When writing JSON to a utf-8 encoded file, disable ASCII escaping (#85)

pdfannots/cli.py

@@ -149,7 +149,9 @@ def main() -> None:
         mdargs = {k: getattr(args, k) for k in MD_FORMAT_ARGS}
         printer = (GroupedMarkdownPrinter if args.group else MarkdownPrinter)(**mdargs)
     elif args.format == "json":
-        printer = JsonPrinter(remove_hyphens=args.remove_hyphens)
+        printer = JsonPrinter(
+            remove_hyphens=args.remove_hyphens,
+            output_codec=args.output.encoding)
 
     def write_if_nonempty(s: str) -> None:
         if s:

pdfannots/printer/json.py

@@ -31,10 +31,19 @@ def annot_to_dict(
 
 
 class JsonPrinter(Printer):
-    def __init__(self, *, remove_hyphens: bool) -> None:
-        self.remove_hyphens = remove_hyphens  # Whether to remove hyphens across a line break
+    def __init__(
+            self,
+            *,
+            remove_hyphens: bool,  # Whether to remove hyphens across a line break
+            output_codec: str      # Text codec in use for output
+    ) -> None:
+        self.remove_hyphens = remove_hyphens
         self.seen_first = False
 
+        # JSON must be represented as UTF-8, UTF-16, or UTF-32. If the output codec is
+        # one of these, we can disable ASCII string escaping in the JSON encoder.
+        self.ensure_ascii = output_codec not in ['utf-8', 'utf-16', 'utf-32']
+
     def end(self) -> str:
         return '\n'
 
@@ -51,4 +60,4 @@ def print_file(
             self.seen_first = True
 
         annots = [annot_to_dict(document, a, self.remove_hyphens) for a in document.iter_annots()]
-        yield from json.JSONEncoder(indent=2).iterencode(annots)
+        yield from json.JSONEncoder(indent=2, ensure_ascii=self.ensure_ascii).iterencode(annots)

tests.py

@@ -307,7 +307,7 @@ def test_multicolorgrouping(self) -> None:
 
 class JsonPrinterTest(PrinterTestBase):
     def test_flat(self) -> None:
-        p = JsonPrinter(remove_hyphens=False)
+        p = JsonPrinter(remove_hyphens=False, output_codec='utf-8')
 
         j = json.loads(
             p.begin()