From e0fe461b3e7e85165a9188d6f92e59f66a93b82f Mon Sep 17 00:00:00 2001
From: Krasimir Angelov <kr.angelov@gmail.com>
Date: Tue, 15 Sep 2026 15:05:01 +0200
Subject: [PATCH] for backwards compatibility, now a string literal is a
 sequence of any characters except space.

---
 src/runtime/c/pgf/expr.cxx   | 29 ++++++++++++++++++++++++++---
 src/runtime/c/pgf/expr.h     |  3 ++-
 src/runtime/c/pgf/parser.cxx |  9 ++++++---
 3 files changed, 34 insertions(+), 7 deletions(-)

diff --git a/src/runtime/c/pgf/expr.cxx b/src/runtime/c/pgf/expr.cxx
index 67d7e866e..107386283 100644
--- a/src/runtime/c/pgf/expr.cxx
+++ b/src/runtime/c/pgf/expr.cxx
@@ -308,14 +308,13 @@ PgfType PgfInternalMarshaller::match_type(PgfUnmarshaller *u, PgfType ty)
 PgfExprParser::PgfExprParser(PgfText *input, size_t byte_pos, PgfUnmarshaller *unmarshaller)
 {
     inp = input;
-    pos = (const char*) &inp->text[byte_pos];
-	ch  = ' ';
     u   = unmarshaller;
     token_pos = NULL;
     token_value = NULL;
     bs = NULL;
 
-	token();
+    reset_pos(byte_pos);
+    token();
 }
 
 PgfExprParser::~PgfExprParser()
@@ -433,6 +432,30 @@ bool PgfExprParser::str_char()
     return getc();
 }
 
+void PgfExprParser::raw_token()
+{
+	token_tag   = PGF_TOKEN_STR;
+    token_pos   = pos;
+	token_value = NULL;
+
+	while (pgf_utf8_is_space(ch)) {
+        token_pos   = pos;
+        if (!getc()) {
+            token_tag = PGF_TOKEN_EOF;
+            return;
+        }
+	}
+
+	while (!pgf_utf8_is_space(ch)) {
+        putc(ch);
+        if (!getc())
+            break;
+	}
+
+    if (token_value == NULL)
+        token_tag   = PGF_TOKEN_EOF;
+}
+
 void PgfExprParser::token()
 {
     if (token_value != NULL)
diff --git a/src/runtime/c/pgf/expr.h b/src/runtime/c/pgf/expr.h
index 63f063683..047539913 100644
--- a/src/runtime/c/pgf/expr.h
+++ b/src/runtime/c/pgf/expr.h
@@ -175,6 +175,7 @@ public:
 
     bool str_char();
     void token();
+    void raw_token();
     bool lookahead(int ch);
 
     bool parse_bind();
@@ -195,7 +196,7 @@ public:
     bool is_str()   { return (token_tag == PGF_TOKEN_STR); }
     bool is_ident() { return (token_tag == PGF_TOKEN_IDENT); }
 
-    void ident2str() { token_tag = PGF_TOKEN_STR; }
+    void reset_pos(size_t pos) { this->pos = &inp->text[pos]; this->ch = ' '; }
 
     const PgfText *get_token_value() { return token_value; }
     const char *get_token_pos() { return token_pos; }
diff --git a/src/runtime/c/pgf/parser.cxx b/src/runtime/c/pgf/parser.cxx
index 86a5e43bf..ca54677ca 100644
--- a/src/runtime/c/pgf/parser.cxx
+++ b/src/runtime/c/pgf/parser.cxx
@@ -1576,9 +1576,12 @@ void PgfParser::suspend(Cont *cont,Item *item,bool do_predict,ref<PgfSymbolCat>
                         bu_literal(cont->state, "Int", &eparser, viterbi_prob);
                     } else if (eparser.is_flt()) {
                         bu_literal(cont->state, "Float", &eparser, viterbi_prob);
-                    } else if (eparser.is_ident()) {
-                        eparser.ident2str();
-                        bu_literal(cont->state, "String", &eparser, viterbi_prob);
+                    } else {
+                        eparser.reset_pos(cont->state->end.byte_pos);
+                        eparser.raw_token();
+                        if (eparser.is_str()) {
+                            bu_literal(cont->state, "String", &eparser, viterbi_prob);
+                        }
                     }
                     bu_predict(concr->phrasetable1, cont->state, viterbi_prob, 1, sentence->size);
                 }