Moving to each_match() API

4 files changed, 131 insertions, 100 deletions

diff --git a/Lua/lbp.c b/Lua/lbp.c
index c1c3f61..60dc9b9 100644
--- a/Lua/lbp.c
+++ b/Lua/lbp.c
@@ -36,7 +36,6 @@ lua_State *cur_state = NULL;
 static void match_error(pat_t *pat, const char *msg)
 {
     (void)pat;
-    recycle_all_matches();
     lua_pushstring(cur_state, msg);
     lua_error(cur_state);
 }
diff --git a/bp.c b/bp.c
index b863e24..ba4473b 100644
--- a/bp.c
+++ b/bp.c
@@ -60,6 +60,9 @@ static const char *usage = (
 #define ALL_CONTEXT -2
 #define NO_CONTEXT -1
 
+#define each_match_file(callback, userdata, f, pat, defs) \
+    each_match(callback, userdata, (f)->start, (f)->end, pat, defs, options.skip, options.ignorecase)
+
 // Flag-configurable options:
 static struct {
     int context_before, context_after;
@@ -88,6 +91,7 @@ const char *LINE_FORMATS[] = {
 // If a file is partly through being modified when the program exits, restore it from backup.
 static FILE *modifying_file = NULL;
 static file_t *backup_file;
+static int files_printed = 0;
 
 //
 // Helper function to reduce code duplication
@@ -207,34 +211,30 @@ static int is_text_file(const char *filename)
 //
 // Print matches in JSON format.
 //
-static int print_matches_as_json(file_t *f, pat_t *pattern, pat_t *defs)
+static bp_match_behavior print_match_as_json(match_t *m, int matchnum, void *data)
 {
-    int nmatches = 0;
-    for (match_t *m = NULL; next_match(&m, f->start, f->end, pattern, defs, options.skip, options.ignorecase); ) {
-        if (++nmatches > 1)
-            printf(",\n");
-        printf("{\"filename\":\"%s\",\"match\":", f->filename);
-        json_match(f->start, m, options.verbose);
-        printf("}");
-    }
-    return nmatches;
+    file_t *f = (file_t*)data;
+    if (++matchnum > 1)
+        printf(",\n");
+    printf("{\"match\":");
+    json_match(f->start, m, options.verbose);
+    printf("}");
+    return BP_CONTINUE;
 }
 
 //
 // Print matches in a visual explanation style
 //
-static int explain_matches(file_t *f, pat_t *pattern, pat_t *defs)
+static bp_match_behavior print_match_explanation(match_t *m, int matchnum, void *data)
 {
-    int nmatches = 0;
-    for (match_t *m = NULL; next_match(&m, f->start, f->end, pattern, defs, options.skip, options.ignorecase); ) {
-        if (++nmatches == 1) {
-            if (options.print_filenames)
-                fprint_filename(stdout, f->filename);
-        } else
-            printf("\n\n");
-        explain_match(m);
-    }
-    return nmatches;
+    file_t *f = (file_t*)data;
+    if (matchnum == 0) {
+        if (options.print_filenames)
+            fprint_filename(stdout, f->filename);
+    } else
+        printf("\n\n");
+    explain_match(m);
+    return BP_CONTINUE;
 }
 
 //
@@ -346,40 +346,54 @@ static void on_nl(FILE *out)
     }
 }
 
+typedef struct {
+    file_t *f;
+    FILE *out;
+    const char *prev;
+    print_options_t print_opts;
+} print_userdata_t;
+
+//
+// Print the text of a single match with the chosen context info
+//
+static bp_match_behavior print_single_match(match_t *m, int matchnum, void *data)
+{
+    print_userdata_t *info = data;
+    if (matchnum == 0 && options.print_filenames) {
+        if (files_printed++ > 0)
+            fprintf(info->out, "\n");
+        fprint_filename(info->out, info->f->filename);
+    }
+    fprint_context(info->out, info->f, info->prev, m->start);
+    if (info->print_opts.normal_color) fprintf(info->out, "%s", info->print_opts.normal_color);
+    fprint_match(info->out, info->f->start, m, &info->print_opts);
+    if (info->print_opts.normal_color) fprintf(info->out, "%s", info->print_opts.normal_color);
+    info->prev = m->end;
+    return BP_CONTINUE;
+}
+
 //
 // Print all the matches in a file.
 //
 static int print_matches(FILE *out, file_t *f, pat_t *pattern, pat_t *defs)
 {
-    static int printed_filenames = 0;
-    int matches = 0;
-    const char *prev = NULL;
+    print_userdata_t userdata = {.out = out, .f = f, .prev = NULL};
+    userdata.print_opts = (print_options_t){.fprint_between = _fprint_between, .on_nl = on_nl};
 
     printing_file = f;
     last_line_num = -1;
 
-    print_options_t print_opts = {.fprint_between = _fprint_between, .on_nl = on_nl};
     if (options.format == FORMAT_FANCY) {
-        print_opts.match_color = "\033[0;31;1m";
-        print_opts.replace_color = "\033[0;34;1m";
-        print_opts.normal_color = "\033[m";
-    }
-    for (match_t *m = NULL; next_match(&m, f->start, f->end, pattern, defs, options.skip, options.ignorecase); ) {
-        if (++matches == 1 && options.print_filenames) {
-            if (printed_filenames++ > 0) printf("\n");
-            fprint_filename(out, f->filename);
-        }
-        fprint_context(out, f, prev, m->start);
-        if (print_opts.normal_color) fprintf(out, "%s", print_opts.normal_color);
-        fprint_match(out, f->start, m, &print_opts);
-        if (print_opts.normal_color) fprintf(out, "%s", print_opts.normal_color);
-        prev = m->end;
+        userdata.print_opts.match_color = "\033[0;31;1m";
+        userdata.print_opts.replace_color = "\033[0;34;1m";
+        userdata.print_opts.normal_color = "\033[m";
     }
+    int matches = each_match_file(print_single_match, &userdata, f, pattern, defs);
     // Print trailing context if needed:
     if (matches > 0) {
-        fprint_context(out, f, prev, NULL);
+        fprint_context(out, f, userdata.prev, NULL);
         if (last_line_num < 0) { // Hacky fix to ensure line number gets printed for `bp -p '$$'`
-            fprint_linenum(out, f, f->nlines, print_opts.normal_color);
+            fprint_linenum(out, f, f->nlines, userdata.print_opts.normal_color);
             fputc('\n', out);
         }
     }
@@ -404,21 +418,14 @@ static int process_file(const char *filename, pat_t *pattern, pat_t *defs)
 
     int matches = 0;
     if (options.mode == MODE_EXPLAIN) {
-        matches += explain_matches(f, pattern, defs);
+        matches += each_match_file(print_match_explanation, f, f, pattern, defs);
     } else if (options.mode == MODE_LISTFILES) {
-        match_t *m = NULL;
-        if (next_match(&m, f->start, f->end, pattern, defs, options.skip, options.ignorecase)) {
-            printf("%s\n", f->filename);
-            matches += 1;
-        }
-        stop_matching(&m);
+        matches += each_match_file((bp_match_callback)(BP_STOP), f, f, pattern, defs);
     } else if (options.mode == MODE_JSON) {
-        matches += print_matches_as_json(f, pattern, defs);
+        matches += each_match_file(print_match_as_json, f, f, pattern, defs);
     } else if (options.mode == MODE_INPLACE) {
-        match_t *m = NULL;
-        bool found = next_match(&m, f->start, f->end, pattern, defs, options.skip, options.ignorecase);
-        stop_matching(&m);
-        if (!found) return 0;
+        if (each_match_file((bp_match_callback)(BP_STOP), f, f, pattern, defs) == 0)
+            return 0;
 
         // Ensure the file is resident in memory:
         if (f->mmapped) {
@@ -441,8 +448,6 @@ static int process_file(const char *filename, pat_t *pattern, pat_t *defs)
     }
     fflush(stdout);
 
-    if (recycle_all_matches() != 0)
-        fprintf(stderr, "\033[33;1mMemory leak: there should no longer be any matches in use at this point.\033[m\n");
     destroy_file(&f);
     (void)fflush(stdout);
     return matches;
@@ -682,7 +687,6 @@ int main(int argc, char *argv[])
     // This code frees up all residual heap-allocated memory. Since the program
     // is about to exit, this step is unnecessary. However, it is useful for
     // tracking down memory leaks.
-    free_all_matches();
     free_all_pats();
     while (loaded_files) {
         file_t *next = loaded_files->next;
diff --git a/match.c b/match.c
index 0ff6243..c07086f 100644
--- a/match.c
+++ b/match.c
@@ -9,6 +9,7 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <setjmp.h>
 
 #include "match.h"
 #include "pattern.h"
@@ -49,12 +50,15 @@ typedef struct match_ctx_s {
 static match_t *unused_matches = NULL;
 static match_t *in_use_matches = NULL;
 
-static void default_error_handler(pat_t *pat, const char *msg) {
-    (void)pat;
+static void default_error_handler(const char *msg) {
     errx(EXIT_FAILURE, "%s", msg);
 }
 
+// In case of errors, jump out of everything, clean up memory, and call the error handler.
+// Errors in this case are things like referencing a rule that isn't defined.
+static jmp_buf error_jump;
 static bp_errhand_t error_handler = default_error_handler;
+static char *error_message = NULL;
 
 #define MATCHES(...) (match_t*[]){__VA_ARGS__, NULL}
 
@@ -62,17 +66,19 @@ __attribute__((hot, nonnull(1,2,3)))
 static match_t *match(match_ctx_t *ctx, const char *str, pat_t *pat);
 __attribute__((returns_nonnull))
 static match_t *new_match(pat_t *pat, const char *start, const char *end, match_t *children[]);
+__attribute__((nonnull))
+static void recycle_match(match_t **at_m);
+static size_t free_all_matches(void);
+static size_t recycle_all_matches(void);
 
-__attribute__((format(printf,2,3)))
-static inline void match_error(pat_t *pat, const char *fmt, ...)
+__attribute__((format(printf,1,2)))
+static inline void match_error(const char *fmt, ...)
 {
     va_list args;
     va_start(args, fmt);
-    char buf[256];
-    vsnprintf(buf, sizeof(buf)-1, fmt, args);
+    vasprintf(&error_message, fmt, args);
     va_end(args);
-    if (error_handler)
-        error_handler(pat, buf);
+    longjmp(error_jump, 1);
 }
 
 static match_t *clone_match(match_t *m)
@@ -226,7 +232,8 @@ static pat_t *_lookup_def(pat_t *defs, const char *name, size_t namelen)
                 return defs->args.def.meaning;
             defs = defs->args.def.next_def;
         } else {
-            match_error(defs, "Invalid pattern type in definitions");
+            free_all_matches();
+            match_error("Invalid pattern type in definitions");
             return NULL;
         }
     }
@@ -667,7 +674,8 @@ static match_t *match(match_ctx_t *ctx, const char *str, pat_t *pat)
 
         pat_t *ref = lookup_ctx(ctx, pat->args.ref.name, pat->args.ref.len);
         if (ref == NULL) {
-            match_error(pat, "Unknown pattern: '%.*s'", (int)pat->args.ref.len, pat->args.ref.name);
+            free_all_matches();
+            match_error("Unknown pattern: '%.*s'", (int)pat->args.ref.len, pat->args.ref.name);
             return NULL;
         }
 
@@ -754,7 +762,8 @@ static match_t *match(match_ctx_t *ctx, const char *str, pat_t *pat)
         return new_match(pat, str, str, NULL);
     }
     default: {
-        match_error(pat, "Unknown pattern type: %u", pat->type);
+        free_all_matches();
+        match_error("Unknown pattern type: %u", pat->type);
         return NULL;
     }
     }
@@ -839,23 +848,11 @@ size_t free_all_matches(void)
 
 //
 // Iterate over matches.
-// Usage: for (match_t *m = NULL; next_match(&m, ...); ) {...}
 //
-bool next_match(match_t **m, const char *start, const char *end, pat_t *pat, pat_t *defs, pat_t *skip, bool ignorecase)
+int each_match(bp_match_callback callback, void *userdata, const char *start, const char *end, pat_t *pat, pat_t *defs, pat_t *skip, bool ignorecase)
 {
-    const char *pos;
-    if (*m) {
-        // Make sure forward progress is occurring, even after zero-width matches:
-        pos = ((*m)->end > (*m)->start) ? (*m)->end : (*m)->end+1;
-        recycle_match(m);
-    } else {
-        pos = start;
-    }
-
-    if (!pat) {
-        error_handler = default_error_handler;
-        return false;
-    }
+    if (!callback || !start || !pat) return -1;
+    if (!end) end = start + strlen(start);
 
     match_ctx_t ctx = {
         .cache = &(cache_t){0},
@@ -864,20 +861,53 @@ bool next_match(match_t **m, const char *start, const char *end, pat_t *pat, pat
         .ignorecase = ignorecase,
         .defs = defs,
     };
-    *m = (pos <= end) ? _next_match(&ctx, pos, pat, skip) : NULL;
+
+    int num_matches = 0;
+    pat_t *first = get_prerequisite(&ctx, pat);
+    // Don't bother looping if this can only match at the start/end:
+    if (first->type == BP_START_OF_FILE || first->type == BP_END_OF_FILE) {
+        match_t *m = match(&ctx, first->type == BP_START_OF_FILE ? start : end, pat);
+        if (m) {
+            if ((size_t)callback > 3)
+                (void)callback(m, num_matches, userdata);
+            ++num_matches;
+        }
+        return num_matches;
+    }
+
+    if (setjmp(error_jump) == 0) {
+        for (const char *str = start; str <= end; ) {
+            match_t *m = _next_match(&ctx, str, pat, skip);
+            if (!m) break;
+            else if (callback == (void*)BP_STOP)
+                break;
+            else if (callback != (void*)BP_CONTINUE && callback(m, num_matches++, userdata) == BP_STOP)
+                break;
+            else if (str == m->start && str == end)
+                break;
+            str = m->end > str ? m->end : next_char(str, end);
+            recycle_all_matches();
+        }
+    } else {
+        if (error_handler)
+            error_handler(error_message ? error_message : "An unknown error occurred");
+    }
+
+    if (error_message) {
+        free(error_message);
+        error_message = NULL;
+    }
     cache_destroy(&ctx);
-    return *m != NULL;
+    free_all_matches();
+
+    return num_matches;
 }
 
-//
-// Wrapper for next_match() that sets an error handler
-//
-bool next_match_safe(match_t **m, const char *start, const char *end, pat_t *pat, pat_t *defs, pat_t *skip, bool ignorecase, bp_errhand_t errhand)
+bp_errhand_t set_match_error_handler(bp_errhand_t errhand)
 {
+    bp_errhand_t old_errhand = errhand;
     error_handler = errhand;
-    bool ret = next_match(m, start, end, pat, defs, skip, ignorecase);
-    error_handler = default_error_handler;
-    return ret;
+    return old_errhand;
 }
 
 //
diff --git a/match.h b/match.h
index cf4f977..9e5d050 100644
--- a/match.h
+++ b/match.h
@@ -25,15 +25,13 @@ typedef struct match_s {
     struct match_s *_children[3];
 } match_t;
 
-typedef void (*bp_errhand_t)(pat_t *pat, const char *err_msg);
+typedef void (*bp_errhand_t)(const char *err_msg);
+
+typedef enum {BP_STOP = 0, BP_CONTINUE} bp_match_behavior;
+typedef bp_match_behavior (*bp_match_callback)(match_t *m, int matchnum, void *userdata);
+int each_match(bp_match_callback fn, void *userdata, const char *start, const char *end, pat_t *pat, pat_t *defs, pat_t *skip, bool ignorecase);
+bp_errhand_t set_match_error_handler(bp_errhand_t errhand);
 
-__attribute__((nonnull))
-void recycle_match(match_t **at_m);
-size_t free_all_matches(void);
-size_t recycle_all_matches(void);
-bool next_match(match_t **m, const char *start, const char *end, pat_t *pat, pat_t *defs, pat_t *skip, bool ignorecase);
-#define stop_matching(m) next_match(m, NULL, NULL, NULL, NULL, NULL, 0)
-bool next_match_safe(match_t **m, const char *start, const char *end, pat_t *pat, pat_t *defs, pat_t *skip, bool ignorecase, bp_errhand_t errhand);
 __attribute__((nonnull))
 match_t *get_numbered_capture(match_t *m, int n);
 __attribute__((nonnull, pure))