From 95b02a948385370092bae845a83fd5f14af82438 Mon Sep 17 00:00:00 2001
From: lemon <lsof@mailbox.org>
Date: Thu, 1 Sep 2022 09:31:49 +0200
Subject: relax block comment lexing requirements

---
 src/cffc.hff  |  1 +
 src/parse.cff | 26 ++++++++++++++++++++++++--
 test/2.cff    | 11 +++++++++++
 3 files changed, 36 insertions(+), 2 deletions(-)

diff --git a/src/cffc.hff b/src/cffc.hff
index 135e8d8..6ae2c6d 100644
--- a/src/cffc.hff
+++ b/src/cffc.hff
@@ -149,6 +149,7 @@ struct Parser {
    curloc Loc,
    eof bool,
    error bool,
+   is_comment bool,
    is_header bool,
    peekchr Option<int>,
    peektok Option<Tok>,
diff --git a/src/parse.cff b/src/parse.cff
index f5beb23..ce32475 100644
--- a/src/parse.cff
+++ b/src/parse.cff
@@ -339,6 +339,10 @@ fn lex(P *Parser) Tok {
    tok.loc = (P.tokloc = P.curloc);
    if isdigit(c = chrpeek(P)) {
       let s [80]u8 = {};
+      if P.is_comment {
+         readtilsep(P, s[0::], #t);
+         return tok;
+      }
       if readtilsep(P, s[0::], #t) < 0 {
          fatal(P, tok.loc, "bad number literal");
       }
@@ -352,6 +356,10 @@ fn lex(P *Parser) Tok {
    }
    if isalpha(c) or c == '_' or c == '$' or c > 0x7F {
       let s [120]u8;
+      if P.is_comment {
+         readtilsep(P, s[0::], #t);
+         return tok;
+      }
       if readtilsep(P, s[0::], #f) < 0 {
          fatal(P, tok.loc, "identifier too long");
       }
@@ -370,12 +378,18 @@ fn lex(P *Parser) Tok {
    }
    if c == '#' {
       let s [100]u8 = {};
+      if P.is_comment {
+         readtilhsep(P, s[0::], #t);
+         return tok;
+      }
       if readtilhsep(P, s[0::], #f) < 0 {
          fatal(P, P.tokloc, "invalid #keyword");
       }
       switch {
       case streq(s, "#") and chrpeek(P) == '{';
          chr(P);
+         let is_comment = P.is_comment;
+         P.is_comment = #t;
          let bal = 1;
          while bal != 0 {
             switch lex(P).t {
@@ -384,6 +398,7 @@ fn lex(P *Parser) Tok {
             case :eof; fatal(P, tok.loc, "unterminated comment");
             }
          }
+         P.is_comment = is_comment;
          return lex(P);
       case streq(s, "#");
          tok.t = '#';
@@ -428,9 +443,12 @@ fn lex(P *Parser) Tok {
                   delim == '"' ? "string" : "character");
          }
          if c != '\\' {
-            str->push(c);
+            if !P.is_comment {
+               str->push(c);
+            }
             continue;
          }
+         if P.is_comment { continue; }
          switch ((c = chr(P))) {
          case 0, '\n';
             fatal(P, P.tokloc, "unterminated %s literal",
@@ -452,6 +470,8 @@ fn lex(P *Parser) Tok {
          }
       }
 
+      if P.is_comment { return tok; }
+
       if delim == '"' {
          tok.t = :str;
          str->push('\0');
@@ -556,7 +576,9 @@ fn lex(P *Parser) Tok {
       tok.t = :eof;
       return tok;
    }
-   fatal(P, tok.loc, "stray %qc in program", c);
+   if !P.is_comment {
+      fatal(P, tok.loc, "stray %qc in program", c);
+   }
 }
 
 fn lexpeek(P *Parser) Tok {
diff --git a/test/2.cff b/test/2.cff
index 43fc148..65165d1 100644
--- a/test/2.cff
+++ b/test/2.cff
@@ -6,6 +6,17 @@ bitfield Foo : u16 {
    num 10 signed,
 }
 
+#{
+   Test block comment,
+   contains tokens " ",
+   this does not end the comment -> '}'
+   8ab
+   #test
+   i`m having to use backtick instead of single quote for the apostrophe in this phrase,
+   so the main use is for commenting out code,
+   and inline comments to annotate e.g. parameter names
+}
+
 extern fn main() int {
    let foo Foo = {};
    printf("0x%X: %d, %d, %d\n", foo.#raw, foo.tag, foo.flag, foo.num);
-- 
cgit v1.2.3