file_lines module, herschrijf omd ermee

4 files changed, 137 insertions, 72 deletions

diff --git a/src/io/lines.c b/src/io/lines.c
new file mode 100644
index 0000000..6367c1d
--- /dev/null
+++ b/src/io/lines.c
@@ -0,0 +1,92 @@
+#include <stdlib.h>
+#include <string.h>
+
+#include "lines.h"
+#include "util/debug.h"
+#include "util/error.h"
+
+
+bool file_lines_open(const char *progname, const char *fname, FILE *f, struct file_lines *dst) {
+  dst->f = f;
+
+  dst->cap = 4096;
+  dst->buffer = malloc(dst->cap);
+  if (!dst->buffer) {
+    print_error_nomem(progname);
+    return false;
+  }
+
+  dst->cursor = 0;
+  dst->readfrom = (size_t)-1;
+  dst->progname = progname;
+  dst->fname = fname;
+  return true;
+}
+
+bool file_lines_read(struct file_lines *fl, struct string_view *dst) {
+  // Return already-read line if we have any
+  if (fl->readfrom != (size_t)-1) {
+    char *p = memchr(fl->buffer + fl->readfrom, '\n', fl->cursor - fl->readfrom);
+    if (p != NULL) {
+      dst->s = fl->buffer + fl->readfrom;
+      dst->len = p - (fl->buffer + fl->readfrom);
+      fl->readfrom = p - fl->buffer + 1;
+      DEBUG("next line from %zu len %zu\n", fl->readfrom, dst->len);
+      return true;
+    }
+
+    DEBUG("no more newlines, moving back from %zu len %zu\n", fl->readfrom, fl->cursor - fl->readfrom);
+    // move back whatever we have left
+    memmove(fl->buffer, fl->buffer + fl->readfrom, fl->cursor - fl->readfrom);
+    fl->cursor -= fl->readfrom;
+  }
+
+  DEBUG("cursor=%zu\n", fl->cursor);
+
+  // read a line
+  size_t linelen = 0;
+  while (true) {
+    // grow buffer when small
+    if (fl->cap - fl->cursor < 1024) {
+      fl->cap *= 2;
+      if (fl->cap == 0) {
+        fprintf(stderr, "%s: Regel te lang\n", fl->progname);
+        exit(1);
+      }
+      DEBUG("realloc to cap %zu\n", fl->cap);
+      fl->buffer = realloc(fl->buffer, fl->cap);
+      if (!fl->buffer) print_error_nomem(fl->progname);
+    }
+
+    size_t nr = fread(fl->buffer + fl->cursor, 1, fl->cap - fl->cursor, fl->f);
+    DEBUG("nr=%zu\n", nr);
+    if (nr == 0) {
+      if (feof(fl->f)) {
+        if (fl->cursor == 0) return false;
+        linelen = fl->cursor;
+        break;
+      }
+      if (fl->fname) fprintf(stderr, "%s: Fout bij lezen van bestand '%s'\n", fl->progname, fl->fname);
+      else fprintf(stderr, "%s: Fout bij lezen van standaardinvoer\n", fl->progname);
+      exit(1);
+    }
+    char *p = memchr(fl->buffer + fl->cursor, '\n', nr);
+    fl->cursor += nr;
+    if (p != NULL) {
+      linelen = p - fl->buffer;
+      break;
+    }
+  }
+
+  DEBUG("line with len %zu (cursor=%zu)\n", linelen, fl->cursor);
+
+  // we have a newline at buffer + linelen; return lines in buffer
+  dst->s = fl->buffer;
+  dst->len = linelen;
+  fl->readfrom = linelen + 1;
+  return true;
+}
+
+void file_lines_close(struct file_lines *fl) {
+  free(fl->buffer);
+}
diff --git a/src/io/lines.h b/src/io/lines.h
new file mode 100644
index 0000000..b09b756
--- /dev/null
+++ b/src/io/lines.h
@@ -0,0 +1,25 @@
+#pragma once
+
+#include <stdio.h>
+#include <stdbool.h>
+
+#include "util/string_view.h"
+
+
+struct file_lines {
+  const char *progname, *fname;
+
+  FILE *f;
+  size_t cap;
+  char *buffer;
+  size_t cursor;
+  size_t readfrom;  // if -1, no newline in buffer
+};
+
+// Returns false on EOF. FILE remains ownership of the caller.
+bool file_lines_open(const char *progname, const char *fname, FILE *f, struct file_lines *dst);
+
+// Returns whether successful
+bool file_lines_read(struct file_lines *fl, struct string_view *dst);
+
+void file_lines_close(struct file_lines *fl);
diff --git a/src/omd.c b/src/omd.c
index b43538a..aa48ee0 100644
--- a/src/omd.c
+++ b/src/omd.c
@@ -3,6 +3,7 @@
 #include <stdlib.h>
 #include <string.h>
 #include <unistd.h>
+#include "io/lines.h"
 #include "util/versie.h"
 #include "util/error.h"
 #include "util/debug.h"
@@ -48,83 +49,20 @@ static void reverse_in_place(char *buf, size_t len) {
 }
 
 static void process(const char *fname, FILE *f) {
-  size_t cap = 4096;
-  char *buffer = malloc(cap);
-  if (!buffer) print_error_nomem("omd");
-
-  size_t cursor = 0;
+  struct file_lines fl;
+  if (!file_lines_open("omd", fname, f, &fl)) exit(1);
 
   while (true) {
-    DEBUG("cursor=%zu\n", cursor);
-
-    // read a line
-    size_t linelen = 0;
-    while (true) {
-      // grow buffer when small
-      if (cap - cursor < 1024) {
-        cap *= 2;
-        if (cap == 0) {
-          fprintf(stderr, "omd: Regel te lang\n");
-          exit(1);
-        }
-        DEBUG("realloc to cap %zu\n", cap);
-        buffer = realloc(buffer, cap);
-        if (!buffer) print_error_nomem("omd");
-      }
+    struct string_view sv;
+    if (!file_lines_read(&fl, &sv)) break;
 
-      size_t nr = fread(buffer + cursor, 1, cap - cursor, f);
-      DEBUG("nr=%zu\n", nr);
-      if (nr == 0) {
-        if (feof(f)) {
-          if (cursor == 0) return;
-          linelen = cursor;
-          break;
-        }
-        if (fname) fprintf(stderr, "omd: Fout bij lezen van bestand '%s'\n", fname);
-        else fprintf(stderr, "omd: Fout bij lezen van standaardinvoer\n");
-        exit(1);
-      }
-      char *p = memchr(buffer + cursor, '\n', nr);
-      cursor += nr;
-      if (p != NULL) {
-        linelen = p - buffer;
-        break;
-      }
-    }
-
-    DEBUG("line with len %zu (cursor=%zu)\n", linelen, cursor);
-
-    // we have a newline at cursor+linelen; print all lines that we already have in full
-    size_t linestart = 0;
-    while (true) {
-      reverse_in_place(buffer + linestart, linelen);
-      if (linestart + linelen < cap && buffer[linestart + linelen] == '\n') {
-        size_t nw = fwrite(buffer + linestart, 1, linelen + 1, stdout);
-        if (nw < linelen + 1) exit(1);
-      } else {
-        size_t nw = fwrite(buffer + linestart, 1, linelen, stdout);
-        if (nw < linelen) exit(1);
-        putchar('\n');
-        // no newline in the buffer but nevertheless a completed line means that input is at EOF
-        goto cleanup;
-      }
-
-      linestart += linelen + 1;
-      char *p = memchr(buffer + linestart, '\n', cursor - linestart);
-      if (p == NULL) {
-        DEBUG("no more newlines, moving back from %zu len %zu\n", linestart, cursor - linestart);
-        // move back whatever we have left
-        memmove(buffer, buffer + linestart, cursor - linestart);
-        cursor -= linestart;
-        break;
-      }
-      linelen = p - (buffer + linestart);
-      DEBUG("next line from %zu len %zu\n", linestart, linelen);
-    }
+    reverse_in_place(sv.s, sv.len);
+    size_t nw = fwrite(sv.s, 1, sv.len, stdout);
+    if (nw < sv.len) exit(1);
+    putchar('\n');
   }
 
-cleanup:
-  free(buffer);
+  file_lines_close(&fl);
 }
 
 int entry_omd(int argc, char **argv) {
diff --git a/src/util/string_view.h b/src/util/string_view.h
new file mode 100644
index 0000000..82f7b0c
--- /dev/null
+++ b/src/util/string_view.h
@@ -0,0 +1,10 @@
+#pragma once
+
+#include <stddef.h>
+
+
+// A non-owned view on a (sub)string.
+struct string_view {
+  char *s;
+  size_t len;
+};