stacksz, sizeof(*stack));
@@ -283,7 +322,6 @@ parse_file(int fd, char *fname)
 int
 main(int argc, char *argv[])
 {
-	char		*fname;
 	size_t		 fsz, i;
 	ssize_t		 rsz;
 	int		 ch, fd, show_all;
@@ -298,6 +336,13 @@ main(int argc, char *argv[])
 			return 1;
 		}
 	}
+	argc -= optind;
+	argv += optind;
+
+	if (argc > 1) {
+		nchild.parent = argv[0];
+		nchild.child = argv[1];
+	}
 
 	/* Exclude relations that are already fully implemented. */
 	if (show_all == 0) {
@@ -329,5 +374,13 @@ main(int argc, char *argv[])
 		if (table[i].count != -1)
 			printf("%d\t%s\t%s\n", table[i].count,
 			    table[i].parent, table[i].child);
+
+	/* Optional parent-child histogram. */
+	if (nchild.parent != NULL) {
+		printf("%s %s", nchild.parent, nchild.child);
+		for (i = 0; i < nchildsz; i++)
+			printf(" %d", nchild.freq[i]);
+		putchar('\n');
+	}
 	return 0;
 }
--
 To unsubscribe send an email to source+unsubscribe@mandoc.bsd.lv