@bentopdf/pymupdf-wasm 0.11.14 → 0.11.16

package/dist/index.js

@@ -2320,12 +2320,88 @@ doc.scrub(
 
 # 2. Image compression
 if ${compressImages ? "True" : "False"}:
+    import math as _math
+
+    _dpi_target = ${dpiTarget}
+    _dpi_threshold = ${dpiThreshold}
+    _set_to_gray = ${convertToGray ? "True" : "False"}
+    _effective_threshold = max(_dpi_threshold or 0, (_dpi_target or 0) + 10) if _dpi_target else None
+
+    # Pass 1: Handle lossless (PNG/Flate) images via page.replace_image()
+    # Calculate DPI for each xref
+    _xref_dpi = {}
+    for _page in doc:
+        for _info in _page.get_image_info(hashes=False, xrefs=True):
+            _xref = _info.get("xref", 0)
+            if _xref <= 0:
+                continue
+            _bbox = _info.get("bbox")
+            _w = _info.get("width", 0)
+            _h = _info.get("height", 0)
+            if _bbox and _w > 0 and _h > 0:
+                _disp_w = abs(_bbox[2] - _bbox[0])
+                _disp_h = abs(_bbox[3] - _bbox[1])
+                if _disp_w > 0 and _disp_h > 0:
+                    _dpi = min(_w / _disp_w * 72, _h / _disp_h * 72)
+                    if _xref not in _xref_dpi or _dpi < _xref_dpi[_xref]:
+                        _xref_dpi[_xref] = _dpi
+
+    _handled = set()
+    for _page in doc:
+        for _img in _page.get_images():
+            _xref = _img[0]
+            if _xref in _handled:
+                continue
+            _handled.add(_xref)
+
+            _mask_xref = _img[1]
+            _xref_obj = doc.xref_object(_xref)
+
+            if "FlateDecode" not in _xref_obj:
+                continue
+
+            _min_dpi = _xref_dpi.get(_xref, float("inf"))
+            _needs_downscale = bool(
+                _dpi_target and _effective_threshold
+                and _min_dpi != float("inf")
+                and _min_dpi > _effective_threshold
+            )
+            if not _needs_downscale and not _set_to_gray:
+                continue
+
+            try:
+                _base = pymupdf.Pixmap(doc, _xref)
+
+                if _base.alpha:
+                    _base = pymupdf.Pixmap(_base, 0)
+
+                if _mask_xref:
+                    _mask = pymupdf.Pixmap(doc, _mask_xref)
+                    _base = pymupdf.Pixmap(_base, _mask)
+
+                if _set_to_gray and _base.colorspace and _base.colorspace.n > 1:
+                    _base = pymupdf.Pixmap(pymupdf.csGRAY, _base)
+                elif _base.colorspace and _base.colorspace.n > 3:
+                    _base = pymupdf.Pixmap(pymupdf.csRGB, _base)
+
+                if _needs_downscale:
+                    _ratio = _min_dpi / _dpi_target
+                    _shrink_n = max(0, min(7, int(_math.log2(_ratio))))
+                    if _shrink_n > 0:
+                        _base.shrink(_shrink_n)
+
+                _page.replace_image(_xref, pixmap=_base)
+                _base = None
+            except Exception as _e:
+                pass
+
+    # Pass 2: Handle lossy (JPEG) images via rewrite_images
     doc.rewrite_images(
         dpi_threshold=${dpiThreshold},
         dpi_target=${dpiTarget},
         quality=${imageQuality},
+        lossless=False,
         lossy=${processLossy ? "True" : "False"},
-        lossless=${processLossless ? "True" : "False"},
         bitonal=${processBitonal ? "True" : "False"},
         color=${processColor ? "True" : "False"},
         gray=${processGray ? "True" : "False"},

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "@bentopdf/pymupdf-wasm",
-    "version": "0.11.14",
+    "version": "0.11.16",
     "description": "PyMuPDF compiled to WebAssembly - Full PDF manipulation in the browser",
     "type": "module",
     "main": "dist/index.js",