Implement quotes in cmd line parser

examples/cmdline.c

@@ -6,7 +6,7 @@
 static const unsigned int ARGV_COUNT_MAX = 100;
 static const unsigned int ARG_SIZE_MAX = 4096;
 
-static const char *const cmdline = "foo bar\\ car";
+static const char *const cmdline = "foo bar\\ baz \"car cdr\"";
 
 int main()
 {
@@ -26,9 +26,10 @@ int main()
     ));
 
     assert(strcmp(error_msg, "") == 0);
-    assert(argc == 2);
+    assert(argc == 3);
     assert(strcmp(argv[0], "foo") == 0);
-    assert(strcmp(argv[1], "bar car") == 0);
+    assert(strcmp(argv[1], "bar baz") == 0);
+    assert(strcmp(argv[2], "car cdr") == 0);
 
     return 0;
 }

src/cmdline.c

@@ -13,6 +13,7 @@ enum State {
     WHITESPACE,
     TOKEN,
     BACKSLASHED,
+    QUOTED,
 };
 
 bool kernaux_cmdline_parse(
@@ -75,6 +76,15 @@ bool kernaux_cmdline_parse(
                 state = BACKSLASHED;
                 argv[(*argc)++] = buffer;
             }
+            else if (cur == '"') {
+                if (*argc >= argv_count_max) {
+                    kernaux_strncpy(error_msg, "too many args", 13);
+                    goto fail;
+                }
+
+                state = QUOTED;
+                argv[(*argc)++] = buffer;
+            }
             else {
                 if (*argc >= argv_count_max) {
                     kernaux_strncpy(error_msg, "too many args", 13);
@@ -108,6 +118,15 @@ bool kernaux_cmdline_parse(
                 state = BACKSLASHED;
                 argv[(*argc)++] = buffer;
             }
+            else if (cur == '"') {
+                if (*argc >= argv_count_max) {
+                    kernaux_strncpy(error_msg, "too many args", 13);
+                    goto fail;
+                }
+
+                state = QUOTED;
+                argv[(*argc)++] = buffer;
+            }
             else {
                 if (*argc >= argv_count_max) {
                     kernaux_strncpy(error_msg, "too many args", 13);
@@ -150,6 +169,10 @@ bool kernaux_cmdline_parse(
             else if (cur == '\\') {
                 state = BACKSLASHED;
             }
+            else if (cur == '"') {
+                kernaux_strncpy(error_msg, "unescaped quotation mark", 24);
+                goto fail;
+            }
             else {
                 if (buffer_size >= arg_size_max) {
                     kernaux_strncpy(error_msg, "arg too long", 12);
@@ -177,6 +200,32 @@ bool kernaux_cmdline_parse(
                 ++buffer_size;
             }
             break;
+
+        case QUOTED:
+            if (cur == '\0') {
+                kernaux_strncpy(error_msg, "EOL inside quote", 16);
+                goto fail;
+            }
+            else if (cur == '"') {
+                if (buffer_size >= arg_size_max) {
+                    kernaux_strncpy(error_msg, "arg too long", 12);
+                    goto fail;
+                }
+
+                state = WHITESPACE;
+                *(buffer++) = '\0';
+                buffer_size = 0;
+            }
+            else {
+                if (buffer_size >= arg_size_max) {
+                    kernaux_strncpy(error_msg, "arg too long", 12);
+                    goto fail;
+                }
+
+                *(buffer++) = cur;
+                ++buffer_size;
+            }
+            break;
         }
 
         if (state == FINAL) {

tests/test_cmdline.c

@@ -25,10 +25,21 @@ static void test(
 
 static const char *const argv0[] = {};
 
+static const char *const argv_empty[] = {""};
+static const char *const argv_empty_X2[] = {"", ""};
+static const char *const argv_empty_X3[] = {"", "", ""};
+
 static const char *const argv_foo[] = {"foo"};
 static const char *const argv_foo_bar[] = {"foo", "bar"};
 static const char *const argv_foo_bar_car[] = {"foo", "bar", "car"};
 
+static const char *const argv_spaceX3_X3[] = {"   ", "   ", "   "};
+static const char *const argv_backslashX3_X3[] = {"\\\\\\", "\\\\\\", "\\\\\\"};
+static const char *const argv_quotmarkX3_X3[] = {"\"\"\"", "\"\"\"", "\"\"\""};
+
+static const char *const argv_foospacebar_car[] = {"foo bar", "car"};
+static const char *const argv_foo_barspacecar[] = {"foo", "bar car"};
+
 static const char *const argv_space[] = {" "};
 static const char *const argv_backslash[] = {"\\"};
 static const char *const argv_quotmark[] = {"\""};
@@ -63,10 +74,6 @@ static const char *const argv_foo_spacebarspace[] = {"foo", " bar "};
 static const char *const argv_foo_backslashbarbackslash[] = {"foo", "\\bar\\"};
 static const char *const argv_foo_quotmarkbarquotmark[] = {"foo", "\"bar\""};
 
-static const char *const argv_spaceX3_X3[] = {"   ", "   ", "   "};
-static const char *const argv_backslashX3_X3[] = {"\\\\\\", "\\\\\\", "\\\\\\"};
-static const char *const argv_quotmarkX3_X3[] = {"\"\"\"", "\"\"\"", "\"\"\""};
-
 int main()
 {
     test("",             0, 0, true, "", 0, argv0);
@@ -82,6 +89,20 @@ int main()
     test("  foo  bar  ", 0, 0, true, "", 2, argv_foo_bar);
     test("foo bar car",  0, 0, true, "", 3, argv_foo_bar_car);
 
+    test("\"\"",                    0, 0, true, "", 1, argv_empty);
+    test("\"\" \"\"",               0, 0, true, "", 2, argv_empty_X2);
+    test("\"\" \"\" \"\"",          0, 0, true, "", 3, argv_empty_X3);
+    test("\"foo\"",                 0, 0, true, "", 1, argv_foo);
+    test("\"foo\" \"bar\"",         0, 0, true, "", 2, argv_foo_bar);
+    test(" \"foo\" \"bar\"",        0, 0, true, "", 2, argv_foo_bar);
+    test("\"foo\" \"bar\" ",        0, 0, true, "", 2, argv_foo_bar);
+    test(" \"foo\" \"bar\" ",       0, 0, true, "", 2, argv_foo_bar);
+    test("\"foo\"  \"bar\"",        0, 0, true, "", 2, argv_foo_bar);
+    test("\"foo\"  \"bar\"  ",      0, 0, true, "", 2, argv_foo_bar);
+    test("  \"foo\"  \"bar\"",      0, 0, true, "", 2, argv_foo_bar);
+    test("  \"foo\"  \"bar\"  ",    0, 0, true, "", 2, argv_foo_bar);
+    test("\"foo\" \"bar\" \"car\"", 0, 0, true, "", 3, argv_foo_bar_car);
+
     test("foo bar car", 3, 0, true, "", 3, argv_foo_bar_car);
     test("foo bar car", 0, 4, true, "", 3, argv_foo_bar_car);
     test("foo bar car", 3, 4, true, "", 3, argv_foo_bar_car);
@@ -90,6 +111,14 @@ int main()
     test("foo bar car", 0, 3, false, "arg too long",  0, argv0);
     test("foo bar car", 2, 3, false, "arg too long",  0, argv0);
 
+    test("\"foo\" \"bar\" \"car\"", 3, 0, true, "", 3, argv_foo_bar_car);
+    test("\"foo\" \"bar\" \"car\"", 0, 4, true, "", 3, argv_foo_bar_car);
+    test("\"foo\" \"bar\" \"car\"", 3, 4, true, "", 3, argv_foo_bar_car);
+
+    test("\"foo\" \"bar\" \"car\"", 2, 0, false, "too many args", 0, argv0);
+    test("\"foo\" \"bar\" \"car\"", 0, 3, false, "arg too long",  0, argv0);
+    test("\"foo\" \"bar\" \"car\"", 2, 3, false, "arg too long",  0, argv0);
+
     test("\\ ",             0, 0, true, "", 1, argv_space);
     test("\\\\",            0, 0, true, "", 1, argv_backslash);
     test("\\\"",            0, 0, true, "", 1, argv_quotmark);
@@ -124,6 +153,13 @@ int main()
     test("foo \\\\bar\\\\", 0, 0, true, "", 2, argv_foo_backslashbarbackslash);
     test("foo \\\"bar\\\"", 0, 0, true, "", 2, argv_foo_quotmarkbarquotmark);
 
+    test("foo\\ bar car",       0, 0, true, "", 2, argv_foospacebar_car);
+    test("\"foo bar\" car",     0, 0, true, "", 2, argv_foospacebar_car);
+    test("\"foo bar\" \"car\"", 0, 0, true, "", 2, argv_foospacebar_car);
+    test("foo bar\\ car",       0, 0, true, "", 2, argv_foo_barspacecar);
+    test("foo \"bar car\"",     0, 0, true, "", 2, argv_foo_barspacecar);
+    test("\"foo\" \"bar car\"", 0, 0, true, "", 2, argv_foo_barspacecar);
+
     test("\\ \\ \\  \\ \\ \\  \\ \\ \\ ",          3, 0, true, "", 3, argv_spaceX3_X3);
     test("\\\\\\\\\\\\ \\\\\\\\\\\\ \\\\\\\\\\\\", 3, 0, true, "", 3, argv_backslashX3_X3);
     test("\\\"\\\"\\\" \\\"\\\"\\\" \\\"\\\"\\\"", 3, 0, true, "", 3, argv_quotmarkX3_X3);