From 5d9b04897107b3fb0a639acc233b8861a73f287f Mon Sep 17 00:00:00 2001
From: Kevin Newton <kddnewton@gmail.com>
Date: Wed, 30 Aug 2023 17:21:50 -0400
Subject: [PATCH] Add a value to numbered references

---
 config.yml                                    |  3 ++
 include/yarp.h                                |  1 +
 src/yarp.c                                    | 35 ++++++++++++++++++-
 .../unparser/corpus/literal/dstr.txt          |  2 +-
 .../unparser/corpus/literal/variables.txt     |  2 +-
 test/yarp/snapshots/whitequark/nth_ref.txt    |  2 +-
 6 files changed, 41 insertions(+), 4 deletions(-)

diff --git a/config.yml b/config.yml
index 32a605c61d9..8ebfb772bf4 100644
--- a/config.yml
+++ b/config.yml
@@ -1703,6 +1703,9 @@ nodes:
                 ^^^^^
           end
   - name: NumberedReferenceReadNode
+    fields:
+      - name: number
+        type: uint32
     comment: |
       Represents reading a numbered reference to a capture in the previous match.
 
diff --git a/include/yarp.h b/include/yarp.h
index 6488d01adcb..378efe0c937 100644
--- a/include/yarp.h
+++ b/include/yarp.h
@@ -16,6 +16,7 @@
 #include "yarp/version.h"
 
 #include <assert.h>
+#include <errno.h>
 #include <stdarg.h>
 #include <stdbool.h>
 #include <stdint.h>
diff --git a/src/yarp.c b/src/yarp.c
index 5381eabe006..7df7ca0653c 100644
--- a/src/yarp.c
+++ b/src/yarp.c
@@ -613,6 +613,38 @@ yp_scope_node_init(yp_node_t *node, yp_scope_node_t *scope) {
 /* Node creation functions                                                    */
 /******************************************************************************/
 
+// Parse the decimal number represented by the range of bytes. returns
+// UINT32_MAX if the number fails to parse. This function assumes that the range
+// of bytes has already been validated to contain only decimal digits.
+static uint32_t
+parse_decimal_number(yp_parser_t *parser, const uint8_t *start, const uint8_t *end) {
+    ptrdiff_t diff = end - start;
+    assert(diff > 0 && ((unsigned long) diff < SIZE_MAX));
+    size_t length = (size_t) diff;
+
+    char *digits = calloc(length + 1, sizeof(char));
+    memcpy(digits, start, length);
+    digits[length] = '\0';
+
+    char *endptr;
+    errno = 0;
+    unsigned long value = strtoul(digits, &endptr, 10);
+
+    if ((digits == endptr) || (*endptr != '\0') || (errno == ERANGE)) {
+        yp_diagnostic_list_append(&parser->error_list, start, end, "invalid decimal number");
+        value = UINT32_MAX;
+    }
+
+    free(digits);
+
+    if (value > UINT32_MAX) {
+        yp_diagnostic_list_append(&parser->error_list, start, end, "invalid decimal number");
+        value = UINT32_MAX;
+    }
+
+    return (uint32_t) value;
+}
+
 // Parse out the options for a regular expression.
 static inline yp_node_flags_t
 yp_regular_expression_flags_create(const yp_token_t *closing) {
@@ -3289,7 +3321,8 @@ yp_numbered_reference_read_node_create(yp_parser_t *parser, const yp_token_t *na
         {
             .type = YP_NODE_NUMBERED_REFERENCE_READ_NODE,
             .location = YP_LOCATION_TOKEN_VALUE(name),
-        }
+        },
+        .number = parse_decimal_number(parser, name->start + 1, name->end)
     };
 
     return node;
diff --git a/test/yarp/snapshots/unparser/corpus/literal/dstr.txt b/test/yarp/snapshots/unparser/corpus/literal/dstr.txt
index facf0171e79..396ebcd3daf 100644
--- a/test/yarp/snapshots/unparser/corpus/literal/dstr.txt
+++ b/test/yarp/snapshots/unparser/corpus/literal/dstr.txt
@@ -57,7 +57,7 @@ ProgramNode(0...299)(
        [StringNode(146...147)(nil, (146...147), nil, "a"),
         EmbeddedVariableNode(147...150)(
           (147...148),
-          NumberedReferenceReadNode(148...150)()
+          NumberedReferenceReadNode(148...150)(1)
         )],
        (150...151)
      ),
diff --git a/test/yarp/snapshots/unparser/corpus/literal/variables.txt b/test/yarp/snapshots/unparser/corpus/literal/variables.txt
index dfb85ec0137..fe98a8cd57f 100644
--- a/test/yarp/snapshots/unparser/corpus/literal/variables.txt
+++ b/test/yarp/snapshots/unparser/corpus/literal/variables.txt
@@ -5,7 +5,7 @@ ProgramNode(0...66)(
      InstanceVariableReadNode(2...4)(:@a),
      ClassVariableReadNode(5...8)(:@@a),
      GlobalVariableReadNode(9...11)(),
-     NumberedReferenceReadNode(12...14)(),
+     NumberedReferenceReadNode(12...14)(1),
      BackReferenceReadNode(15...17)(),
      ConstantReadNode(18...23)(),
      ConstantPathNode(24...37)(
diff --git a/test/yarp/snapshots/whitequark/nth_ref.txt b/test/yarp/snapshots/whitequark/nth_ref.txt
index 3bfe4c9a5c3..ca131a208ce 100644
--- a/test/yarp/snapshots/whitequark/nth_ref.txt
+++ b/test/yarp/snapshots/whitequark/nth_ref.txt
@@ -1,4 +1,4 @@
 ProgramNode(0...3)(
   [],
-  StatementsNode(0...3)([NumberedReferenceReadNode(0...3)()])
+  StatementsNode(0...3)([NumberedReferenceReadNode(0...3)(10)])
 )