html: Split html parsing cache and state into html_context.

html_document is now a simple client of html_context.

4 files changed, 173 insertions, 125 deletions

diff --git a/source/html/css-apply.c b/source/html/css-apply.c
index 67814440..bbfc7ffa 100644
--- a/source/html/css-apply.c
+++ b/source/html/css-apply.c
@@ -990,7 +990,7 @@ default_computed_style(struct computed_style *style)
 }
 
 void
-compute_style(html_document *doc, struct computed_style *style, struct style *node)
+compute_style(fz_context *ctx, html_context *htx, struct computed_style *style, struct style *node)
 {
 	struct value *value;
 
@@ -1087,7 +1087,7 @@ compute_style(html_document *doc, struct computed_style *style, struct style *no
 		const char *font_variant = get_style_property_string(node, "font-variant", "normal");
 		const char *font_style = get_style_property_string(node, "font-style", "normal");
 		const char *font_weight = get_style_property_string(node, "font-weight", "normal");
-		style->font = html_load_font(doc, font_family, font_variant, font_style, font_weight);
+		style->font = html_load_font(ctx, htx, font_family, font_variant, font_style, font_weight);
 	}
 }
 
diff --git a/source/html/font.c b/source/html/font.c
index 9c408367..51654ae9 100644
--- a/source/html/font.c
+++ b/source/html/font.c
@@ -9,10 +9,9 @@ static const char *font_names[16] = {
 };
 
 fz_font *
-html_load_font(html_document *doc,
+html_load_font(fz_context *ctx, html_context *htx,
 	const char *family, const char *variant, const char *style, const char *weight)
 {
-	fz_context *ctx = doc->ctx;
 	unsigned char *data;
 	unsigned int size;
 
@@ -22,15 +21,15 @@ html_load_font(html_document *doc,
 	int is_italic = !strcmp(style, "italic") || !strcmp(style, "oblique");
 
 	int idx = is_mono * 8 + is_sans * 4 + is_bold * 2 + is_italic;
-	if (!doc->fonts[idx])
+	if (!htx->fonts[idx])
 	{
 		data = pdf_lookup_builtin_font(font_names[idx], &size);
 		if (!data) {
 		printf("data=%p idx=%d s=%s\n", data, idx, font_names[idx]);
 			abort();
 		}
-		doc->fonts[idx] = fz_new_font_from_memory(ctx, font_names[idx], data, size, 0, 1);
+		htx->fonts[idx] = fz_new_font_from_memory(ctx, font_names[idx], data, size, 0, 1);
 	}
 
-	return doc->fonts[idx];
+	return htx->fonts[idx];
 }
diff --git a/source/html/handler.c b/source/html/handler.c
index c89e82ae..626ed443 100644
--- a/source/html/handler.c
+++ b/source/html/handler.c
@@ -2,125 +2,141 @@
 
 #define DEFW (450)
 #define DEFH (600)
+#define DEFEM (12)
 
-void
-html_close_document(html_document *doc)
+typedef struct html_document_s html_document;
+
+struct html_document_s
+{
+	fz_document super;
+	fz_context *ctx;
+	html_context htx;
+	struct box *box;
+};
+
+static void
+htdoc_close_document(html_document *doc)
 {
 	fz_context *ctx = doc->ctx;
+	html_fini(ctx, &doc->htx);
 	fz_free(ctx, doc);
 }
 
-int
-html_count_pages(html_document *doc)
+static int
+htdoc_count_pages(html_document *doc)
 {
 	int count;
 
-	if (!doc->box) html_layout_document(doc, DEFW, DEFH, 12);
+	// TODO: reflow
 
-	count = ceilf(doc->box->h / doc->page_h);
-printf("count pages! %g / %g = %d\n", doc->box->h, doc->page_h, count);
+	count = ceilf(doc->box->h / doc->htx.page_h);
+printf("count pages! %g / %g = %d\n", doc->box->h, doc->htx.page_h, count);
 	return count;
 }
 
-html_page *
-html_load_page(html_document *doc, int number)
+static void *
+htdoc_load_page(html_document *doc, int number)
 {
 printf("load page %d\n", number);
-	if (!doc->box) html_layout_document(doc, DEFW, DEFH, 12);
+	// TODO: reflow
 	return (void*)((intptr_t)number + 1);
 }
 
-void
-html_free_page(html_document *doc, html_page *page)
+static void
+htdoc_free_page(html_document *doc, void *page)
+{
+}
+
+static void
+htdoc_layout(html_document *doc, float w, float h, float em)
 {
+	html_layout(doc->ctx, &doc->htx, doc->box, w, h, em);
 }
 
-fz_rect *
-html_bound_page(html_document *doc, html_page *page, fz_rect *bbox)
+static fz_rect *
+htdoc_bound_page(html_document *doc, void *page, fz_rect *bbox)
 {
-	if (!doc->box) html_layout_document(doc, DEFW, DEFH, 12);
+	// TODO: reflow
 	printf("html: bound page\n");
 	bbox->x0 = bbox->y0 = 0;
-	bbox->x1 = doc->page_w;
-	bbox->y1 = doc->page_h;
+	bbox->x1 = doc->htx.page_w;
+	bbox->y1 = doc->htx.page_h;
 	return bbox;
 }
 
-void
-html_run_page(html_document *doc, html_page *page, fz_device *dev, const fz_matrix *ctm, fz_cookie *cookie)
+static void
+htdoc_run_page(html_document *doc, void *page, fz_device *dev, const fz_matrix *ctm, fz_cookie *cookie)
 {
 	int n = ((intptr_t)page) - 1;
 	printf("html: run page %d\n", n);
-	html_run_box(doc->ctx, doc->box, n * doc->page_h, (n+1) * doc->page_h, dev, ctm);
+	html_draw(doc->ctx, &doc->htx, doc->box, n * doc->htx.page_h, (n+1) * doc->htx.page_h, dev, ctm);
 }
 
-html_document *
-html_open_document_with_stream(fz_context *ctx, fz_stream *file)
+
+static html_document *
+htdoc_open_document_with_stream(fz_context *ctx, fz_stream *file)
 {
 	html_document *doc;
+	fz_archive *zip;
 	fz_buffer *buf;
-	fz_xml *xml;
 
-	buf = fz_read_all(file, 0);
-	fz_write_buffer_byte(ctx, buf, 0);
-
-printf("html: parsing XHTML.\n");
-	xml = fz_parse_xml(ctx, buf->data, buf->len, 1);
-	fz_drop_buffer(ctx, buf);
+	zip = fz_open_directory(ctx, ".");
 
 	doc = fz_malloc_struct(ctx, html_document);
 	doc->ctx = ctx;
-	doc->dirname = NULL;
+	html_init(ctx, &doc->htx, zip);
 
-	doc->super.close = (void*)html_close_document;
-	doc->super.layout = (void*)html_layout_document;
-	doc->super.count_pages = (void*)html_count_pages;
-	doc->super.load_page = (void*)html_load_page;
-	doc->super.bound_page = (void*)html_bound_page;
-	doc->super.run_page_contents = (void*)html_run_page;
-	doc->super.free_page = (void*)html_free_page;
+	doc->super.close = (void*)htdoc_close_document;
+	doc->super.layout = (void*)htdoc_layout;
+	doc->super.count_pages = (void*)htdoc_count_pages;
+	doc->super.load_page = (void*)htdoc_load_page;
+	doc->super.bound_page = (void*)htdoc_bound_page;
+	doc->super.run_page_contents = (void*)htdoc_run_page;
+	doc->super.free_page = (void*)htdoc_free_page;
 
-	doc->xml = xml;
-	doc->box = NULL;
+	buf = fz_read_all(file, 0);
+	fz_write_buffer_byte(ctx, buf, 0);
+	doc->box = html_generate(ctx, &doc->htx, ".", buf);
+	fz_drop_buffer(ctx, buf);
 
 	return doc;
 }
 
-html_document *
-html_open_document(fz_context *ctx, const char *filename)
+static html_document *
+htdoc_open_document(fz_context *ctx, const char *filename)
 {
-	fz_stream *file;
+	char dirname[2048];
+	fz_archive *zip;
+	fz_buffer *buf;
 	html_document *doc;
-	char *s;
 
-	file = fz_open_file(ctx, filename);
-	if (!file)
-		fz_throw(ctx, FZ_ERROR_GENERIC, "cannot open file '%s': %s", filename, strerror(errno));
+	fz_dirname(dirname, filename, sizeof dirname);
+	zip = fz_open_directory(ctx, dirname);
 
-	fz_try(ctx)
-	{
-		doc = html_open_document_with_stream(ctx, file);
-	}
-	fz_always(ctx)
-	{
-		fz_close(file);
-	}
-	fz_catch(ctx)
-	{
-		fz_rethrow(ctx);
-	}
+	doc = fz_malloc_struct(ctx, html_document);
+	doc->ctx = ctx;
+	html_init(ctx, &doc->htx, zip);
+
+	doc->super.close = (void*)htdoc_close_document;
+	doc->super.layout = (void*)htdoc_layout;
+	doc->super.count_pages = (void*)htdoc_count_pages;
+	doc->super.load_page = (void*)htdoc_load_page;
+	doc->super.bound_page = (void*)htdoc_bound_page;
+	doc->super.run_page_contents = (void*)htdoc_run_page;
+	doc->super.free_page = (void*)htdoc_free_page;
+
+	buf = fz_read_file(ctx, filename);
+	fz_write_buffer_byte(ctx, buf, 0);
+	doc->box = html_generate(ctx, &doc->htx, ".", buf);
+	fz_drop_buffer(ctx, buf);
 
-	doc->dirname = fz_strdup(ctx, filename);
-	s = strrchr(doc->dirname, '/');
-	if (!s) s = strrchr(doc->dirname, '\\');
-	if (s) s[1] = 0;
-	else doc->dirname[0] = 0;
+	htdoc_layout(doc, DEFW, DEFH, DEFEM);
 
 	return doc;
 }
 
 static int
-html_recognize(fz_context *doc, const char *magic)
+htdoc_recognize(fz_context *doc, const char *magic)
 {
 	char *ext = strrchr(magic, '.');
 
@@ -137,7 +153,7 @@ html_recognize(fz_context *doc, const char *magic)
 
 fz_document_handler html_document_handler =
 {
-	(fz_document_recognize_fn *)&html_recognize,
-	(fz_document_open_fn *)&html_open_document,
-	(fz_document_open_with_stream_fn *)&html_open_document_with_stream
+	(fz_document_recognize_fn *)&htdoc_recognize,
+	(fz_document_open_fn *)&htdoc_open_document,
+	(fz_document_open_with_stream_fn *)&htdoc_open_document_with_stream
 };
diff --git a/source/html/layout.c b/source/html/layout.c
index 822e2937..f07108b5 100644
--- a/source/html/layout.c
+++ b/source/html/layout.c
@@ -100,33 +100,30 @@ static void generate_text(fz_context *ctx, struct box *box, const char *text)
 	}
 }
 
-static void generate_image(html_document *doc, struct box *box, const char *src)
+static void generate_image(fz_context *ctx, html_context *htx, const char *base_uri, struct box *box, const char *src)
 {
-	fz_context *ctx = doc->ctx;
 	fz_image *img;
 	fz_buffer *buf;
-	char filename[2048];
+	char path[2048];
 
 	struct box *flow = box;
 	while (flow->type != BOX_FLOW)
 		flow = flow->up;
 
-	fz_strlcpy(filename, doc->dirname, sizeof filename);
-	fz_strlcat(filename, src, sizeof filename);
+	fz_strlcpy(path, base_uri, sizeof path);
+	fz_strlcat(path, "/", sizeof path);
+	fz_strlcat(path, src, sizeof path);
+	fz_cleanname(path);
 
-	buf = fz_read_file(ctx, filename);
+	buf = fz_read_archive_entry(ctx, htx->zip, path);
 	img = fz_new_image_from_buffer(ctx, buf);
 	fz_drop_buffer(ctx, buf);
 
 	add_flow_image(ctx, flow, &box->style, img);
 }
 
-struct box *new_box(fz_context *ctx, fz_xml *node)
+static void init_box(fz_context *ctx, struct box *box, fz_xml *node)
 {
-	struct box *box;
-
-	box = fz_malloc_struct(ctx, struct box);
-
 	box->type = BOX_BLOCK;
 	box->x = box->y = 0;
 	box->w = box->h = 0;
@@ -142,11 +139,16 @@ struct box *new_box(fz_context *ctx, fz_xml *node)
 	box->flow_tail = &box->flow_head;
 
 	default_computed_style(&box->style);
+}
 
+static struct box *new_box(fz_context *ctx, fz_xml *node)
+{
+	struct box *box = fz_malloc_struct(ctx, struct box);
+	init_box(ctx, box, node);
 	return box;
 }
 
-void insert_box(fz_context *ctx, struct box *box, int type, struct box *top)
+static void insert_box(fz_context *ctx, struct box *box, int type, struct box *top)
 {
 	box->type = type;
 
@@ -234,9 +236,9 @@ static void insert_inline_box(fz_context *ctx, struct box *box, struct box *top)
 	}
 }
 
-static void generate_boxes(html_document *doc, fz_xml *node, struct box *top, struct rule *rule, struct style *up_style)
+static void generate_boxes(fz_context *ctx, html_context *htx, const char *base_uri,
+	fz_xml *node, struct box *top, struct rule *rule, struct style *up_style)
 {
-	fz_context *ctx = doc->ctx;
 	struct style style;
 	struct box *box;
 	const char *tag;
@@ -257,7 +259,7 @@ static void generate_boxes(html_document *doc, fz_xml *node, struct box *top, st
 			if (!strcmp(tag, "br"))
 			{
 				box = new_box(ctx, node);
-				compute_style(doc, &box->style, &style);
+				compute_style(ctx, htx, &box->style, &style);
 				top = insert_break_box(ctx, box, top);
 			}
 
@@ -267,16 +269,16 @@ static void generate_boxes(html_document *doc, fz_xml *node, struct box *top, st
 				if (src)
 				{
 					box = new_box(ctx, node);
-					compute_style(doc, &box->style, &style);
+					compute_style(ctx, htx, &box->style, &style);
 					insert_inline_box(ctx, box, top);
-					generate_image(doc, box, src);
+					generate_image(ctx, htx, base_uri, box, src);
 				}
 			}
 
 			else if (display != DIS_NONE)
 			{
 				box = new_box(ctx, node);
-				compute_style(doc, &box->style, &style);
+				compute_style(ctx, htx, &box->style, &style);
 
 				if (display == DIS_BLOCK)
 				{
@@ -297,7 +299,7 @@ static void generate_boxes(html_document *doc, fz_xml *node, struct box *top, st
 				}
 
 				if (fz_xml_down(node))
-					generate_boxes(doc, fz_xml_down(node), box, rule, &style);
+					generate_boxes(ctx, htx, base_uri, fz_xml_down(node), box, rule, &style);
 
 				// TODO: remove empty flow boxes
 			}
@@ -804,7 +806,7 @@ draw_block_box(fz_context *ctx, struct box *box, float page_top, float page_bot,
 }
 
 void
-html_run_box(fz_context *ctx, struct box *box, float page_top, float page_bot, fz_device *dev, const fz_matrix *inctm)
+html_draw(fz_context *ctx, html_context *htx, struct box *box, float page_top, float page_bot, fz_device *dev, const fz_matrix *inctm)
 {
 	fz_matrix ctm = *inctm;
 	fz_pre_translate(&ctm, 0, -page_top);
@@ -836,16 +838,15 @@ static char *concat_text(fz_context *ctx, fz_xml *root)
 	return s;
 }
 
-static struct rule *load_css(html_document *doc, struct rule *css, fz_xml *root)
+static struct rule *html_load_css(fz_context *ctx, html_context *htx, const char *base_uri, struct rule *css, fz_xml *root)
 {
-	fz_context *ctx = doc->ctx;
 	fz_xml *node;
-	char filename[2048];
+	fz_buffer *buf;
+	char path[2048];
 
 	for (node = root; node; node = fz_xml_next(node))
 	{
 		const char *tag = fz_xml_tag(node);
-#if 1
 		if (tag && !strcmp(tag, "link"))
 		{
 			char *rel = fz_xml_att(node, "rel");
@@ -855,13 +856,21 @@ static struct rule *load_css(html_document *doc, struct rule *css, fz_xml *root)
 				if ((type && !strcmp(type, "text/css")) || !type)
 				{
 					char *href = fz_xml_att(node, "href");
-					fz_strlcpy(filename, doc->dirname, sizeof filename);
-					fz_strlcat(filename, href, sizeof filename);
-					css = fz_parse_css_file(ctx, css, filename);
+					if (href)
+					{
+						fz_strlcpy(path, base_uri, sizeof path);
+						fz_strlcat(path, "/", sizeof path);
+						fz_strlcat(path, href, sizeof path);
+						fz_cleanname(path);
+
+						buf = fz_read_archive_entry(ctx, htx->zip, path);
+						fz_write_buffer_byte(ctx, buf, 0);
+						css = fz_parse_css(ctx, css, (char*)buf->data);
+						fz_drop_buffer(ctx, buf);
+					}
 				}
 			}
 		}
-#endif
 		if (tag && !strcmp(tag, "style"))
 		{
 			char *s = concat_text(ctx, node);
@@ -869,46 +878,70 @@ static struct rule *load_css(html_document *doc, struct rule *css, fz_xml *root)
 			fz_free(ctx, s);
 		}
 		if (fz_xml_down(node))
-			css = load_css(doc, css, fz_xml_down(node));
+			css = html_load_css(ctx, htx, base_uri, css, fz_xml_down(node));
 	}
 	return css;
 }
 
 void
-html_layout_document(html_document *doc, float page_w, float page_h, float em)
+html_layout(fz_context *ctx, html_context *htx, struct box *box, float w, float h, float em)
+{
+	struct box page_box;
+
+	htx->page_w = w;
+	htx->page_h = h;
+
+	printf("html: laying out text.\n");
+
+	init_box(ctx, &page_box, NULL);
+	page_box.w = w;
+	page_box.h = 0;
+
+	layout_block(ctx, box, &page_box, em, 0, h);
+
+	printf("html: finished.\n");
+}
+
+struct box *
+html_generate(fz_context *ctx, html_context *htx, const char *base_uri, fz_buffer *buf)
 {
-	struct rule *css = NULL;
-	struct box *root_box;
-	struct box *page_box;
+	fz_xml *xml;
+	struct rule *css;
+	struct box *box;
 	struct style style;
 
-	doc->page_w = page_w;
-	doc->page_h = page_h;
+	printf("html: parsing XHTML.\n");
+	xml = fz_parse_xml(ctx, buf->data, buf->len, 1);
 
-printf("html: parsing style sheets.\n");
-	css = fz_parse_css(doc->ctx, NULL, default_css);
-	css = load_css(doc, css, doc->xml);
+	printf("html: parsing style sheets.\n");
+	css = fz_parse_css(ctx, NULL, default_css);
+	css = html_load_css(ctx, htx, base_uri, css, xml);
 
 	// print_rules(css);
 
+	printf("html: applying styles and generating boxes.\n");
+	box = new_box(ctx, NULL);
+
 	style.up = NULL;
 	style.count = 0;
 
-	root_box = new_box(doc->ctx, NULL);
-
-	page_box = new_box(doc->ctx, NULL);
-	page_box->w = page_w;
-	page_box->h = 0;
+	generate_boxes(ctx, htx, base_uri, xml, box, css, &style);
 
-	// TODO: split generate and layout
+	return box;
+}
 
-printf("html: applying styles and generating boxes.\n");
-	generate_boxes(doc, doc->xml, root_box, css, &style);
-printf("html: laying out text.\n");
-	layout_block(doc->ctx, root_box, page_box, em, 0, page_h);
-printf("html: finished.\n");
+void html_init(fz_context *ctx, html_context *html, fz_archive *zip)
+{
+	memset(html, 0, sizeof *html);
+	html->zip = zip;
+}
 
-	// print_box(doc->ctx, root_box, 0);
+void html_rebind(html_context *html, fz_context *ctx)
+{
+	fz_rebind_archive(html->zip, ctx);
+}
 
-	doc->box = root_box;
+void html_fini(fz_context *ctx, html_context *html)
+{
+	fz_close_archive(ctx, html->zip);
 }