feat(expansion): parse_var_name and tests

src/expansion/expansion.c

@@ -1,3 +1,4 @@
+#define _POSIX_C_SOURCE 200809L
 #include <ctype.h>
 #include <stdbool.h>
 #include <stdio.h>
@@ -6,13 +7,75 @@
 
 #include "../utils/ast/ast.h"
 
-// static size_t var_len(char *start)
-// {
-//     char *iter = start;
-//     while (*iter != ' ' && *iter != 0)
-//         *iter++;
-//     return iter - start;
-// }
+static bool is_var_start_char(char c)
+{
+    return isalpha(c) || c == '_';
+}
+
+static bool is_var_char(char c)
+{
+    return isalnum(c) || c == '_';
+}
+
+static bool is_special_var_char(char c)
+{
+    return c == '@' || c == '*' || c == '?' || c == '$' || isdigit(c)
+        || c == '#';
+}
+
+size_t parse_var_name(char *str, char **res)
+{
+    char *brace = NULL;
+    size_t i = 1; // skip the '$'
+
+    if (str[i] == '{' && str[i + 1] != 0 && str[i + 1] != '}')
+    {
+        if (is_special_var_char(str[i + 1]) && str[i + 2] == '}')
+        {
+            // Special variable like ${1}, ${?}
+            *res = strndup(str + i + 1, 1);
+            return 4; // length of ${X}
+        }
+
+        brace = str + i;
+        i++; // skip the '{'
+    }
+    else if (is_special_var_char(str[i]))
+    {
+        *res = strndup(str + i, 1);
+        return 2; // length of $X
+    }
+
+    if (!is_var_start_char(str[i]))
+    {
+        // Not a valid variable start
+        *res = NULL;
+        return 0;
+    }
+
+    while (1)
+    {
+        if (str[i] == '}' && *brace == '{')
+        {
+            *res = strndup(str + 2, i - 2);
+            return i + 1;
+        }
+        else if (!is_var_char(str[i]))
+        {
+            if (brace != NULL)
+            {
+                // Missing closing '}'
+                *res = NULL;
+                return 0;
+            }
+            break;
+        }
+        i++;
+    }
+
+    *res = strndup(str + 1, i - 1);
+    return i;
+}
 
 struct ast_command *expand(struct ast_command *command)
 {
@@ -34,21 +97,18 @@ struct ast_command *expand(struct ast_command *command)
         {
             if (in_quotes)
             {
-                // do nothing
+                continue; // do nothing
             }
             else if (str[i] == '\'')
             {
+                // remove quote
                 in_quotes = !in_quotes;
                 memmove(&str[i], &str[i + 1], strlen(&str[i + 1]) + 1);
+                i--;
             }
             else if (str[i] == '$' && str[i + 1] != 0 && !isspace(str[i + 1]))
             {
-                // size_t len = var_len(str + i + 1);
-                // char *end = str + i + len + 1;
-                // char c = *end;
-                // *end = 0;
-                // printf("var: %s\n", str + i + 1);
-                // *end = c;
+                // variable expansion
             }
         }
 
@@ -71,17 +131,3 @@ struct ast_command *expand(struct ast_command *command)
     }
     return command;
 }
-
-// int main()
-// {
-//     printf("Expansion module test\n");
-//     struct ast_command ast_command;
-//     // char str[] = "echo Hello $?";
-//     char str[] = "echo Hello $AE86";
-//     ast_command.command = list_append(NULL, str);
-
-//     struct ast_command *command2 = expand(&ast_command);
-//     printf("command2: %s\n", (char *)command2->command->data);
-
-//     return 0;
-// }

src/expansion/expansion.h

@@ -1,4 +1,15 @@
 #ifndef EXPANSION_H
 #define EXPANSION_H
 
+#include <stddef.h>
+
+/**
+ * Parse a variable from a string starting with '$'.
+ * @param str The input string starting with '$'. It must start with '$'.
+ * @param res Pointer to a char pointer that will be set to the extracted
+ * variable name.
+ * @return The number of characters processed in the input string.
+ */
+size_t parse_var_name(char *str, char **res);
+
 #endif /* ! EXPANSION_H */

tests/unit/expansion/parse_var.c

@@ -0,0 +1,138 @@
+#include <criterion/criterion.h>
+#include <criterion/new/assert.h>
+#include <criterion/redirect.h>
+
+#include "../../../src/expansion/expansion.h"
+
+TestSuite(parse_var_name);
+
+// char *input = "$MY$VAR";
+// char *input = "$MY$VAR$";
+// char *input = "$MY$VAR${}";
+// char *input = "$MY$VAR${1}";
+
+Test(parse_var_name, basic_variable)
+{
+    char *input = "$MY_VAR";
+    char *extracted_var = NULL;
+    size_t r = parse_var_name(input, &extracted_var);
+
+    cr_expect(r == 7);
+    cr_expect_str_eq(extracted_var, "MY_VAR");
+    free(extracted_var);
+}
+
+Test(parse_var_name, multi_basic_variable)
+{
+    char *input = "$MY$VAR";
+    char *extracted_var = NULL;
+    size_t r = parse_var_name(input, &extracted_var);
+
+    cr_expect(r == 3);
+    cr_expect_str_eq(extracted_var, "MY");
+    free(extracted_var);
+
+    input += r;
+    r = parse_var_name(input, &extracted_var);
+
+    cr_expect(r == 4);
+    cr_expect_str_eq(extracted_var, "VAR");
+    free(extracted_var);
+}
+
+Test(parse_var_name, variable_with_braces)
+{
+    char *input = "${MY_VAR}";
+    char *extracted_var = NULL;
+    size_t r = parse_var_name(input, &extracted_var);
+
+    cr_expect(r == 9);
+    cr_expect_str_eq(extracted_var, "MY_VAR");
+    free(extracted_var);
+}
+
+Test(parse_var_name, special_variable)
+{
+    char *input = "$1";
+    char *extracted_var = NULL;
+    size_t r = parse_var_name(input, &extracted_var);
+
+    cr_expect(r == 2);
+    cr_expect_str_eq(extracted_var, "1");
+    free(extracted_var);
+}
+
+Test(parse_var_name, incomplete_braces)
+{
+    char *input = "${MY_VAR";
+    char *extracted_var = NULL;
+    size_t r = parse_var_name(input, &extracted_var);
+
+    cr_expect(r == 0);
+    cr_expect(extracted_var == NULL);
+}
+
+Test(parse_var_name, empty_braces)
+{
+    char *input = "${}";
+    char *extracted_var = NULL;
+    size_t r = parse_var_name(input, &extracted_var);
+
+    cr_expect(r == 0);
+    cr_expect(extracted_var == NULL);
+}
+
+Test(parse_var_name, dollar_sign_only)
+{
+    char *input = "$";
+    char *extracted_var = NULL;
+    size_t r = parse_var_name(input, &extracted_var);
+
+    cr_expect(r == 0);
+    cr_expect(extracted_var == NULL);
+}
+
+Test(parse_var_name, variable_followed_by_dollar)
+{
+    char *input = "$MY$VAR$";
+    char *extracted_var = NULL;
+    size_t r = parse_var_name(input, &extracted_var);
+
+    cr_expect(r == 3);
+    cr_expect_str_eq(extracted_var, "MY");
+    free(extracted_var);
+
+    input += r;
+    r = parse_var_name(input, &extracted_var);
+
+    cr_expect(r == 4);
+    cr_expect_str_eq(extracted_var, "VAR");
+    free(extracted_var);
+
+    input += r;
+    r = parse_var_name(input, &extracted_var);
+
+    cr_expect(r == 0);
+    cr_expect(extracted_var == NULL);
+}
+
+Test(parse_var_name, special_variable_followed_by_text)
+{
+    char *input = "$1VAR";
+    char *extracted_var = NULL;
+    size_t r = parse_var_name(input, &extracted_var);
+
+    cr_expect(r == 2);
+    cr_expect_str_eq(extracted_var, "1");
+    free(extracted_var);
+}
+
+Test(parse_var_name, bad_variable_with_braces)
+{
+    char *input = "${1VAR}";
+    char *extracted_var = NULL;
+    size_t r = parse_var_name(input, &extracted_var);
+
+    cr_expect(r == 0);
+    cr_expect(extracted_var == NULL);
+}