First pass at formatter

5 files changed, 116 insertions, 28 deletions

diff --git a/src/formatter.c b/src/formatter.c
index 81a5e126..83d1b7a0 100644
--- a/src/formatter.c
+++ b/src/formatter.c
@@ -1,23 +1,87 @@
 // This code defines functions for transforming ASTs back into Tomo source text
 
+#include <gc.h>
+#include <setjmp.h>
+#include <string.h>
+
 #include "ast.h"
+#include "formatter.h"
+#include "parse/context.h"
+#include "parse/files.h"
+#include "parse/utils.h"
 #include "stdlib/datatypes.h"
 #include "stdlib/optionals.h"
+#include "stdlib/tables.h"
 #include "stdlib/text.h"
 
-Text_t code_format(ast_t *ast, Table_t *comments) {
+OptionalText_t next_comment(Table_t comments, const char **pos, const char *end) {
+    for (const char *p = *pos; p < end; p++) {
+        const char **comment_end = Table$get(comments, &p, parse_comments_info);
+        if (comment_end) {
+            *pos = *comment_end;
+            return Text$from_strn(p, (int64_t)(*comment_end - p));
+        }
+    }
+    return NONE_TEXT;
+}
+
+Text_t format_code(ast_t *ast, Table_t comments) {
     (void)comments;
     switch (ast->tag) {
-    default: return Text$from_strn(ast->start, (int64_t)(ast->end - ast->start));
+    default: {
+        Text_t code = Text$from_strn(ast->start, (int64_t)(ast->end - ast->start));
+        return Text$replace(code, Text("\t"), Text("    "));
+    }
     }
 }
 
-OptionalText_t code_format_inline(ast_t *ast, Table_t *comments) {
-    (void)comments;
+OptionalText_t format_inline_code(ast_t *ast, Table_t comments) {
     for (const char *p = ast->start; p < ast->end; p++) {
         if (*p == '\n' || *p == '\r') return NONE_TEXT;
     }
+    const char *pos = ast->start;
+    OptionalText_t comment = next_comment(comments, &pos, ast->end);
+    if (comment.length >= 0) return NONE_TEXT;
     switch (ast->tag) {
     default: return Text$from_strn(ast->start, (int64_t)(ast->end - ast->start));
     }
 }
+
+Text_t format_file(const char *path) {
+    file_t *file = load_file(path);
+    if (!file) return EMPTY_TEXT;
+
+    jmp_buf on_err;
+    if (setjmp(on_err) != 0) {
+        return Text$from_str(file->text);
+    }
+    parse_ctx_t ctx = {
+        .file = file,
+        .on_err = &on_err,
+        .comments = {},
+    };
+
+    const char *pos = file->text;
+    if (match(&pos, "#!")) // shebang
+        some_not(&pos, "\r\n");
+
+    whitespace(&ctx, &pos);
+    ast_t *ast = parse_file_body(&ctx, pos);
+    if (!ast) return Text$from_str(file->text);
+    pos = ast->end;
+    whitespace(&ctx, &pos);
+    if (pos < file->text + file->len && *pos != '\0') {
+        return Text$from_str(file->text);
+    }
+
+    const char *fmt_pos = file->text;
+    Text_t code = EMPTY_TEXT;
+    for (OptionalText_t comment; (comment = next_comment(ctx.comments, &fmt_pos, ast->start)).length > 0;) {
+        code = Texts(code, comment, "\n");
+    }
+    code = Texts(code, format_code(ast, ctx.comments));
+    for (OptionalText_t comment; (comment = next_comment(ctx.comments, &fmt_pos, ast->start)).length > 0;) {
+        code = Texts(code, comment, "\n");
+    }
+    return code;
+}
diff --git a/src/formatter.h b/src/formatter.h
index c3941de9..dd4c7815 100644
--- a/src/formatter.h
+++ b/src/formatter.h
@@ -5,5 +5,7 @@
 #include "ast.h"
 #include "stdlib/datatypes.h"
 
-Text_t code_format(ast_t *ast, Table_t *comments);
-OptionalText_t code_format_inline(ast_t *ast, Table_t *comments);
+Text_t format_file(const char *path);
+Text_t format_code(ast_t *ast, Table_t comments);
+OptionalText_t format_inline_code(ast_t *ast, Table_t comments);
+OptionalText_t next_comment(Table_t comments, const char **pos, const char *end);
diff --git a/src/parse/context.c b/src/parse/context.c
index 689fa1a4..cd8d16bc 100644
--- a/src/parse/context.c
+++ b/src/parse/context.c
@@ -5,4 +5,4 @@
 #include "../stdlib/tables.h"
 #include "../stdlib/types.h"
 
-TypeInfo_t *parse_comments_info = Table$info(Pointer$info("@", &Memory$info), Pointer$info("@", &Memory$info));
+const TypeInfo_t *parse_comments_info = Table$info(Pointer$info("@", &Memory$info), Pointer$info("@", &Memory$info));
diff --git a/src/parse/context.h b/src/parse/context.h
index 4d519a83..f1e3be2f 100644
--- a/src/parse/context.h
+++ b/src/parse/context.h
@@ -8,7 +8,7 @@
 #include "../stdlib/files.h"
 #include "../stdlib/types.h"
 
-const TypeInfo_t *parse_comments_info;
+extern const TypeInfo_t *parse_comments_info;
 
 typedef struct {
     file_t *file;
diff --git a/src/tomo.c b/src/tomo.c
index d02cce74..36a47b7f 100644
--- a/src/tomo.c
+++ b/src/tomo.c
@@ -19,6 +19,7 @@
 #include "compile/files.h"
 #include "compile/headers.h"
 #include "config.h"
+#include "formatter.h"
 #include "modules.h"
 #include "naming.h"
 #include "parse/files.h"
@@ -74,9 +75,9 @@ static const char *paths_str(List_t paths) {
 
 static OptionalList_t files = NONE_LIST, args = NONE_LIST, uninstall = NONE_LIST, libraries = NONE_LIST;
 static OptionalBool_t verbose = false, quiet = false, show_version = false, show_parse_tree = false,
-                      show_prefix = false, stop_at_transpile = false, stop_at_obj_compilation = false,
-                      compile_exe = false, should_install = false, clean_build = false, source_mapping = true,
-                      show_changelog = false;
+                      do_format_code = false, show_prefix = false, stop_at_transpile = false,
+                      stop_at_obj_compilation = false, compile_exe = false, should_install = false, clean_build = false,
+                      source_mapping = true, show_changelog = false;
 
 static OptionalText_t show_codegen = NONE_TEXT,
                       cflags = Text("-Werror -fdollars-in-identifiers -std=c2x -Wno-trigraphs "
@@ -185,23 +186,38 @@ int main(int argc, char *argv[]) {
                         "  --optimization|-O <level>: set optimization level\n"
                         "  --run|-r: run a program from " TOMO_PREFIX "/share/tomo_" TOMO_VERSION "/installed\n");
     Text_t help = Texts(Text("\x1b[1mtomo\x1b[m: a compiler for the Tomo programming language"), Text("\n\n"), usage);
-    tomo_parse_args(
-        argc, argv, usage, help, TOMO_VERSION, {"files", true, List$info(&Path$info), &files},
-        {"args", true, List$info(&Text$info), &args}, {"verbose", false, &Bool$info, &verbose},
-        {"v", false, &Bool$info, &verbose}, {"version", false, &Bool$info, &show_version},
-        {"parse", false, &Bool$info, &show_parse_tree}, {"p", false, &Bool$info, &show_parse_tree},
-        {"prefix", false, &Bool$info, &show_prefix}, {"quiet", false, &Bool$info, &quiet},
-        {"q", false, &Bool$info, &quiet}, {"transpile", false, &Bool$info, &stop_at_transpile},
-        {"t", false, &Bool$info, &stop_at_transpile}, {"compile-obj", false, &Bool$info, &stop_at_obj_compilation},
-        {"c", false, &Bool$info, &stop_at_obj_compilation}, {"compile-exe", false, &Bool$info, &compile_exe},
-        {"e", false, &Bool$info, &compile_exe}, {"uninstall", false, List$info(&Text$info), &uninstall},
-        {"u", false, List$info(&Text$info), &uninstall}, {"library", false, List$info(&Path$info), &libraries},
-        {"L", false, List$info(&Path$info), &libraries}, {"show-codegen", false, &Text$info, &show_codegen},
-        {"C", false, &Text$info, &show_codegen}, {"install", false, &Bool$info, &should_install},
-        {"I", false, &Bool$info, &should_install}, {"optimization", false, &Text$info, &optimization},
-        {"O", false, &Text$info, &optimization}, {"force-rebuild", false, &Bool$info, &clean_build},
-        {"f", false, &Bool$info, &clean_build}, {"source-mapping", false, &Bool$info, &source_mapping},
-        {"m", false, &Bool$info, &source_mapping}, {"changelog", false, &Bool$info, &show_changelog}, );
+    tomo_parse_args(argc, argv, usage, help, TOMO_VERSION, //
+                    {"files", true, List$info(&Path$info), &files}, //
+                    {"args", true, List$info(&Text$info), &args}, //
+                    {"verbose", false, &Bool$info, &verbose}, //
+                    {"v", false, &Bool$info, &verbose}, //
+                    {"version", false, &Bool$info, &show_version}, //
+                    {"parse", false, &Bool$info, &show_parse_tree}, //
+                    {"p", false, &Bool$info, &show_parse_tree}, //
+                    {"format", false, &Bool$info, &do_format_code}, //
+                    {"prefix", false, &Bool$info, &show_prefix}, //
+                    {"quiet", false, &Bool$info, &quiet}, //
+                    {"q", false, &Bool$info, &quiet}, //
+                    {"transpile", false, &Bool$info, &stop_at_transpile}, //
+                    {"t", false, &Bool$info, &stop_at_transpile}, //
+                    {"compile-obj", false, &Bool$info, &stop_at_obj_compilation}, //
+                    {"c", false, &Bool$info, &stop_at_obj_compilation}, //
+                    {"compile-exe", false, &Bool$info, &compile_exe}, //
+                    {"e", false, &Bool$info, &compile_exe}, //
+                    {"uninstall", false, List$info(&Text$info), &uninstall}, //
+                    {"u", false, List$info(&Text$info), &uninstall}, //
+                    {"library", false, List$info(&Path$info), &libraries}, //
+                    {"L", false, List$info(&Path$info), &libraries}, //
+                    {"show-codegen", false, &Text$info, &show_codegen}, //
+                    {"C", false, &Text$info, &show_codegen}, //
+                    {"install", false, &Bool$info, &should_install}, //
+                    {"I", false, &Bool$info, &should_install}, //
+                    {"optimization", false, &Text$info, &optimization}, //
+                    {"O", false, &Text$info, &optimization}, //
+                    {"force-rebuild", false, &Bool$info, &clean_build}, {"f", false, &Bool$info, &clean_build}, //
+                    {"source-mapping", false, &Bool$info, &source_mapping},
+                    {"m", false, &Bool$info, &source_mapping}, //
+                    {"changelog", false, &Bool$info, &show_changelog}, );
 
     if (show_prefix) {
         print(TOMO_PREFIX);
@@ -299,6 +315,12 @@ int main(int argc, char *argv[]) {
             continue;
         }
 
+        if (do_format_code) {
+            Text_t formatted = format_file(Path$as_c_string(path));
+            print(formatted);
+            continue;
+        }
+
         Path_t exe_path = compile_exe ? Path$with_extension(path, Text(""), true)
                                       : build_file(Path$with_extension(path, Text(""), true), "");