read_file utils

2 files changed, 113 insertions, 0 deletions

diff --git a/src/io/read_file.c b/src/io/read_file.c
new file mode 100644
index 0000000..b2f4604
--- /dev/null
+++ b/src/io/read_file.c
@@ -0,0 +1,94 @@
+#include <fcntl.h>
+#include <stdlib.h>
+#include <sys/mman.h>
+#include <sys/stat.h>
+#include <unistd.h>
+
+#include "read_file.h"
+
+#define CHECK_OOM(ptr) \
+  if (ptr == NULL) { \
+    fprintf(stderr, "geheugen is op"); \
+    /* I think I should free the original memory here, but whatever */ \
+    return NULL; \
+  }
+
+struct filebuf *stream_to_filebuf(FILE *restrict stream) {
+  size_t sz = 0;
+  size_t cap = 4096;
+  char *buf = malloc(cap);
+  CHECK_OOM(buf);
+
+  while (!feof(stream)) {
+    if (cap-sz == 0) {
+      cap *= 2;
+      buf = realloc(buf, cap);
+      CHECK_OOM(buf);
+    }
+
+    const size_t amount = cap-sz;
+    const size_t n = fread(buf+sz, 1, amount, stream);
+    sz += n;
+
+    if (n != amount && !feof(stream)) {
+      fprintf(stderr, "fout tijdens lezen van standaard invoer.\n");
+      free(buf);
+      return NULL;
+    }
+  }
+
+  buf = realloc(buf, sz);
+
+  struct filebuf *res = calloc(1, sizeof(struct filebuf));
+  res->buf = buf;
+  res->sz = sz;
+  res->mapping_type = MT_MMAP;
+  return res;
+}
+
+static void *fd_to_mmap(const int fd, struct stat sb) {
+  void *addr = mmap(NULL, sb.st_size, PROT_READ, MAP_SHARED, fd, 0);
+  if (addr == MAP_FAILED) {
+    return NULL;
+  }
+
+  close(fd);
+  return addr;
+}
+
+struct filebuf *file_to_filebuf(char *fname) {
+  const int fd = open(fname, O_RDONLY);
+
+  struct stat sb;
+  if (fstat(fd, &sb) == -1) {
+    goto stream;
+  }
+
+  void *addr;
+  if ((addr = fd_to_mmap(fd, sb)) != NULL) {
+    struct filebuf *res = calloc(1, sizeof(struct filebuf));
+    res->buf = addr;
+    res->sz = sb.st_size;
+    res->mapping_type = MT_MMAP;
+    return res;
+  }
+
+stream: {
+  FILE *stream = fdopen(fd, "rb");
+  struct filebuf *res = stream_to_filebuf(stream);
+  fclose(stream);
+  return res;
+}
+}
+
+void free_filebuf(struct filebuf *fb) {
+  switch (fb->mapping_type) {
+    case MT_MMAP:
+      munmap(fb->buf, fb->sz);
+      break;
+    case MT_OWNED:
+      free(fb->buf);
+      break;
+  }
+  free(fb);
+}
diff --git a/src/io/read_file.h b/src/io/read_file.h
new file mode 100644
index 0000000..201ae5b
--- /dev/null
+++ b/src/io/read_file.h
@@ -0,0 +1,19 @@
+#pragma once
+
+#include <stddef.h>
+#include <stdio.h>
+
+enum mapping_type {
+  MT_MMAP,
+  MT_OWNED,
+};
+
+struct filebuf {
+  char *buf;
+  size_t sz;
+  enum mapping_type mapping_type;
+};
+
+struct filebuf *stream_to_filebuf(FILE *restrict stream);
+struct filebuf *file_to_filebuf(char *fname);
+void free_filebuf(struct filebuf *filebuf);