URL rewritten completely. Error handling not implemented in public

1999-04-24  bertrand <Bertrand.Guiheneuf@inria.fr>

	* camel/url-util.c (new_g_url): URL
	rewritten completely. Error handling not
	implemented in public functions.
	But URL scan works pretty well :)))

svn path=/trunk/; revision=874

2 files changed, 242 insertions, 52 deletions

diff --git a/camel/url-util.c b/camel/url-util.c
index d1ea09c6c9..45b5c0691d 100644
--- a/camel/url-util.c
+++ b/camel/url-util.c
@@ -31,20 +31,96 @@
    Uniform Ressource Locators 
    
      Bertrand. */
+/*
+  XXX TODO: recover  the words between #'s or ?'s after the path */
 
 
-#include <ctype.h> /* for isalpha */
-#include <stdlib.h> /* for atoi */
-
 #include "url-util.h"
 
 
+typedef gboolean find_item_func(GString *url, GString **item, guint *position, gboolean *error);
+
+typedef struct {
+	char *item_name;
+	GString **item_value;
+	find_item_func *find_func;
+} FindStepStruct;
+
+static gboolean find_protocol(GString *url, GString **item, guint *position, gboolean *error);
+static gboolean find_user(GString *url, GString **item, guint *position, gboolean *error);
+static gboolean find_passwd(GString *url, GString **item, guint *position, gboolean *error);
+static gboolean find_host(GString *url, GString **item, guint *position, gboolean *error);
+static gboolean find_port(GString *url, GString **item, guint *position, gboolean *error);
+static gboolean find_path(GString *url, GString **item, guint *position, gboolean *error);
+
+
+
+/**
+ * new_g_url: create an Gurl object from a string
+ * @url_string: The string containing the URL to scan
+ * 
+ * This routine takes a GString and parses it as an
+ * URL of the form:
+ * protocol://user:password@host:port/path
+ * there is no test on the values. For example,
+ * "port" can be a string, not only a number !
+ * The Gurl structure fields ar filled with
+ * the scan results. When a member of the 
+ * general URL can not be found, the corresponding
+ * Gurl member is NULL  
+ * 
+ * Return value: a Gurl structure containng the URL items.
+ **/
+Gurl *new_g_url(GString* url_string)
+{
+	Gurl *g_url;
+	
+	GString *protocol;
+	GString *user;
+	GString *passwd;
+	GString *host;
+	GString *port;
+	GString *path;
+	
+	guint position=0;
+	gboolean error;
+	gboolean found;
+	guint i;
+	
+	g_url = g_new(Gurl,1);
+
+#define NB_STEP_URL  6
+	{
+		FindStepStruct step[NB_STEP_URL] = {
+			{ "protocol", &(g_url->protocol), find_protocol},
+			{ "user", &(g_url->user), find_user},
+			{ "password", &(g_url->passwd), find_passwd},
+			{ "host", &(g_url->host), find_host},
+			{ "port", &(g_url->port), find_port},
+			{ "path", &(g_url->path), find_path}
+		};
+		
+		for (i=0; i<NB_STEP_URL; i++) {
+			found = step[i].find_func(url_string, 
+						  step[i].item_value, 
+						  &position, 
+						  &error);
+		}
+	}
+	
+	return g_url;
+}
+
+
+
 
 
+/** So, yes, I must admit there would have been more elegant
+    ways to do this, but it works, and quite well :)  */
 
 
 static gboolean 
-find_protocol(GString *url, GString **protocol, guint *position, gboolean *error)
+find_protocol(GString *url, GString **item, guint *position, gboolean *error)
 {
 
 	guint i;
@@ -55,7 +131,7 @@ find_protocol(GString *url, GString **protocol, guint *position, gboolean *error
 	str_url = url->str;
 	len_url = url->len;
 	
-	*protocol = NULL;
+	*item = NULL;
 	*error = FALSE;
 	i=*position;
 	
@@ -71,7 +147,7 @@ find_protocol(GString *url, GString **protocol, guint *position, gboolean *error
 		{
 			
 			str_protocol = g_strndup(str_url, i-3);
-			*protocol = g_string_new(str_protocol);
+			*item = g_string_new(str_protocol);
 			*position=i;
 			return TRUE;
 		}
@@ -83,7 +159,7 @@ find_protocol(GString *url, GString **protocol, guint *position, gboolean *error
 
 
 static gboolean
-find_user(GString *url, GString **user, guint *position, gboolean *error)
+find_user(GString *url, GString **item, guint *position, gboolean *error)
 {
 	guint i;
 	guint at_pos;
@@ -95,7 +171,7 @@ find_user(GString *url, GString **user, guint *position, gboolean *error)
 	str_url = url->str;
 	len_url = url->len;
 	
-	*user = NULL;
+	*item = NULL;
 	i=*position;
 	
 
@@ -110,21 +186,21 @@ find_user(GString *url, GString **user, guint *position, gboolean *error)
 	while ( (i<at_pos) && (str_url[i] != ':') ) i++;
 
 	/* now if i has not been incremented at all, there is no user */
-	if (i == *position) return FALSE;
+	if (i == *position) {
+		(*position)++;
+		return FALSE;
+	}
 	
 	str_user = g_strndup(str_url+ *position, i - *position);
-	*user = g_string_new(str_user);
+	*item = g_string_new(str_user);
 	if (i<at_pos) *position=i+1; /* there was a ':', skip it */
 	else *position=i;
 	
-	return TRUE;
-
-	
-	
+	return TRUE;	
 }
 
 static gboolean
-find_passwd(GString *url, GString **passwd, guint *position, gboolean *error)
+find_passwd(GString *url, GString **item, guint *position, gboolean *error)
 {
 	guint i;
 	
@@ -135,7 +211,7 @@ find_passwd(GString *url, GString **passwd, guint *position, gboolean *error)
 	str_url = url->str;
 	len_url = url->len;
 	
-	*passwd = NULL;
+	*item = NULL;
 	i=*position;
 	
 
@@ -150,24 +226,132 @@ find_passwd(GString *url, GString **passwd, guint *position, gboolean *error)
 	}
 	
 	str_passwd = g_strndup(str_url+ *position, i - *position);
-	*passwd = g_string_new(str_passwd);
+	*item = g_string_new(str_passwd);
 	*position=i+1; /* skip it the '@' */
 	
 	return TRUE;
+}
+
+
+
+static gboolean
+find_host(GString *url, GString **item, guint *position, gboolean *error)
+{
+	guint i;
+	guint slash_pos;
+	
+	gchar *str_url;
+	gint len_url;
+	gchar *str_host;
+	
+	str_url = url->str;
+	len_url = url->len;
+	
+	*item = NULL;
+	i=*position;
+	
+
+	/* find a '/' */
+	while ((i<len_url) && (str_url[i] != '/')) i++;
+	
+	slash_pos = i;
+	i = *position;
+
+	/* find a ':' */
+	while ( (i<slash_pos) && (str_url[i] != ':') ) i++;
+
+	/* at this point if i has not been incremented at all, 
+	   there is no host */
+	if (i == *position) {
+		(*position)++;
+		return FALSE;
+	}
+	
+	str_host = g_strndup(str_url+ *position, i - *position);
+	*item = g_string_new(str_host);
+	if (i<slash_pos) *position=i+1; /* there was a ':', skip it */
+	else *position=i;
+	
+	return TRUE;
+}
+
 
+static gboolean
+find_port(GString *url, GString **item, guint *position, gboolean *error)
+{
+	guint i;
+	guint slash_pos;
+	
+	gchar *str_url;
+	gint len_url;
+	gchar *str_port;
 	
+	str_url = url->str;
+	len_url = url->len;
+	
+	*item = NULL;
+	i=*position;
 	
+
+	/* find a '/' */
+	while ((i<len_url) && (str_url[i] != '/')) i++;
+	
+	slash_pos = i;
+	i = *position;
+
+	/* find a ':' */
+	while ( (i<slash_pos) && (str_url[i] != ':') ) i++;
+
+	/* at this point if i has not been incremented at all, 
+	   there is no port */
+	if (i == *position) return FALSE;
+
+	str_port = g_strndup(str_url+ *position, i - *position);
+	*item = g_string_new(str_port);
+	*position = i;
+	return TRUE;
 }
 
 
+static gboolean
+find_path(GString *url, GString **item, guint *position, gboolean *error)
+{
+	guint i;
+	
+	gchar *str_url;
+	gint len_url;
+	gchar *str_path;
+	
+	str_url = url->str;
+	len_url = url->len;
+	
+	*item = NULL;
+	i=*position;
+	
+
+	/* find a '#' */
+	while ((i<len_url) && (str_url[i] != '#') && (str_url[i] != '?')) i++;
+	
+	/*i has not been incremented at all, there is no path */
+	if (i == *position) return FALSE;
+	
+	str_path = g_strndup(str_url+ *position, i - *position);
+	*item = g_string_new(str_path);
+	*position=i;
+	
+	
+	return TRUE;
+}
 
 
 /* to tests this file :
-   gcc -o test_url_util `glib-config --cflags`  -DTEST_URL_UTIL url-util.c `glib-config --libs
+   gcc -o test_url_util `glib-config --cflags`  -DTEST_URL_UTIL url-util.c `glib-config --libs`
    ./test_url_util URL
 */
 #ifdef TEST_URL_UTIL
 
+
+
 int 
 main (int argc, char **argv)
 {
@@ -176,47 +360,43 @@ main (int argc, char **argv)
 	GString *protocol;
 	GString *user;
 	GString *passwd;
+	GString *host;
+	GString *port;
+	GString *path;
 	guint position=0;
 	gboolean error;
 	gboolean found;
 	guint i;
-
+	guint i_pos;
+
+#define NB_STEP_TEST  6
+	FindStepStruct test_step[NB_STEP_TEST] = {
+		{ "protocol", &protocol, find_protocol},
+		{ "user", &user, find_user},
+		{ "password", &passwd, find_passwd},
+		{ "host", &host, find_host},
+		{ "port", &port, find_port},
+		{ "path", &path, find_path}
+	};
 	url = g_string_new(argv[1]);
 	printf("URL to test : %s\n\n", url->str);
-	
-	/* Try to find the protocol */
-	found = find_protocol(url, &protocol, &position, &error);
-	if (found) {
-		printf("protocol found : %s\n", protocol->str);
-	} else printf("protocol not found in URL\n\n");
-	printf("posistion of the next item:\n");
-	printf("%s\n", url->str);
-	for(i=0; i<position; i++) printf(" ");
-	printf("^\n");
+	for (i=0; i<NB_STEP_TEST; i++) {
+		found = test_step[i].find_func(url, 
+					  test_step[i].item_value, 
+					  &position, 
+					  &error);
+		if (found) {
+			printf("\t\t\t\t** %s found : %s\n",
+			       test_step[i].item_name,
+			       (*test_step[i].item_value)->str);
+		} else printf("** %s not found in URL\n", test_step[i].item_name);
+		printf("next item position:\n");
+		printf("%s\n", url->str);
+		for(i_pos=0; i_pos<position; i_pos++) printf(" ");
+		printf("^\n");
 		
-	/* Try to find the user name */
-	found = find_user(url, &user, &position, &error);
-	if (found) {
-		printf("name found : %s\n", user->str);
-	} else printf("user name not found in URL\n");
-	printf("posistion of the next item:\n");
-	printf("%s\n", url->str);
-	for(i=0; i<position; i++) printf(" ");
-	printf("^\n");
-	
-	/* Try to find the password */
-	found = find_passwd(url, &passwd, &position, &error);
-	if (found) {
-		printf("passwd found : %s\n", passwd->str);
-		printf("\n");
-	} else printf("passwd not found in URL\n");
-	printf("posistion of the next item:\n");
-	printf("%s\n", url->str);
-	for(i=0; i<position; i++) printf(" ");
-	printf("^\n");
-	
-	
-	return 0;
+	}
+	
 }
 
 #endif /* TEST_URL_UTIL */
diff --git a/camel/url-util.h b/camel/url-util.h
index 2c9218ea78..fbacbe6da6 100644
--- a/camel/url-util.h
+++ b/camel/url-util.h
@@ -31,6 +31,16 @@ extern "C" {
 #pragma }
 #endif /* __cplusplus */
 
+typedef struct {
+	GString *protocol;
+	GString *user;
+	GString *passwd;
+	GString *host;
+	GString *port;
+	GString *path;
+} Gurl;
+
+Gurl *new_g_url(GString* url_string);
 
 #ifdef __cplusplus
 }