PyMuPDF 1.23.14__tar.gz → 1.23.16__tar.gz

{PyMuPDF-1.23.14 → PyMuPDF-1.23.16}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: PyMuPDF
-Version: 1.23.14
+Version: 1.23.16
 Summary: A high performance Python library for data extraction, analysis, conversion & manipulation of PDF (and other) documents.
 Description-Content-Type: text/markdown
 Author: Artifex

{PyMuPDF-1.23.14 → PyMuPDF-1.23.16}/changes.txt

@@ -2,6 +2,30 @@ Change Log
 ==========
 
 
+**Changes in version 1.23.16 (2024-01-18)**
+
+* Bug fixes:
+
+  * **Fixed** `3058 <https://github.com/pymupdf/PyMuPDF/issues/3058>`_: Pixmap created from CMYK JPEG delivers RGB format
+
+* Other:
+
+  * In table detection strategy "lines_strict", exclude fill-only vector graphics.
+  * Fixed sysinstall test failure.
+  * In documentation, update feature matrix with item about text writing.
+
+
+**Changes in version 1.23.15 (2024-01-16)**
+
+* Bug fixes:
+
+  * **Fixed** `3050 <https://github.com/pymupdf/PyMuPDF/issues/3050>`_: python3.9 pix.set_pixel has something wrong in c.append( ord(i))
+
+* Other:
+
+  * Improved docs for Page.find_tables().
+
+
 **Changes in version 1.23.14 (2024-01-15)**
 
 * Bug fixes:

{PyMuPDF-1.23.14 → PyMuPDF-1.23.16}/scripts/sysinstall.py

@@ -291,7 +291,7 @@ def main():
     # PyMuPDF.
     command = f'. {test_venv}/bin/activate'
     command += f' && LD_LIBRARY_PATH={root_prefix}/lib PYTHONPATH={pythonpath}'
-    command += f' pytest -k "not test_color_count" {pymupdf_dir}'
+    command += f' pytest -k "not test_color_count and not test_3050" {pymupdf_dir}'
     run(command)
 
 

{PyMuPDF-1.23.14 → PyMuPDF-1.23.16}/setup.py

@@ -1072,7 +1072,7 @@ with open( f'{g_root}/READMErb.md', encoding='utf-8') as f:
 # We generate different wheels depending on g_flavour.
 #
 
-version = '1.23.14'
+version = '1.23.16'
 version_b = '1.23.9'
 
 tag_python = None

{PyMuPDF-1.23.14 → PyMuPDF-1.23.16}/src/__init__.py

@@ -2396,7 +2396,9 @@ class Colorspace:
 
     def __init__(self, type_):
         """Supported are GRAY, RGB and CMYK."""
-        if type_ == CS_GRAY:
+        if isinstance( type_, mupdf.FzColorspace):
+            self.this = type_
+        elif type_ == CS_GRAY:
             self.this = mupdf.FzColorspace(mupdf.FzColorspace.Fixed_GRAY)
         elif type_ == CS_CMYK:
             self.this = mupdf.FzColorspace(mupdf.FzColorspace.Fixed_CMYK)
@@ -7759,8 +7761,8 @@ class Page:
                 else:
                     pm = mupdf.fz_convert_pixmap(
                             arg_pix,
-                            mupdf.FzColorspace(0),
-                            mupdf.FzColorspace(0),
+                            mupdf.FzColorspace(),
+                            mupdf.FzColorspace(),
                             mupdf.FzDefaultColorspaces(None),
                             mupdf.FzColorParams(),
                             1,
@@ -9429,7 +9431,7 @@ class Pixmap:
                 self.this = mupdf.fz_convert_pixmap(
                         spix,
                         cs,
-                        mupdf.FzColorspace(0),
+                        mupdf.FzColorspace(),
                         mupdf.FzDefaultColorspaces(None),
                         mupdf.FzColorParams(),
                         1
@@ -9647,6 +9649,11 @@ class Pixmap:
                 text += f'    {type(arg)}: {arg}\n'
             raise Exception( text)
 
+        # 2024-01-16: Experimental support for a memory-view of the underlying
+        # data.  Doesn't seem to make much difference to Pixmap.set_pixel() so
+        # not currently used.
+        self._memory_view = None
+
     def __len__(self):
         return self.size
 
@@ -10117,20 +10124,28 @@ class Pixmap:
 
     def set_pixel(self, x, y, color):
         """Set color of pixel (x, y)."""
+        if g_use_extra:
+            return extra.set_pixel(self.this.m_internal, x, y, color)
         pm = self.this
         if not _INRANGE(x, 0, pm.w() - 1) or not _INRANGE(y, 0, pm.h() - 1):
             raise ValueError( MSG_PIXEL_OUTSIDE)
         n = pm.n()
-        c = list()
         for j in range(n):
             i = color[j]
             if not _INRANGE(i, 0, 255):
                 raise ValueError( MSG_BAD_COLOR_SEQ)
-            c.append( ord(i))
         stride = mupdf.fz_pixmap_stride( pm)
         i = stride * y + n * x
-        for j in range(n):
-            pm.m_internal.samples[i + j] = c[j]
+        if 0:
+            # Using a cached self._memory_view doesn't actually make much
+            # difference to speed.
+            if not self._memory_view:
+                self._memory_view = self.samples_mv
+            for j in range(n):
+                self._memory_view[i + j] = color[j]
+        else:
+            for j in range(n):
+                pm.fz_samples_set(i + j, color[j])
 
     def set_rect(self, bbox, color):
         """Set color of all pixels in bbox."""
@@ -10154,6 +10169,8 @@ class Pixmap:
             JM_Warning("ignoring shrink factor < 1")
             return
         mupdf.fz_subsample_pixmap( self.this, factor)
+        # Pixmap has changed so clear our memory view.
+        self._memory_view = None
 
     @property
     def size(self):
@@ -21739,8 +21756,8 @@ def int_rc(text):
     return int(text)
 
 VersionFitz = "1.23.9" # MuPDF version.
-VersionBind = "1.23.14" # PyMuPDF version.
-VersionDate = "2024-01-15 00:00:01"
+VersionBind = "1.23.16" # PyMuPDF version.
+VersionDate = "2024-01-18 00:00:01"
 VersionDate2 = VersionDate.replace('-', '').replace(' ', '').replace(':', '')
 version = (VersionBind, VersionFitz, VersionDate2)
 pymupdf_version_tuple = tuple( [int_rc(i) for i in VersionBind.split('.')])

{PyMuPDF-1.23.14 → PyMuPDF-1.23.16}/src/extra.i

@@ -3670,6 +3670,7 @@ PyObject* extractBLOCKS(mupdf::FzStextPage& self)
             mupdf::fz_clear_buffer(res);  // set text buffer to empty
             int line_n = -1;
             int last_char = 0;
+            (void) line_n;  /* Not actually used, but keeping in the code for now. */
             for (fz_stext_line* line = block->u.t.first_line; line; line = line->next)
             {
                 line_n++;
@@ -3954,6 +3955,62 @@ int pixmap_n(mupdf::FzPixmap& pixmap)
     return mupdf::fz_pixmap_components( pixmap);
 }
 
+static int
+JM_INT_ITEM(PyObject *obj, Py_ssize_t idx, int *result)
+{
+    PyObject *temp = PySequence_ITEM(obj, idx);
+    if (!temp) return 1;
+    if (PyLong_Check(temp)) {
+        *result = (int) PyLong_AsLong(temp);
+        Py_DECREF(temp);
+    } else if (PyFloat_Check(temp)) {
+        *result = (int) PyFloat_AsDouble(temp);
+        Py_DECREF(temp);
+    } else {
+        Py_DECREF(temp);
+        return 1;
+    }
+    if (PyErr_Occurred()) {
+        PyErr_Clear();
+        return 1;
+    }
+    return 0;
+}
+
+PyObject *set_pixel(fz_pixmap* pm, int x, int y, PyObject *color)
+{
+    fz_context* ctx = mupdf::internal_context_get();
+    if (0
+            || x < 0
+            || x >= pm->w
+            || y < 0
+            || y >= pm->h
+            )
+    {
+        throw std::range_error( MSG_PIXEL_OUTSIDE);
+    }
+    int n = pm->n;
+    if (!PySequence_Check(color) || PySequence_Size(color) != n) {
+        throw std::range_error(MSG_BAD_COLOR_SEQ);
+    }
+    int i, j;
+    unsigned char c[5];
+    for (j = 0; j < n; j++) {
+        if (JM_INT_ITEM(color, j, &i) == 1) {
+            throw std::range_error(MSG_BAD_COLOR_SEQ);
+        }
+        if (i < 0 or i >= 256) {
+            throw std::range_error(MSG_BAD_COLOR_SEQ);
+        }
+        c[j] = (unsigned char) i;
+    }
+    int stride = fz_pixmap_stride(ctx, pm);
+    i = stride * y + n * x;
+    for (j = 0; j < n; j++) {
+        pm->samples[i + j] = c[j];
+    }
+    Py_RETURN_NONE;
+}
 //-------------------------------------------
 // make a buffer from an stext_page's text
 //-------------------------------------------
@@ -4409,3 +4466,5 @@ PyObject *pixmap_pixel(fz_pixmap* pm, int x, int y);
 int pixmap_n(mupdf::FzPixmap& pixmap);
 
 PyObject* JM_search_stext_page(fz_stext_page *page, const char *needle);
+
+PyObject *set_pixel(fz_pixmap* pm, int x, int y, PyObject *color);

{PyMuPDF-1.23.14 → PyMuPDF-1.23.16}/src/table.py

@@ -1886,8 +1886,24 @@ def make_edges(page, clip=None, tset=None):
 
         parea = abs(page.rect) * 0.8  # area of the full page (80%)
 
-        # exclude graphics that are too large
-        paths = [p for p in page.get_drawings() if abs(p["rect"]) < parea]
+        # exclude irrelevant graphics
+        paths = []
+        for p in page.get_drawings():
+            if abs(p["rect"]) >= parea:
+                continue
+            if "s" in p["type"]:
+                paths.append(p)
+                continue
+            if (
+                p["rect"].width > 3
+                and p["rect"].height > 3
+                and (
+                    tset.vertical_strategy == "lines_strict"
+                    or tset.horizontal_strategy == "lines_strict"
+                )
+            ):
+                continue
+            paths.append(p)
 
         # make a list of vector graphics rectangles (IRects are sufficient)
         prects = sorted([p["rect"] for p in paths], key=lambda r: (r.y1, r.x0))
@@ -1998,14 +2014,6 @@ def make_edges(page, clip=None, tset=None):
         return line_dict
 
     for p in paths:
-        if p["type"] == "f" and p["fill"] == (1, 1, 1):
-            continue
-        if p["type"] == "f" and p["rect"].width > 3 and p["rect"].height > 3:
-            if (
-                tset.vertical_strategy == "lines_strict"
-                or tset.horizontal_strategy == "lines_strict"
-            ):
-                continue
         items = p["items"]  # items in this path
 
         # if 'closePath', add a line from last to first point

PyMuPDF-1.23.16/src_classic/version.i

@@ -0,0 +1,7 @@
+%pythoncode %{
+VersionFitz = "1.23.9" # MuPDF version.
+VersionBind = "1.23.16" # PyMuPDF version.
+VersionDate = "2024-01-18 00:00:01"
+version = (VersionBind, VersionFitz, "20240118000001")
+pymupdf_version_tuple = tuple( [int(i) for i in VersionFitz.split('.')])
+%}

{PyMuPDF-1.23.14 → PyMuPDF-1.23.16}/tests/test_pixmap.py

@@ -94,6 +94,10 @@ def test_setalpha():
     assert t == alphas
 
 def test_color_count():
+    '''
+    This is known to fail if MuPDF is built without PyMuPDF's custom config.h,
+    e.g. in Linux system installs.
+    '''
     pm = fitz.Pixmap(imgfile)
     assert pm.color_count() == 40624
 
@@ -152,3 +156,54 @@ def test_3020():
     pm2 = fitz.Pixmap(pm, 20, 30, None)
     pm3 = fitz.Pixmap(fitz.csGRAY, pm)
     pm4 = fitz.Pixmap(pm, pm3)
+
+def test_3050():
+    '''
+    This is known to fail if MuPDF is built without it's default third-party
+    libraries, e.g. in Linux system installs.
+    '''
+    pdf_file = fitz.open(pdf)
+    for page_no, page in enumerate(pdf_file):
+        zoom_x = 4.0
+        zoom_y = 4.0
+        matrix = fitz.Matrix(zoom_x, zoom_y)
+        pix = page.get_pixmap(matrix=matrix)
+        digest0 = pix.digest
+        print(f'{pix.width=} {pix.height=}')
+        def product(x, y):
+            for yy in y:
+                for xx in x:
+                    yield (xx, yy)
+        n = 0
+        # We use a small subset of the image because non-optimised rebase gets
+        # very slow.
+        for pos in product(range(100), range(100)):
+            if sum(pix.pixel(pos[0], pos[1])) >= 600:
+                n += 1
+                pix.set_pixel(pos[0], pos[1], (255, 255, 255))
+        digest1 = pix.digest
+        print(f'{page_no=} {n=} {digest0=} {digest1=}')
+        digest_expected = b'\xd7x\x94_\x98\xa1<-/\xf3\xf9\x04\xec#\xaa\xee'
+        pix.save(os.path.abspath(f'{__file__}/../../tests/test_3050_out.png'))
+        assert digest1 != digest0
+        assert digest1 == digest_expected
+
+def test_3058():
+    doc = fitz.Document(os.path.abspath(f'{__file__}/../../tests/resources/test_3058.pdf'))
+    images = doc[0].get_images(full=True)
+    pix = fitz.Pixmap(doc, 17)
+    
+    # First bug was that `pix.colorspace` was DeviceRGB.
+    assert str(pix.colorspace) == 'Colorspace(CS_CMYK) - DeviceCMYK'
+    
+    pix = fitz.Pixmap(fitz.csRGB, pix)
+    assert str(pix.colorspace) == 'Colorspace(CS_RGB) - DeviceRGB'
+    
+    # Second bug was that the image was converted to RGB via greyscale proofing
+    # color space, so image contained only shades of grey. This compressed
+    # easily to a .png file, so we crudely check the bug is fixed by looking at
+    # size of .png file.
+    path = os.path.abspath(f'{__file__}/../../tests/test_3058_out.png')
+    pix.save(path)
+    s = os.path.getsize(path)
+    assert 1800000 < s < 2600000, f'Unexpected size of {path}: {s}'

PyMuPDF-1.23.14/src_classic/version.i

@@ -1,7 +0,0 @@
-%pythoncode %{
-VersionFitz = "1.23.9" # MuPDF version.
-VersionBind = "1.23.14" # PyMuPDF version.
-VersionDate = "2024-01-15 00:00:01"
-version = (VersionBind, VersionFitz, "20240115000001")
-pymupdf_version_tuple = tuple( [int(i) for i in VersionFitz.split('.')])
-%}