String: Usar delimitadores de más de un caracter en split (#142)

* string: split with long delimitator

* string: split with empty string as separator

* string: split with null separator

* string: string array usage in string split

* string: split empty text and separator

* string: split code style

src/commons/string.c

@@ -26,7 +26,7 @@ static void _string_do(char *text, void (*closure)(char*));
 static void _string_lower_element(char* ch);
 static void _string_upper_element(char* ch);
 void _string_append_with_format_list(const char* format, char** original, va_list arguments);
-char** _string_split(char* text, char* separator, bool(*condition)(char*, int));
+char** _string_split(char* text, char* separator, bool(*is_last_token)(int));
 static void _string_array_push(char*** array, char* text, int size);
 static char* _string_array_replace(char** array, int pos, char* text);
 
@@ -145,15 +145,15 @@ bool string_equals_ignore_case(char *actual, char *expected) {
 }
 
 char **string_split(char *text, char *separator) {
-	bool _is_last_token(char* next, int _) {
-		return next != NULL;
+	bool _is_last_token(int _) {
+		return false;
 	}
 	return _string_split(text, separator, _is_last_token);
 }
 
 char** string_n_split(char *text, int n, char* separator) {
-	bool _is_last_token(char* next, int index) {
-		return next != NULL && index < (n - 1);
+	bool _is_last_token(int index) {
+		return index == (n - 1);
 	}
 	return _string_split(text, separator, _is_last_token);
 }
@@ -286,27 +286,26 @@ void _string_append_with_format_list(const char* format, char** original, va_lis
 	free(temporal);
 }
 
-char** _string_split(char* text, char* separator, bool(*condition)(char*, int)) {
+char** _string_split(char* text, char* separator, bool(*is_last_token)(int)) {
 	char **substrings = string_array_new();
-	int size = 0;
-
-	char *text_to_iterate = string_duplicate(text);
-	char *next = text_to_iterate;
-
-	while(condition(next, size)) {
-		char* token = strsep(&next, separator);
-		if(token == NULL) {
-			break;
+	int index = 0;
+
+	char *end, *start = text;
+	if (separator != NULL) {
+		while ((end = strstr(start, separator)) != NULL && !is_last_token(index)) {
+			if (string_is_empty(separator)) {
+				if (string_length(start) > 1)
+					end = start + 1;
+			 	else
+					break;
+			}
+			_string_array_push(&substrings, string_substring_until(start, end - start), index++);
+			start = end + string_length(separator);
 		}
-		_string_array_push(&substrings, string_duplicate(token), size);
-		size++;
-	};
-
-	if (next != NULL) {
-		_string_array_push(&substrings, string_duplicate(next), size);
 	}
 
-	free(text_to_iterate);
+	_string_array_push(&substrings, string_duplicate(start), index);
+
 	return substrings;
 }
 

tests/unit-tests/test_string.c

@@ -182,9 +182,9 @@ context (test_string) {
 
         describe("Split") {
 
-            it("split_with_delimitators") {
-                char *line = "path/to/file";
-                char** substrings = string_split(line, "/");
+            it("split_with_separators") {
+                char *line = "path//to//file";
+                char** substrings = string_split(line, "//");
 
                 should_ptr(substrings) not be null;
                 should_string(substrings[0]) be equal to ("path");
@@ -195,7 +195,35 @@ context (test_string) {
                 string_array_destroy(substrings);
             } end
 
-            it("split_starting_with_delimitator") {
+            it("split_with_empty_string_as_separator") {
+                char *line = "hello";
+                char** substrings = string_split(line, "");
+
+                should_ptr(substrings) not be null;
+                should_string(substrings[0]) be equal to ("h");
+                should_string(substrings[1]) be equal to ("e");
+                should_string(substrings[2]) be equal to ("l");
+                should_string(substrings[3]) be equal to ("l");
+                should_string(substrings[4]) be equal to ("o");
+                should_ptr(substrings[5]) be null;
+
+                string_iterate_lines(substrings, (void*) free);
+                free(substrings);
+            } end
+
+            it("split_with_null_separator") {
+                char *line = "path/to/file";
+                char** substrings = string_split(line, NULL);
+
+                should_ptr(substrings) not be null;
+                should_string(substrings[0]) be equal to ("path/to/file");
+                should_ptr(substrings[1]) be null;
+
+                string_iterate_lines(substrings, (void*) free);
+                free(substrings);
+            } end
+
+            it("split_starting_with_separator") {
                 char* line = "/path/to/file";
                 char** substrings = string_split(line, "/");
 
@@ -209,7 +237,7 @@ context (test_string) {
                 string_array_destroy(substrings);
             } end
 
-            it("split_ending_with_delimitator") {
+            it("split_ending_with_separator") {
                 char* line = "path/to/file/";
                 char** substrings = string_split(line, "/");
 
@@ -223,7 +251,7 @@ context (test_string) {
                 string_array_destroy(substrings);
             } end
 
-            it("split_having_delimitators_in_between") {
+            it("split_having_separators_in_between") {
                 char* line = "path/to//file";
                 char** substrings = string_split(line, "/");
 
@@ -249,6 +277,18 @@ context (test_string) {
 
             } end
 
+            it("split_is_empty_with_empty_separator") {
+                char* line = "";
+                char** substrings = string_split(line, "");
+
+                should_ptr(substrings) not be null;
+                should_string(substrings[0]) be equal to("");
+                should_ptr(substrings[1]) be null;
+
+                string_array_destroy(substrings);
+
+            } end
+
             it("n_split_when_n_is_less_than_splitted_elements") {
                 char *line = "Hola planeta tierra";
                 char** substrings = string_n_split(line, 2, " ");