bpo-35766: Merge typed_ast back into CPython (GH-11645)

1 files changed, 56 insertions, 1 deletions

diff --git a/Parser/tokenizer.c b/Parser/tokenizer.c
index 3e3cf2c..1ded9ad 100644
--- a/Parser/tokenizer.c
+++ b/Parser/tokenizer.c
@@ -48,6 +48,10 @@ static int tok_nextc(struct tok_state *tok);
 static void tok_backup(struct tok_state *tok, int c);
 
 
+/* Spaces in this constant are treated as "zero or more spaces or tabs" when
+   tokenizing. */
+static const char* type_comment_prefix = "# type: ";
+
 /* Create and initialize a new tok_state structure */
 
 static struct tok_state *
@@ -82,6 +86,7 @@ tok_new(void)
     tok->decoding_readline = NULL;
     tok->decoding_buffer = NULL;
 #endif
+    tok->type_comments = 0;
 
     return tok;
 }
@@ -1245,11 +1250,61 @@ tok_get(struct tok_state *tok, char **p_start, char **p_end)
     /* Set start of current token */
     tok->start = tok->cur - 1;
 
-    /* Skip comment */
+    /* Skip comment, unless it's a type comment */
     if (c == '#') {
+        const char *prefix, *p, *type_start;
+
         while (c != EOF && c != '\n') {
             c = tok_nextc(tok);
         }
+
+        if (tok->type_comments) {
+            p = tok->start;
+            prefix = type_comment_prefix;
+            while (*prefix && p < tok->cur) {
+                if (*prefix == ' ') {
+                    while (*p == ' ' || *p == '\t') {
+                        p++;
+                    }
+                } else if (*prefix == *p) {
+                    p++;
+                } else {
+                    break;
+                }
+
+                prefix++;
+            }
+
+            /* This is a type comment if we matched all of type_comment_prefix. */
+            if (!*prefix) {
+                int is_type_ignore = 1;
+                tok_backup(tok, c);  /* don't eat the newline or EOF */
+
+                type_start = p;
+
+                is_type_ignore = tok->cur >= p + 6 && memcmp(p, "ignore", 6) == 0;
+                p += 6;
+                while (is_type_ignore && p < tok->cur) {
+                    if (*p == '#')
+                        break;
+                    is_type_ignore = is_type_ignore && (*p == ' ' || *p == '\t');
+                    p++;
+                }
+
+                if (is_type_ignore) {
+                    /* If this type ignore is the only thing on the line, consume the newline also. */
+                    if (blankline) {
+                        tok_nextc(tok);
+                        tok->atbol = 1;
+                    }
+                    return TYPE_IGNORE;
+                } else {
+                    *p_start = (char *) type_start;  /* after type_comment_prefix */
+                    *p_end = tok->cur;
+                    return TYPE_COMMENT;
+                }
+            }
+        }
     }
 
     /* Check for EOF and errors now */