-/* $Id: html.c,v 1.196 2017/01/18 19:22:21 schwarze Exp $ */
+/* $Id: html.c,v 1.220 2017/09/06 16:24:25 schwarze Exp $ */
/*
* Copyright (c) 2008-2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
* Copyright (c) 2011-2015, 2017 Ingo Schwarze <schwarze@openbsd.org>
#include <string.h>
#include <unistd.h>
-#include "mandoc.h"
#include "mandoc_aux.h"
+#include "mandoc.h"
+#include "roff.h"
#include "out.h"
#include "html.h"
#include "manconf.h"
{"br", HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL},
{"a", 0},
{"table", HTML_NLALL | HTML_INDENT},
- {"tbody", HTML_NLALL | HTML_INDENT},
+ {"colgroup", HTML_NLALL | HTML_INDENT},
{"col", HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL},
{"tr", HTML_NLALL | HTML_INDENT},
{"td", HTML_NLAROUND},
{"dl", HTML_NLALL | HTML_INDENT},
{"dt", HTML_NLAROUND},
{"dd", HTML_NLAROUND | HTML_INDENT},
- {"blockquote", HTML_NLALL | HTML_INDENT},
{"pre", HTML_NLALL | HTML_NOINDENT},
+ {"var", 0},
+ {"cite", 0},
{"b", 0},
{"i", 0},
{"code", 0},
{"math", HTML_NLALL | HTML_INDENT},
{"mrow", 0},
{"mi", 0},
+ {"mn", 0},
{"mo", 0},
{"msup", 0},
{"msub", 0},
};
static void a2width(const char *, struct roffsu *);
-static void html_endline(struct html *);
-static void html_indent(struct html *);
+static void print_byte(struct html *, char);
+static void print_endword(struct html *);
+static void print_indent(struct html *);
+static void print_word(struct html *, const char *);
+
static void print_ctag(struct html *, struct tag *);
-static int print_escape(char);
+static int print_escape(struct html *, char);
static int print_encode(struct html *, const char *, const char *, int);
static void print_href(struct html *, const char *, const char *, int);
static void print_metaf(struct html *, enum mandoc_esc);
h = mandoc_calloc(1, sizeof(struct html));
- h->tags.head = NULL;
+ h->tag = NULL;
h->style = outopts->style;
h->base_man = outopts->man;
h->base_includes = outopts->includes;
h = (struct html *)p;
- while ((tag = h->tags.head) != NULL) {
- h->tags.head = tag->next;
+ while ((tag = h->tag) != NULL) {
+ h->tag = tag->next;
free(tag);
}
t = print_otag(h, TAG_STYLE, "");
print_text(h, "table.head, table.foot { width: 100%; }");
- html_endline(h);
+ print_endline(h);
print_text(h, "td.head-rtitle, td.foot-os { text-align: right; }");
- html_endline(h);
+ print_endline(h);
print_text(h, "td.head-vol { text-align: center; }");
- html_endline(h);
- print_text(h, "table.foot td { width: 50%; }");
- html_endline(h);
- print_text(h, "table.head td { width: 33%; }");
- html_endline(h);
- print_text(h, "div.spacer { margin: 1em 0; }");
+ print_endline(h);
+ print_text(h, "div.Pp { margin: 1ex 0ex; }");
print_tagq(h, t);
if (h->style)
}
}
+char *
+html_make_id(const struct roff_node *n)
+{
+ const struct roff_node *nch;
+ char *buf, *cp;
+
+ for (nch = n->child; nch != NULL; nch = nch->next)
+ if (nch->type != ROFFT_TEXT)
+ return NULL;
+
+ buf = NULL;
+ deroff(&buf, n);
+ if (buf == NULL)
+ return NULL;
+
+ /* http://www.w3.org/TR/html5/dom.html#the-id-attribute */
+
+ for (cp = buf; *cp != '\0'; cp++)
+ if (*cp == ' ')
+ *cp = '_';
+
+ return buf;
+}
+
int
html_strlen(const char *cp)
{
}
static int
-print_escape(char c)
+print_escape(struct html *h, char c)
{
switch (c) {
case '<':
- printf("<");
+ print_word(h, "<");
break;
case '>':
- printf(">");
+ print_word(h, ">");
break;
case '&':
- printf("&");
+ print_word(h, "&");
break;
case '"':
- printf(""");
+ print_word(h, """);
break;
case ASCII_NBRSP:
- printf(" ");
+ print_word(h, " ");
break;
case ASCII_HYPH:
- putchar('-');
+ print_byte(h, '-');
break;
case ASCII_BREAK:
break;
static int
print_encode(struct html *h, const char *p, const char *pend, int norecurse)
{
- size_t sz;
- int c, len, nospace;
+ char numbuf[16];
+ struct tag *t;
const char *seq;
+ size_t sz;
+ int c, len, breakline, nospace;
enum mandoc_esc esc;
- static const char rejs[9] = { '\\', '<', '>', '&', '"',
+ static const char rejs[10] = { ' ', '\\', '<', '>', '&', '"',
ASCII_NBRSP, ASCII_HYPH, ASCII_BREAK, '\0' };
if (pend == NULL)
pend = strchr(p, '\0');
+ breakline = 0;
nospace = 0;
while (p < pend) {
continue;
}
- sz = strcspn(p, rejs);
- if (p + sz > pend)
- sz = pend - p;
-
- fwrite(p, 1, sz, stdout);
- p += (int)sz;
+ for (sz = strcspn(p, rejs); sz-- && p < pend; p++)
+ print_byte(h, *p);
+
+ if (breakline &&
+ (p >= pend || *p == ' ' || *p == ASCII_NBRSP)) {
+ t = print_otag(h, TAG_DIV, "");
+ print_text(h, "\\~");
+ print_tagq(h, t);
+ breakline = 0;
+ while (p < pend && (*p == ' ' || *p == ASCII_NBRSP))
+ p++;
+ continue;
+ }
if (p >= pend)
break;
- if (print_escape(*p++))
+ if (*p == ' ') {
+ print_endword(h);
+ p++;
+ continue;
+ }
+
+ if (print_escape(h, *p++))
continue;
esc = mandoc_escape(&p, &seq, &len);
if (c <= 0)
continue;
break;
+ case ESCAPE_BREAK:
+ breakline = 1;
+ continue;
case ESCAPE_NOSPACE:
if ('\0' == *p)
nospace = 1;
if ((c < 0x20 && c != 0x09) ||
(c > 0x7E && c < 0xA0))
c = 0xFFFD;
- if (c > 0x7E)
- printf("&#%d;", c);
- else if ( ! print_escape(c))
- putchar(c);
+ if (c > 0x7E) {
+ (void)snprintf(numbuf, sizeof(numbuf), "&#x%.4X;", c);
+ print_word(h, numbuf);
+ } else if (print_escape(h, c) == 0)
+ print_byte(h, c);
}
return nospace;
print_encode(h, pp, p, 1);
if (man && p[1] == 'S') {
if (sec == NULL)
- putchar('1');
+ print_byte(h, '1');
else
print_encode(h, sec, NULL, 1);
} else if ((man && p[1] == 'N') ||
{
va_list ap;
struct roffsu mysu, *su;
+ char numbuf[16];
struct tag *t;
const char *attr;
- char *s;
+ char *arg1, *arg2;
double v;
int i, have_style, tflags;
tflags = htmltags[tag].flags;
- /* Push this tags onto the stack of open scopes. */
+ /* Push this tag onto the stack of open scopes. */
if ((tflags & HTML_NOSTACK) == 0) {
t = mandoc_malloc(sizeof(struct tag));
t->tag = tag;
- t->next = h->tags.head;
- h->tags.head = t;
+ t->next = h->tag;
+ h->tag = t;
} else
t = NULL;
if (tflags & HTML_NLBEFORE)
- html_endline(h);
- if (h->flags & HTML_NLDONE)
- html_indent(h);
+ print_endline(h);
+ if (h->col == 0)
+ print_indent(h);
else if ((h->flags & HTML_NOSPACE) == 0) {
if (h->flags & HTML_KEEP)
- printf(" ");
+ print_word(h, " ");
else {
if (h->flags & HTML_PREKEEP)
h->flags |= HTML_KEEP;
- putchar(' ');
+ print_endword(h);
}
}
/* Print out the tag name and attributes. */
- printf("<%s", htmltags[tag].name);
+ print_byte(h, '<');
+ print_word(h, htmltags[tag].name);
va_start(ap, fmt);
have_style = 0;
while (*fmt != '\0') {
if (*fmt == 's') {
- printf(" style=\"");
have_style = 1;
fmt++;
break;
}
- s = va_arg(ap, char *);
+
+ /* Parse a non-style attribute and its arguments. */
+
+ arg1 = va_arg(ap, char *);
switch (*fmt++) {
case 'c':
attr = "class";
attr = "id";
break;
case '?':
- attr = s;
- s = va_arg(ap, char *);
+ attr = arg1;
+ arg1 = va_arg(ap, char *);
break;
default:
abort();
}
- printf(" %s=\"", attr);
+ arg2 = NULL;
+ if (*fmt == 'M')
+ arg2 = va_arg(ap, char *);
+ if (arg1 == NULL)
+ continue;
+
+ /* Print the non-style attributes. */
+
+ print_byte(h, ' ');
+ print_word(h, attr);
+ print_byte(h, '=');
+ print_byte(h, '"');
switch (*fmt) {
- case 'M':
- print_href(h, s, va_arg(ap, char *), 1);
+ case 'I':
+ print_href(h, arg1, NULL, 0);
fmt++;
break;
- case 'I':
- print_href(h, s, NULL, 0);
+ case 'M':
+ print_href(h, arg1, arg2, 1);
fmt++;
break;
case 'R':
- putchar('#');
+ print_byte(h, '#');
+ print_encode(h, arg1, NULL, 1);
+ fmt++;
+ break;
+ case 'T':
+ print_encode(h, arg1, NULL, 1);
+ print_word(h, "\" title=\"");
+ print_encode(h, arg1, NULL, 1);
fmt++;
- /* FALLTHROUGH */
+ break;
default:
- print_encode(h, s, NULL, 1);
+ print_encode(h, arg1, NULL, 1);
break;
}
- putchar('"');
+ print_byte(h, '"');
}
/* Print out styles. */
- s = NULL;
- su = &mysu;
while (*fmt != '\0') {
+ arg1 = NULL;
+ su = NULL;
/* First letter: input argument type. */
switch (*fmt++) {
case 'h':
i = va_arg(ap, int);
+ su = &mysu;
SCALE_HS_INIT(su, i);
break;
case 's':
- s = va_arg(ap, char *);
+ arg1 = va_arg(ap, char *);
break;
case 'u':
su = va_arg(ap, struct roffsu *);
break;
- case 'v':
- i = va_arg(ap, int);
- SCALE_VS_INIT(su, i);
- break;
case 'w':
- s = va_arg(ap, char *);
- a2width(s, su);
+ if ((arg2 = va_arg(ap, char *)) != NULL) {
+ su = &mysu;
+ a2width(arg2, su);
+ }
+ if (*fmt == '*') {
+ if (su != NULL && su->unit == SCALE_EN &&
+ su->scale > 5.9 && su->scale < 6.1)
+ su = NULL;
+ fmt++;
+ }
+ if (*fmt == '+') {
+ if (su != NULL) {
+ /* Make even bold text fit. */
+ su->scale *= 1.2;
+ /* Add padding. */
+ su->scale += 3.0;
+ }
+ fmt++;
+ }
+ if (*fmt == '-') {
+ if (su != NULL)
+ su->scale *= -1.0;
+ fmt++;
+ }
break;
default:
abort();
/* Second letter: style name. */
switch (*fmt++) {
- case 'b':
- attr = "margin-bottom";
- break;
case 'h':
attr = "height";
break;
case 'l':
attr = "margin-left";
break;
- case 't':
- attr = "margin-top";
- break;
case 'w':
attr = "width";
break;
attr = "min-width";
break;
case '?':
- printf("%s: %s;", s, va_arg(ap, char *));
- continue;
+ attr = arg1;
+ arg1 = va_arg(ap, char *);
+ break;
default:
abort();
}
- v = su->scale;
- if (su->unit == SCALE_MM && (v /= 100.0) == 0.0)
- v = 1.0;
- else if (su->unit == SCALE_BU)
- v /= 24.0;
- printf("%s: %.2f%s;", attr, v, roffscales[su->unit]);
+ if (su == NULL && arg1 == NULL)
+ continue;
+
+ if (have_style == 1)
+ print_word(h, " style=\"");
+ else
+ print_byte(h, ' ');
+ print_word(h, attr);
+ print_byte(h, ':');
+ print_byte(h, ' ');
+ if (su != NULL) {
+ v = su->scale;
+ if (su->unit == SCALE_MM && (v /= 100.0) == 0.0)
+ v = 1.0;
+ else if (su->unit == SCALE_BU)
+ v /= 24.0;
+ (void)snprintf(numbuf, sizeof(numbuf), "%.2f", v);
+ print_word(h, numbuf);
+ print_word(h, roffscales[su->unit]);
+ } else
+ print_word(h, arg1);
+ print_byte(h, ';');
+ have_style = 2;
}
- if (have_style)
- putchar('"');
+ if (have_style == 2)
+ print_byte(h, '"');
va_end(ap);
/* Accommodate for "well-formed" singleton escaping. */
if (HTML_AUTOCLOSE & htmltags[tag].flags)
- putchar('/');
+ print_byte(h, '/');
- putchar('>');
+ print_byte(h, '>');
if (tflags & HTML_NLBEGIN)
- html_endline(h);
+ print_endline(h);
else
h->flags |= HTML_NOSPACE;
if (tflags & HTML_NOINDENT)
h->noindent--;
if (tflags & HTML_NLEND)
- html_endline(h);
- html_indent(h);
- printf("</%s>", htmltags[tag->tag].name);
+ print_endline(h);
+ print_indent(h);
+ print_byte(h, '<');
+ print_byte(h, '/');
+ print_word(h, htmltags[tag->tag].name);
+ print_byte(h, '>');
if (tflags & HTML_NLAFTER)
- html_endline(h);
+ print_endline(h);
- h->tags.head = tag->next;
+ h->tag = tag->next;
free(tag);
}
void
print_gen_decls(struct html *h)
{
-
- puts("<!DOCTYPE html>");
- h->flags |= HTML_NLDONE;
+ print_word(h, "<!DOCTYPE html>");
+ print_endline(h);
}
void
print_text(struct html *h, const char *word)
{
- if ((h->flags & (HTML_NLDONE | HTML_NOSPACE)) == 0) {
+ if (h->col && (h->flags & HTML_NOSPACE) == 0) {
if ( ! (HTML_KEEP & h->flags)) {
if (HTML_PREKEEP & h->flags)
h->flags |= HTML_KEEP;
- putchar(' ');
+ print_endword(h);
} else
- printf(" ");
+ print_word(h, " ");
}
assert(NULL == h->metaf);
print_otag(h, TAG_I, "");
break;
default:
- html_indent(h);
+ print_indent(h);
break;
}
{
struct tag *tag;
- while ((tag = h->tags.head) != NULL) {
+ while ((tag = h->tag) != NULL) {
print_ctag(h, tag);
if (until && tag == until)
return;
{
struct tag *tag;
- while ((tag = h->tags.head) != NULL) {
+ while ((tag = h->tag) != NULL) {
if (suntil && tag == suntil)
return;
print_ctag(h, tag);
{
struct tag *t;
- t = print_otag(h, TAG_DIV, "c", "spacer");
+ t = print_otag(h, TAG_DIV, "c", "Pp");
print_tagq(h, t);
}
+
+/***********************************************************************
+ * Low level output functions.
+ * They implement line breaking using a short static buffer.
+ ***********************************************************************/
+
/*
- * If something was printed on the current output line, end it.
- * Not to be called right after html_indent().
+ * Buffer one HTML output byte.
+ * If the buffer is full, flush and deactivate it and start a new line.
+ * If the buffer is inactive, print directly.
*/
static void
-html_endline(struct html *h)
+print_byte(struct html *h, char c)
{
- if (h->flags & HTML_NLDONE)
+ if ((h->flags & HTML_BUFFER) == 0) {
+ putchar(c);
+ h->col++;
+ return;
+ }
+
+ if (h->col + h->bufcol < sizeof(h->buf)) {
+ h->buf[h->bufcol++] = c;
return;
+ }
putchar('\n');
- h->flags |= HTML_NLDONE | HTML_NOSPACE;
+ h->col = 0;
+ print_indent(h);
+ putchar(' ');
+ putchar(' ');
+ fwrite(h->buf, h->bufcol, 1, stdout);
+ putchar(c);
+ h->col = (h->indent + 1) * 2 + h->bufcol + 1;
+ h->bufcol = 0;
+ h->flags &= ~HTML_BUFFER;
+}
+
+/*
+ * If something was printed on the current output line, end it.
+ * Not to be called right after print_indent().
+ */
+void
+print_endline(struct html *h)
+{
+ if (h->col == 0)
+ return;
+
+ if (h->bufcol) {
+ putchar(' ');
+ fwrite(h->buf, h->bufcol, 1, stdout);
+ h->bufcol = 0;
+ }
+ putchar('\n');
+ h->col = 0;
+ h->flags |= HTML_NOSPACE;
+ h->flags &= ~HTML_BUFFER;
+}
+
+/*
+ * Flush the HTML output buffer.
+ * If it is inactive, activate it.
+ */
+static void
+print_endword(struct html *h)
+{
+ if (h->noindent) {
+ print_byte(h, ' ');
+ return;
+ }
+
+ if ((h->flags & HTML_BUFFER) == 0) {
+ h->col++;
+ h->flags |= HTML_BUFFER;
+ } else if (h->bufcol) {
+ putchar(' ');
+ fwrite(h->buf, h->bufcol, 1, stdout);
+ h->col += h->bufcol + 1;
+ }
+ h->bufcol = 0;
}
/*
* but do not use print_otag() for producing it.
*/
static void
-html_indent(struct html *h)
+print_indent(struct html *h)
{
- int i;
+ size_t i;
- if ((h->flags & HTML_NLDONE) == 0)
+ if (h->col)
return;
- if (h->noindent == 0)
- for (i = 0; i < h->indent * 2; i++)
+ if (h->noindent == 0) {
+ h->col = h->indent * 2;
+ for (i = 0; i < h->col; i++)
putchar(' ');
- h->flags &= ~(HTML_NLDONE | HTML_NOSPACE);
+ }
+ h->flags &= ~HTML_NOSPACE;
+}
+
+/*
+ * Print or buffer some characters
+ * depending on the current HTML output buffer state.
+ */
+static void
+print_word(struct html *h, const char *cp)
+{
+ while (*cp != '\0')
+ print_byte(h, *cp++);
}
/*
static void
a2width(const char *p, struct roffsu *su)
{
- if (a2roffsu(p, su, SCALE_MAX) < 2) {
+ const char *end;
+
+ end = a2roffsu(p, su, SCALE_MAX);
+ if (end == NULL || *end != '\0') {
su->unit = SCALE_EN;
su->scale = html_strlen(p);
} else if (su->scale < 0.0)
git.ckatri.com / mandoc.git / blobdiff