feat: concatenate adjacent string literals

4 files changed, 45 insertions, 1 deletions

diff --git a/src/main.c b/src/main.c
index e8e1030..44a317c 100644
--- a/src/main.c
+++ b/src/main.c
@@ -32,6 +32,7 @@ int main(int argc, char** argv) {
         return 0;
     }
 
+    concat_adjacent_string_literals(pp_tokens);
     TokenArray* tokens = convert_pp_tokens_to_tokens(pp_tokens);
     Program* prog = parse(tokens);
 
diff --git a/src/preprocess.c b/src/preprocess.c
index 9605f69..b06de6d 100644
--- a/src/preprocess.c
+++ b/src/preprocess.c
@@ -1177,6 +1177,34 @@ TokenArray* preprocess(InFile* src, StrArray* included_files, StrArray* user_inc
     return do_preprocess(src, 0, macros, included_files, user_include_dirs);
 }
 
+void concat_adjacent_string_literals(TokenArray* pp_tokens) {
+    size_t last_nonempty_token_pos = 0;
+    TokenKind last_nonempty_token_kind = TokenKind_eof;
+    for (size_t pos = 0; pos < pp_tokens->len; ++pos) {
+        Token* pp_tok = &pp_tokens->data[pos];
+        TokenKind k = pp_tok->kind;
+        if (k == TokenKind_removed || k == TokenKind_whitespace || k == TokenKind_newline) {
+            continue;
+        }
+        if (k == TokenKind_literal_str && last_nonempty_token_kind == TokenKind_literal_str) {
+            // Concatenate adjacent string literals.
+            Token* last_pp_tok = &pp_tokens->data[last_nonempty_token_pos];
+            const char* s1 = last_pp_tok->value.string;
+            size_t l1 = strlen(s1);
+            const char* s2 = pp_tok->value.string;
+            size_t l2 = strlen(s2);
+            char* buf = calloc(l1 + l2 + 1, sizeof(char));
+            memcpy(buf, s1, l1);
+            memcpy(buf + l1, s2, l2);
+            last_pp_tok->value.string = buf;
+            pp_tok->kind = TokenKind_removed;
+        } else {
+            last_nonempty_token_pos = pos;
+            last_nonempty_token_kind = k;
+        }
+    }
+}
+
 void print_token_to_file(FILE* out, TokenArray* pp_tokens) {
     for (size_t i = 0; i < pp_tokens->len; ++i) {
         Token* tok = &pp_tokens->data[i];
diff --git a/src/preprocess.h b/src/preprocess.h
index a39f01c..4bf9834 100644
--- a/src/preprocess.h
+++ b/src/preprocess.h
@@ -6,6 +6,7 @@
 #include "token.h"
 
 TokenArray* preprocess(InFile* src, StrArray* included_files, StrArray* user_include_dirs);
+void concat_adjacent_string_literals(TokenArray* pp_tokens);
 void print_token_to_file(FILE* output_file, TokenArray* pp_tokens);
 
 #endif
diff --git a/tests/string_operations.sh b/tests/string_operations.sh
index 0c9a07e..2466145 100644
--- a/tests/string_operations.sh
+++ b/tests/string_operations.sh
@@ -23,6 +23,21 @@ int main() {
 EOF
 
 cat <<'EOF' > expected
+abc
+defghijkl
+EOF
+
+test_diff <<'EOF'
+int printf();
+
+int main() {
+    printf("abc\n");
+    printf("def" "ghi"
+    "jkl\n");
+}
+EOF
+
+cat <<'EOF' > expected
 h
 l
 ,
@@ -54,4 +69,3 @@ int main() {
     }
 }
 EOF
-