[path-] fix undercounted progress for multibyte chars

saulpw · Feb 19, 2024 · 2a45fd5 · 2a45fd5
1 parent be295d8
commit 2a45fd5
Showing 1 changed file with 9 additions and 8 deletions.
diff --git a/visidata/path.py b/visidata/path.py
@@ -95,7 +95,7 @@ def peek(self, size=-1):
 
 
 class FileProgress:
-    'Open file in binary mode and track read() progress.'
+    'Open file as text or binary, then track read() progress.'
     def __init__(self, path, fp, mode='r', **kwargs):
         self.path = path
         self.fp = fp
@@ -130,13 +130,13 @@ def read(self, size=-1):
         r = self.fp_orig_read(size)
         if self.prog:
             if r:
-                self.prog.addProgress(len(r))
+                self.prog.addProgress(len(r.encode(self.encoding)) if self.encoding else len(r))
         return r
 
     def readline(self, size=-1):
         r = self.fp_orig_readline(size)
         if self.prog:
-            self.prog.addProgress(len(r))
+            self.prog.addProgress(len(r.encode(self.encoding)) if self.encoding else len(r))
         return r
 
     def __getattr__(self, k):
@@ -148,15 +148,15 @@ def __enter__(self):
 
     def __next__(self):
         r = next(self.fp)
-        self.prog.addProgress(len(r))
+        self.prog.addProgress(len(r.encode(self.encoding)) if self.encoding else len(r))
         return r
 
     def __iter__(self):
         if not self.prog:
             yield from self.fp
         else:
             for line in self.fp:
-                self.prog.addProgress(len(line))
+                self.prog.addProgress(len(line.encode(self.encoding)) if self.encoding else len(line))
                 yield line
 
     def __exit__(self, type, value, tb):
@@ -326,9 +326,10 @@ def _open(self, *args, **kwargs):
 
     def __iter__(self):
         with Progress(total=filesize(self)) as prog:
-            with self.open(encoding=vd.options.encoding) as fd:
-                for i, line in enumerate(fd):
-                    prog.addProgress(len(line))
+            enc = vd.options.encoding
+            with self.open(encoding=enc) as fd:
+                for line in fd:
+                    prog.addProgress(len(line.encode(enc)))
                     yield line.rstrip('\n')
 
     def open_bytes(self, mode='rb'):