WIP

13 files changed, 225 insertions, 132 deletions

diff --git a/Makefile b/Makefile
index 2eb7f56..e914372 100644
--- a/Makefile
+++ b/Makefile
@@ -6,7 +6,7 @@ CWARN=-Wall -Wpedantic -Wextra -Wno-unknown-pragmas -Wno-missing-field-initializ
 G ?=
 O ?= -O3
 
-CFILES=compiler.c grammar.c utils.c vm.c
+CFILES=compiler.c grammar.c utils.c vm.c file_loader.c
 OBJFILES=$(CFILES:.c=.o)
 
 all: $(NAME)
diff --git a/bpeg.c b/bpeg.c
index 7e89dba..b8b4c0d 100644
--- a/bpeg.c
+++ b/bpeg.c
@@ -12,6 +12,7 @@
 #include <unistd.h>
 
 #include "compiler.h"
+#include "file_loader.h"
 #include "grammar.h"
 #include "utils.h"
 #include "vm.h"
@@ -50,25 +51,18 @@ static char *getflag(const char *flag, char *argv[], int *i)
 
 static int run_match(grammar_t *g, const char *filename, vm_op_t *pattern, unsigned int flags)
 {
-    char *input;
-    if (filename == NULL || streq(filename, "-")) {
-        input = readfile(STDIN_FILENO);
-    } else {
-        int fd = open(filename, O_RDONLY);
-        check(fd >= 0, "Couldn't open file: %s", filename);
-        input = readfile(fd);
-    }
-    match_t *m = match(g, input, pattern, flags);
+    file_t *f = load_file(filename);
+    match_t *m = match(g, f, f->contents, pattern, flags);
     if (m != NULL && m->end > m->start + 1) {
         if (filename != NULL) {
             if (isatty(STDOUT_FILENO)) printf("\033[1;4;33m%s\033[0m\n", filename);
             else printf("%s\n", filename);
         }
         print_match(m, isatty(STDOUT_FILENO) ? "\033[0m" : NULL, (flags & BPEG_VERBOSE) != 0);
-        freefile(input);
+        destroy_file(&f);
         return 0;
     } else {
-        freefile(input);
+        destroy_file(&f);
         return 1;
     }
 }
@@ -85,12 +79,11 @@ int main(int argc, char *argv[])
     grammar_t *g = new_grammar();
 
     // Load builtins:
-    int fd;
-    if ((fd=open("/etc/xdg/bpeg/builtins.bpeg", O_RDONLY)) >= 0)
-        load_grammar(g, readfile(fd)); // Keep in memory for debugging output
+    if (access("/etc/xdg/bpeg/builtins.bpeg", R_OK) != -1)
+        load_grammar(g, load_file("/etc/xdg/bpeg/builtins.bpeg")); // Keep in memory for debugging output
     sprintf(path, "%s/.config/bpeg/builtins.bpeg", getenv("HOME"));
-    if ((fd=open(path, O_RDONLY)) >= 0)
-        load_grammar(g, readfile(fd)); // Keep in memory for debugging output
+    if (access(path, R_OK) != -1)
+        load_grammar(g, load_file(path)); // Keep in memory for debugging output
 
     int i, npatterns = 0;
     check(argc > 1, "%s", usage);
@@ -106,63 +99,58 @@ int main(int argc, char *argv[])
         } else if (streq(argv[i], "--ignore-case") || streq(argv[i], "-i")) {
             flags |= BPEG_IGNORECASE;
         } else if (FLAG("--replace") || FLAG("-r")) {
-            vm_op_t *p = bpeg_replacement(bpeg_pattern("pattern"), flag);
+            vm_op_t *p = bpeg_replacement(bpeg_pattern(NULL, "pattern"), flag);
             check(p, "Replacement failed to compile");
-            add_def(g, flag, "replacement", p);
+            add_def(g, NULL, flag, "replacement", p);
             rule = "replace-all";
         } else if (FLAG("--grammar") || FLAG("-g")) {
-            int fd;
-            if (streq(flag, "-")) {
-                fd = STDIN_FILENO;
-            } else {
-                fd = open(flag, O_RDONLY);
-                if (fd < 0) {
-                    sprintf(path, "%s/.config/bpeg/%s.bpeg", getenv("HOME"), flag);
-                    fd = open(path, O_RDONLY);
-                }
-                if (fd < 0) {
-                    sprintf(path, "/etc/xdg/bpeg/%s.bpeg", flag);
-                    fd = open(path, O_RDONLY);
-                }
-                check(fd >= 0, "Couldn't find grammar: %s", flag);
+            file_t *f = load_file(flag);
+            if (f == NULL) {
+                sprintf(path, "%s/.config/bpeg/%s.bpeg", getenv("HOME"), flag);
+                f = load_file(path);
+            }
+            if (f == NULL) {
+                sprintf(path, "/etc/xdg/bpeg/%s.bpeg", flag);
+                f = load_file(path);
             }
-            load_grammar(g, readfile(fd)); // Keep in memory for debug output
+            check(f != NULL, "Couldn't find grammar: %s", flag);
+            load_grammar(g, f); // Keep in memory for debug output
         } else if (FLAG("--define") || FLAG("-d")) {
             char *def = flag;
             char *eq = strchr(def, '=');
             check(eq, "Rule definitions must include an '='\n\n%s", usage);
             *eq = '\0';
             char *src = ++eq;
-            vm_op_t *pat = bpeg_pattern(src);
+            vm_op_t *pat = bpeg_pattern(NULL, src);
             check(pat, "Failed to compile pattern");
-            add_def(g, src, def, pat);
+            add_def(g, NULL, src, def, pat);
         } else if (FLAG("--define-string") || FLAG("-D")) {
             char *def = flag;
             char *eq = strchr(def, '=');
             check(eq, "Rule definitions must include an '='\n\n%s", usage);
             *eq = '\0';
             char *src = ++eq;
-            vm_op_t *pat = bpeg_stringpattern(src);
+            vm_op_t *pat = bpeg_stringpattern(NULL, src);
             check(pat, "Failed to compile pattern");
-            add_def(g, src, def, pat);
+            add_def(g, NULL, src, def, pat);
         } else if (FLAG("--pattern") || FLAG("-p")) {
             check(npatterns == 0, "Cannot define multiple patterns");
-            vm_op_t *p = bpeg_pattern(flag);
+            vm_op_t *p = bpeg_pattern(NULL, flag);
             check(p, "Pattern failed to compile: '%s'", flag);
-            add_def(g, flag, "pattern", p);
+            add_def(g, NULL, flag, "pattern", p);
             ++npatterns;
         } else if (FLAG("--pattern-string") || FLAG("-P")) {
-            vm_op_t *p = bpeg_stringpattern(flag);
+            vm_op_t *p = bpeg_stringpattern(NULL, flag);
             check(p, "Pattern failed to compile");
-            add_def(g, flag, "pattern", p);
+            add_def(g, NULL, flag, "pattern", p);
             ++npatterns;
         } else if (FLAG("--mode") || FLAG("-m")) {
             rule = flag;
         } else if (argv[i][0] != '-') {
             if (npatterns > 0) break;
-            vm_op_t *p = bpeg_stringpattern(argv[i]);
+            vm_op_t *p = bpeg_stringpattern(NULL, argv[i]);
             check(p, "Pattern failed to compile");
-            add_def(g, argv[i], "pattern", p);
+            add_def(g, NULL, argv[i], "pattern", p);
             ++npatterns;
         } else {
             printf("Unrecognized flag: %s\n\n%s\n", argv[i], usage);
diff --git a/compiler.c b/compiler.c
index b27717b..fe7ad3f 100644
--- a/compiler.c
+++ b/compiler.c
@@ -5,8 +5,8 @@
 #include "compiler.h"
 #include "utils.h"
 
-static vm_op_t *expand_chain(vm_op_t *first);
-static vm_op_t *expand_choices(vm_op_t *first);
+static vm_op_t *expand_chain(file_t *f, vm_op_t *first);
+static vm_op_t *expand_choices(file_t *f, vm_op_t *first);
 static vm_op_t *chain_together(vm_op_t *first, vm_op_t *second);
 static void set_range(vm_op_t *op, ssize_t min, ssize_t max, vm_op_t *pat, vm_op_t *sep);
 
@@ -31,11 +31,11 @@ static void set_range(vm_op_t *op, ssize_t min, ssize_t max, vm_op_t *pat, vm_op
  * followed by any patterns (e.g. "`x `y"), otherwise return
  * the original input.
  */
-static vm_op_t *expand_chain(vm_op_t *first)
+static vm_op_t *expand_chain(file_t *f, vm_op_t *first)
 {
-    vm_op_t *second = bpeg_simplepattern(first->end);
+    vm_op_t *second = bpeg_simplepattern(f, first->end);
     if (second == NULL) return first;
-    second = expand_chain(second);
+    second = expand_chain(f, second);
     check(second->end > first->end, "No forward progress in chain!");
     return chain_together(first, second);
 }
@@ -45,14 +45,14 @@ static vm_op_t *expand_chain(vm_op_t *first)
  * followed by any "/"-separated patterns (e.g. "`x/`y"), otherwise
  * return the original input.
  */
-static vm_op_t *expand_choices(vm_op_t *first)
+static vm_op_t *expand_choices(file_t *f, vm_op_t *first)
 {
-    first = expand_chain(first);
+    first = expand_chain(f, first);
     const char *str = first->end;
     if (!matchchar(&str, '/')) return first;
-    vm_op_t *second = bpeg_simplepattern(str);
+    vm_op_t *second = bpeg_simplepattern(f, str);
     check(second, "Expected pattern after '/'");
-    second = expand_choices(second);
+    second = expand_choices(f, second);
     vm_op_t *choice = calloc(sizeof(vm_op_t), 1);
     choice->op = VM_OTHERWISE;
     choice->start = first->start;
@@ -84,7 +84,7 @@ static vm_op_t *chain_together(vm_op_t *first, vm_op_t *second)
 /*
  * Compile a string of BPEG code into virtual machine opcodes
  */
-vm_op_t *bpeg_simplepattern(const char *str)
+vm_op_t *bpeg_simplepattern(file_t *f, const char *str)
 {
     if (!*str) return NULL;
     str = after_spaces(str);
@@ -100,7 +100,7 @@ vm_op_t *bpeg_simplepattern(const char *str)
             if (matchchar(&str, '.')) { // ".."
                 if (matchchar(&str, '.')) // "..."
                     op->multiline = 1;
-                vm_op_t *till = bpeg_simplepattern(str);
+                vm_op_t *till = bpeg_simplepattern(f, str);
                 op->op = VM_UPTO_AND;
                 op->len = -1;
                 op->args.pat = till;
@@ -177,7 +177,7 @@ vm_op_t *bpeg_simplepattern(const char *str)
         }
         // Not <pat>
         case '!': {
-            vm_op_t *p = bpeg_simplepattern(str);
+            vm_op_t *p = bpeg_simplepattern(f, str);
             check(p, "Expected pattern after '!'\n");
             str = p->end;
             op->op = VM_NOT;
@@ -202,13 +202,13 @@ vm_op_t *bpeg_simplepattern(const char *str)
             } else {
                 min = n1, max = n1;
             }
-            vm_op_t *pat = bpeg_simplepattern(str);
+            vm_op_t *pat = bpeg_simplepattern(f, str);
             check(pat, "Expected pattern after repetition count");
             str = pat->end;
             str = after_spaces(str);
             vm_op_t *sep = NULL;
             if (matchchar(&str, '%')) {
-                sep = bpeg_simplepattern(str);
+                sep = bpeg_simplepattern(f, str);
                 check(sep, "Expected pattern for separator after '%%'");
                 str = sep->end;
             } else {
@@ -219,7 +219,7 @@ vm_op_t *bpeg_simplepattern(const char *str)
         }
         // Lookbehind
         case '<': {
-            vm_op_t *pat = bpeg_simplepattern(str);
+            vm_op_t *pat = bpeg_simplepattern(f, str);
             check(pat, "Expected pattern after <");
             str = pat->end;
             check(pat->len != -1, "Lookbehind patterns must have a fixed length");
@@ -231,7 +231,7 @@ vm_op_t *bpeg_simplepattern(const char *str)
         }
         // Lookahead
         case '>': {
-            vm_op_t *pat = bpeg_simplepattern(str);
+            vm_op_t *pat = bpeg_simplepattern(f, str);
             check(pat, "Expected pattern after >");
             str = pat->end;
             op->op = VM_BEFORE;
@@ -242,9 +242,9 @@ vm_op_t *bpeg_simplepattern(const char *str)
         // Parentheses
         case '(': {
             free(op);
-            op = bpeg_simplepattern(str);
+            op = bpeg_simplepattern(f, str);
             check(op, "Expected pattern inside parentheses");
-            op = expand_choices(op);
+            op = expand_choices(f, op);
             str = op->end;
             str = after_spaces(str);
             check(matchchar(&str, ')'), "Expected closing ')' instead of \"%s\"", str);
@@ -261,7 +261,7 @@ vm_op_t *bpeg_simplepattern(const char *str)
                 str = closing;
                 check(matchchar(&str, ']'), "Expected closing ']'");
             }
-            vm_op_t *pat = bpeg_simplepattern(str);
+            vm_op_t *pat = bpeg_simplepattern(f, str);
             check(pat, "Expected pattern after @");
             str = pat->end;
             op->args.capture.capture_pat = pat;
@@ -275,9 +275,9 @@ vm_op_t *bpeg_simplepattern(const char *str)
             if (strncmp(str, "=>", 2) == 0) {
                 str += strlen("=>");
             } else {
-                pat = bpeg_simplepattern(str);
+                pat = bpeg_simplepattern(f, str);
                 check(pat, "Invalid pattern after '{'");
-                pat = expand_choices(pat);
+                pat = expand_choices(f, pat);
                 str = pat->end;
                 str = after_spaces(str);
                 check(matchchar(&str, '=') && matchchar(&str, '>'),
@@ -360,7 +360,7 @@ vm_op_t *bpeg_simplepattern(const char *str)
     if (strncmp(after_spaces(str), "==", 2) == 0) {
         str = after_spaces(str)+2;
         vm_op_t *first = op;
-        vm_op_t *second = bpeg_simplepattern(str);
+        vm_op_t *second = bpeg_simplepattern(f, str);
         check(second, "Expected pattern after '=='");
         check(first->len == -1 || second->len == -1 || first->len == second->len,
               "Two patterns cannot possibly match the same (different lengths: %ld != %ld)",
@@ -382,7 +382,7 @@ vm_op_t *bpeg_simplepattern(const char *str)
 /*
  * Similar to bpeg_simplepattern, except that the pattern begins with an implicit, unclosable quote.
  */
-vm_op_t *bpeg_stringpattern(const char *str)
+vm_op_t *bpeg_stringpattern(file_t *f, const char *str)
 {
     vm_op_t *ret = NULL;
     while (*str) {
@@ -395,7 +395,7 @@ vm_op_t *bpeg_stringpattern(const char *str)
         for (; *str; str++) {
             if (*str == '\\') {
                 check(str[1], "Expected more string contents after backslash");
-                interp = bpeg_simplepattern(str + 1);
+                interp = bpeg_simplepattern(f, str + 1);
                 check(interp != NULL, "No valid BPEG pattern detected after backslash");
                 break;
             }
@@ -448,10 +448,10 @@ vm_op_t *bpeg_replacement(vm_op_t *pat, const char *replacement)
     return op;
 }
 
-vm_op_t *bpeg_pattern(const char *str)
+vm_op_t *bpeg_pattern(file_t *f, const char *str)
 {
-    vm_op_t *op = bpeg_simplepattern(str);
-    if (op != NULL) op = expand_choices(op);
+    vm_op_t *op = bpeg_simplepattern(f, str);
+    if (op != NULL) op = expand_choices(f, op);
     return op;
 }
 
diff --git a/compiler.h b/compiler.h
index 855b05d..005eacc 100644
--- a/compiler.h
+++ b/compiler.h
@@ -7,11 +7,12 @@
 #include <stdlib.h>
 
 #include "types.h"
+#include "file_loader.h"
 
-vm_op_t *bpeg_simplepattern(const char *str);
-vm_op_t *bpeg_stringpattern(const char *str);
+vm_op_t *bpeg_simplepattern(file_t *f, const char *str);
+vm_op_t *bpeg_stringpattern(file_t *f, const char *str);
 vm_op_t *bpeg_replacement(vm_op_t *pat, const char *replacement);
-vm_op_t *bpeg_pattern(const char *str);
+vm_op_t *bpeg_pattern(file_t *f, const char *str);
 
 #endif
 // vim: ts=4 sw=0 et cino=L2,l1,(0,W4,m1
diff --git a/file_loader.c b/file_loader.c
new file mode 100644
index 0000000..b7c3817
--- /dev/null
+++ b/file_loader.c
@@ -0,0 +1,102 @@
+/*
+ * file_loader.c - Implementation of some file loading functionality.
+ */
+
+#include <ctype.h>
+#include <fcntl.h>
+#include <limits.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+
+#include "file_loader.h"
+
+/*
+ * Read an entire file into memory.
+ */
+file_t *load_file(const char *filename)
+{
+    if (filename == NULL) filename = "-";
+    int fd = strcmp(filename, "-") != 0 ? open(filename, O_RDONLY) : STDIN_FILENO;
+    if (fd < 0) return NULL;
+    file_t *f = calloc(sizeof(file_t), 1);
+    f->filename = strdup(filename);
+    // TODO: use mmap when possible
+    f->mmapped = 0;
+    size_t capacity = 1000;
+    f->length = 0;
+    f->contents = calloc(sizeof(char), capacity+1);
+    ssize_t just_read;
+    while ((just_read=read(fd, &f->contents[f->length], capacity - f->length)) > 0) {
+        f->length += (size_t)just_read;
+        if (f->length >= capacity)
+            f->contents = realloc(f->contents, sizeof(char)*(capacity *= 2) + 1);
+    }
+    f->contents[f->length] = '\0';
+    close(fd);
+    
+    // Calculate line numbers:
+    size_t linecap = 10;
+    f->lines = calloc(sizeof(const char*), linecap);
+    f->nlines = 1;
+    char *p = f->contents;
+    for (size_t n = 0; p && *p; ++n) {
+        if (n >= linecap)
+            f->lines = realloc(f->lines, sizeof(const char*)*(linecap *= 2));
+        f->lines[n] = p;
+        p = strchr(p, '\n');
+        if (p) ++p;
+    }
+
+    return f;
+}
+
+void destroy_file(file_t **f)
+{
+    if ((*f)->filename) {
+        free((char*)(*f)->filename);
+        (*f)->filename = NULL;
+    }
+    if ((*f)->lines) {
+        free((*f)->lines);
+        (*f)->lines = NULL;
+    }
+    if ((*f)->contents) {
+        free((*f)->contents);
+        (*f)->contents = NULL;
+    }
+    free(*f);
+    *f = NULL;
+}
+
+size_t get_line_number(file_t *f, const char *p)
+{
+    // TODO: binary search
+    for (size_t n = 1; n < f->nlines; n++) {
+        if (f->lines[n] > p)
+            return n;
+    }
+    return 0;
+}
+
+const char *get_line(file_t *f, size_t line_number)
+{
+    if (line_number == 0 || line_number > f->nlines) return NULL;
+    return f->lines[line_number - 1];
+}
+
+void fprint_line(FILE *dest, file_t *f, const char *start, const char *end, const char *msg)
+{
+    size_t linenum = get_line_number(f, start);
+    const char *line = get_line(f, linenum);
+    size_t charnum = 1 + (size_t)(start - line);
+    fprintf(dest, "\033[1m%s:%ld:%ld:\033[0m %s\n",
+            f->filename, linenum, charnum, msg);
+    const char *eol = linenum == f->nlines ? strchr(line, '\0') : strchr(line, '\n');
+    if (end == NULL || end > eol) end = eol;
+    fprintf(dest, "\033[2m% 5ld |\033[0m %.*s\033[31;4;1m%.*s\033[0m%.*s\n",
+            linenum,
+            (int)charnum - 1, line,
+            (int)(end - &line[charnum-1]), &line[charnum-1],
+            (int)(eol - end), end);
+}
diff --git a/file_loader.h b/file_loader.h
new file mode 100644
index 0000000..cb49373
--- /dev/null
+++ b/file_loader.h
@@ -0,0 +1,22 @@
+/*
+ * file_loader.h - Definitions of an API for loading files.
+ */
+#ifndef FILE_LOADER__H
+#define FILE_LOADER__H
+
+#include <stdio.h>
+
+typedef struct {
+    const char *filename;
+    char *contents, **lines;
+    size_t length, nlines;
+    unsigned int mmapped:1;
+} file_t;
+
+file_t *load_file(const char *filename);
+void destroy_file(file_t **f);
+size_t get_line_number(file_t *f, const char *p);
+const char *get_line(file_t *f, size_t line_number);
+void fprint_line(FILE *dest, file_t *f, const char *start, const char *end, const char *msg);
+
+#endif
diff --git a/grammar.c b/grammar.c
index 61c82c9..b4621ee 100644
--- a/grammar.c
+++ b/grammar.c
@@ -4,6 +4,7 @@
 
 #include "grammar.h"
 #include "compiler.h"
+#include "file_loader.h"
 #include "utils.h"
 
 grammar_t *new_grammar(void)
@@ -13,12 +14,13 @@ grammar_t *new_grammar(void)
     return g;
 }
 
-void add_def(grammar_t *g, const char *src, const char *name, vm_op_t *op)
+void add_def(grammar_t *g, file_t *f, const char *src, const char *name, vm_op_t *op)
 {
     if (g->defcount >= g->defcapacity) {
         g->definitions = realloc(g->definitions, sizeof(&g->definitions[0])*(g->defcapacity += 32));
     }
     int i = g->defcount;
+    g->definitions[i].file = f;
     g->definitions[i].source = src;
     g->definitions[i].name = name;
     g->definitions[i].op = op;
@@ -29,9 +31,10 @@ void add_def(grammar_t *g, const char *src, const char *name, vm_op_t *op)
  * Load the given grammar (semicolon-separated definitions)
  * and return the first rule defined.
  */
-vm_op_t *load_grammar(grammar_t *g, const char *src)
+vm_op_t *load_grammar(grammar_t *g, file_t *f)
 {
     vm_op_t *ret = NULL;
+    const char *src = f->contents;
     do {
         src = after_spaces(src);
         if (!*src) break;
@@ -41,9 +44,9 @@ vm_op_t *load_grammar(grammar_t *g, const char *src)
         name = strndup(name, (size_t)(name_end-name));
         src = after_spaces(name_end);
         check(matchchar(&src, '='), "Expected '=' in definition");
-        vm_op_t *op = bpeg_pattern(src);
+        vm_op_t *op = bpeg_pattern(f, src);
         check(op, "Couldn't load definition");
-        add_def(g, src, name, op);
+        add_def(g, f, src, name, op);
         if (ret == NULL) {
             ret = op;
         }
diff --git a/grammar.h b/grammar.h
index f75b111..0f57616 100644
--- a/grammar.h
+++ b/grammar.h
@@ -7,13 +7,14 @@
 #include <stdlib.h>
 #include <string.h>
 
+#include "file_loader.h"
 #include "types.h"
 
 grammar_t *new_grammar(void);
-void add_def(grammar_t *g, const char *src, const char *name, vm_op_t *op);
+void add_def(grammar_t *g, file_t *f, const char *src, const char *name, vm_op_t *op);
 void push_backref(grammar_t *g, const char *name, match_t *capture);
 void pop_backrefs(grammar_t *g, size_t count);
-vm_op_t *load_grammar(grammar_t *g, const char *source);
+vm_op_t *load_grammar(grammar_t *g, file_t *f);
 vm_op_t *lookup(grammar_t *g, const char *name);
 
 #endif
diff --git a/types.h b/types.h
index 3749156..de408cc 100644
--- a/types.h
+++ b/types.h
@@ -6,6 +6,8 @@
 
 #include <sys/types.h>
 
+#include "file_loader.h"
+
 enum BPEGFlag {
     BPEG_VERBOSE    = 1 << 0,
     BPEG_IGNORECASE = 1 << 1,
@@ -85,6 +87,7 @@ typedef struct match_s {
 typedef struct {
     const char *name;
     const char *source;
+    file_t *file;
     vm_op_t *op;
 } def_t;
 
diff --git a/utils.c b/utils.c
index 7844934..6b9af5c 100644
--- a/utils.c
+++ b/utils.c
@@ -153,28 +153,5 @@ size_t unescape_string(char *dest, const char *src, size_t bufsize)
 #undef PUT
 }
 
-/*
- * Read an entire file into memory. (Guaranteeing that ret[-1] == '\0')
- */
-char *readfile(int fd)
-{
-    size_t capacity = 1000, len = 0;
-    char *buf = calloc(sizeof(char), capacity+1);
-    buf[len++] = '\0';
-    ssize_t just_read;
-    while ((just_read=read(fd, &buf[len], capacity-len)) > 0) {
-        len += (size_t)just_read;
-        if (len >= capacity)
-            buf = realloc(buf, sizeof(char)*(capacity *= 2));
-    }
-    buf[len] = '\0';
-    close(fd);
-    return &buf[1];
-}
-
-void freefile(char *f)
-{
-    free(&f[-1]);
-}
 
 // vim: ts=4 sw=0 et cino=L2,l1,(0,W4,m1
diff --git a/utils.h b/utils.h
index 8588d24..28bfe05 100644
--- a/utils.h
+++ b/utils.h
@@ -17,8 +17,6 @@
 #define check(cond, ...) do { if (!(cond)) { fprintf(stderr, __VA_ARGS__); fwrite("\n", 1, 1, stderr); _exit(1); } } while(0)
 #define debug(...) do { if (verbose) fprintf(stderr, __VA_ARGS__); } while(0)
 
-char *readfile(int fd);
-void freefile(char *f);
 char unescapechar(const char *escaped, const char **end);
 const char *after_name(const char *str);
 const char *after_spaces(const char *str);
diff --git a/vm.c b/vm.c
index 7b6a04b..637c4b8 100644
--- a/vm.c
+++ b/vm.c
@@ -77,7 +77,7 @@ typedef struct recursive_ref_s {
  * a match struct, or NULL if no match is found.
  * The returned value should be free()'d to avoid memory leaking.
  */
-static match_t *_match(grammar_t *g, const char *str, vm_op_t *op, unsigned int flags, recursive_ref_t *rec)
+static match_t *_match(grammar_t *g, file_t *f, const char *str, vm_op_t *op, unsigned int flags, recursive_ref_t *rec)
 {
     switch (op->op) {
         case VM_EMPTY: {
@@ -116,7 +116,7 @@ static match_t *_match(grammar_t *g, const char *str, vm_op_t *op, unsigned int
             return m;
         }
         case VM_NOT: {
-            match_t *m = _match(g, str, op->args.pat, flags, rec);
+            match_t *m = _match(g, f, str, op->args.pat, flags, rec);
             if (m != NULL) {
                 destroy_match(&m);
                 return NULL;
@@ -134,7 +134,7 @@ static match_t *_match(grammar_t *g, const char *str, vm_op_t *op, unsigned int
             if (op->args.pat) {
                 for (const char *prev = NULL; prev < str; ) {
                     prev = str;
-                    match_t *p = _match(g, str, op->args.pat, flags, rec);
+                    match_t *p = _match(g, f, str, op->args.pat, flags, rec);
                     if (p) {
                         m->child = p;
                         m->end = p->end;
@@ -168,11 +168,11 @@ static match_t *_match(grammar_t *g, const char *str, vm_op_t *op, unsigned int
                 // Separator
                 match_t *sep = NULL;
                 if (op->args.repetitions.sep != NULL && reps > 0) {
-                    sep = _match(g, str, op->args.repetitions.sep, flags, rec);
+                    sep = _match(g, f, str, op->args.repetitions.sep, flags, rec);
                     if (sep == NULL) break;
                     str = sep->end;
                 }
-                match_t *p = _match(g, str, op->args.repetitions.repeat_pat, flags, rec);
+                match_t *p = _match(g, f, str, op->args.repetitions.repeat_pat, flags, rec);
                 if (p == NULL || (p->end == prev && reps > 0)) { // Prevent infinite loops
                     destroy_match(&sep);
                     destroy_match(&p);
@@ -204,11 +204,8 @@ static match_t *_match(grammar_t *g, const char *str, vm_op_t *op, unsigned int
         case VM_AFTER: {
             ssize_t backtrack = op->args.pat->len;
             check(backtrack != -1, "'<' is only allowed for fixed-length operations");
-            // Check for necessary space:
-            for (int i = 0; i < backtrack; i++) {
-                if (str[-i] == '\0') return NULL;
-            }
-            match_t *before = _match(g, str - backtrack, op->args.pat, flags, rec);
+            if (str - backtrack < f->contents) return NULL;
+            match_t *before = _match(g, f, str - backtrack, op->args.pat, flags, rec);
             if (before == NULL) return NULL;
             match_t *m = calloc(sizeof(match_t), 1);
             m->start = str;
@@ -218,7 +215,7 @@ static match_t *_match(grammar_t *g, const char *str, vm_op_t *op, unsigned int
             return m;
         }
         case VM_BEFORE: {
-            match_t *after = _match(g, str, op->args.pat, flags, rec);
+            match_t *after = _match(g, f, str, op->args.pat, flags, rec);
             if (after == NULL) return NULL;
             match_t *m = calloc(sizeof(match_t), 1);
             m->start = str;
@@ -228,7 +225,7 @@ static match_t *_match(grammar_t *g, const char *str, vm_op_t *op, unsigned int
             return m;
         }
         case VM_CAPTURE: {
-            match_t *p = _match(g, str, op->args.pat, flags, rec);
+            match_t *p = _match(g, f, str, op->args.pat, flags, rec);
             if (p == NULL) return NULL;
             match_t *m = calloc(sizeof(match_t), 1);
             m->start = str;
@@ -241,16 +238,16 @@ static match_t *_match(grammar_t *g, const char *str, vm_op_t *op, unsigned int
             return m;
         }
         case VM_OTHERWISE: {
-            match_t *m = _match(g, str, op->args.multiple.first, flags, rec);
-            if (m == NULL) m = _match(g, str, op->args.multiple.second, flags, rec);
+            match_t *m = _match(g, f, str, op->args.multiple.first, flags, rec);
+            if (m == NULL) m = _match(g, f, str, op->args.multiple.second, flags, rec);
             return m;
         }
         case VM_CHAIN: {
-            match_t *m1 = _match(g, str, op->args.multiple.first, flags, rec);
+            match_t *m1 = _match(g, f, str, op->args.multiple.first, flags, rec);
             if (m1 == NULL) return NULL;
 
             size_t nbackrefs = push_backrefs(g, m1);
-            match_t *m2 = _match(g, m1->end, op->args.multiple.second, flags, rec);
+            match_t *m2 = _match(g, f, m1->end, op->args.multiple.second, flags, rec);
             pop_backrefs(g, nbackrefs);
             if (m2 == NULL) {
                 destroy_match(&m1);
@@ -265,11 +262,11 @@ static match_t *_match(grammar_t *g, const char *str, vm_op_t *op, unsigned int
             return m;
         }
         case VM_EQUAL: {
-            match_t *m1 = _match(g, str, op->args.multiple.first, flags, rec);
+            match_t *m1 = _match(g, f, str, op->args.multiple.first, flags, rec);
             if (m1 == NULL) return NULL;
 
             // <p1>==<p2> matches iff both have the same start and end point:
-            match_t *m2 = _match(g, str, op->args.multiple.second, flags, rec);
+            match_t *m2 = _match(g, f, str, op->args.multiple.second, flags, rec);
             if (m2 == NULL || m2->end != m1->end) {
                 destroy_match(&m1);
                 destroy_match(&m2);
@@ -288,7 +285,7 @@ static match_t *_match(grammar_t *g, const char *str, vm_op_t *op, unsigned int
             m->start = str;
             m->op = op;
             if (op->args.replace.replace_pat) {
-                match_t *p = _match(g, str, op->args.replace.replace_pat, flags, rec);
+                match_t *p = _match(g, f, str, op->args.replace.replace_pat, flags, rec);
                 if (p == NULL) return NULL;
                 m->child = p;
                 m->end = p->end;
@@ -320,7 +317,7 @@ static match_t *_match(grammar_t *g, const char *str, vm_op_t *op, unsigned int
             };
             match_t *best = NULL;
           left_recursive:;
-            match_t *p = _match(g, str, r, flags, &wrap);
+            match_t *p = _match(g, f, str, r, flags, &wrap);
             if (p == NULL) return best;
             if (wrap.hit && (best == NULL || p->end > best->end)) {
                 best = p;
@@ -343,7 +340,10 @@ static match_t *_match(grammar_t *g, const char *str, vm_op_t *op, unsigned int
             return match_backref(str, op, (match_t*)op->args.backref, flags);
         }
         case VM_NODENT: {
-            if (str[-1] == '\0') { // First line
+            size_t linenum = get_line_number(f, str);
+            if (linenum == 1) { // First line
+                if (str > f->contents)
+                    return NULL;
                 match_t *m = calloc(sizeof(match_t), 1);
                 m->start = str;
                 m->end = str;
@@ -352,9 +352,7 @@ static match_t *_match(grammar_t *g, const char *str, vm_op_t *op, unsigned int
             } else if (str[-1] != '\n') {
                 return NULL; // Not at beginning of line
             }
-            const char *p = &str[-1];
-            while (*p == '\n') --p; // Skip blank lines
-            while (p[-1] && p[-1] != '\n') --p; // Backtrack to start of last (nonblank) line
+            const char *p = get_line(f, linenum - 1);
             // Count indentation:
             char denter = *p;
             int dents = 0;
@@ -686,9 +684,9 @@ static match_t *match_backref(const char *str, vm_op_t *op, match_t *cap, unsign
     return ret;
 }
 
-match_t *match(grammar_t *g, const char *str, vm_op_t *op, unsigned int flags)
+match_t *match(grammar_t *g, file_t *f, const char *str, vm_op_t *op, unsigned int flags)
 {
-    return _match(g, str, op, flags, NULL);
+    return _match(g, f, str, op, flags, NULL);
 }
 
 // vim: ts=4 sw=0 et cino=L2,l1,(0,W4,m1
diff --git a/vm.h b/vm.h
index ac0269f..feb8eeb 100644
--- a/vm.h
+++ b/vm.h
@@ -12,7 +12,7 @@
 #include "types.h"
 
 const char *opcode_name(enum VMOpcode o);
-match_t *match(grammar_t *g, const char *str, vm_op_t *op, unsigned int flags);
+match_t *match(grammar_t *g, file_t *f, const char *str, vm_op_t *op, unsigned int flags);
 void destroy_match(match_t **m);
 void print_pattern(vm_op_t *op);
 void print_match(match_t *m, const char *color, int verbose);