fixup! Allow to parse macro identifiers in variable decls

Author: d-alonso

codee/patches/0020-Allow-to-parse-macro-identifiers-in-variable-decls.patch

@@ -0,0 +1,233 @@
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: Diego Alonso <[email protected]>
+Date: Thu, 26 Jun 2025 14:04:51 +0200
+Subject: Allow to parse macro identifiers in variable decls
+
+---
+ grammar.js    |   2 +
+ src/scanner.c | 128 +++++++++++++++++++++++++++++++++++++++-----------
+ 2 files changed, 103 insertions(+), 27 deletions(-)
+
+diff --git a/grammar.js b/grammar.js
+index 6e79004..40ac8b7 100644
+--- a/grammar.js
++++ b/grammar.js
+@@ -67,6 +67,7 @@ module.exports = grammar({
+     $._external_end_of_statement,
+     $._preproc_unary_operator,
+     $.hollerith_constant,
++    $.macro_identifier,
+   ],
+ 
+   extras: $ => [
+@@ -870,6 +871,7 @@ module.exports = grammar({
+         $.derived_type,
+         alias($.procedure_declaration, $.procedure),
+         $.declared_type,
++        $.macro_identifier,
+       )),
+       optional(seq(',',
+         commaSep1(
+diff --git a/src/scanner.c b/src/scanner.c
+index b768d99..e477df4 100644
+--- a/src/scanner.c
++++ b/src/scanner.c
+@@ -1,4 +1,5 @@
+ #include "tree_sitter/alloc.h"
++#include "tree_sitter/array.h"
+ #include "tree_sitter/parser.h"
+ #include <ctype.h>
+ #include <wctype.h>
+@@ -13,10 +14,12 @@ enum TokenType {
+     END_OF_STATEMENT,
+     PREPROC_UNARY_OPERATOR,
+     HOLLERITH_CONSTANT,
++    MACRO_IDENTIFIER,
+ };
+ 
+ typedef struct {
+     bool in_line_continuation;
++    Array(char *) MacroIdentifiers;
+ } Scanner;
+ 
+ typedef enum {
+@@ -301,31 +304,44 @@ static bool scan_end_line_continuation(Scanner *scanner, TSLexer *lexer) {
+     return true;
+ }
+ 
+-static bool scan_string_literal_kind(TSLexer *lexer) {
+-  // Strictly, it's allowed for the kind to be an integer literal, in
+-  // practice I've not seen it
++typedef Array(char) String;
++
++// Returns NULL on error, otherwise an allocated char array for an identifier
++static String *scan_identifier(TSLexer *lexer) {
+   if (!iswalpha(lexer->lookahead)) {
++    return NULL;
++  }
++  String *possible_identifier = ts_calloc(1, sizeof(String));
++  while (is_identifier_char(lexer->lookahead) && !lexer->eof(lexer)) {
++    array_push(possible_identifier, lexer->lookahead);
++    // Don't capture the trailing underscore as part of the kind identifier
++    // If another user of this function wants to mark the end again after
++    // the identifier they're free to do so
++    if (lexer->lookahead == '_') {
++      lexer->mark_end(lexer);
++    }
++    advance(lexer);
++  }
++  if (possible_identifier->size == 0) {
++    array_delete(possible_identifier);
++    ts_free(possible_identifier);
++    return NULL;
++  }
++  return possible_identifier;
++}
++
++static bool scan_string_literal_kind(TSLexer *lexer, String *identifier) {
++  if (identifier->size == 0) {
++    return false;
++  }
++
++  char last_char = identifier->contents[identifier->size - 1];
++  if ((last_char != '_') ||
++      (lexer->lookahead != '"' && lexer->lookahead != '\'')) {
+     return false;
+   }
+ 
+   lexer->result_symbol = STRING_LITERAL_KIND;
+-
+-  // We need two characters of lookahead to see `_"`
+-  char current_char = '\0';
+-
+-  while (is_identifier_char(lexer->lookahead) && !lexer->eof(lexer)) {
+-      current_char = lexer->lookahead;
+-      // Don't capture the trailing underscore as part of the kind identifier
+-      if (lexer->lookahead == '_') {
+-          lexer->mark_end(lexer);
+-      }
+-      advance(lexer);
+-  }
+-
+-  if ((current_char != '_') || (lexer->lookahead != '"' && lexer->lookahead != '\'')) {
+-    return false;
+-  }
+-
+   return true;
+ }
+ 
+@@ -393,6 +409,28 @@ static bool scan_string_literal(TSLexer *lexer) {
+     return false;
+ }
+ 
++static bool scan_macro_identifier(Scanner *scanner, TSLexer *lexer,
++                                  String *identifier) {
++  unsigned num_macro_ids = scanner->MacroIdentifiers.size;
++  if (num_macro_ids == 0) {
++    return false;
++  }
++
++  for (size_t i = 0, end = scanner->MacroIdentifiers.size; i < end; ++i) {
++    char *macro_id = *array_get(&scanner->MacroIdentifiers, i);
++    unsigned macro_id_len = strlen(macro_id);
++    if (identifier->size != macro_id_len) {
++      continue;
++    }
++    if (strncmp(macro_id, identifier->contents, identifier->size) == 0) {
++      lexer->mark_end(lexer);
++      lexer->result_symbol = MACRO_IDENTIFIER;
++      return true;
++    }
++  }
++  return false;
++}
++
+ /// Need an external scanner to catch '!' before its parsed as a comment
+ static bool scan_preproc_unary_operator(TSLexer *lexer) {
+   const char next_char = lexer->lookahead;
+@@ -467,19 +505,50 @@ static bool scan(Scanner *scanner, TSLexer *lexer, const bool *valid_symbols) {
+         return true;
+     }
+ 
+-    if (valid_symbols[STRING_LITERAL_KIND]) {
++    // These symbols both scan for an identifier, we need to combine the logic
++    // and they always need to be the last to look for since we can't backtrack
++    if (valid_symbols[STRING_LITERAL_KIND] || valid_symbols[MACRO_IDENTIFIER]) {
++      String *identifier = scan_identifier(lexer);
++      bool identifier_result = false;
+       // This may need a lot of lookahead, so should (probably) always
+       // be the last token to look for
+-      if (scan_string_literal_kind(lexer)) {
++      if (identifier && valid_symbols[STRING_LITERAL_KIND]) {
++        if (scan_string_literal_kind(lexer, identifier)) {
++          identifier_result = true;
++        }
++      }
++      if (!identifier_result && identifier && valid_symbols[MACRO_IDENTIFIER]) {
++        if (scan_macro_identifier(scanner, lexer, identifier)) {
++          identifier_result = true;
++        }
++      }
++      if (identifier) {
++        ts_free(identifier);
++      }
++      if (identifier_result) {
+         return true;
+       }
+     }
+-
+     return false;
+ }
+ 
+ void *tree_sitter_fortran_external_scanner_create() {
+-    return ts_calloc(1, sizeof(bool));
++  Scanner *result = (Scanner *)ts_calloc(1, sizeof(Scanner));
++  char *macro_ids = getenv("CODEE_TS_MACRO_IDS");
++  if (!macro_ids) {
++    return result;
++  }
++  char *macro_id = strtok(macro_ids, ":");
++  Array(char *) *macroIdsResult = &result->MacroIdentifiers;
++  while (macro_id) {
++    int length = strlen(macro_id);
++    char *new_str = (char *)ts_malloc((length + 1) * sizeof(char));
++    strncpy(new_str, macro_id, length);
++    array_push(macroIdsResult, new_str);
++    // Keep splitting
++    macro_id = strtok(NULL, ":");
++  }
++  return result;
+ }
+ 
+ bool tree_sitter_fortran_external_scanner_scan(void *payload, TSLexer *lexer,
+@@ -491,8 +560,9 @@ bool tree_sitter_fortran_external_scanner_scan(void *payload, TSLexer *lexer,
+ unsigned tree_sitter_fortran_external_scanner_serialize(void *payload,
+                                                         char *buffer) {
+     Scanner *scanner = (Scanner *)payload;
+-    buffer[0] = (char)scanner->in_line_continuation;
+-    return 1;
++    unsigned size = sizeof(*scanner);
++    memcpy(buffer, scanner, size);
++    return size;
+ }
+ 
+ void tree_sitter_fortran_external_scanner_deserialize(void *payload,
+@@ -500,11 +570,17 @@ void tree_sitter_fortran_external_scanner_deserialize(void *payload,
+                                                       unsigned length) {
+     Scanner *scanner = (Scanner *)payload;
+     if (length > 0) {
+-        scanner->in_line_continuation = buffer[0];
++      unsigned size = sizeof(*scanner);
++      memcpy(scanner, buffer, size);
+     }
+ }
+ 
+ void tree_sitter_fortran_external_scanner_destroy(void *payload) {
+     Scanner *scanner = (Scanner *)payload;
++    for (size_t i = 0, end = scanner->MacroIdentifiers.size; i < end; ++i) {
++      char *str = *array_get(&scanner->MacroIdentifiers, i);
++      ts_free(str);
++    }
++    array_delete(&scanner->MacroIdentifiers);
+     ts_free(scanner);
+ }