From cd86cd6bdf35ff7224015fc359bb82f9d99adf40 Mon Sep 17 00:00:00 2001
From: anon <anon@anon.anon>
Date: Fri, 6 Sep 2024 16:22:52 +0200
Subject: [PATCH] +tslex
---
C_C++/bison/tslex/Makefile | 3 ++
C_C++/bison/tslex/README.md | 37 ++++++++++++++++++++++
C_C++/bison/tslex/main.y | 46 +++++++++++++++++++++++++++
C_C++/bison/tslex/tslex.c | 63 +++++++++++++++++++++++++++++++++++++
C_C++/bison/tslex/tslex.h | 16 ++++++++++
5 files changed, 165 insertions(+)
create mode 100644 C_C++/bison/tslex/Makefile
create mode 100644 C_C++/bison/tslex/README.md
create mode 100644 C_C++/bison/tslex/main.y
create mode 100644 C_C++/bison/tslex/tslex.c
create mode 100644 C_C++/bison/tslex/tslex.h
diff --git a/C_C++/bison/tslex/Makefile b/C_C++/bison/tslex/Makefile
new file mode 100644
index 0000000..5d5f0a3
--- /dev/null
+++ b/C_C++/bison/tslex/Makefile
@@ -0,0 +1,3 @@
+main:
+ bison --header=main.tab.h main.y
+ gcc main.tab.c tslex.c $$(pkg-config --cflags --libs tree-sitter tree-sitter-c) -ggdb
diff --git a/C_C++/bison/tslex/README.md b/C_C++/bison/tslex/README.md
new file mode 100644
index 0000000..938239c
--- /dev/null
+++ b/C_C++/bison/tslex/README.md
@@ -0,0 +1,37 @@
+a few days ago i saw this pet project on https://git.peppe.rs/languages/tbsp/tree/examples/code-overview/overview.tbsp
+being a DSL fan, i really like the idea.
+except seemingly the guy is making a full general purpose interpreter too,
+which sentences the project to death.
+i have email him about this (response pending)
+
+as i was thinking however i started wondering if bison could do it
+given an adaquate lex()
+
+well, the answer is kinda.
+if you look at the code, well it works.
+
+problems:
+ + bison doesnt know includes so it would have to be preprocessed
+ (to provide tokens as -perversily- the scanner defines them)
+ + the tree sitter query language is much more...
+ well lets say verboser than i imagined, so our code has to be much
+ more complexed than i hoped so;
+ either the bison programmer has to type a lot or lex()
+ define a bunch of wrappers which would be language specific
+ + we are not really using the stack are we?
+ (unless we do stuff like `BEGIN FUNCTION`/`END FUNCTION`
+ to mimic the "original" grammar (which is not bad))
+ so its a hack to begin with,
+ and because of the above reasons and ugly one of that
+
+so, instead, i do think a DSL is the answer.
+only issue is that if the guy disagrees on making it "supplementary"
+then i will have to make my own.
+
+one more thing, a ts node query language would also be nice.
+something along the lines of
+```
+$$->declarator
+```
+expanding to a node object
+(or erroring out safely if the "declarator" named field does not exist)
diff --git a/C_C++/bison/tslex/main.y b/C_C++/bison/tslex/main.y
new file mode 100644
index 0000000..6163bf2
--- /dev/null
+++ b/C_C++/bison/tslex/main.y
@@ -0,0 +1,46 @@
+%{
+ #include <stdio.h>
+ #include <string.h>
+ #include <tree_sitter/api.h>
+ #include "tslex.h"
+%}
+%token UNDEF
+%token<nodeval> FUNCTION
+%union {
+ TSNode nodeval;
+}
+%%
+document
+ : any
+ ;
+
+any
+ : %empty
+ | UNDEF any
+ | FUNCTION any {
+ printf("-- %s\n",
+ ts_node_string($1)
+ );
+
+ char * s = "declarator";
+ TSNode identifier = ts_node_child_by_field_name($1, s, strlen(s));
+
+ if (ts_node_is_null(identifier)) {
+ break;
+ }
+
+ printf("-- %.*s\n",
+ ts_node_text_len(identifier),
+ ts_node_text(identifier)
+ );
+ }
+ ;
+%%
+
+signed main() {
+ yyinit();
+ yyparse();
+ yydeinit();
+
+ return 0;
+}
diff --git a/C_C++/bison/tslex/tslex.c b/C_C++/bison/tslex/tslex.c
new file mode 100644
index 0000000..13cf8ac
--- /dev/null
+++ b/C_C++/bison/tslex/tslex.c
@@ -0,0 +1,63 @@
+#include <stdio.h>
+#include <string.h>
+#include <tree_sitter/api.h>
+
+#include "main.tab.h"
+
+extern const TSLanguage * tree_sitter_c(void);
+
+TSParser * parser;
+TSTree * tree;
+TSTreeCursor cursor;
+
+TSNode current_node;
+
+const char * source_code = "int main() { return 0; }";
+
+int yyinit() {
+ parser = ts_parser_new();
+
+ ts_parser_set_language(parser, tree_sitter_c());
+ tree = ts_parser_parse_string(parser, NULL, source_code, strlen(source_code));
+ cursor = ts_tree_cursor_new(ts_tree_root_node(tree));
+ current_node = ts_tree_root_node(tree);
+}
+
+int yydeinit() {
+ ts_tree_delete(tree);
+ ts_parser_delete(parser);
+ ts_tree_cursor_delete(&cursor);
+}
+
+#define RETURN(x) do {\
+ r = x; \
+ goto END; \
+} while(0);
+
+int yylex() {
+ int r = UNDEF;
+
+ static int end = 0; // XXX
+ if (end) {
+ return 0;
+ }
+
+ if (!strcmp("function_definition", ts_node_type(current_node))) {
+ yylval.nodeval = current_node;
+ RETURN(FUNCTION);
+ }
+
+ END:
+ if (ts_tree_cursor_goto_first_child(&cursor)
+ && ts_tree_cursor_goto_next_sibling(&cursor)
+ && ts_tree_cursor_goto_parent(&cursor)) {
+ end = 1;
+ }
+
+ current_node = ts_tree_cursor_current_node(&cursor);
+ return r;
+}
+
+void yyerror() {
+ puts("error -- this should never happen");
+}
diff --git a/C_C++/bison/tslex/tslex.h b/C_C++/bison/tslex/tslex.h
new file mode 100644
index 0000000..4de24bb
--- /dev/null
+++ b/C_C++/bison/tslex/tslex.h
@@ -0,0 +1,16 @@
+#include <tree_sitter/api.h>
+
+extern void yyerror();
+extern int yyinit();
+extern int yydeinit();
+extern int yylex();
+
+extern const char * source_code;
+
+const char * ts_node_text(const TSNode node) {
+ return source_code + ts_node_start_byte(node);
+}
+
+int ts_node_text_len(const TSNode node) {
+ return ts_node_end_byte(node) - ts_node_start_byte(node);
+}