feat(lexer): quote handling

src/lexer/lexer.c

@@ -224,6 +224,36 @@ char *stream_init(void)
     return stream;
 }
 
+/*
+ * @brief: Updates the lexing_mode to LEXER_NORMAL
+ *         if the SECOND quote is found at stream[i].
+ *         Updates the lexing_mode to the corresponding quote type
+ *         if the FIRST quote of any type is found.
+ *
+ * @return: true if an update was done. false otherwise.
+ */
+static bool update_lexing_mode(char *stream, ssize_t i,
+                               enum lexing_mode *lexing_mode)
+{
+    enum lexing_mode mode_before_update = *lexing_mode;
+    // SECOND quote
+    if (*lexing_mode == LEXER_QUOTE && stream[i] == '\'')
+        *lexing_mode = LEXER_NORMAL;
+    if (*lexing_mode == LEXER_DOUBLE_QUOTE && stream[i] == '"')
+        *lexing_mode = LEXER_NORMAL;
+
+    // FIRST quote
+    if (*lexing_mode == LEXER_NORMAL)
+    {
+        if (stream[i] == '"')
+            *lexing_mode = LEXER_DOUBLE_QUOTE;
+
+        if (stream[i] == '\'')
+            *lexing_mode = LEXER_QUOTE;
+    }
+    return *lexing_mode != mode_before_update;
+}
+
 struct token *peek_token(void)
 {
     // we already created the upcoming token during the previous call to peek()
@@ -233,10 +263,15 @@ struct token *peek_token(void)
     }
 
     char *stream = stream_init();
-
     ssize_t i = 0;
 
+    // Usefull to know if we are inside a quote or double quote
+    enum lexing_mode lexing_mode = LEXER_NORMAL;
+
     while (i < remaining_chars)
+    {
+        // true if we didn't encounter a quotes of any type at stream[i]
+        if (!update_lexing_mode(stream, i, &lexing_mode))
         {
             if (is_special_char(stream[i]))
             {
@@ -248,6 +283,7 @@ struct token *peek_token(void)
             {
                 break;
             }
+        }
         i++;
     }
 
@@ -266,10 +302,15 @@ struct token *pop_token(void)
         return NULL;
     }
     char *stream = stream_init();
-
     ssize_t i = 0;
 
+    // Usefull to know if we are inside a quote or double quote
+    enum lexing_mode lexing_mode = LEXER_NORMAL;
+
     while (i < remaining_chars)
+    {
+        // true if we didn't encounter a quotes of any type at stream[i]
+        if (!update_lexing_mode(stream, i, &lexing_mode))
         {
             if (is_special_char(stream[i]))
             {
@@ -281,6 +322,7 @@ struct token *pop_token(void)
             {
                 break;
             }
+        }
         i++;
     }
 

src/lexer/lexer.h

@@ -3,6 +3,13 @@
 
 #include <sys/types.h>
 
+enum lexing_mode
+{
+    LEXER_NORMAL,
+    LEXER_QUOTE,
+    LEXER_DOUBLE_QUOTE
+};
+
 enum token_type
 {
     // Special characters
@@ -10,8 +17,12 @@ enum token_type
     TOKEN_EOF,
     TOKEN_WORD,
     TOKEN_NEWLINE,
+
+    // WARNING: quote and double quote should never be used inside a token.
+    // Its only use is to know if we are inside a quote, and which type of quote
     TOKEN_QUOTE,
     TOKEN_DOUBLE_QUOTE,
+
     TOKEN_GRAVE,
     TOKEN_SEMICOLON,
     TOKEN_COMMENT,
@@ -43,8 +54,6 @@ struct token
 
 /*
  * @brief: returns the next (newly allocated) token without consuming it.
- *         if end of input is reached, enters in EOF looping node,
- *         returning only the same token of type TOKEN_EOF.
  *         if end of input is reached, returns a token of type TOKEN_EOF.
  */
 struct token *peek_token(void);
@@ -57,7 +66,6 @@ struct token *peek_token(void);
  * @warning: if the last returned token was a token EOF, it frees it
  *          and returns NULL. This means that after peeking a token EOF
  *          in the parser, there must be EXACTLY ONE call to pop_token().
- *
  */
 struct token *pop_token(void);