Fixed problem with %recover

3 years ago · 8ad71c4adc
parent 0d005fb3da
commit 8ad71c4adc
5 changed files with 68 additions and 46 deletions
--- a/docs/native.js
+++ b/docs/native.js
--- a/docs/native.wasm
+++ b/docs/native.wasm
--- a/lint/peglint.cc
+++ b/lint/peglint.cc
@ -144,7 +144,7 @@ int main(int argc, const char **argv) {
            name = peg::TraceOpeName::get(const_cast<peg::Ope &>(ope));

            auto lit = dynamic_cast<const peg::LiteralString *>(&ope);
-            if (lit) { name += " '" + lit->lit_ + "'"; }
+            if (lit) { name += " '" + peg::escape_characters(lit->lit_) + "'"; }
          }
          std::cout << "E " << pos << backtrack << "\t" << indent << "┌" << name
                    << " #" << c.trace_ids.back() << std::endl;
--- a/peglib.h
+++ b/peglib.h
@ -201,6 +201,10 @@ inline std::string escape_characters(const char *s, size_t n) {
  return str;
 }

+inline std::string escape_characters(std::string_view sv) {
+  return escape_characters(sv.data(), sv.size());
+}
+
 /*-----------------------------------------------------------------------------
 *  resolve_escape_sequence
 *---------------------------------------------------------------------------*/
@ -685,11 +689,7 @@ private:
    if (len) {
      size_t i = 0;
      int c = error_pos[i++];
-      if (std::ispunct(c)) {
-        while (i < len && std::ispunct(error_pos[i])) {
-          i++;
-        }
-      } else {
+      if (!std::ispunct(c) && !std::isspace(c)) {
        while (i < len && !std::ispunct(error_pos[i]) &&
               !std::isspace(error_pos[i])) {
          i++;
@ -2121,7 +2121,6 @@ private:
 static const char *WHITESPACE_DEFINITION_NAME = "%whitespace";
 static const char *WORD_DEFINITION_NAME = "%word";
 static const char *RECOVER_DEFINITION_NAME = "%recover";
-static const char *RECOVER_TO_DEFINITION_NAME = "%recover_to";

 /*
 * Definition
@ -3213,8 +3212,6 @@ private:
        auto ope = ref(*data.grammar, ident, vs.sv().data(), is_macro, args);
        if (ident == RECOVER_DEFINITION_NAME) {
          ope = rec(ope);
-        } else if (ident == RECOVER_TO_DEFINITION_NAME) {
-          ope = rec(ope);
        }

        if (ignore) {
@ -3401,17 +3398,6 @@ private:
        rule.is_macro = true;
        rule.params = {"x"};
      }
-
-      // `%recover_to`
-      {
-        auto &rule = grammar[RECOVER_TO_DEFINITION_NAME];
-        rule <= oom(seq(npd(ref(grammar, "x", "", false, {})), dot()));
-        rule.name = RECOVER_TO_DEFINITION_NAME;
-        rule.s_ = "[native]";
-        rule.ignoreSemanticValue = true;
-        rule.is_macro = true;
-        rule.params = {"x"};
-      }
    }

    std::any dt = &data;
--- a/test/test2.cc
+++ b/test/test2.cc
@ -1083,19 +1083,19 @@ TEST_CASE("Error recovery 1", "[error]") {

    HEADER     <- '[' _ CATEGORY (':' _  ATTRIBUTES)? ']'

-    CATEGORY   <- < [-_a-zA-Z0-9 ]+ > _
+    CATEGORY   <- < [-_a-zA-Z0-9\u0080-\uFFFF ]+ > _
    ATTRIBUTES <- ATTRIBUTE (',' _ ATTRIBUTE)*
-    ATTRIBUTE  <- < [-_a-zA-Z0-9]+ > _
+    ATTRIBUTE  <- < [-_a-zA-Z0-9\u0080-\uFFFF]+ > _

    ENTRIES    <- (ENTRY (__ ENTRY)*)?

-    ENTRY      <- ONE_WAY PHRASE ('|' _ PHRASE)*
-                / PHRASE ('|' _ PHRASE)
-                / %recover_to(__ / HEADER)
+    ENTRY      <- ONE_WAY PHRASE ('|' _ PHRASE)* !'='
+                / PHRASE ('|' _ PHRASE)+ !'='
+                / %recover((!(__ / HEADER) .)+)

    ONE_WAY    <- PHRASE '=' _
    PHRASE     <- WORD (' ' WORD)* _
-    WORD       <- < (![ \t\r\n=|[#] .)+ >
+    WORD       <- < (![ \t\r\n=|[\]#] .)+ >

    ~__        <- _ (comment? nl _)+
    ~_         <- [ \t]*
@ -1107,9 +1107,10 @@ TEST_CASE("Error recovery 1", "[error]") {
  REQUIRE(!!pg); // OK

  std::vector<std::string> errors{
-    R"(2:6: syntax error, unexpected '|', expecting <WORD>.)",
-    R"(4:4: syntax error, unexpected '\n', expecting <WORD>.)",
-    R"(8:4: syntax error, unexpected '\n', expecting <WORD>.)"
+    R"(3:6: syntax error, unexpected '|', expecting <WORD>.)",
+    R"(7:4: syntax error, unexpected '\n', expecting <WORD>.)",
+    R"(10:1: syntax error, unexpected '[', expecting <WORD>.)",
+    R"(18:17: syntax error, unexpected '=', expecting <ENTRY>, <WORD>.)",
  };

  size_t i = 0;
@ -1122,16 +1123,27 @@ TEST_CASE("Error recovery 1", "[error]") {
  pg.enable_ast();

  std::shared_ptr<Ast> ast;
-  REQUIRE_FALSE(pg.parse(R"([Section1]
+  REQUIRE_FALSE(pg.parse(R"([Section 1]
+111 = 222 | 333
 aaa || bbb
 ccc = ddd
+
+[Section 2]
 eee
+fff | ggg
+
+[Section 3
+hhh | iii

-[Section2]
+[Section 日本語]
+ppp | qqq

-fff
+[Section 4]
+jjj | kkk
+lll = mmm | nnn = ooo

-ggg hhh | iii
+[Section 5]
+rrr | sss

  )", ast));

@ -1140,30 +1152,54 @@ ggg hhh | iii
  REQUIRE(ast_to_s(ast) ==
 R"(+ START
  + SECTION
-    - HEADER/0[CATEGORY] (Section1)
+    - HEADER/0[CATEGORY] (Section 1)
    + ENTRIES
+      + ENTRY/0
+        - ONE_WAY/0[WORD] (111)
+        - PHRASE/0[WORD] (222)
+        - PHRASE/0[WORD] (333)
      + ENTRY/2
      + ENTRY/0
        - ONE_WAY/0[WORD] (ccc)
        - PHRASE/0[WORD] (ddd)
-      + ENTRY/2
  + SECTION
-    - HEADER/0[CATEGORY] (Section2)
+    - HEADER/0[CATEGORY] (Section 2)
    + ENTRIES
      + ENTRY/2
      + ENTRY/1
-        + PHRASE
-          - WORD (ggg)
-          - WORD (hhh)
+        - PHRASE/0[WORD] (fff)
+        - PHRASE/0[WORD] (ggg)
+      + ENTRY/2
+      + ENTRY/1
+        - PHRASE/0[WORD] (hhh)
        - PHRASE/0[WORD] (iii)
+  + SECTION
+    - HEADER/0[CATEGORY] (Section 日本語)
+    + ENTRIES
+      + ENTRY/1
+        - PHRASE/0[WORD] (ppp)
+        - PHRASE/0[WORD] (qqq)
+  + SECTION
+    - HEADER/0[CATEGORY] (Section 4)
+    + ENTRIES
+      + ENTRY/1
+        - PHRASE/0[WORD] (jjj)
+        - PHRASE/0[WORD] (kkk)
+      + ENTRY/2
+  + SECTION
+    - HEADER/0[CATEGORY] (Section 5)
+    + ENTRIES
+      + ENTRY/1
+        - PHRASE/0[WORD] (rrr)
+        - PHRASE/0[WORD] (sss)
 )");
 }

 TEST_CASE("Error recovery 2", "[error]") {
  parser pg(R"(
-    START <- ENTRY ((',' ENTRY) / %recover_to(',' / Space))* (_ / %recover_to('!.'))
+    START <- ENTRY ((',' ENTRY) / %recover((!(',' / Space) .)+))* (_ / %recover((!'!.' .)+))
    ENTRY <- '[' ITEM (',' ITEM)* ']'
-    ITEM  <- WORD / NUM / %recover_to(',' / ']')
+    ITEM  <- WORD / NUM / %recover((!(',' / ']') .)+)
    NUM   <- [0-9]+ ![a-z]
    WORD  <- '"' [a-z]+ '"'

@ -1174,14 +1210,14 @@ TEST_CASE("Error recovery 2", "[error]") {
  REQUIRE(!!pg); // OK

  std::vector<std::string> errors{
-    R"(1:6: syntax error, unexpected '],['.)",
+    R"(1:6: syntax error, unexpected ']'.)",
    R"(1:18: syntax error, unexpected 'z', expecting <NUM>.)",
-    R"(1:24: syntax error, unexpected ',"', expecting <WORD>.)",
+    R"(1:24: syntax error, unexpected ',', expecting <WORD>.)",
    R"(1:31: syntax error, unexpected 'ccc', expecting <NUM>.)",
    R"(1:38: syntax error, unexpected 'ddd', expecting <NUM>.)",
-    R"(1:55: syntax error, unexpected '],[', expecting <WORD>.)",
+    R"(1:55: syntax error, unexpected ']', expecting <WORD>.)",
    R"(1:58: syntax error, unexpected '\n', expecting <NUM>.)",
-    R"(1:56: syntax error, unexpected ',[', expecting <Space>.)",
+    R"(1:56: syntax error, unexpected ',', expecting <Space>.)",
  };

  size_t i = 0;