Jules - fix #4994 and #5112

src/__init__.py

@@ -21544,13 +21544,18 @@ def JM_new_buffer_from_stext_page(page):
     for block in page:
         if block.m_internal.type == mupdf.FZ_STEXT_BLOCK_TEXT:
             for line in block:
+                break_line = True
                 for ch in line:
                     if (not JM_rects_overlap(rect, JM_char_bbox(line, ch))
                             and not mupdf.fz_is_infinite_rect(rect)
                             ):
                         continue
-                    mupdf.fz_append_rune(buf, ch.m_internal.c)
-                mupdf.fz_append_byte(buf, ord('\n'))
+                    if not ch.m_internal.next and (line.m_internal.flags & mupdf.FZ_STEXT_LINE_FLAGS_JOINED):
+                        break_line = 0
+                    else:
+                        mupdf.fz_append_rune(buf, ch.m_internal.c)
+                if break_line:
+                    mupdf.fz_append_byte(buf, ord('\n'))
             mupdf.fz_append_byte(buf, ord('\n'))
     return buf
 
@@ -22031,9 +22036,8 @@ def __str__(self):
                     break
                 haystack += 1
                 #next_char:;
-            assert haystack_string[haystack] == '\n', \
-                    f'{haystack=} {haystack_string[haystack]=}'
-            haystack += 1
+            if haystack_string[haystack] == '\n':
+                haystack += 1
         assert haystack_string[haystack] == '\n', \
                 f'{haystack=} {haystack_string[haystack]=}'
         haystack += 1

src/extra.i

@@ -788,6 +788,10 @@ static void JM_set_annot_callout_line(mupdf::PdfAnnot& annot, PyObject *callout,
 {
     fz_point points[3];
     mupdf::FzPoint p;
+    if (count > sizeof(points)/sizeof(points[0]))
+    {
+        throw std::runtime_error(MSG_BAD_ARG_POINTS);
+    }
     for (int i = 0; i < count; i++)
     {
         p = JM_point_from_py(PyTuple_GetItem(callout, (Py_ssize_t) i));
@@ -2301,16 +2305,23 @@ void _as_text(fz_stext_block *block, mupdf::FzBuffer& res, mupdf::FzStextPage& p
                 last_char = 0;
                 for (line = block->u.t.first_line; line; line = line->next)
                 {
+                    int break_line = 1;
                     for (ch = line->first_char; ch; ch = ch->next)
                     {
                         fz_rect chbbox = JM_char_bbox( line, ch);
-                        if (mupdf::ll_fz_is_infinite_rect(rect) || JM_rects_overlap(rect, chbbox))
+                        if (!ch->next && (line->flags & FZ_STEXT_LINE_FLAGS_JOINED)) {
+                            break_line = 0;
+                        }
+                        else if (mupdf::ll_fz_is_infinite_rect(rect) || JM_rects_overlap(rect, chbbox))
                         {
                             last_char = ch->c;
                             JM_append_rune(res.m_internal, last_char);
                         }
                     }
-                    if (last_char != 10 && last_char > 0)
+                    if (!break_line)
+                    {
+                    }
+                    else if (last_char != 10 && last_char > 0)
                     {
                         mupdf::ll_fz_append_string(res.m_internal, "\n");
                         last_char = 10;
@@ -3973,11 +3984,11 @@ PyObject *set_pixel(fz_pixmap* pm, int x, int y, PyObject *color)
         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];
+    if (!PySequence_Check(color) || PySequence_Size(color) != n || n > sizeof(c)) {
+        throw std::range_error(MSG_BAD_COLOR_SEQ);
+    }
     for (j = 0; j < n; j++) {
         if (JM_INT_ITEM(color, j, &i) == 1) {
             throw std::range_error(MSG_BAD_COLOR_SEQ);
@@ -4013,14 +4024,22 @@ JM_new_buffer_from_stext_page(fz_stext_page *page)
         for (block = page->first_block; block; block = block->next) {
             if (block->type == FZ_STEXT_BLOCK_TEXT) {
                 for (line = block->u.t.first_line; line; line = line->next) {
+                    int break_line = 1;
                     for (ch = line->first_char; ch; ch = ch->next) {
                         if (!JM_rects_overlap(rect, JM_char_bbox(line, ch)) &&
                             !fz_is_infinite_rect(rect)) {
                             continue;
                         }
-                        fz_append_rune(ctx, buf, ch->c);
+                        if (!ch->next && (line->flags & FZ_STEXT_LINE_FLAGS_JOINED)) {
+                            break_line = 0;
+                        }
+                        else {
+                            fz_append_rune(ctx, buf, ch->c);
+                        }
+                    }
+                    if (break_line) {
+                        fz_append_byte(ctx, buf, '\n');
                     }
-                    fz_append_byte(ctx, buf, '\n');
                 }
                 fz_append_byte(ctx, buf, '\n');
             }
@@ -4261,8 +4280,9 @@ try_new_match:
                     haystack += fz_chartorune(&c, haystack);
 next_char:;
                 }
-                assert(*haystack == '\n');
-                ++haystack;
+                if (*haystack == '\n') {
+                    ++haystack;
+                }
             }
             assert(*haystack == '\n');
             ++haystack;
@@ -4347,7 +4367,10 @@ PyObject* ll_JM_color_count(fz_pixmap *pm, PyObject *clip)
     unsigned char* s = pm->samples + stride * (irect.y0 - pm->y) + n * (irect.x0 - pm->x);
     // Cache previous pixel.
     char oldpix[10];
-    assert(n <= sizeof(oldpix));
+    if (n > sizeof(oldpix))
+    {
+        throw std::range_error(MSG_PIXEL_OUTSIDE);
+    }
     memcpy(oldpix, s, n);
     long cnt = 0;
     for (size_t i = 0; i < height; i++)

tests/test_annots.py

@@ -794,3 +794,21 @@ def test_4936():
         drawings = page.get_drawings()
     print(f'{len(drawings)=}')
     assert len(drawings) == 0
+
+
+def test_5112():
+    print()
+    with pymupdf.open() as document:
+        page = document.new_page()
+        rect = pymupdf.Rect(50, 50, 200, 150)
+        callout = [(10, 10)] * 100
+        try:
+            print(f'test_5112(): calling page.add_freetext_annot().', flush=1)
+            e = page.add_freetext_annot(rect, "test", callout=callout)
+        except Exception as e:
+            print(f'Received exception: {e}')
+            assert isinstance(e, RuntimeError)
+            assert str(e) == 'bad seq of points'
+        else:
+            assert 0, f'Expected exception from page.add_freetext_annot().'
+

tests/test_textextract.py

@@ -948,3 +948,26 @@ def test_4503():
 
     assert strikeout, f'Expected bit 0 (FZ_STEXT_STRIKEOUT) to be set in {span_0["char_flags"]=:#x}.'
     assert text_0 == 'the right to request the state to review and, if appropriate,'
+
+
+def test_4994():
+    print()
+    print(f'{pymupdf.pymupdf_version=}.')
+    print(f'{pymupdf.pymupdf_version_tuple=}.')
+    sys.stdout.flush()
+    path = os.path.normpath(f'{__file__}/../../tests/resources/test_4994.pdf')
+    with pymupdf.open(path) as document:
+        for page in document:
+            text = page.get_text()
+            print(f'test_4994(): {page.get_text()=}')
+            print(f'test_4994(): {page.get_text(flags=pymupdf.TEXTFLAGS_TEXT | pymupdf.TEXT_DEHYPHENATE)=}')
+            print(f'test_4994(): {page.get_text(flags=pymupdf.TEXT_DEHYPHENATE)=}')
+            sys.stdout.flush()
+            search_results = page.search_for('climate')
+            print(f'test_4994(): {search_results=}')
+            print(f'test_4994(): {len(search_results)=}')
+            print(f'test_4994(): {search_results=}')
+            sys.stdout.flush()
+            assert len(search_results) == 2
+            #assert text == 'planet-warming gas driving climate change.'
+    sys.stdout.flush()