diff options
Diffstat (limited to 'libbuild2/cc/lexer.cxx')
-rw-r--r-- | libbuild2/cc/lexer.cxx | 34 |
1 files changed, 25 insertions, 9 deletions
diff --git a/libbuild2/cc/lexer.cxx b/libbuild2/cc/lexer.cxx index d2be3d8..123a41e 100644 --- a/libbuild2/cc/lexer.cxx +++ b/libbuild2/cc/lexer.cxx @@ -138,10 +138,13 @@ namespace build2 using type = token_type; void lexer:: - next (token& t, xchar c, bool ignore_pp) + next (token& t, pair<xchar, bool> cf, bool ignore_pp) { - for (;; c = skip_spaces ()) + for (;; cf = skip_spaces ()) { + xchar c (cf.first); + + t.first = cf.second; t.file = &log_file_; t.line = log_line_ ? *log_line_ : c.line; t.column = c.column; @@ -197,7 +200,7 @@ namespace build2 { // Note that we keep using the passed token for buffers. // - c = skip_spaces (false); // Stop at newline. + c = skip_spaces (false).first; // Stop at newline. if (eos (c) || c == '\n') break; @@ -215,7 +218,7 @@ namespace build2 // if (!(c >= '0' && c <= '9')) { - next (t, c, false); + next (t, make_pair (c, false), false); if (t.type == type::identifier) { @@ -230,7 +233,7 @@ namespace build2 if (t.type != type::identifier || t.value != "line") continue; - c = skip_spaces (false); + c = skip_spaces (false).first; if (!(c >= '0' && c <= '9')) fail (c) << "line number expected after #line directive"; @@ -242,7 +245,7 @@ namespace build2 continue; // Parse the tail, if any. } - next (t, c, false); + next (t, make_pair (c, false), false); } break; } @@ -823,7 +826,7 @@ namespace build2 // See if we have the file. // - c = skip_spaces (false); + c = skip_spaces (false).first; if (c == '\"') { @@ -1007,16 +1010,24 @@ namespace build2 } auto lexer:: - skip_spaces (bool nl) -> xchar + skip_spaces (bool nl) -> pair<xchar, bool> { xchar c (get ()); + // Besides the first character, we also need to take into account any + // newlines that we are skipping. For example, the first character may + // be a space at the end of the line which we will skip along with the + // following newline. + // + bool first (c.column == 1); + for (; !eos (c); c = get ()) { switch (c) { case '\n': if (!nl) break; + first = true; // Fall through. case ' ': case '\t': @@ -1072,11 +1083,16 @@ namespace build2 if (!nl) break; + first = true; continue; } // C comment. // + // Note that for the first logic we consider a C comment to be + // entirely part of the same logical line even if there are + // newlines inside. + // if (p == '*') { get (p); @@ -1132,7 +1148,7 @@ namespace build2 break; } - return c; + return make_pair (c, first); } ostream& |