guestfs_read_lines: Reimplement to avoid protocol limits.

This also makes a larger test suite for this command.

TODO

@@ -567,7 +567,6 @@ the p.o.v of the API and ABI.
 
  - guestfs_lstatlist
  - guestfs_lxattrlist
- - guestfs_read_lines
  - guestfs_readlinklist
  - guestfs_write
  - guestfs_write_append

daemon/file.c

@@ -85,54 +85,6 @@ do_touch (const char *path)
   return 0;
 }
 
-char **
-do_read_lines (const char *path)
-{
-  DECLARE_STRINGSBUF (r);
-  FILE *fp;
-  char *line = NULL;
-  size_t len = 0;
-  ssize_t n;
-
-  CHROOT_IN;
-  fp = fopen (path, "r");
-  CHROOT_OUT;
-
-  if (!fp) {
-    reply_with_perror ("fopen: %s", path);
-    return NULL;
-  }
-
-  while ((n = getline (&line, &len, fp)) != -1) {
-    /* Remove either LF or CRLF. */
-    if (n >= 2 && line[n-2] == '\r' && line[n-1] == '\n')
-      line[n-2] = '\0';
-    else if (n >= 1 && line[n-1] == '\n')
-      line[n-1] = '\0';
-
-    if (add_string (&r, line) == -1) {
-      free (line);
-      fclose (fp);
-      return NULL;
-    }
-  }
-
-  free (line);
-
-  if (end_stringsbuf (&r) == -1) {
-    fclose (fp);
-    return NULL;
-  }
-
-  if (fclose (fp) == EOF) {
-    reply_with_perror ("fclose: %s", path);
-    free_stringslen (r.argv, r.size);
-    return NULL;
-  }
-
-  return r.argv;
-}
-
 int
 do_rm (const char *path)
 {

generator/generator_actions.ml

@@ -2059,6 +2059,51 @@ buffer.
 Unlike C<guestfs_cat>, this function can correctly
 handle files that contain embedded ASCII NUL characters." };
 
+  { defaults with
+    name = "read_lines";
+    style = RStringList "lines", [Pathname "path"], [];
+    tests = [
+      InitISOFS, Always, TestOutputList (
+        [["read_lines"; "/known-4"]], ["abc"; "def"; "ghi"]);
+      InitISOFS, Always, TestOutputList (
+        [["read_lines"; "/empty"]], []);
+      InitScratchFS, Always, TestOutputList (
+        [["write"; "/read_lines1"; "\n"];
+         ["read_lines"; "/read_lines1"]], [""]);
+      InitScratchFS, Always, TestOutputList (
+        [["write"; "/read_lines2"; "\r\n"];
+         ["read_lines"; "/read_lines2"]], [""]);
+      InitScratchFS, Always, TestOutputList (
+        [["write"; "/read_lines3"; "\n\r\n"];
+         ["read_lines"; "/read_lines3"]], [""; ""]);
+      InitScratchFS, Always, TestOutputList (
+        [["write"; "/read_lines4"; "a"];
+         ["read_lines"; "/read_lines4"]], ["a"]);
+      InitScratchFS, Always, TestOutputList (
+        [["write"; "/read_lines5"; "a\nb"];
+         ["read_lines"; "/read_lines5"]], ["a"; "b"]);
+      InitScratchFS, Always, TestOutputList (
+        [["write"; "/read_lines6"; "a\nb\n"];
+         ["read_lines"; "/read_lines6"]], ["a"; "b"]);
+      InitScratchFS, Always, TestOutputList (
+        [["write"; "/read_lines7"; "a\nb\r\n"];
+         ["read_lines"; "/read_lines7"]], ["a"; "b"]);
+      InitScratchFS, Always, TestOutputList (
+        [["write"; "/read_lines8"; "a\nb\r\n\n"];
+         ["read_lines"; "/read_lines8"]], ["a"; "b"; ""]);
+    ];
+    shortdesc = "read file as lines";
+    longdesc = "\
+Return the contents of the file named C<path>.
+
+The file contents are returned as a list of lines.  Trailing
+C<LF> and C<CRLF> character sequences are I<not> returned.
+
+Note that this function cannot correctly handle binary files
+(specifically, files containing C<\\0> character which is treated
+as end of string).  For those you need to use the C<guestfs_read_file>
+function and split the buffer into lines yourself." };
+
 ]
 
 (* daemon_functions are any functions which cause some action
@@ -2332,28 +2377,6 @@ of the L<vgs(8)> command.  The \"full\" version includes all fields." };
 List all the logical volumes detected.  This is the equivalent
 of the L<lvs(8)> command.  The \"full\" version includes all fields." };
 
-  { defaults with
-    name = "read_lines";
-    style = RStringList "lines", [Pathname "path"], [];
-    proc_nr = Some 15;
-    tests = [
-      InitISOFS, Always, TestOutputList (
-        [["read_lines"; "/known-4"]], ["abc"; "def"; "ghi"]);
-      InitISOFS, Always, TestOutputList (
-        [["read_lines"; "/empty"]], [])
-    ];
-    shortdesc = "read file as lines";
-    longdesc = "\
-Return the contents of the file named C<path>.
-
-The file contents are returned as a list of lines.  Trailing
-C<LF> and C<CRLF> character sequences are I<not> returned.
-
-Note that this function cannot correctly handle binary files
-(specifically, files containing C<\\0> character which is treated
-as end of line).  For those you need to use the C<guestfs_read_file>
-function which has a more complex interface." };
-
   { defaults with
     name = "aug_init";
     style = RErr, [Pathname "root"; Int "flags"], [];

src/file.c

@@ -122,6 +122,72 @@ guestfs__read_file (guestfs_h *g, const char *path, size_t *size_r)
   return NULL;
 }
 
+char **
+guestfs__read_lines (guestfs_h *g, const char *file)
+{
+  size_t i, count, size, len;
+  char *buf = NULL;
+  char **ret = NULL;
+
+  /* Read the whole file into memory. */
+  buf = guestfs__read_file (g, file, &size);
+  if (buf == NULL)
+    return NULL;
+
+  /* 'buf' contains the list of strings, separated by LF or CRLF
+   * characters.  Convert this to a list of lines.  Note we have to
+   * handle the cases where the buffer is zero length and where the
+   * final string is not terminated.
+   */
+  count = 0;
+  for (i = 0; i < size; ++i)
+    if (buf[i] == '\n')
+      count++;
+  if (size > 0 && buf[size-1] != '\n')
+    count++;
+
+  ret = malloc ((count + 1) * sizeof (char *));
+  if (!ret) {
+    perrorf (g, "malloc");
+    goto err;
+  }
+
+  count = 0;
+  if (size > 0) {
+    ret[count++] = buf;
+    for (i = 0; i < size; ++i) {
+      if (buf[i] == '\n') {
+        buf[i] = '\0';
+        if (i+1 < size)
+          ret[count++] = &buf[i+1];
+      }
+    }
+  }
+  ret[count] = NULL;
+
+  /* Duplicate the strings, and remove the trailing \r characters if any. */
+  for (i = 0; ret[i] != NULL; ++i) {
+    ret[i] = strdup (ret[i]);
+    if (ret[i] == NULL) {
+      perrorf (g, "strdup");
+      while (i > 0)
+        free (ret[--i]);
+      goto err;
+    }
+    len = strlen (ret[i]);
+    if (len > 0 && ret[i][len-1] == '\r')
+      ret[i][len-1] = '\0';
+  }
+  free (buf);
+
+  return ret;
+
+ err:
+  free (buf);
+  free (ret);
+  return NULL;
+}
+
 char **
 guestfs__find (guestfs_h *g, const char *directory)
 {