[project @ 2005-07-23 20:43:37 by bursa]

Rewrite and simplify url_nice() to improve suggested filenames. Add option to keep extensions (no UI currently).

svn path=/import/netsurf/; revision=1814

6 files changed, 189 insertions, 121 deletions

diff --git a/riscos/download.c b/riscos/download.c
index 015ca6fe8..56612075b 100644
--- a/riscos/download.c
+++ b/riscos/download.c
@@ -32,6 +32,7 @@
 #include "oslib/wimpspriteop.h"
 #include "netsurf/content/fetch.h"
 #include "netsurf/desktop/gui.h"
+#include "netsurf/riscos/options.h"
 #include "netsurf/riscos/gui.h"
 #include "netsurf/riscos/query.h"
 #include "netsurf/riscos/wimp.h"
@@ -177,6 +178,7 @@ struct gui_download_window *gui_download_window_create(const char *url,
 	url_func_result res;
 	char *local_path;
 	utf8_convert_ret err;
+	size_t i;
 
 	dw = malloc(sizeof *dw);
 	if (!dw) {
@@ -242,8 +244,16 @@ struct gui_download_window *gui_download_window_create(const char *url,
 	download_template->icons[ICON_DOWNLOAD_ICON].data.indirected_sprite.id =
 			(osspriteop_id) dw->sprite_name;
 
-	if ((res = url_nice(url, &nice)) == URL_FUNC_OK) {
-		strcpy(dw->path, nice);
+	if ((res = url_nice(url, &nice, option_strip_extensions)) ==
+			URL_FUNC_OK) {
+		for (i = 0; nice[i]; i++) {
+			if (nice[i] == '.')
+				nice[i] = '/';
+			else if (nice[i] <= ' ' ||
+					strchr(":*#$&@^%\\", nice[i]))
+				nice[i] = '_';
+		}
+		strncpy(dw->path, nice, sizeof dw->path);
 		free(nice);
 	}
 	else
@@ -296,7 +306,8 @@ struct gui_download_window *gui_download_window_create(const char *url,
 
 	ro_gui_dialog_open(dw->window);
 
-	/* issue the warning now, so that it appears in front of the download window! */
+	/* issue the warning now, so that it appears in front of the download
+	 * window! */
 	if (space_warning)
 		warn_user("DownloadWarn", messages_get("NoDiscSpace"));
 
diff --git a/riscos/options.h b/riscos/options.h
index b97ddc6e2..7428fa39b 100644
--- a/riscos/options.h
+++ b/riscos/options.h
@@ -61,6 +61,7 @@ extern bool option_block_popups;
 extern bool option_url_suggestion;
 extern int option_image_memory_direct;	/* -1 means auto-detect */
 extern int option_image_memory_compressed;	/* -1 means auto-detect */
+extern bool option_strip_extensions;
 
 #define EXTRA_OPTION_DEFINE \
 bool option_use_mouse_gestures = false;\
@@ -102,7 +103,8 @@ int option_font_default = CSS_FONT_FAMILY_SANS_SERIF; \
 bool option_block_popups = false; \
 bool option_url_suggestion = true; \
 int option_image_memory_direct = -1; \
-int option_image_memory_compressed = -1;
+int option_image_memory_compressed = -1; \
+bool option_strip_extensions = true;
 
 #define EXTRA_OPTION_TABLE \
 { "use_mouse_gestures",     OPTION_BOOL,    &option_use_mouse_gestures },\
@@ -144,6 +146,7 @@ int option_image_memory_compressed = -1;
 { "block_popups",           OPTION_BOOL,    &option_block_popups }, \
 { "url_suggestion",         OPTION_BOOL,    &option_url_suggestion }, \
 { "image_memory_direct",    OPTION_INTEGER, &option_image_memory_direct }, \
-{ "image_memory_compressed",OPTION_INTEGER, &option_image_memory_compressed }
+{ "image_memory_compressed",OPTION_INTEGER, &option_image_memory_compressed }, \
+{ "strip_extensions",       OPTION_BOOL,    &option_strip_extensions }
 
 #endif
diff --git a/riscos/save.c b/riscos/save.c
index 807b45549..03224fcb9 100644
--- a/riscos/save.c
+++ b/riscos/save.c
@@ -29,6 +29,7 @@
 #include "netsurf/image/bitmap.h"
 #include "netsurf/riscos/gui.h"
 #include "netsurf/riscos/menus.h"
+#include "netsurf/riscos/options.h"
 #include "netsurf/riscos/save_complete.h"
 #include "netsurf/riscos/save_draw.h"
 #include "netsurf/riscos/thumbnail.h"
@@ -50,7 +51,8 @@ static bool using_dragasprite = true;
 static bool saving_from_dialog = true;
 static osspriteop_area *saveas_area = NULL;
 static wimp_w gui_save_sourcew = (wimp_w)-1;
-static char save_leafname[32];
+#define LEAFNAME_MAX 200
+static char save_leafname[LEAFNAME_MAX];
 
 typedef enum { LINK_ACORN, LINK_ANT, LINK_TEXT } link_format;
 
@@ -58,7 +60,8 @@ static bool ro_gui_save_complete(struct content *c, char *path);
 static bool ro_gui_save_content(struct content *c, char *path);
 static void ro_gui_save_object_native(struct content *c, char *path);
 static bool ro_gui_save_link(struct content *c, link_format format, char *path);
-static void ro_gui_save_set_state(struct content *c, gui_save_type save_type, char *leaf_buf, char *icon_buf);
+static void ro_gui_save_set_state(struct content *c, gui_save_type save_type,
+		char *leaf_buf, char *icon_buf);
 static bool ro_gui_save_create_thumbnail(struct content *c, const char *name);
 
 
@@ -182,7 +185,7 @@ void ro_gui_saveas_quit(void)
 
 void ro_gui_save_prepare(gui_save_type save_type, struct content *c)
 {
-	char name_buf[64];
+	char name_buf[LEAFNAME_MAX];
 	char icon_buf[20];
 
 	assert((save_type == GUI_SAVE_HOTLIST_EXPORT_HTML) ||
@@ -329,7 +332,8 @@ void gui_drag_save_selection(struct selection *s, struct gui_window *g)
 
 	gui_save_selection = s;
 
-	ro_gui_save_set_state(NULL, GUI_SAVE_TEXT_SELECTION, save_leafname, icon_buf);
+	ro_gui_save_set_state(NULL, GUI_SAVE_TEXT_SELECTION, save_leafname,
+			icon_buf);
 
 	gui_current_drag_type = GUI_DRAG_SAVE;
 
@@ -873,19 +877,22 @@ bool ro_gui_save_link(struct content *c, link_format format, char *path)
  *
  * \param  c          content being saved
  * \param  save_type  type of save operation being performed
- * \param  leaf_buf   buffer to receive suggested leafname
- * \param  icon_buf   buffer to receive sprite name
+ * \param  leaf_buf   buffer to receive suggested leafname, length at least
+ *                    LEAFNAME_MAX
+ * \param  icon_buf   buffer to receive sprite name, length at least 13
  */
 
-void ro_gui_save_set_state(struct content *c, gui_save_type save_type, char *leaf_buf, char *icon_buf)
+void ro_gui_save_set_state(struct content *c, gui_save_type save_type,
+		char *leaf_buf, char *icon_buf)
 {
 	/* filename */
 	const char *name = gui_save_table[save_type].name;
 	url_func_result res;
 	bool done = false;
-	char *nice = NULL;
+	char *nice;
 	utf8_convert_ret err;
 	char *local_name;
+	size_t i;
 
 	/* parameters that we need to remember */
 	gui_save_current_type = save_type;
@@ -897,13 +904,23 @@ void ro_gui_save_set_state(struct content *c, gui_save_type save_type, char *lea
 		gui_save_filetype = ro_content_filetype(c);
 
 	/* leafname */
-	if (c && (res = url_nice(c->url, (char **)&nice)) == URL_FUNC_OK)
+	if (c && (res = url_nice(c->url, &nice, option_strip_extensions)) ==
+			URL_FUNC_OK) {
+		for (i = 0; nice[i]; i++) {
+			if (nice[i] == '.')
+				nice[i] = '/';
+			else if (nice[i] <= ' ' ||
+					strchr(":*#$&@^%\\", nice[i]))
+				nice[i] = '_';
+		}
 		name = nice;
-	else
+	} else {
 		name = messages_get(name);
+	}
 
 	/* filename is utf8 */
-	strcpy(leaf_buf, name);
+	strncpy(leaf_buf, name, LEAFNAME_MAX);
+	leaf_buf[LEAFNAME_MAX - 1] = 0;
 
 	err = utf8_to_local_encoding(name, 0, &local_name);
 	if (err != UTF8_CONVERT_OK) {
@@ -944,15 +961,17 @@ void ro_gui_save_set_state(struct content *c, gui_save_type save_type, char *lea
 		error = ro_gui_wimp_get_sprite(icon_buf, &sprite);
 		if (error) {
 			LOG(("ro_gui_wimp_get_sprite: 0x%x: %s",
-				error->errnum, error->errmess));
+					error->errnum, error->errmess));
 			warn_user("MiscError", error->errmess);
 		} else {
-			/* the sprite area should always be large enough for file_xxx sprites */
-			assert(sprite->size <= saveas_area->size - saveas_area->first);
+			/* the sprite area should always be large enough for
+			 * file_xxx sprites */
+			assert(sprite->size <= saveas_area->size -
+					saveas_area->first);
 
 			memcpy((byte*)saveas_area + saveas_area->first,
-				sprite,
-				sprite->size);
+					sprite,
+					sprite->size);
 
 			saveas_area->sprite_count = 1;
 			saveas_area->used = saveas_area->first + sprite->size;
diff --git a/riscos/theme_install.c b/riscos/theme_install.c
index 93ef0a97d..9c483ba8b 100644
--- a/riscos/theme_install.c
+++ b/riscos/theme_install.c
@@ -203,7 +203,7 @@ void theme_install_install(bool apply)
 
 	assert(theme_install_content);
 
-	if (url_nice(theme_install_descriptor.name, &theme_file) !=
+	if (url_nice(theme_install_descriptor.name, &theme_file, true) !=
 			URL_FUNC_OK) {
 		warn_user("ThemeInstallErr", 0);
 		theme_install_close();
diff --git a/utils/url.c b/utils/url.c
index c602af912..1eb0f6f5e 100644
--- a/utils/url.c
+++ b/utils/url.c
@@ -15,6 +15,7 @@
 #include <stdbool.h>
 #include <stdlib.h>
 #include <string.h>
+#include <strings.h>
 #include <sys/types.h>
 #include <regex.h>
 #include "netsurf/utils/log.h"
@@ -22,7 +23,7 @@
 #include "netsurf/utils/utils.h"
 
 
-regex_t url_re, url_up_re, url_nice_re;
+regex_t url_re, url_up_re;
 
 /**
  * Initialise URL routines.
@@ -49,18 +50,15 @@ void url_init(void)
 			"/([^/]|[.][^./]|[^./][.]|[^./][^./]|[^/][^/][^/]+)?"
 			"/[.][.](/|$)",
 			REG_EXTENDED);
-	regcomp_wrapper(&url_nice_re,
-			"^([^.]{0,4}[.])?([^.][^.][.])?([^/?&;.=]*)"
-			"(=[^/?&;.]*)?[/?&;.]",
-			REG_EXTENDED);
 }
 
 
 /**
  * Normalize a URL.
  *
- * \param  url  an absolute URL
- * \return  cleaned up url, allocated on the heap, or 0 on failure
+ * \param  url     an absolute URL
+ * \param  result  pointer to pointer to buffer to hold cleaned up url
+ * \return  URL_FUNC_OK on success
  *
  * If there is no scheme, http:// is added. The scheme and host are
  * lower-cased. Default ports are removed (http only). An empty path is
@@ -208,9 +206,10 @@ url_func_result url_normalize(const char *url, char **result)
 /**
  * Resolve a relative URL to absolute form.
  *
- * \param  rel   relative URL
- * \param  base  base URL, must be absolute and cleaned as by url_normalize()
- * \return  an absolute URL, allocated on the heap, or 0 on failure
+ * \param  rel     relative URL
+ * \param  base    base URL, must be absolute and cleaned as by url_normalize()
+ * \param  result  pointer to pointer to buffer to hold absolute url
+ * \return  URL_FUNC_OK on success
  */
 
 url_func_result url_join(const char *rel, const char *base, char **result)
@@ -433,8 +432,9 @@ step7:	/* 7) */
 /**
  * Return the host name from an URL.
  *
- * \param  url  an absolute URL
- * \returns  host name allocated on heap, or 0 on failure
+ * \param  url     an absolute URL
+ * \param  result  pointer to pointer to buffer to hold host name
+ * \return  URL_FUNC_OK on success
  */
 
 url_func_result url_host(const char *url, char **result)
@@ -459,20 +459,23 @@ url_func_result url_host(const char *url, char **result)
 		return URL_FUNC_NOMEM;
 	}
 	strncpy((*result), url + match[URL_RE_AUTHORITY].rm_so,
-			match[URL_RE_AUTHORITY].rm_eo - match[4].rm_so);
+			match[URL_RE_AUTHORITY].rm_eo -
+			match[URL_RE_AUTHORITY].rm_so);
 	(*result)[match[URL_RE_AUTHORITY].rm_eo -
 			match[URL_RE_AUTHORITY].rm_so] = 0;
 
 	return URL_FUNC_OK;
 }
 
+
 /**
- * Return the scheme name from an URL
+ * Return the scheme name from an URL.
  *
- * \param url     an absolute URL
- * \param result  pointer to pointer to buffer to hold scheme name
- * \return URL_FUNC_OK on success
+ * \param  url     an absolute URL
+ * \param  result  pointer to pointer to buffer to hold scheme name
+ * \return  URL_FUNC_OK on success
  */
+
 url_func_result url_scheme(const char *url, char **result)
 {
 	int m;
@@ -494,7 +497,6 @@ url_func_result url_scheme(const char *url, char **result)
 		LOG(("malloc failed"));
 		return URL_FUNC_NOMEM;
 	}
-
 	strncpy((*result), url + match[URL_RE_SCHEME].rm_so,
 			match[URL_RE_SCHEME].rm_eo -
 			match[URL_RE_SCHEME].rm_so);
@@ -503,106 +505,126 @@ url_func_result url_scheme(const char *url, char **result)
 	return URL_FUNC_OK;
 }
 
+
 /**
  * Attempt to find a nice filename for a URL.
  *
- * \param  url  an absolute URL
- * \returns  filename allocated on heap, or 0 on memory exhaustion
+ * \param  url     an absolute URL
+ * \param  result  pointer to pointer to buffer to hold filename
+ * \param  remove_extensions  remove any extensions from the filename
+ * \return  URL_FUNC_OK on success
  */
 
-url_func_result url_nice(const char *url, char **result)
+url_func_result url_nice(const char *url, char **result,
+		bool remove_extensions)
 {
-	unsigned int i, j, k = 0, so;
-	unsigned int len;
-	const char *colon;
-	char buf[40];
-	char *rurl;
 	int m;
 	regmatch_t match[10];
+	regoff_t start, end;
+	size_t i;
+	char *dot;
 
-	/* just in case */
-	(*result) = 0;
+	*result = 0;
 
-	(*result) = malloc(40);
-	if (!(*result))
-		return URL_FUNC_NOMEM;
-
-	len = strlen(url);
-	assert(len != 0);
-	rurl = malloc(len + 1);
-	if (!rurl) {
-		free((*result));
-		return URL_FUNC_NOMEM;
+	m = regexec(&url_re, url, 10, match, 0);
+	if (m) {
+		LOG(("url '%s' failed to match regex", url));
+		return URL_FUNC_FAILED;
 	}
 
-	/* reverse url into rurl */
-	for (i = 0, j = len - 1; i != len; i++, j--)
-		rurl[i] = url[j];
-	rurl[len] = 0;
-
-	/* prepare a fallback: always succeeds */
-	colon = strchr(url, ':');
-	if (colon)
-		url = colon + 1;
-	strncpy((*result), url, 15);
-	(*result)[15] = 0;
-	for (i = 0; (*result)[i]; i++)
-		if (!isalnum((*result)[i]))
-			(*result)[i] = '_';
-
-	/* append nice pieces */
-	j = 0;
-	do {
-		m = regexec(&url_nice_re, rurl + j, 10, match, 0);
-		if (m)
-			break;
-
-		if (match[3].rm_so != match[3].rm_eo) {
-			so = match[3].rm_so;
-			i = match[3].rm_eo - so;
-			if (15 < i) {
-				so = match[3].rm_eo - 15;
-				i = 15;
-			}
-			if (15 < k + i)
-				break;
-			if (k)
-				k+=2;
-			strncpy(buf + k, rurl + j + so, i);
-			k += i;
-			buf[k] = 160;	/* nbsp */
-			buf[k+1] = 0xc2;	/* as UTF-8 */
-		}
+	/* extract the last component of the path, if possible */
+	if (match[URL_RE_PATH].rm_so == -1 || match[URL_RE_PATH].rm_so ==
+			match[URL_RE_PATH].rm_eo)
+		goto no_path;  /* no path, or empty */
+	for (end = match[URL_RE_PATH].rm_eo - 1;
+			end != match[URL_RE_PATH].rm_so && url[end] == '/';
+			end--)
+		;
+	if (end == match[URL_RE_PATH].rm_so)
+		goto no_path;  /* path is a string of '/' */
+	end++;
+	for (start = end - 1;
+			start != match[URL_RE_PATH].rm_so && url[start] != '/';
+			start--)
+		;
+	if (url[start] == '/')
+		start++;
+
+	if (!strncasecmp(url + start, "index.", 6) ||
+			!strncasecmp(url + start, "default.", 8)) {
+		/* try again */
+		if (start == match[URL_RE_PATH].rm_so)
+			goto no_path;
+		for (end = start - 1;
+				end != match[URL_RE_PATH].rm_so &&
+				url[end] == '/';
+				end--)
+			;
+		if (end == match[URL_RE_PATH].rm_so)
+			goto no_path;
+		end++;
+		for (start = end - 1;
+				start != match[URL_RE_PATH].rm_so &&
+				url[start] != '/';
+				start--)
+		;
+		if (url[start] == '/')
+			start++;
+	}
 
-		j += match[0].rm_eo;
-	} while (j != len);
+	*result = malloc(end - start + 1);
+	if (!*result) {
+		LOG(("malloc failed"));
+		return URL_FUNC_NOMEM;
+	}
+	strncpy(*result, url + start, end - start);
+	(*result)[end - start] = 0;
 
-	if (k == 0) {
-		free(rurl);
-		return URL_FUNC_OK;
+	if (remove_extensions) {
+		dot = strchr(*result, '.');
+		if (dot && dot != *result)
+			*dot = 0;
 	}
 
-	/* reverse back */
-	for (i = 0, j = k - 1; i != k; i++, j--)
-		(*result)[i] = buf[j];
-	(*result)[k] = 0;
+	return URL_FUNC_OK;
 
-	for (i = 0; i != k; i++)
-		if ((*result)[i] != (char) 0xa0 && !isalnum((*result)[i]))
-			(*result)[i] = '_';
+no_path:
 
-	free(rurl);
+	/* otherwise, use the host name, with '.' replaced by '_' */
+	if (match[URL_RE_AUTHORITY].rm_so != -1 &&
+			match[URL_RE_AUTHORITY].rm_so !=
+			match[URL_RE_AUTHORITY].rm_eo) {
+		*result = malloc(match[URL_RE_AUTHORITY].rm_eo -
+				match[URL_RE_AUTHORITY].rm_so + 1);
+		if (!*result) {
+			LOG(("malloc failed"));
+			return URL_FUNC_NOMEM;
+		}
+		strncpy(*result, url + match[URL_RE_AUTHORITY].rm_so,
+				match[URL_RE_AUTHORITY].rm_eo -
+				match[URL_RE_AUTHORITY].rm_so);
+		(*result)[match[URL_RE_AUTHORITY].rm_eo -
+				match[URL_RE_AUTHORITY].rm_so] = 0;
 
-	return URL_FUNC_OK;
+		for (i = 0; (*result)[i]; i++)
+			if ((*result)[i] == '.')
+				(*result)[i] = '_';
+
+		return URL_FUNC_OK;
+	}	
+	
+	return URL_FUNC_FAILED;
 }
 
+
 /**
- * Escape a string suitable for inclusion in an URI
+ * Escape a string suitable for inclusion in an URL.
  *
- * \param unescaped  The unescaped string
- * \param result  Pointer to location to store escaped string
- * \return URL_FUNC_OK on success
+ * \param  unescaped  the unescaped string
+ * \param  result     pointer to pointer to buffer to hold escaped string
+ * \return  URL_FUNC_OK on success
  */
+
 url_func_result url_escape(const char *unescaped, char **result)
 {
 	int len;
@@ -649,6 +671,7 @@ url_func_result url_escape(const char *unescaped, char **result)
 	return URL_FUNC_OK;
 }
 
+
 #ifdef TEST
 
 int main(int argc, char *argv[])
@@ -670,19 +693,30 @@ int main(int argc, char *argv[])
 			printf("<== '%s'\n", s);
 			free(s);
 		}*/
-		if (1 != i) {
+/*		if (1 != i) {
 			res = url_join(argv[i], argv[1], &s);
 			if (res == URL_FUNC_OK) {
 				printf("'%s' + '%s' \t= '%s'\n", argv[1],
 						argv[i], s);
 				free(s);
 			}
+		}*/
+		printf("'%s' => ", argv[i]);
+		res = url_nice(argv[i], &s, true);
+		if (res == URL_FUNC_OK) {
+			printf("'%s', ", s);
+			free(s);
+		} else {
+			printf("failed %u, ", res);
 		}
-/*		res = url_nice(argv[i], &s);
+		res = url_nice(argv[i], &s, false);
 		if (res == URL_FUNC_OK) {
-			printf("'%s'\n", s);
+			printf("'%s', ", s);
 			free(s);
-		}*/
+		} else {
+			printf("failed %u, ", res);
+		}
+		printf("\n");
 	}
 	return 0;
 }
diff --git a/utils/url.h b/utils/url.h
index d803eafeb..94aa626ca 100644
--- a/utils/url.h
+++ b/utils/url.h
@@ -24,7 +24,8 @@ url_func_result url_normalize(const char *url, char **result);
 url_func_result url_join(const char *rel, const char *base, char **result);
 url_func_result url_host(const char *url, char **result);
 url_func_result url_scheme(const char *url, char **result);
-url_func_result url_nice(const char *url, char **result);
+url_func_result url_nice(const char *url, char **result,
+		bool remove_extensions);
 url_func_result url_escape(const char *unescaped, char **result);
 
 #endif