Fix an issue of capturing variable substitution

arithy · arithy · commit 4a8bafcf9768 · 2025-03-15T14:37:22.000+09:00
The issue was reported in #92.
diff --git a/src/packcc.c b/src/packcc.c
@@ -1275,10 +1275,20 @@ static bool_t can_character_follow_identifier_(char ch) {
 
 static bool_t stream__write_text(
     stream_t *obj, const char *str, size_t len, bool_t lhead, size_t dedent, size_t indent,
-    code_block_kind_t kind, size_t ncapts, const string_array_t *mvars, const subst_map_t *subst
+    code_block_kind_t kind, const node_const_array_t *capts, const string_array_t *mvars, const subst_map_t *subst
 ) {
     size_t i = 0;
+    size_t mc = 0;
     if (len == VOID_VALUE) return lhead; /* for safety */
+    if (capts) {
+        size_t ic;
+        for (ic = 0; ic < capts->n; ic++) {
+            size_t kc;
+            assert(capts->p[ic]->type == NODE_CAPTURE);
+            kc = capts->p[ic]->data.capture.index + 1;
+            if (mc < kc) mc = kc;
+        }
+    }
     while (i < len) {
         if (str[i] == '\0') break;
         if (match_eol(str, i, len, &i)) {
@@ -1324,15 +1334,21 @@ static bool_t stream__write_text(
                         c = str[j];
                         if (!(c >= '0' && c <= '9')) break;
                         k = (c - '0') + 10 * k;
-                        if (k > ncapts) k = ncapts + 1; /* to avoid overflow */
+                        if (k > mc) k = mc + 1; /* to avoid overflow */
                         j++;
                     }
                     if (j < len && (str[j] == 's' || str[j] == 'e')) j++;
-                    if ((j >= len || can_character_follow_identifier_(str[j])) && k <= ncapts) {
-                        stream__puts(obj, VARNAME_CAPTURE_PREFIX);
-                        i++;
-                        while (i < j) stream__putc(obj, str[i++]);
-                        b = TRUE;
+                    if (j >= len || can_character_follow_identifier_(str[j])) {
+                        size_t ic;
+                        for (ic = 0; ic < capts->n; ic++) {
+                            if (k == capts->p[ic]->data.capture.index + 1) break;
+                        }
+                        if (ic < capts->n) {
+                            stream__puts(obj, VARNAME_CAPTURE_PREFIX);
+                            i++;
+                            while (i < j) stream__putc(obj, str[i++]);
+                            b = TRUE;
+                        }
                     }
                 }
                 else if (c == '{') { /* ${...} */
@@ -1475,7 +1491,7 @@ static bool_t stream__write_text(
 
 static bool_t stream__write_code(
     stream_t *obj, const char *str, size_t len, bool_t lhead, size_t dedent, size_t indent,
-    code_block_kind_t kind, size_t ncapts, const string_array_t *mvars, const subst_map_t *subst, bool_t dircpp
+    code_block_kind_t kind, const node_const_array_t *capts, const string_array_t *mvars, const subst_map_t *subst, bool_t dircpp
 ) {
     bool_t b = TRUE; /* TRUE if succeeding leading spaces in a line */
     size_t h = 0;
@@ -1486,7 +1502,7 @@ static bool_t stream__write_code(
         if (!dircpp && b && match_directive_c(str, i, len, &k)) {
             {
                 const size_t j = find_back_preceding_spaces_in_line(str, i);
-                stream__write_text(obj, str + h, j - h, lhead, dedent, indent, kind, ncapts, mvars, subst);
+                stream__write_text(obj, str + h, j - h, lhead, dedent, indent, kind, capts, mvars, subst);
             }
             {
                 size_t md = VOID_VALUE; /* the minimum indent length of the successive lines */
@@ -1502,29 +1518,29 @@ static bool_t stream__write_code(
                     }
                 }
                 if (md == VOID_VALUE) md = 0;
-                lhead = stream__write_code(obj, str + i, k - i, FALSE, md, INDENT_UNIT, kind, ncapts, mvars, subst, TRUE);
+                lhead = stream__write_code(obj, str + i, k - i, FALSE, md, INDENT_UNIT, kind, capts, mvars, subst, TRUE);
             }
             h = i = k;
             /* b == TRUE */
         }
         else if (match_comment_cxx(str, i, len, &k)) {
-            lhead = stream__write_text(obj, str + h, i - h, lhead, dedent, indent, kind, ncapts, mvars, subst);
-            lhead = stream__write_text(obj, str + i, k - i, lhead, dedent, indent, CODE_BLOCK_KIND__NONE, ncapts, mvars, subst);
+            lhead = stream__write_text(obj, str + h, i - h, lhead, dedent, indent, kind, capts, mvars, subst);
+            lhead = stream__write_text(obj, str + i, k - i, lhead, dedent, indent, CODE_BLOCK_KIND__NONE, capts, mvars, subst);
             h = i = k;
             b = TRUE;
         }
         else if (match_comment_c(str, i, len, &k)) {
-            lhead = stream__write_text(obj, str + h, i - h, lhead, dedent, indent, kind, ncapts, mvars, subst);
-            lhead = stream__write_text(obj, str + i, k - i, lhead, dedent, indent, CODE_BLOCK_KIND__NONE, ncapts, mvars, subst);
+            lhead = stream__write_text(obj, str + h, i - h, lhead, dedent, indent, kind, capts, mvars, subst);
+            lhead = stream__write_text(obj, str + i, k - i, lhead, dedent, indent, CODE_BLOCK_KIND__NONE, capts, mvars, subst);
             h = i = k;
             /* do not change `b` because a C-style comment is dealt as a white space */
         }
         else if (
             match_quotation(str, i, len, '\'', &k) ||
             match_quotation(str, i, len, '\"', &k)
         ) {
-            lhead = stream__write_text(obj, str + h, i - h, lhead, dedent, indent, kind, ncapts, mvars, subst);
-            lhead = stream__write_text(obj, str + i, k - i, lhead, dedent, indent, CODE_BLOCK_KIND__NONE, ncapts, mvars, subst);
+            lhead = stream__write_text(obj, str + h, i - h, lhead, dedent, indent, kind, capts, mvars, subst);
+            lhead = stream__write_text(obj, str + i, k - i, lhead, dedent, indent, CODE_BLOCK_KIND__NONE, capts, mvars, subst);
             h = i = k;
             b = FALSE;
         }
@@ -1538,12 +1554,12 @@ static bool_t stream__write_code(
             }
         }
     }
-    return stream__write_text(obj, str + h, i - h, lhead, dedent, indent, kind, ncapts, mvars, subst);
+    return stream__write_text(obj, str + h, i - h, lhead, dedent, indent, kind, capts, mvars, subst);
 }
 
 static void stream__write_code_block(
     stream_t *obj, const char *str, size_t len, size_t indent, const char *path, size_t lineno,
-    code_block_kind_t kind, size_t ncapts, const string_array_t *mvars, const subst_map_t *subst
+    code_block_kind_t kind, const node_const_array_t *capts, const string_array_t *mvars, const subst_map_t *subst
 ) {
     size_t m = VOID_VALUE; /* the minimum indent length except for the start line of the code block */
     size_t b = TRUE; /* TRUE if the code starts at the first line */
@@ -1584,7 +1600,7 @@ static void stream__write_code_block(
         assert(b || d >= m);
         stream__write_characters(obj, ' ', indent + (b ? 0 : d - m));
     }
-    stream__write_code(obj, str + h, len - h, FALSE, m, indent, kind, ncapts, mvars, subst, FALSE);
+    stream__write_code(obj, str + h, len - h, FALSE, m, indent, kind, capts, mvars, subst, FALSE);
     stream__putc(obj, '\n');
     if (obj->line != VOID_VALUE)
         stream__write_line_directive(obj, obj->path, obj->line + 1);
@@ -1593,7 +1609,7 @@ static void stream__write_code_block(
 static void stream__write_footer(
     stream_t *obj, const char *str, size_t len, const char *path, size_t lineno, const subst_map_t *subst
 ) {
-    stream__write_code_block(obj, str, len, 0, path, lineno, CODE_BLOCK_KIND__NORMAL, 0, NULL, subst);
+    stream__write_code_block(obj, str, len, 0, path, lineno, CODE_BLOCK_KIND__NORMAL, NULL, NULL, subst);
 }
 
 static char *get_home_directory(void) {
@@ -4590,7 +4606,7 @@ static bool_t generate(context_t *ctx) {
             for (i = 0; i < ctx->eheader.n; i++) {
                 stream__write_code_block(
                     &hstream, ctx->eheader.p[i].text, ctx->eheader.p[i].len, 0,
-                    ctx->eheader.p[i].fpos.path, ctx->eheader.p[i].fpos.line, CODE_BLOCK_KIND__NORMAL, 0, NULL, &(ctx->subst)
+                    ctx->eheader.p[i].fpos.path, ctx->eheader.p[i].fpos.line, CODE_BLOCK_KIND__NORMAL, NULL, NULL, &(ctx->subst)
                 );
                 stream__puts(&hstream, "\n");
             }
@@ -4607,7 +4623,7 @@ static bool_t generate(context_t *ctx) {
             for (i = 0; i < ctx->header.n; i++) {
                 stream__write_code_block(
                     &hstream, ctx->header.p[i].text, ctx->header.p[i].len, 0,
-                    ctx->header.p[i].fpos.path, ctx->header.p[i].fpos.line, CODE_BLOCK_KIND__NORMAL, 0, NULL, &(ctx->subst)
+                    ctx->header.p[i].fpos.path, ctx->header.p[i].fpos.line, CODE_BLOCK_KIND__NORMAL, NULL, NULL, &(ctx->subst)
                 );
                 stream__puts(&hstream, "\n");
             }
@@ -4619,7 +4635,7 @@ static bool_t generate(context_t *ctx) {
             for (i = 0; i < ctx->esource.n; i++) {
                 stream__write_code_block(
                     &sstream, ctx->esource.p[i].text, ctx->esource.p[i].len, 0,
-                    ctx->esource.p[i].fpos.path, ctx->esource.p[i].fpos.line, CODE_BLOCK_KIND__NORMAL, 0, NULL, &(ctx->subst)
+                    ctx->esource.p[i].fpos.path, ctx->esource.p[i].fpos.line, CODE_BLOCK_KIND__NORMAL, NULL, NULL, &(ctx->subst)
                 );
                 stream__puts(&sstream, "\n");
             }
@@ -4667,7 +4683,7 @@ static bool_t generate(context_t *ctx) {
             for (i = 0; i < ctx->source.n; i++) {
                 stream__write_code_block(
                     &sstream, ctx->source.p[i].text, ctx->source.p[i].len, 0,
-                    ctx->source.p[i].fpos.path, ctx->source.p[i].fpos.line, CODE_BLOCK_KIND__NORMAL, 0, NULL, &(ctx->subst)
+                    ctx->source.p[i].fpos.path, ctx->source.p[i].fpos.line, CODE_BLOCK_KIND__NORMAL, NULL, NULL, &(ctx->subst)
                 );
                 stream__puts(&sstream, "\n");
             }
@@ -6423,7 +6439,7 @@ static bool_t generate(context_t *ctx) {
                     }
                     stream__write_code_block(
                         &sstream, b->text, b->len, INDENT_UNIT, b->fpos.path, b->fpos.line,
-                        CODE_BLOCK_KIND__PROGPRED, c->n, &(ctx->mvars), &(ctx->subst)
+                        CODE_BLOCK_KIND__PROGPRED, c, &(ctx->mvars), &(ctx->subst)
                     );
                     k = c->n;
                     while (k > 0) {
@@ -6537,7 +6553,7 @@ static bool_t generate(context_t *ctx) {
                     }
                     stream__write_code_block(
                         &sstream, b->text, b->len, INDENT_UNIT, b->fpos.path, b->fpos.line,
-                        CODE_BLOCK_KIND__ACTION, c->n, &(ctx->mvars), &(ctx->subst)
+                        CODE_BLOCK_KIND__ACTION, c, &(ctx->mvars), &(ctx->subst)
                     );
                     k = c->n;
                     while (k > 0) {
diff --git a/tests/captures.d/expected.txt b/tests/captures.d/expected.txt
@@ -1,7 +1,11 @@
+predicate $6 = 132,137:capt6
+predicate $8 = 138,143:capt8
 0,11:capture all
 12,19:capture
 34,40:second
 59,64:words
 83,92:something
 93,97:more 98,106:captures
 116,121:again
+action $6 = 132,137:capt6
+action $8 = 138,143:capt8
diff --git a/tests/captures.d/input.peg b/tests/captures.d/input.peg
@@ -1,5 +1,5 @@
 FILE <- (LINE '\n')*
-LINE <- ALL / FIRST / SECOND / MULTI1 / MULTI2 / MORE / CAPTURED / UNKNOWN
+LINE <- ALL / FIRST / SECOND / MULTI1 / MULTI2 / MORE / CAPTURED / ALTERNATE / UNKNOWN
 
 ALL <- <'capture' _ 'all'> { printf("%d,%d:%s\n", (int)$1s, (int)$1e, $1); }
 FIRST <- <'capture'> _ 'first' { printf("%d,%d:%s\n", (int)$1s, (int)$1e, $1); }
@@ -8,6 +8,16 @@ MULTI1 <- 'capture' _ 'multiple1' (_ <[a-z]+>)+ { printf("%d,%d:%s\n", (int)$1s,
 MULTI2 <- 'capture' _ 'multiple2' (_ <[a-z]+> { printf("%d,%d:%s\n", (int)$1s, (int)$1e, $1); })+
 MORE <- <'more'> _ <'captures'> { printf("%d,%d:%s %d,%d:%s\n", (int)$1s, (int)$1e, $1, (int)$2s, (int)$2e, $2); }
 CAPTURED <- "captured" _ <[a-z]+> _ 'and' _ $1 { printf("%d,%d:%s\n", (int)$1s, (int)$1e, $1); }
+ALTERNATE <- < 'capt1' > / ( < 'capt2' > _ ( < 'capt3' > / < 'capt4' > ) _ < 'capt5' > / < 'capt6' > _ ( < 'capt7' > / < 'capt8' >
+            &{
+                printf("predicate $6 = %d,%d:%s\n", (int)$6s, (int)$6e, $6 ? $6 : "(null)");
+                printf("predicate $8 = %d,%d:%s\n", (int)$8s, (int)$8e, $8 ? $8 : "(null)");
+            }
+            {
+                printf("action $6 = %d,%d:%s\n", (int)$6s, (int)$6e, $6 ? $6 : "(null)");
+                printf("action $8 = %d,%d:%s\n", (int)$8s, (int)$8e, $8 ? $8 : "(null)");
+            }
+        ) _ < 'capt9' > )
 
 UNKNOWN <- [^\n]+ { printf("ERROR: %s\n", $0); }
 _ <- " "
diff --git a/tests/captures.d/input.txt b/tests/captures.d/input.txt
@@ -5,3 +5,4 @@ capture multiple1 words
 capture multiple2 something
 more captures
 captured again and again
+capt6 capt8 capt9
diff --git a/tests/substitution.d/expected.c.txt b/tests/substitution.d/expected.c.txt
@@ -197,6 +197,16 @@ TEST COMMON @var.restore()        " @var.restore()        " /* @var.restore()
     TEST PROGPRED5 $123  " $123  " /* $123  */
     TEST PROGPRED6 $123s " $123s " /* $123s */
     TEST PROGPRED7 $123e " $123e " /* $123e */
+    TEST PROGPRED8 pcc_capture__0s:pcc_capture__0e=pcc_capture__0 " $0s:$0e=$0 " /* $0s:$0e=$0 */
+    TEST PROGPRED8 $1s:$1e=$1 " $1s:$1e=$1 " /* $1s:$1e=$1 */
+    TEST PROGPRED8 $2s:$2e=$2 " $2s:$2e=$2 " /* $2s:$2e=$2 */
+    TEST PROGPRED8 $3s:$3e=$3 " $3s:$3e=$3 " /* $3s:$3e=$3 */
+    TEST PROGPRED8 $4s:$4e=$4 " $4s:$4e=$4 " /* $4s:$4e=$4 */
+    TEST PROGPRED8 $5s:$5e=$5 " $5s:$5e=$5 " /* $5s:$5e=$5 */
+    TEST PROGPRED8 pcc_capture__6s:pcc_capture__6e=pcc_capture__6 " $6s:$6e=$6 " /* $6s:$6e=$6 */
+    TEST PROGPRED8 $7s:$7e=$7 " $7s:$7e=$7 " /* $7s:$7e=$7 */
+    TEST PROGPRED8 pcc_capture__8s:pcc_capture__8e=pcc_capture__8 " $8s:$8e=$8 " /* $8s:$8e=$8 */
+    TEST PROGPRED8 $9s:$9e=$9 " $9s:$9e=$9 " /* $9s:$9e=$9 */
     TEST ACTION0 pcc_action_out.        " $$.        " /* $$.        */ // $$.
     TEST ACTION0 $pcc_action_out        " $$$        " /* $$$        */ // $$$
     TEST ACTION0 $pcc_capture__0        " $$0        " /* $$0        */ // $$0
@@ -272,6 +282,16 @@ TEST COMMON @var.restore()        " @var.restore()        " /* @var.restore()
     TEST ACTION5 $123  " $123  " /* $123  */
     TEST ACTION6 $123s " $123s " /* $123s */
     TEST ACTION7 $123e " $123e " /* $123e */
+    TEST ACTION8 pcc_capture__0s:pcc_capture__0e=pcc_capture__0 " $0s:$0e=$0 " /* $0s:$0e=$0 */
+    TEST ACTION8 $1s:$1e=$1 " $1s:$1e=$1 " /* $1s:$1e=$1 */
+    TEST ACTION8 $2s:$2e=$2 " $2s:$2e=$2 " /* $2s:$2e=$2 */
+    TEST ACTION8 $3s:$3e=$3 " $3s:$3e=$3 " /* $3s:$3e=$3 */
+    TEST ACTION8 $4s:$4e=$4 " $4s:$4e=$4 " /* $4s:$4e=$4 */
+    TEST ACTION8 $5s:$5e=$5 " $5s:$5e=$5 " /* $5s:$5e=$5 */
+    TEST ACTION8 pcc_capture__6s:pcc_capture__6e=pcc_capture__6 " $6s:$6e=$6 " /* $6s:$6e=$6 */
+    TEST ACTION8 $7s:$7e=$7 " $7s:$7e=$7 " /* $7s:$7e=$7 */
+    TEST ACTION8 pcc_capture__8s:pcc_capture__8e=pcc_capture__8 " $8s:$8e=$8 " /* $8s:$8e=$8 */
+    TEST ACTION8 $9s:$9e=$9 " $9s:$9e=$9 " /* $9s:$9e=$9 */
 TEST FOOTER $$         " $$         " /* $$         */ // $$
 TEST FOOTER $0         " $0         " /* $0         */ // $0
 TEST FOOTER $0s        " $0s        " /* $0s        */ // $0s
diff --git a/tests/substitution.d/input.peg b/tests/substitution.d/input.peg
@@ -360,6 +360,32 @@ FILE
     / '123'  &{ TEST PROGPRED5 $123  " $123  " /* $123  */ } { TEST ACTION5 $123  " $123  " /* $123  */ }
     / '123s' &{ TEST PROGPRED6 $123s " $123s " /* $123s */ } { TEST ACTION6 $123s " $123s " /* $123s */ }
     / '123e' &{ TEST PROGPRED7 $123e " $123e " /* $123e */ } { TEST ACTION7 $123e " $123e " /* $123e */ }
+    / ( < 'capt2' > ( < 'capt3' > / < 'capt4' > ) < 'capt5' > / < 'capt6' > ( < 'capt7' > / < 'capt8' >
+            &{
+                TEST PROGPRED8 $0s:$0e=$0 " $0s:$0e=$0 " /* $0s:$0e=$0 */
+                TEST PROGPRED8 $1s:$1e=$1 " $1s:$1e=$1 " /* $1s:$1e=$1 */
+                TEST PROGPRED8 $2s:$2e=$2 " $2s:$2e=$2 " /* $2s:$2e=$2 */
+                TEST PROGPRED8 $3s:$3e=$3 " $3s:$3e=$3 " /* $3s:$3e=$3 */
+                TEST PROGPRED8 $4s:$4e=$4 " $4s:$4e=$4 " /* $4s:$4e=$4 */
+                TEST PROGPRED8 $5s:$5e=$5 " $5s:$5e=$5 " /* $5s:$5e=$5 */
+                TEST PROGPRED8 $6s:$6e=$6 " $6s:$6e=$6 " /* $6s:$6e=$6 */
+                TEST PROGPRED8 $7s:$7e=$7 " $7s:$7e=$7 " /* $7s:$7e=$7 */
+                TEST PROGPRED8 $8s:$8e=$8 " $8s:$8e=$8 " /* $8s:$8e=$8 */
+                TEST PROGPRED8 $9s:$9e=$9 " $9s:$9e=$9 " /* $9s:$9e=$9 */
+            }
+            {
+                TEST ACTION8 $0s:$0e=$0 " $0s:$0e=$0 " /* $0s:$0e=$0 */
+                TEST ACTION8 $1s:$1e=$1 " $1s:$1e=$1 " /* $1s:$1e=$1 */
+                TEST ACTION8 $2s:$2e=$2 " $2s:$2e=$2 " /* $2s:$2e=$2 */
+                TEST ACTION8 $3s:$3e=$3 " $3s:$3e=$3 " /* $3s:$3e=$3 */
+                TEST ACTION8 $4s:$4e=$4 " $4s:$4e=$4 " /* $4s:$4e=$4 */
+                TEST ACTION8 $5s:$5e=$5 " $5s:$5e=$5 " /* $5s:$5e=$5 */
+                TEST ACTION8 $6s:$6e=$6 " $6s:$6e=$6 " /* $6s:$6e=$6 */
+                TEST ACTION8 $7s:$7e=$7 " $7s:$7e=$7 " /* $7s:$7e=$7 */
+                TEST ACTION8 $8s:$8e=$8 " $8s:$8e=$8 " /* $8s:$8e=$8 */
+                TEST ACTION8 $9s:$9e=$9 " $9s:$9e=$9 " /* $9s:$9e=$9 */
+            }
+        ) < 'capt9' > )
 
 %%
 
