scanner: handle missing newline at EOF

[ocean] / csrc / scanner.mdc

diff --git a/csrc/scanner.mdc b/csrc/scanner.mdc
index 5480faaf6dddf21105fcafd383962acd6ddb703b..15306dd17494245986dbd014ebd2a2c8693c977d 100644 (file)
--- a/csrc/scanner.mdc
+++ b/csrc/scanner.mdc
@@ -106,7 +106,7 @@ token numbers from `TK_reserved` upwards.
 ### Numbers
 
 Numbers are the messiest tokens to parse, primarily because they can
-contain characters that also have meaning outside of number and,
+contain characters that also have meaning outside of numbers and,
 particularly, immediately after numbers.
 
 The obvious example is the '`-`' sign.  It can come inside a number for
@@ -260,7 +260,7 @@ and the length of the list must be given (`known_count`).
 Tokens matching these known words are reported as the index of the
 list added to `TK_reserved`.
 
-If identifiers are ignored, then any work which is not listed as a
+If identifiers are ignored, then any word which is not listed as a
 known word results in an error.
 
 ###### token config parameters
@@ -324,10 +324,17 @@ below before giving up and assuming an unknown mark.
 
 If an unknown mark contains a quote character or a comment marker, and
 that token is not being ignored, then we terminate the unknown mark
-before that quote or comment.  This ensure that an unknown mark
+before that quote or comment.  This ensures that an unknown mark
 immediately before a string is handled correctly.
 
-If `TK_mark` is ignored, then unknown marks as returned as an error.
+If the first character of a comment marker (i.e. '/') is a known mark,
+the above rules would suggest that the start of a comment would be
+parsed as that mark, which is not what is wanted.  So the introductory
+sequences for a comment ("//" and "/*") are treated as
+partially-known.  They prevent the leading "/" from being a mark by
+itself, but do not actually constitute a stand-alone mark.
+
+If `TK_mark` is ignored, then unknown marks are returned as errors.
 
 ###### token types
        TK_mark,
@@ -407,7 +414,7 @@ and continue until a matching character on the same line.  Any of
 these characters can be included in the list of known marks and then
 they will not be used for identifying strings.
 
-Immediately following the close quote one or two ASCII letters may
+Immediately following the close quote, one or two ASCII letters may
 appear.  These are somewhat like the arbitrary letters allowed in
 "Numbers" above.  They can be used by the language in various ways.
 
@@ -513,6 +520,10 @@ If `TK_string` is ignored, then quote characters will appear as `TK_mark`s.
                                break;
                        }
                }
+               while (!at_eon(state) && (ch = get_char(state)) &&
+                                         iswalpha(ch))
+                       ;
+               unget_char(state);
                close_token(state, &tk);
                return tk;
        }
@@ -894,19 +905,23 @@ a flag that tells us whether or not we need to strip.
 
 ###### internal functions
 
-       static void do_strip(struct token_state *state)
+       static int do_strip(struct token_state *state)
        {
+               int indent = 0;
                if (state->node->needs_strip) {
                        int n = 4;
                        while (n && state->node->code.txt[state->offset] == ' ') {
+                               indent += 1;
                                state->offset += 1;
                                n -= 1;
                        }
                        while (n == 4 && state->node->code.txt[state->offset] == '\t') {
+                               indent = indent_tab(indent);
                                state->offset += 1;
                                n -= 4;
                        }
                }
+               return indent;
        }
 
        static wint_t get_char(struct token_state *state)
@@ -924,9 +939,8 @@ a flag that tells us whether or not we need to strip.
                        state->offset = 0;
                        if (state->node == NULL)
                                return WEOF;
-                       do_strip(state);
                        state->line = state->node->line_no;
-                       state->col = state->node->indent;
+                       state->col = do_strip(state);
                }
 
                ## before get_char
@@ -951,8 +965,7 @@ a flag that tells us whether or not we need to strip.
                        state->col += 1;
                } else if (is_newline(next)) {
                        state->line += 1;
-                       state->col = state->node->indent;
-                       do_strip(state);
+                       state->col = do_strip(state);
                } else if (next == '\t') {
                        state->col = indent_tab(state->col);
                }
@@ -1055,8 +1068,11 @@ parsed too much already.  For that there is `reset_token`.
        static void close_token(struct token_state *state,
                                struct token *tk)
        {
-               tk->txt.len = (state->node->code.txt + state->offset)
-                             - tk->txt.txt;
+               if (state->node != tk->node)
+                       tk->txt.len = tk->node->code.len - (tk->txt.txt - tk->node->code.txt);
+               else
+                       tk->txt.len = (state->node->code.txt + state->offset)
+                                     - tk->txt.txt;
        }
 
        static void reset_token(struct token_state *state, struct token *tok)
@@ -1162,9 +1178,8 @@ As well as getting tokens, we need to be able to create the
                memset(state, 0, sizeof(*state));
                state->node = code;
                state->line = code->line_no;
-               state->col = code->indent;
+               state->col = do_strip(state);
                state->conf = conf;
-               do_strip(state);
                return state;
        }
        void token_close(struct token_state *state)
@@ -1774,7 +1789,7 @@ required indent is found.
                if (c == ' ')
                        skipped += 1;
                else if (c == '\t')
-                       skipped = indent_tab(c);
+                       skipped = indent_tab(skipped);
                else
                        break;
                i+= 1;
@@ -2022,4 +2037,3 @@ the tokens one per line.
                        libmdcode.o libnumber.o libstring.o -licuuc -lgmp
        scanner.o : scanner.c
                $(CC) $(CFLAGS) -c scanner.c
-