vim-patch:9.0.0795: readblob() always reads the whole file

Problem:    readblob() always reads the whole file.
Solution:   Add arguments to read part of the file. (Ken Takata,
            closes vim/vim#11402)

11df3aeee5

Remove trailing whitespace in test as done in patch 9.0.1257.
Move the help for rand() before range().

Co-authored-by: K.Takata <kentkt@csc.jp>

runtime/doc/builtin.txt

@@ -359,7 +359,8 @@ pyxeval({expr})			any	evaluate |python_x| expression
 rand([{expr}])			Number	get pseudo-random number
 range({expr} [, {max} [, {stride}]])
 				List	items from {expr} to {max}
-readblob({fname})		Blob	read a |Blob| from {fname}
+readblob({fname} [, {offset} [, {size}]])
+				Blob	read a |Blob| from {fname}
 readdir({dir} [, {expr}])	List	file names in {dir} selected by {expr}
 readfile({fname} [, {type} [, {max}]])
 				List	get list of lines from file {fname}
@@ -6110,6 +6111,25 @@ pyxeval({expr})						*pyxeval()*
 		Can also be used as a |method|: >
 			GetExpr()->pyxeval()
 <
+rand([{expr}])						*rand()*
+		Return a pseudo-random Number generated with an xoshiro128**
+		algorithm using seed {expr}.  The returned number is 32 bits,
+		also on 64 bits systems, for consistency.
+		{expr} can be initialized by |srand()| and will be updated by
+		rand().  If {expr} is omitted, an internal seed value is used
+		and updated.
+		Returns -1 if {expr} is invalid.
+
+		Examples: >
+			:echo rand()
+			:let seed = srand()
+			:echo rand(seed)
+			:echo rand(seed) % 16  " random number 0 - 15
+<
+		Can also be used as a |method|: >
+			seed->rand()
+<
+
 							*E726* *E727*
 range({expr} [, {max} [, {stride}]])				*range()*
 		Returns a |List| with Numbers:
@@ -6132,29 +6152,22 @@ range({expr} [, {max} [, {stride}]])				*range()*
 		Can also be used as a |method|: >
 			GetExpr()->range()
 <
-rand([{expr}])						*rand()*
-		Return a pseudo-random Number generated with an xoshiro128**
-		algorithm using seed {expr}.  The returned number is 32 bits,
-		also on 64 bits systems, for consistency.
-		{expr} can be initialized by |srand()| and will be updated by
-		rand().  If {expr} is omitted, an internal seed value is used
-		and updated.
-		Returns -1 if {expr} is invalid.
 
-		Examples: >
+readblob({fname} [, {offset} [, {size}]])			*readblob()*
-			:echo rand()
-			:let seed = srand()
-			:echo rand(seed)
-			:echo rand(seed) % 16  " random number 0 - 15
-<
-		Can also be used as a |method|: >
-			seed->rand()
-<
-
-readblob({fname})					*readblob()*
 		Read file {fname} in binary mode and return a |Blob|.
+		If {offset} is specified, read the file from the specified
+		offset.  If it is a negative value, it is used as an offset
+		from the end of the file.  E.g., to read the last 12 bytes: >
+			readblob('file.bin', -12)
+<		If {size} is specified, only the specified size will be read.
+		E.g. to read the first 100 bytes of a file: >
+			readblob('file.bin', 0, 100)
+<		If {size} is -1 or omitted, the whole data starting from
+		{offset} will be read.
 		When the file can't be opened an error message is given and
 		the result is an empty |Blob|.
+		When trying to read bytes beyond the end of the file the
+		result is an empty blob.
 		Also see |readfile()| and |writefile()|.
 
 

src/nvim/eval.c

@@ -5859,27 +5859,57 @@ write_blob_error:
   return false;
 }
 
-/// Read a blob from a file `fd`.
+/// Read blob from file "fd".
+/// Caller has allocated a blob in "rettv".
 ///
 /// @param[in]  fd  File to read from.
-/// @param[in,out]  blob  Blob to write to.
+/// @param[in,out]  rettv  Blob to write to.
+/// @param[in]  offset  Read the file from the specified offset.
+/// @param[in]  size  Read the specified size, or -1 if no limit.
 ///
-/// @return true on success, or false on failure.
+/// @return  OK on success, or FAIL on failure.
-bool read_blob(FILE *const fd, blob_T *const blob)
+int read_blob(FILE *const fd, typval_T *rettv, off_T offset, off_T size_arg)
   FUNC_ATTR_NONNULL_ALL
 {
+  blob_T *const blob = rettv->vval.v_blob;
   FileInfo file_info;
   if (!os_fileinfo_fd(fileno(fd), &file_info)) {
-    return false;
+    return FAIL;  // can't read the file, error
   }
-  const int size = (int)os_fileinfo_size(&file_info);
+
-  ga_grow(&blob->bv_ga, size);
+  int whence;
-  blob->bv_ga.ga_len = size;
+  off_T size = size_arg;
+  if (offset >= 0) {
+    if (size == -1) {
+      // size may become negative, checked below
+      size = (off_T)os_fileinfo_size(&file_info) - offset;
+    }
+    whence = SEEK_SET;
+  } else {
+    if (size == -1) {
+      size = -offset;
+    }
+    whence = SEEK_END;
+  }
+  // Trying to read bytes that aren't there results in an empty blob, not an
+  // error.
+  if (size < 0 || size > (off_T)os_fileinfo_size(&file_info)) {
+    return OK;
+  }
+  if (vim_fseek(fd, offset, whence) != 0) {
+    return OK;
+  }
+
+  ga_grow(&blob->bv_ga, (int)size);
+  blob->bv_ga.ga_len = (int)size;
   if (fread(blob->bv_ga.ga_data, 1, (size_t)blob->bv_ga.ga_len, fd)
       < (size_t)blob->bv_ga.ga_len) {
-    return false;
+    // An empty blob is returned on error.
+    tv_blob_free(rettv->vval.v_blob);
+    rettv->vval.v_blob = NULL;
+    return FAIL;
   }
-  return true;
+  return OK;
 }
 
 /// Saves a typval_T as a string.

src/nvim/eval.lua

@@ -296,7 +296,7 @@ return {
     perleval={args=1, base=1},
     rand={args={0, 1}, base=1},
     range={args={1, 3}, base=1},
-    readblob={args=1, base=1},
+    readblob={args={1, 3}, base=1},
     readdir={args={1, 2}, base=1},
     readfile={args={1, 3}, base=1},
     reduce={args={2, 3}, base=1},

src/nvim/eval/funcs.c

@@ -5592,15 +5592,24 @@ static void read_file_or_blob(typval_T *argvars, typval_T *rettv, bool always_bl
   ptrdiff_t prevlen  = 0;               // length of data in prev
   ptrdiff_t prevsize = 0;               // size of prev buffer
   int64_t maxline  = MAXLNUM;
+  off_T offset = 0;
+  off_T size = -1;
 
   if (argvars[1].v_type != VAR_UNKNOWN) {
-    if (strcmp(tv_get_string(&argvars[1]), "b") == 0) {
+    if (always_blob) {
-      binary = true;
+      offset = (off_T)tv_get_number(&argvars[1]);
-    } else if (strcmp(tv_get_string(&argvars[1]), "B") == 0) {
+      if (argvars[2].v_type != VAR_UNKNOWN) {
-      blob = true;
+        size = (off_T)tv_get_number(&argvars[2]);
-    }
+      }
-    if (argvars[2].v_type != VAR_UNKNOWN) {
+    } else {
-      maxline = tv_get_number(&argvars[2]);
+      if (strcmp(tv_get_string(&argvars[1]), "b") == 0) {
+        binary = true;
+      } else if (strcmp(tv_get_string(&argvars[1]), "B") == 0) {
+        blob = true;
+      }
+      if (argvars[2].v_type != VAR_UNKNOWN) {
+        maxline = tv_get_number(&argvars[2]);
+      }
     }
   }
 
@@ -5619,11 +5628,8 @@ static void read_file_or_blob(typval_T *argvars, typval_T *rettv, bool always_bl
 
   if (blob) {
     tv_blob_alloc_ret(rettv);
-    if (!read_blob(fd, rettv->vval.v_blob)) {
+    if (read_blob(fd, rettv, offset, size) == FAIL) {
       semsg(_(e_notread), fname);
-      // An empty blob is returned on error.
-      tv_blob_free(rettv->vval.v_blob);
-      rettv->vval.v_blob = NULL;
     }
     fclose(fd);
     return;

src/nvim/testdir/test_blob.vim

@@ -439,10 +439,29 @@ func Test_blob_read_write()
       call writefile(b, 'Xblob')
       VAR br = readfile('Xblob', 'B')
       call assert_equal(b, br)
+      VAR br2 = readblob('Xblob')
+      call assert_equal(b, br2)
+      VAR br3 = readblob('Xblob', 1)
+      call assert_equal(b[1 :], br3)
+      VAR br4 = readblob('Xblob', 1, 2)
+      call assert_equal(b[1 : 2], br4)
+      VAR br5 = readblob('Xblob', -3)
+      call assert_equal(b[-3 :], br5)
+      VAR br6 = readblob('Xblob', -3, 2)
+      call assert_equal(b[-3 : -2], br6)
+
+      VAR br1e = readblob('Xblob', 10000)
+      call assert_equal(0z, br1e)
+      VAR br2e = readblob('Xblob', -10000)
+      call assert_equal(0z, br2e)
+
       call delete('Xblob')
   END
   call CheckLegacyAndVim9Success(lines)
 
+  call assert_fails("call readblob('notexist')", 'E484:')
+  " TODO: How do we test for the E485 error?
+
   " This was crashing when calling readfile() with a directory.
   call assert_fails("call readfile('.', 'B')", 'E17: "." is a directory')
 endfunc