Factored debug visualization into its own file

7 files changed, 199 insertions, 179 deletions

diff --git a/Makefile b/Makefile
index bce9d00..24f9ca0 100644
--- a/Makefile
+++ b/Makefile
@@ -4,25 +4,25 @@ PREFIX=/usr/local
 SYSCONFDIR=/etc
 CFLAGS=-std=c99 -Werror -D_XOPEN_SOURCE=700 -D_POSIX_C_SOURCE=200809L -flto
 CWARN=-Wall -Wextra
-#   -Wpedantic -Wsign-conversion -Wtype-limits -Wunused-result -Wnull-dereference \
-# 	-Waggregate-return -Walloc-zero -Walloca -Warith-conversion -Wcast-align -Wcast-align=strict \
-# 	-Wdangling-else -Wdate-time -Wdisabled-optimization -Wdouble-promotion -Wduplicated-branches \
-# 	-Wduplicated-cond -Wexpansion-to-defined -Wfloat-conversion -Wfloat-equal -Wformat-nonliteral \
-# 	-Wformat-security -Wformat-signedness -Wframe-address -Winline -Winvalid-pch -Wjump-misses-init \
-# 	-Wlogical-op -Wlong-long -Wmissing-format-attribute -Wmissing-include-dirs -Wmissing-noreturn \
-# 	-Wnull-dereference -Woverlength-strings -Wpacked -Wpacked-not-aligned -Wpointer-arith \
-# 	-Wredundant-decls -Wshadow -Wshadow=compatible-local -Wshadow=global -Wshadow=local \
-# 	-Wsign-conversion -Wstack-protector -Wsuggest-attribute=const -Wswitch-default -Wswitch-enum \
-# 	-Wsync-nand -Wtrampolines -Wundef -Wunsuffixed-float-constants -Wunused -Wunused-but-set-variable \
-# 	-Wunused-const-variable -Wunused-local-typedefs -Wunused-macros -Wvariadic-macros -Wvector-operation-performance \
-# 	-Wvla -Wwrite-strings
+  # -Wpedantic -Wsign-conversion -Wtype-limits -Wunused-result -Wnull-dereference \
+	# -Waggregate-return -Walloc-zero -Walloca -Warith-conversion -Wcast-align -Wcast-align=strict \
+	# -Wdangling-else -Wdate-time -Wdisabled-optimization -Wdouble-promotion -Wduplicated-branches \
+	# -Wduplicated-cond -Wexpansion-to-defined -Wfloat-conversion -Wfloat-equal -Wformat-nonliteral \
+	# -Wformat-security -Wformat-signedness -Wframe-address -Winline -Winvalid-pch -Wjump-misses-init \
+	# -Wlogical-op -Wlong-long -Wmissing-format-attribute -Wmissing-include-dirs -Wmissing-noreturn \
+	# -Wnull-dereference -Woverlength-strings -Wpacked -Wpacked-not-aligned -Wpointer-arith \
+	# -Wredundant-decls -Wshadow -Wshadow=compatible-local -Wshadow=global -Wshadow=local \
+	# -Wsign-conversion -Wstack-protector -Wsuggest-attribute=const -Wswitch-default -Wswitch-enum \
+	# -Wsync-nand -Wtrampolines -Wundef -Wunsuffixed-float-constants -Wunused -Wunused-but-set-variable \
+	# -Wunused-const-variable -Wunused-local-typedefs -Wunused-macros -Wvariadic-macros -Wvector-operation-performance \
+	# -Wvla -Wwrite-strings
 OSFLAGS != case $$(uname -s) in *BSD|Darwin) echo '-D_BSD_SOURCE';; Linux) echo '-D_GNU_SOURCE';; *) echo '-D_DEFAULT_SOURCE';; esac
 EXTRA=
 G=
 O=-O3
 ALL_FLAGS=$(CFLAGS) $(OSFLAGS) -DBP_NAME="\"$(NAME)\"" $(EXTRA) $(CWARN) $(G) $(O)
 
-CFILES=pattern.c definitions.c utils.c match.c files.c print.c json.c utf8.c
+CFILES=pattern.c definitions.c utils.c match.c files.c print.c matchviz.c json.c utf8.c
 OBJFILES=$(CFILES:.c=.o)
 
 all: $(NAME) bp.1
diff --git a/bp.c b/bp.c
index 5a0da1b..8bc9283 100644
--- a/bp.c
+++ b/bp.c
@@ -24,6 +24,7 @@
 #include "files.h"
 #include "json.h"
 #include "match.h"
+#include "matchviz.h"
 #include "pattern.h"
 #include "print.h"
 #include "utils.h"
diff --git a/match.c b/match.c
index 514d8ca..2ab3a90 100644
--- a/match.c
+++ b/match.c
@@ -288,7 +288,7 @@ static match_t *match(def_t *defs, file_t *f, const char *str, pat_t *pat, bool
                     // Temporarily add a rule that the backref name matches the
                     // exact string of the original match (no replacements)
                     ssize_t len = (ssize_t)(m1->end - m1->start);
-                    pat_t *backref = new_pat(f, m1->start, m1->end, len, len, BP_STRING);
+                    pat_t *backref = new_pat(f, m1->start, m1->end, (size_t)len, len, BP_STRING);
                     backref->args.string = m1->start;
                     defs2 = with_def(defs, pat->args.ref.len, pat->args.ref.name, backref);
                 }
diff --git a/matchviz.c b/matchviz.c
new file mode 100644
index 0000000..ebf0ad9
--- /dev/null
+++ b/matchviz.c
@@ -0,0 +1,171 @@
+//
+// debugviz.c - Debug visualization of pattern matches.
+//
+
+#include <stdio.h>
+#include <string.h>
+
+#include "matchviz.h"
+#include "types.h"
+#include "utils.h"
+
+typedef struct match_node_s {
+    match_t *m;
+    struct match_node_s *next;
+} match_node_t;
+
+__attribute__((nonnull, pure))
+static int height_of_match(match_t *m);
+__attribute__((nonnull))
+static void _visualize_matches(match_node_t *firstmatch, int depth, const char *text, size_t textlen);
+
+//
+// Return the height of a match object (i.e. the number of descendents of the
+// structure).
+//
+static int height_of_match(match_t *m)
+{
+    int height = 0;
+    for (match_t *c = m->child; c; c = c->nextsibling) {
+        int childheight = height_of_match(c);
+        if (childheight > height) height = childheight;
+    }
+    return 1 + height;
+}
+
+//
+// Print a visual explanation for the as-yet-unprinted matches provided.
+//
+static void _visualize_matches(match_node_t *firstmatch, int depth, const char *text, size_t textlen)
+{
+    const char *V = "│"; // Vertical bar
+    const char *H = "─"; // Horizontal bar
+    const char *color = (depth % 2 == 0) ? "34" : "33";
+
+    match_t *viz = firstmatch->m;
+    // This is a heuristic: print matches first if they have more submatches.
+    // In general, this helps reduce the height of the final output by allowing
+    // for more rows that show the same rule matching in multiple places.
+    // TODO: there may be a better heuristic that optimizes for this factor
+    // while also printing earlier matches first when it doesn't affect overall
+    // output height.
+    for (match_node_t *p = firstmatch; p; p = p->next)
+        if (height_of_match(p->m) > height_of_match(viz))
+            viz = p->m;
+    const char *viz_type = viz->pat->start;
+    size_t viz_typelen = (size_t)(viz->pat->end - viz->pat->start);
+
+    // Backrefs use added dim quote marks to indicate that the pattern is a
+    // literal string being matched. (Backrefs have start/end inside the text
+    // input, instead of something the user typed in)
+    if (viz_type >= text && viz_type <= &text[textlen])
+        printf("\033[%luG\033[0;2m\"\033[%s;1m", 2*textlen+3, color);
+    else
+        printf("\033[%luG\033[%s;1m", 2*textlen+3, color);
+
+    for (size_t i = 0; i < viz_typelen; i++) {
+        switch (viz_type[i]) {
+            case '\n': printf("↵"); break;
+            case '\t': printf("⇥"); break;
+            default: printf("%c", viz_type[i]); break;
+        }
+    }
+
+    if (viz_type >= text && viz_type <= &text[textlen])
+        printf("\033[0;2m\"");
+
+    printf("\033[0m");
+
+    match_node_t *children = NULL;
+    match_node_t **nextchild = &children;
+
+#define RIGHT_TYPE(m) (m->m->pat->end == m->m->pat->start + viz_typelen && strncmp(m->m->pat->start, viz_type, viz_typelen) == 0)
+    // Print nonzero-width first:
+    for (match_node_t *m = firstmatch; m; m = m->next) {
+        if (RIGHT_TYPE(m)) {
+            for (match_t *c = m->m->child; c; c = c->nextsibling) {
+                *nextchild = new(match_node_t);
+                (*nextchild)->m = c;
+                nextchild = &((*nextchild)->next);
+            }
+            if (m->m->end == m->m->start) continue;
+            printf("\033[%ldG\033[0;2m%s\033[0;7;%sm", 1+2*(m->m->start - text), V, color);
+            for (const char *c = m->m->start; c < m->m->end; ++c) {
+                // TODO: newline
+                if (c > m->m->start) printf(" ");
+                // TODO: utf8
+                //while ((*c & 0xC0) != 0x80) printf("%c", *(c++));
+                if (*c == '\n')
+                    printf("↵");
+                else if (*c == '\t')
+                    printf("⇥");
+                else
+                    printf("%c", *c);
+            }
+            printf("\033[0;2m%s\033[0m", V);
+        } else {
+            *nextchild = new(match_node_t);
+            (*nextchild)->m = m->m;
+            nextchild = &((*nextchild)->next);
+            printf("\033[%ldG\033[0;2m%s", 1+2*(m->m->start - text), V);
+            for (ssize_t i = (ssize_t)(2*(m->m->end - m->m->start)-1); i > 0; i--)
+                printf(" ");
+            if (m->m->end > m->m->start)
+                printf("\033[0;2m%s", V);
+            printf("\033[0m");
+        }
+    }
+
+    // Print stars for zero-width:
+    for (match_node_t *m = firstmatch; m; m = m->next) {
+        if (m->m->end > m->m->start) continue;
+        if (RIGHT_TYPE(m)) {
+            printf("\033[%ldG\033[7;%sm▒\033[0m", 1+2*(m->m->start - text), color);
+        } else {
+            printf("\033[%ldG\033[0;2m%s\033[0m", 1+2*(m->m->start - text), V);
+        }
+    }
+
+    printf("\n");
+
+    for (match_node_t *m = firstmatch; m; m = m->next) {
+        if (m->m->end == m->m->start) {
+            if (!RIGHT_TYPE(m))
+                printf("\033[%ldG\033[0;2m%s", 1 + 2*(m->m->start - text), V);
+        } else {
+            const char *l = "└";
+            const char *r = "┘";
+            for (match_node_t *c = children; c; c = c->next) {
+                if (c->m->start == m->m->start || c->m->end == m->m->start) l = V;
+                if (c->m->start == m->m->end   || c->m->end == m->m->end) r = V;
+            }
+            printf("\033[%ldG\033[0;2m%s", 1 + 2*(m->m->start - text), l);
+            const char *h = RIGHT_TYPE(m) ? H : " ";
+            for (ssize_t n = (ssize_t)(2*(m->m->end - m->m->start) - 1); n > 0; n--)
+                printf("%s", h);
+            printf("%s\033[0m", r);
+        }
+    }
+#undef RIGHT_TYPE
+
+    printf("\n");
+
+    if (children)
+        _visualize_matches(children, depth+1, text, textlen);
+
+    for (match_node_t *c = children, *next = NULL; c; c = next) {
+        next = c->next;
+        xfree(&c);
+    }
+}
+
+//
+// Print a visualization of a match object.
+//
+void visualize_match(match_t *m)
+{
+    printf("\033[?7l"); // Disable line wrapping
+    match_node_t first = {.m = m};
+    _visualize_matches(&first, 0, m->start, (size_t)(m->end - m->start));
+    printf("\033[?7h"); // Re-enable line wrapping
+}
diff --git a/matchviz.h b/matchviz.h
new file mode 100644
index 0000000..eb87774
--- /dev/null
+++ b/matchviz.h
@@ -0,0 +1,13 @@
+//
+// Debug visualization of matches
+//
+#ifndef DEBUGVIZ__H
+#define DEBUGVIZ__H
+
+#include "types.h"
+
+__attribute__((nonnull))
+void visualize_match(match_t *m);
+
+#endif
+// vim: ts=4 sw=0 et cino=L2,l1,(0,W4,m1
diff --git a/print.c b/print.c
index 8638876..eb5f67d 100644
--- a/print.c
+++ b/print.c
@@ -12,174 +12,11 @@
 #include "types.h"
 #include "utils.h"
 
-typedef struct match_node_s {
-    match_t *m;
-    struct match_node_s *next;
-} match_node_t;
-
 static const char *color_match = "\033[0;31;1m";
 static const char *color_replace = "\033[0;34;1m";
 static const char *color_normal = "\033[0m";
 static const char *current_color = NULL;
 
-__attribute__((nonnull, pure))
-static int height_of_match(match_t *m);
-__attribute__((nonnull))
-static void _visualize_matches(match_node_t *firstmatch, int depth, const char *text, size_t textlen);
-__attribute__((nonnull(1,2)))
-static inline void print_line_number(FILE *out, printer_t *pr, size_t line_number, const char *color);
-
-//
-// Return the height of a match object (i.e. the number of descendents of the
-// structure).
-//
-static int height_of_match(match_t *m)
-{
-    int height = 0;
-    for (match_t *c = m->child; c; c = c->nextsibling) {
-        int childheight = height_of_match(c);
-        if (childheight > height) height = childheight;
-    }
-    return 1 + height;
-}
-
-//
-// Print a visual explanation for the as-yet-unprinted matches provided.
-//
-static void _visualize_matches(match_node_t *firstmatch, int depth, const char *text, size_t textlen)
-{
-    const char *V = "│"; // Vertical bar
-    const char *H = "─"; // Horizontal bar
-    const char *color = (depth % 2 == 0) ? "34" : "33";
-
-    match_t *viz = firstmatch->m;
-    // This is a heuristic: print matches first if they have more submatches.
-    // In general, this helps reduce the height of the final output by allowing
-    // for more rows that show the same rule matching in multiple places.
-    // TODO: there may be a better heuristic that optimizes for this factor
-    // while also printing earlier matches first when it doesn't affect overall
-    // output height.
-    for (match_node_t *p = firstmatch; p; p = p->next)
-        if (height_of_match(p->m) > height_of_match(viz))
-            viz = p->m;
-    const char *viz_type = viz->pat->start;
-    size_t viz_typelen = (size_t)(viz->pat->end - viz->pat->start);
-
-    // Backrefs use added dim quote marks to indicate that the pattern is a
-    // literal string being matched. (Backrefs have start/end inside the text
-    // input, instead of something the user typed in)
-    if (viz_type >= text && viz_type <= &text[textlen])
-        printf("\033[%luG\033[0;2m\"\033[%s;1m", 2*textlen+3, color);
-    else
-        printf("\033[%luG\033[%s;1m", 2*textlen+3, color);
-
-    for (size_t i = 0; i < viz_typelen; i++) {
-        switch (viz_type[i]) {
-            case '\n': printf("↵"); break;
-            case '\t': printf("⇥"); break;
-            default: printf("%c", viz_type[i]); break;
-        }
-    }
-
-    if (viz_type >= text && viz_type <= &text[textlen])
-        printf("\033[0;2m\"");
-
-    printf("\033[0m");
-
-    match_node_t *children = NULL;
-    match_node_t **nextchild = &children;
-
-#define RIGHT_TYPE(m) (m->m->pat->end == m->m->pat->start + viz_typelen && strncmp(m->m->pat->start, viz_type, viz_typelen) == 0)
-    // Print nonzero-width first:
-    for (match_node_t *m = firstmatch; m; m = m->next) {
-        if (RIGHT_TYPE(m)) {
-            for (match_t *c = m->m->child; c; c = c->nextsibling) {
-                *nextchild = new(match_node_t);
-                (*nextchild)->m = c;
-                nextchild = &((*nextchild)->next);
-            }
-            if (m->m->end == m->m->start) continue;
-            printf("\033[%ldG\033[0;2m%s\033[0;7;%sm", 1+2*(m->m->start - text), V, color);
-            for (const char *c = m->m->start; c < m->m->end; ++c) {
-                // TODO: newline
-                if (c > m->m->start) printf(" ");
-                // TODO: utf8
-                //while ((*c & 0xC0) != 0x80) printf("%c", *(c++));
-                if (*c == '\n')
-                    printf("↵");
-                else if (*c == '\t')
-                    printf("⇥");
-                else
-                    printf("%c", *c);
-            }
-            printf("\033[0;2m%s\033[0m", V);
-        } else {
-            *nextchild = new(match_node_t);
-            (*nextchild)->m = m->m;
-            nextchild = &((*nextchild)->next);
-            printf("\033[%ldG\033[0;2m%s", 1+2*(m->m->start - text), V);
-            for (ssize_t i = (ssize_t)(2*(m->m->end - m->m->start)-1); i > 0; i--)
-                printf(" ");
-            if (m->m->end > m->m->start)
-                printf("\033[0;2m%s", V);
-            printf("\033[0m");
-        }
-    }
-
-    // Print stars for zero-width:
-    for (match_node_t *m = firstmatch; m; m = m->next) {
-        if (m->m->end > m->m->start) continue;
-        if (RIGHT_TYPE(m)) {
-            printf("\033[%ldG\033[7;%sm▒\033[0m", 1+2*(m->m->start - text), color);
-        } else {
-            printf("\033[%ldG\033[0;2m%s\033[0m", 1+2*(m->m->start - text), V);
-        }
-    }
-
-    printf("\n");
-
-    for (match_node_t *m = firstmatch; m; m = m->next) {
-        if (m->m->end == m->m->start) {
-            if (!RIGHT_TYPE(m))
-                printf("\033[%ldG\033[0;2m%s", 1 + 2*(m->m->start - text), V);
-        } else {
-            const char *l = "└";
-            const char *r = "┘";
-            for (match_node_t *c = children; c; c = c->next) {
-                if (c->m->start == m->m->start || c->m->end == m->m->start) l = V;
-                if (c->m->start == m->m->end   || c->m->end == m->m->end) r = V;
-            }
-            printf("\033[%ldG\033[0;2m%s", 1 + 2*(m->m->start - text), l);
-            const char *h = RIGHT_TYPE(m) ? H : " ";
-            for (ssize_t n = (ssize_t)(2*(m->m->end - m->m->start) - 1); n > 0; n--)
-                printf("%s", h);
-            printf("%s\033[0m", r);
-        }
-    }
-#undef RIGHT_TYPE
-
-    printf("\n");
-
-    if (children)
-        _visualize_matches(children, depth+1, text, textlen);
-
-    for (match_node_t *c = children, *next = NULL; c; c = next) {
-        next = c->next;
-        xfree(&c);
-    }
-}
-
-//
-// Print a visualization of a match object.
-//
-void visualize_match(match_t *m)
-{
-    printf("\033[?7l"); // Disable line wrapping
-    match_node_t first = {.m = m};
-    _visualize_matches(&first, 0, m->start, (size_t)(m->end - m->start));
-    printf("\033[?7h"); // Re-enable line wrapping
-}
-
 //
 // Print a line number, if it needs to be printed.
 // line number of 0 means "just print an empty space for the number"
diff --git a/print.h b/print.h
index 9a0439b..5ad645c 100644
--- a/print.h
+++ b/print.h
@@ -17,8 +17,6 @@ typedef struct {
     bool print_line_numbers:1;
 } printer_t;
 
-__attribute__((nonnull))
-void visualize_match(match_t *m);
 __attribute__((nonnull(1,2)))
 void print_match(FILE *out, printer_t *pr, match_t *m);
 __attribute__((nonnull))