/* This merely extracts, doesn't do XML or anything. */
+#include <ccan/talloc/talloc.h>
+#include <ccan/str/str.h>
+#include <ccan/str_talloc/str_talloc.h>
#include <err.h>
#include <stdio.h>
#include <stdlib.h>
#include <fcntl.h>
#include <stdbool.h>
#include <ctype.h>
-#include <ccan/talloc/talloc.h>
-#include <ccan/str/str.h>
-#include <ccan/str_talloc/str_talloc.h>
#include "doc_extract.h"
#include "tools.h"
-static char **grab_doc(char **lines, unsigned int num, unsigned int **linemap)
+static char **grab_doc(char **lines, unsigned int **linemap)
{
char **ret;
- unsigned int i;
+ unsigned int i, num;
bool printing = false;
- ret = talloc_array(NULL, char *, num+1);
- *linemap = talloc_array(ret, unsigned int, num);
+ ret = talloc_array(NULL, char *, talloc_array_length(lines));
+ *linemap = talloc_array(ret, unsigned int, talloc_array_length(lines));
num = 0;
for (i = 0; lines[i]; i++) {
curr->lines[curr->num_lines++] = talloc_strdup(curr->lines, line);
}
-struct list_head *extract_doc_sections(char **rawlines, unsigned int num)
+/* We convert tabs to spaces here. */
+static void add_detabbed_line(struct doc_section *curr, const char *rawline)
+{
+ unsigned int i, eff_i, len, off = 0;
+ char *line;
+
+ /* Worst-case alloc: 8 spaces per tab. */
+ line = talloc_array(curr, char, strlen(rawline) +
+ strcount(rawline, "\t") * 7 + 1);
+ len = 0;
+
+ /* We keep track of the *effective* offset of i. */
+ for (i = eff_i = 0; i < strlen(rawline); i++) {
+ if (rawline[i] == '\t') {
+ do {
+ line[len++] = ' ';
+ eff_i++;
+ } while (eff_i % 8 != 0);
+ } else {
+ line[len++] = rawline[i];
+ if (off == 0 && rawline[i] == '*')
+ off = i + 1;
+ eff_i++;
+ }
+ }
+ line[len] = '\0';
+
+ add_line(curr, line + off);
+ talloc_free(line);
+}
+
+/* Not very efficient: we could track prefix length while doing
+ * add_detabbed_line */
+static void trim_lines(struct doc_section *curr)
+{
+ unsigned int i, trim = -1;
+ int last_non_empty = -1;
+
+ /* Get minimum whitespace prefix. */
+ for (i = 0; i < curr->num_lines; i++) {
+ unsigned int prefix = strspn(curr->lines[i], " ");
+ /* Ignore blank lines */
+ if (curr->lines[i][prefix] == '\0')
+ continue;
+ if (prefix < trim)
+ trim = prefix;
+ }
+
+ /* Now trim it. */
+ for (i = 0; i < curr->num_lines; i++) {
+ unsigned int prefix = strspn(curr->lines[i], " ");
+ if (prefix < trim)
+ curr->lines[i] += prefix;
+ else
+ curr->lines[i] += trim;
+
+ /* All blank? Potential to trim. */
+ if (curr->lines[i][strspn(curr->lines[i], " \t")] != '\0')
+ last_non_empty = i;
+ }
+
+ /* Remove trailing blank lines. */
+ curr->num_lines = last_non_empty + 1;
+}
+
+struct list_head *extract_doc_sections(char **rawlines)
{
unsigned int *linemap;
- char **lines = grab_doc(rawlines, num, &linemap);
+ char **lines = grab_doc(rawlines, &linemap);
const char *function = NULL;
struct doc_section *curr = NULL;
unsigned int i;
}
} else {
if (curr)
- add_line(curr, lines[i]);
+ add_detabbed_line(curr, rawlines[linemap[i]]);
}
}
+
+ list_for_each(list, curr, list)
+ trim_lines(curr);
+
talloc_free(lines);
return list;
}