[project @ 2003-12-27 20:15:22 by bursa]

Use charset from Content-Type header.

svn path=/import/netsurf/; revision=460

content/content.c

@@ -57,7 +57,7 @@ static const struct mime_entry mime_map[] = {
 
 /** An entry in handler_map. */
 struct handler_entry {
-	void (*create)(struct content *c);
+	void (*create)(struct content *c, const char *params[]);
 	void (*process_data)(struct content *c, char *data, unsigned long size);
 	int (*convert)(struct content *c, unsigned int width, unsigned int height);
 	void (*revive)(struct content *c, unsigned int width, unsigned int height);
@@ -169,9 +169,15 @@ struct content * content_create(char *url)
  * status is changed to CONTENT_STATUS_LOADING. CONTENT_MSG_LOADING is sent to
  * all users. The create function for the type is called to initialise the type
  * specific parts of the content structure.
+ *
+ * \param c content structure
+ * \param type content_type to initialise to
+ * \param mime_type MIME-type string for this content
+ * \param params array of strings, ordered attribute, value, attribute, ..., 0
  */
 
-void content_set_type(struct content *c, content_type type, char* mime_type)
+void content_set_type(struct content *c, content_type type, char* mime_type,
+		const char *params[])
 {
 	assert(c != 0);
 	assert(c->status == CONTENT_STATUS_TYPE_UNKNOWN);
@@ -180,7 +186,7 @@ void content_set_type(struct content *c, content_type type, char* mime_type)
 	c->type = type;
 	c->mime_type = xstrdup(mime_type);
 	c->status = CONTENT_STATUS_LOADING;
-	handler_map[type].create(c);
+	handler_map[type].create(c, params);
 	content_broadcast(c, CONTENT_MSG_LOADING, 0);
 	/* c may be destroyed at this point as a result of
 	 * CONTENT_MSG_LOADING, so must not be accessed */

content/content.h

@@ -139,7 +139,8 @@ struct browser_window;
 
 content_type content_lookup(const char *mime_type);
 struct content * content_create(char *url);
-void content_set_type(struct content *c, content_type type, char *mime_type);
+void content_set_type(struct content *c, content_type type, char *mime_type,
+		const char *params[]);
 void content_process_data(struct content *c, char *data, unsigned long size);
 void content_convert(struct content *c, unsigned long width, unsigned long height);
 void content_revive(struct content *c, unsigned long width, unsigned long height);

content/fetchcache.c

@@ -15,6 +15,8 @@
 
 #include <assert.h>
 #include <string.h>
+#include <sys/types.h>
+#include <regex.h>
 #include "netsurf/content/cache.h"
 #include "netsurf/content/content.h"
 #include "netsurf/content/fetchcache.h"
@@ -23,7 +25,9 @@
 #include "netsurf/utils/utils.h"
 
 
+static regex_t re_content_type;
 static void fetchcache_callback(fetch_msg msg, void *p, char *data, unsigned long size);
+static char *fetchcache_parse_type(char *s, char **params[]);
 
 
 /**
@@ -95,20 +99,21 @@ void fetchcache_callback(fetch_msg msg, void *p, char *data, unsigned long size)
 {
 	struct content *c = p;
 	content_type type;
-	char *mime_type;
+	char *mime_type, *url;
-	char *semic;
+	char **params;
-	char *url;
+	unsigned int i;
 
 	switch (msg) {
 		case FETCH_TYPE:
 			c->total_size = size;
-			mime_type = xstrdup(data);
+			mime_type = fetchcache_parse_type(data, &params);
-			if ((semic = strchr(mime_type, ';')) != 0)
-				*semic = 0;	/* remove "; charset=..." */
 			type = content_lookup(mime_type);
 			LOG(("FETCH_TYPE, type %u", type));
-			content_set_type(c, type, mime_type);
+			content_set_type(c, type, mime_type, params);
 			free(mime_type);
+			for (i = 0; params[i]; i++)
+				free(params[i]);
+			free(params);
 			break;
 
 		case FETCH_DATA:
@@ -172,6 +177,62 @@ void fetchcache_callback(fetch_msg msg, void *p, char *data, unsigned long size)
 }
 
 
+/**
+ * Initialise the fetchcache module.
+ */
+
+void fetchcache_init(void)
+{
+	regcomp_wrapper(&re_content_type,
+			"^([-0-9a-zA-Z_.]+/[-0-9a-zA-Z_.]+)[ \t]*"
+			"(;[ \t]*([-0-9a-zA-Z_.]+)="
+			"([-0-9a-zA-Z_.]+|\"([^\"]|[\\].)*\")[ \t]*)*$",
+			REG_EXTENDED);
+}
+
+
+/**
+ * Parse a Content-Type header.
+ *
+ * \param s a Content-Type header
+ * \param params updated to point to an array of strings, ordered attribute,
+ *   value, attribute, ..., 0
+ * \return a new string containing the MIME-type
+ */
+
+#define MAX_ATTRS 10
+
+char *fetchcache_parse_type(char *s, char **params[])
+{
+	char *type;
+	unsigned int i;
+	int r;
+	regmatch_t pmatch[2 + MAX_ATTRS * 3];
+	*params = xcalloc(MAX_ATTRS * 2 + 2, sizeof (*params)[0]);
+
+	r = regexec(&re_content_type, s, 2 + MAX_ATTRS * 3, pmatch, 0);
+	if (r) {
+		LOG(("failed to parse content-type '%s'", s));
+		return xstrdup(s);
+	}
+
+	type = strndup(s + pmatch[1].rm_so, pmatch[1].rm_eo - pmatch[1].rm_so);
+	assert(type);
+
+	/* parameters */
+	for (i = 0; i != MAX_ATTRS && pmatch[2 + 3 * i].rm_so != -1; i++) {
+		(*params)[2 * i] = strndup(s + pmatch[2 + 3 * i + 1].rm_so,
+				pmatch[2 + 3 * i + 1].rm_eo - pmatch[2 + 3 * i + 1].rm_so);
+		(*params)[2 * i + 1] = strndup(s + pmatch[2 + 3 * i + 2].rm_so,
+				pmatch[2 + 3 * i + 2].rm_eo - pmatch[2 + 3 * i + 2].rm_so);
+		assert((*params)[2 * i] && (*params)[2 * i + 1]);
+	}
+	(*params)[2 * i] = 0;
+
+	return type;
+}
+
+
 #ifdef TEST
 
 #include <unistd.h>

content/fetchcache.h

@@ -26,5 +26,6 @@ struct content * fetchcache(const char *url, char *referer,
 		void *p1, void *p2, unsigned long width, unsigned long height,
 		bool only_2xx, char *post_urlenc,
 		struct form_successful_control *post_multipart, bool cookies);
+void fetchcache_init(void);
 
 #endif

content/other.c

@@ -17,7 +17,7 @@
 #include "netsurf/utils/utils.h"
 
 
-void other_create(struct content *c)
+void other_create(struct content *c, char *params[])
 {
 	c->data.other.data = xcalloc(0, 1);
 	c->data.other.length = 0;

content/other.h

@@ -24,7 +24,7 @@ struct content_other_data {
 	unsigned long length;	/**< Current length of stored data. */
 };
 
-void other_create(struct content *c);
+void other_create(struct content *c, char *params[]);
 void other_process_data(struct content *c, char *data, unsigned long size);
 int other_convert(struct content *c, unsigned int width, unsigned int height);
 void other_revive(struct content *c, unsigned int width, unsigned int height);

css/css.c

@@ -91,7 +91,7 @@ const struct css_style css_blank_style = {
 
 
 
-void css_create(struct content *c)
+void css_create(struct content *c, const char *params[])
 {
 	unsigned int i;
 	LOG(("content %p", c));

css/css.h

@@ -5,6 +5,21 @@
  * Copyright 2003 James Bursa <bursa@users.sourceforge.net>
  */
 
+/** \file
+ * CSS handling (interface).
+ *
+ * This module aims to implement CSS 2.1.
+ *
+ * CSS stylesheets are held in a struct ::content with type CONTENT_CSS.
+ * Creation and parsing should be carried out via the content_* functions.
+ *
+ * Styles are stored in a struct ::css_style, which can be retrieved from a
+ * content using css_get_style().
+ *
+ * css_parse_property_list() constructs a struct ::css_style from a CSS
+ * property list, as found in HTML style attributes.
+ */
+
 #ifndef _NETSURF_CSS_CSS_H_
 #define _NETSURF_CSS_CSS_H_
 
@@ -12,7 +27,7 @@
 #include "libxml/HTMLparser.h"
 #include "css_enum.h"
 
-/**
+/*
  * structures and typedefs
  */
 
@@ -21,6 +36,7 @@ typedef unsigned long colour;  /* 0xbbggrr */
 #define CSS_COLOR_INHERIT 0x2000000
 #define CSS_COLOR_NONE 0x3000000
 
+/** Representation of a CSS 2 length. */
 struct css_length {
 	float value;
 	css_unit unit;
@@ -36,6 +52,7 @@ typedef enum {
 	CSS_TEXT_DECORATION_UNKNOWN = 0x1000
 } css_text_decoration;
 
+/** Representation of a complete CSS 2 style. */
 struct css_style {
 	colour background_color;
 	css_clear clear;
@@ -98,19 +115,14 @@ struct css_style {
 
 struct css_stylesheet;
 
-struct css_selector {
+/** Data specific to CONTENT_CSS. */
-	const char *element;
-	char *class;
-	char *id;
-};
-
 struct content_css_data {
-	struct css_stylesheet *css;
+	struct css_stylesheet *css;	/**< Opaque stylesheet data. */
-	unsigned int import_count;
+	unsigned int import_count;	/**< Number of entries in import_url. */
-	char **import_url;
+	char **import_url;		/**< Imported stylesheet urls. */
-	struct content **import_content;
+	struct content **import_content; /**< Imported stylesheet contents. */
-	char *data;
+	char *data;			/**< Source data. */
-	unsigned int length;
+	unsigned int length;		/**< Current length of data. */
 };
 
 
@@ -191,13 +203,13 @@ struct parse_params {
 
 #endif
 
-/**
+/*
  * interface
  */
 
 struct content;
 
-void css_create(struct content *c);
+void css_create(struct content *c, const char *params[]);
 void css_process_data(struct content *c, char *data, unsigned long size);
 int css_convert(struct content *c, unsigned int width, unsigned int height);
 void css_revive(struct content *c, unsigned int width, unsigned int height);
@@ -218,7 +230,7 @@ void css_add_declarations(struct css_style *style, struct css_node *declaration)
 unsigned int css_hash(const char *s);
 
 void css_parser_Trace(FILE *TraceFILE, char *zTracePrompt);
-void *css_parser_Alloc(void *(*mallocProc)(int));
+void *css_parser_alloc(void *(*mallocProc)(size_t));
 void css_parser_Free(void *p, void (*freeProc)(void*));
 void css_parser_(void *yyp, int yymajor, char* yyminor,
 		struct parse_params *param);

debug/netsurfd.c

@@ -39,6 +39,7 @@ int main(int argc, char *argv[])
 
 	fetch_init();
 	cache_init();
+	fetchcache_init();
 
 	while (1) {
 		puts("=== URL:");
@@ -46,7 +47,7 @@ int main(int argc, char *argv[])
 			return 0;
 		url[strlen(url) - 1] = 0;
 		destroyed = 0;
-		c = fetchcache(url, 0, callback, 0, 0, 100, 1000, false, 0, 0);
+		c = fetchcache(url, 0, callback, 0, 0, 100, 1000, false, 0, 0, true);
 		if (c) {
 			done = c->status == CONTENT_STATUS_DONE;
 			while (!done)

desktop/netsurf.c

@@ -52,6 +52,7 @@ void netsurf_init(int argc, char** argv)
   gui_init(argc, argv);
   fetch_init();
   cache_init();
+  fetchcache_init();
   nspng_init();
   nsgif_init();
 }

render/html.c

@@ -29,9 +29,21 @@ static void html_object_callback(content_msg msg, struct content *object,
 		void *p1, void *p2, const char *error);
 
 
-void html_create(struct content *c)
+void html_create(struct content *c, const char *params[])
 {
-	c->data.html.parser = htmlCreatePushParserCtxt(0, 0, "", 0, 0, XML_CHAR_ENCODING_8859_1);
+	unsigned int i;
+	xmlCharEncoding encoding = XML_CHAR_ENCODING_8859_1;
+
+	for (i = 0; params[i]; i += 2) {
+		if (strcasecmp(params[i], "charset") == 0) {
+			encoding = xmlParseCharEncoding(params[i + 1]);
+			if (encoding == XML_CHAR_ENCODING_ERROR)
+				encoding = XML_CHAR_ENCODING_8859_1;
+			break;
+		}
+	}
+
+	c->data.html.parser = htmlCreatePushParserCtxt(0, 0, "", 0, 0, encoding);
 	c->data.html.layout = NULL;
 	c->data.html.style = NULL;
 	c->data.html.fonts = NULL;
@@ -341,9 +353,10 @@ void html_find_stylesheets(struct content *c, xmlNode *head)
 			/* create stylesheet */
 			LOG(("style element"));
 			if (c->data.html.stylesheet_content[1] == 0) {
+				const char *params[] = { 0 };
 				c->data.html.stylesheet_content[1] =
 						content_create(c->data.html.base_url);
-				content_set_type(c->data.html.stylesheet_content[1], CONTENT_CSS, "text/css");
+				content_set_type(c->data.html.stylesheet_content[1], CONTENT_CSS, "text/css", params);
 			}
 
 			/* can't just use xmlNodeGetContent(node), because that won't give

render/html.h

@@ -51,7 +51,7 @@ struct content_html_data {
 	} *object;
 };
 
-void html_create(struct content *c);
+void html_create(struct content *c, const char *params[]);
 void html_process_data(struct content *c, char *data, unsigned long size);
 int html_convert(struct content *c, unsigned int width, unsigned int height);
 void html_revive(struct content *c, unsigned int width, unsigned int height);

render/textplain.c

@@ -18,9 +18,9 @@ static const char header[] = "<html><body><pre>";
 static const char footer[] = "</pre></body></html>";
 
 
-void textplain_create(struct content *c)
+void textplain_create(struct content *c, const char *params[])
 {
-	html_create(c);
+	html_create(c, params);
 	htmlParseChunk(c->data.html.parser, header, sizeof(header) - 1, 0);
 }
 

render/textplain.h

@@ -10,7 +10,7 @@
 
 #include "netsurf/content/content.h"
 
-void textplain_create(struct content *c);
+void textplain_create(struct content *c, const char *params[]);
 void textplain_process_data(struct content *c, char *data, unsigned long size);
 int textplain_convert(struct content *c, unsigned int width, unsigned int height);
 void textplain_revive(struct content *c, unsigned int width, unsigned int height);

riscos/draw.c

@@ -14,7 +14,7 @@
 #include "netsurf/utils/log.h"
 #include "oslib/drawfile.h"
 
-void draw_create(struct content *c)
+void draw_create(struct content *c, const char *params[])
 {
 	c->data.draw.data = xcalloc(0, 1);
 	c->data.draw.length = 0;

riscos/draw.h

@@ -16,7 +16,7 @@ struct content_draw_data {
 };
 
 void draw_init(void);
-void draw_create(struct content *c);
+void draw_create(struct content *c, const char *params[]);
 void draw_process_data(struct content *c, char *data, unsigned long size);
 int draw_convert(struct content *c, unsigned int width, unsigned int height);
 void draw_revive(struct content *c, unsigned int width, unsigned int height);

riscos/gif.c

@@ -26,7 +26,7 @@ void nsgif_init(void)
 {
 }
 
-void nsgif_create(struct content*c)
+void nsgif_create(struct content *c, const char *params[])
 {
   c->data.gif.sprite_area = 0;
   c->data.gif.data = xcalloc(0, 1);

riscos/gif.h

@@ -21,7 +21,7 @@ struct content_gif_data {
 };
 
 void nsgif_init(void);
-void nsgif_create(struct content *c);
+void nsgif_create(struct content *c, const char *params[]);
 void nsgif_process_data(struct content *c, char *data, unsigned long size);
 int nsgif_convert(struct content *c, unsigned int width, unsigned int height);
 void nsgif_revive(struct content *c, unsigned int width, unsigned int height);

riscos/jpeg.c

@@ -19,7 +19,7 @@
 #include "oslib/jpeg.h"
 
 
-void jpeg_create(struct content *c)
+void jpeg_create(struct content *c, const char *params[])
 {
 	c->data.jpeg.data = xcalloc(0, 1);
 	c->data.jpeg.length = 0;

riscos/jpeg.h

@@ -15,7 +15,7 @@ struct content_jpeg_data {
 	unsigned long length;
 };
 
-void jpeg_create(struct content *c);
+void jpeg_create(struct content *c, const char *params[]);
 void jpeg_process_data(struct content *c, char *data, unsigned long size);
 int jpeg_convert(struct content *c, unsigned int width, unsigned int height);
 void jpeg_revive(struct content *c, unsigned int width, unsigned int height);

riscos/png.c

@@ -64,7 +64,7 @@ void nspng_init(void)
 }
 
 
-void nspng_create(struct content *c)
+void nspng_create(struct content *c, const char *params[])
 {
 #ifndef NO_IFC
 	if (imagefileconvert) {

riscos/png.h

@@ -27,7 +27,7 @@ struct content_png_data {
 };
 
 void nspng_init(void);
-void nspng_create(struct content *c);
+void nspng_create(struct content *c, const char *params[]);
 void nspng_process_data(struct content *c, char *data, unsigned long size);
 int nspng_convert(struct content *c, unsigned int width, unsigned int height);
 void nspng_revive(struct content *c, unsigned int width, unsigned int height);

riscos/sprite.c

@@ -15,7 +15,7 @@
 #include "oslib/colourtrans.h"
 #include "oslib/osspriteop.h"
 
-void sprite_create(struct content *c)
+void sprite_create(struct content *c, const char *params[])
 {
 	c->data.sprite.data = xcalloc(4, 1);
 	c->data.sprite.length = 4;

riscos/sprite.h

@@ -18,7 +18,7 @@ struct content_sprite_data {
 };
 
 void sprite_init(void);
-void sprite_create(struct content *c);
+void sprite_create(struct content *c, const char *params[]);
 void sprite_process_data(struct content *c, char *data, unsigned long size);
 int sprite_convert(struct content *c, unsigned int width, unsigned int height);
 void sprite_revive(struct content *c, unsigned int width, unsigned int height);

utils/utils.c

@@ -13,6 +13,8 @@
 #include <string.h>
 #include <sys/stat.h>
 #include <uri.h>
+#include <sys/types.h>
+#include <regex.h>
 #include "libxml/encoding.h"
 #include "libxml/uri.h"
 #include "netsurf/utils/log.h"
@@ -256,3 +258,23 @@ bool is_dir(const char *path)
 
 	return S_ISDIR(s.st_mode) ? true : false;
 }
+
+
+/**
+ * Compile a regular expression, handling errors.
+ *
+ * Parameters as for regcomp(), see man regex.
+ */
+
+void regcomp_wrapper(regex_t *preg, const char *regex, int cflags)
+{
+	char errbuf[200];
+	int r;
+	r = regcomp(preg, regex, cflags);
+	if (r) {
+		regerror(r, preg, errbuf, sizeof errbuf);
+		fprintf(stderr, "Failed to compile regexp '%s'\n", regex);
+		die(errbuf);
+	}
+}
+

utils/utils.h

@@ -10,6 +10,8 @@
 
 #include <stdbool.h>
 #include <stdlib.h>
+#include <sys/types.h>
+#include <regex.h>
 #include "libxml/encoding.h"
 
 void die(const char * const error);
@@ -27,5 +29,6 @@ char *squash_tolat1(xmlChar *s);
 char *url_join(char *rel_url, char *base_url);
 char *get_host_from_url(char* url);
 bool is_dir(const char *path);
+void regcomp_wrapper(regex_t *preg, const char *regex, int cflags);
 
 #endif