From a2e5f9e1c9f36d9e85577095e1b13fea2dfb827d Mon Sep 17 00:00:00 2001
From: Jan Lindemann <jan@janware.com>
Date: Sun, 5 Nov 2017 18:30:57 +0100
Subject: [PATCH] grammar.py: Fix duplicate and missing rules

- unroll_lists() sometimes introduced identical lists into a rule set
- unroll_options() sometimes chopped off a production's last rule

Signed-off-by: Jan Lindemann <jan@janware.com>
---
 tools/python/jwutils/grammar.py | 34 +++++++++++++++++++++++++++++++--
 1 file changed, 32 insertions(+), 2 deletions(-)

diff --git a/tools/python/jwutils/grammar.py b/tools/python/jwutils/grammar.py
index e604383..51e83dc 100644
--- a/tools/python/jwutils/grammar.py
+++ b/tools/python/jwutils/grammar.py
@@ -418,6 +418,18 @@ def split_list_by_regex(l_, regex):
     l = copy.deepcopy(l_)
     return [list(x[1]) for x in itertools.groupby(l, lambda x: re.match(regex, x)) if not x[0]]
 
+def remove_duplicate_rules(rules):
+    r = []
+    for rule in rules:
+        if rule in r:
+            continue
+        r.append(rule)
+    slog(DEBUG, "rules after removing duplicates >")
+    for rule in rules:
+        slog(DEBUG, "-> " + format_rule(rule))
+    slog(DEBUG, "rules after removing duplicates <")
+    return r
+
 def grammar_tokenize_ebnf(content):
     r = []
     c = ''
@@ -627,6 +639,7 @@ def grammar_fix_extensions(grammar, mode):
 
 def grammar_unroll_lists(grammar):
     delimiters = [ '","', '";"', '"|"' ] # TODO: this could be a function parameter to make it generic
+    newrule = None
     for tok, p in grammar.iteritems():
         newrules = []
         for rule in p.rules:
@@ -677,13 +690,22 @@ def grammar_unroll_lists(grammar):
                     listrule.append(c)
                     continue
                 newrule.append(c)
+            slog(DEBUG, "appending " + format_rule(newrule))
             newrules.append(newrule)
-        grammar[tok].rules = newrules
+            newrule = None
+        else:
+            if newrule is not None:
+                slog(DEBUG, "appending " + format_rule(newrule))
+                newrules.append(newrule)
+                newrule = None
+        slog(DEBUG, "done processing rules for " + tok)
+        grammar[tok].rules = remove_duplicate_rules(newrules)
     return grammar
 
 def rules_unroll_options(rules):
     r = []
     found = False
+    newrule = None
     slog(DEBUG, "unrolling", format_rules(rules))
     for rule in rules:
         square = 0
@@ -733,13 +755,19 @@ def rules_unroll_options(rules):
             break
         if not found:
             r.append(newrule)
+            newrule = None
+    else:
+        if newrule is not None:
+            slog(DEBUG, "appending " + format_rule(newrule))
+            r.append(newrule)
+            newrule = None
     if found:
         return rules_unroll_options(r)
     return r
 
 def grammar_unroll_options(grammar):
     for tok, p in grammar.iteritems():
-        grammar[tok].rules = rules_unroll_options(p.rules)
+        grammar[tok].rules = remove_duplicate_rules(rules_unroll_options(p.rules))
     return grammar
 
 def rules_unroll_alternatives(rules):
@@ -1208,6 +1236,7 @@ def grammar_create_y(grammar, opts):
     out += textwrap.dedent("""\
 
         %define parse.error verbose
+        // %define lr.type ielr
         %define api.pure full
         %param { struct context *context } { void *scanner }
     """)
@@ -1673,6 +1702,7 @@ class GrammarCmd(jwutils.Cmd):
             grammar = grammar_unroll_alternatives(grammar)
         if args.unroll_options:
             grammar = grammar_unroll_options(grammar)
+        #grammar['logical_expression'].dump(ERR)
         if len(args.check_symbols):
             check_symbols = []
             if args.check_symbols == 'all':