Fixes #2035

.github/workflows/test_1.21.yml

@@ -0,0 +1,48 @@
+name: Test branch 1.21.
+
+on:
+  schedule:
+    - cron: '13 5 * * *'
+  workflow_dispatch:
+
+jobs:
+
+  test_branch_121:
+    # Simple build+test on single platform using latest MuPDF from git. This is
+    # a cut-down version of `build_wheels` except that we use latest MuPDF from
+    # git and use a single platform and python version.
+    #
+    name: Test branch 1.21
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        os: [ubuntu-latest]
+
+    steps:
+
+      - uses: actions/checkout@v2
+        with:
+          ref: 1.21
+      - uses: actions/setup-python@v2
+
+      # Set up cibuildwheel.
+      #
+      - name: cibuildwheel
+        uses: pypa/cibuildwheel@v2.11.2
+
+        env:
+            # We test building with the default hard-coded mupdf URL.
+
+            # Build on single cpu.
+            CIBW_ARCHS_LINUX: x86_64
+
+            # Build for single python version.
+            CIBW_BUILD: "cp311*"
+
+            # Don't build for unsupported platforms.
+            CIBW_SKIP: "pp* *i686 *-musllinux_* cp36*"
+
+            # Get cibuildwheel to run pytest with each wheel.
+            CIBW_TEST_REQUIRES: "fontTools pytest"
+            CIBW_TEST_COMMAND: "pytest -s {project}/tests"
+            CIBW_BUILD_VERBOSITY: 3

docs/document.rst

@@ -519,7 +519,7 @@ For details on **embedded files** refer to Appendix 3.
 
           - ``startpage``: (int) the first page number (0-based) to apply the label rule. This key **must be present**. The rule is applied to all subsequent pages until either end of document or superseded by the rule with the next larger page number.
           - ``prefix``: (str) an arbitrary string to start the label with, e.g. "A-". Default is "".
-          - ``style``: (str) the numbering style. Available are "D" (decimal), "r"/"R" (Roman numbers, lower / upper case), and "a"/"A" (lower / upper case alphabetical numbering: "a" through "z", then "aa" through "az", etc.). Default is "". If "", no numbering will take place and the pages in that range will receive the same label consisting of the ``prefix`` value. If prefix is also omitted, then the label will be "".
+          - ``style``: (str) the numbering style. Available are "D" (decimal), "r"/"R" (Roman numbers, lower / upper case), and "a"/"A" (lower / upper case alphabetical numbering: "a" through "z", then "aa" through "zz", etc.). Default is "". If "", no numbering will take place and the pages in that range will receive the same label consisting of the ``prefix`` value. If prefix is also omitted, then the label will be "".
           - ``firstpagenum``: (int) start numbering with this value. Default is 1, smaller values are ignored.
 
        For example::

docs/xml-class.rst

@@ -103,17 +103,20 @@ There is no need to ever directly construct an :ref:`Xml` object: after creating
        :arg width: if provided, either an absolute (int) value, or a percentage string like "30%". A percentage value refers to the width of the specified ``where`` rectangle in :meth:`Story.place`. If this value is provided and ``height`` is omitted, the image will be included keeping its aspect ratio.
        :arg height: if provided, either an absolute (int) value, or a percentage string like "30%". A percentage value refers to the height of the specified ``where`` rectangle in :meth:`Story.place`. If this value is provided and ``width`` is omitted, the image's aspect ratio will be honored.
 
-    .. method:: add_link(link)
+    .. method:: add_link(href, text=None)
 
-       Add an :htmlTag:`a` tag.
+       Add an :htmlTag:`a` tag - inline element, treated like text.
+
+       :arg str href: the URL target.
+       :arg str text: the text to display. If omitted, the ``href`` text is shown instead.
 
     .. method:: add_number_list
 
        Add an :htmlTag:`ol` tag, context manager.
 
     .. method:: add_paragraph
 
-       Add a :htmlTag:`p` tag.
+       Add a :htmlTag:`p` tag, context manager.
 
     .. method:: add_span
 

fitz/fitz.i

@@ -2773,15 +2773,17 @@ if len(pyliste) == 0 or min(pyliste) not in range(len(self)) or max(pyliste) not
 
                 if (pdf_is_jpx_image(gctx, obj)) {
                     img_type = FZ_IMAGE_JPX;
+                    res = pdf_load_stream(gctx, obj);
                     ext = "jpx";
                 }
                 if (JM_is_jbig2_image(gctx, obj)) {
                     img_type = FZ_IMAGE_JBIG2;
+                    res = pdf_load_stream(gctx, obj);
                     ext = "jb2";
                 }
-                res = pdf_load_raw_stream(gctx, obj);
                 if (img_type == FZ_IMAGE_UNKNOWN) {
-                    unsigned char *c = NULL;
+                    res = pdf_load_raw_stream(gctx, obj);
+                     unsigned char *c = NULL;
                     fz_buffer_storage(gctx, res, &c);
                     img_type = fz_recognize_image_format(gctx, c);
                     ext = JM_image_extension(img_type);
@@ -2793,9 +2795,10 @@ if len(pyliste) == 0 or min(pyliste) not in range(len(self)) or max(pyliste) not
                     res = fz_new_buffer_from_image_as_png(gctx, img,
                                 fz_default_color_params);
                     ext = "png";
-                } else /*if (smask == 0)*/ {
+                } else {
                     img = fz_new_image_from_buffer(gctx, res);
                 }
+
                 fz_image_resolution(img, &xres, &yres);
                 width = img->w;
                 height = img->h;
@@ -2833,7 +2836,8 @@ if len(pyliste) == 0 or min(pyliste) not in range(len(self)) or max(pyliste) not
 
             fz_catch(gctx) {
                 Py_CLEAR(rc);
-                Py_RETURN_NONE;
+                fz_warn(gctx, fz_caught_message(gctx));
+                Py_RETURN_FALSE;
             }
             if (!rc)
                 Py_RETURN_NONE;
@@ -13122,11 +13126,13 @@ struct Xml
             self.append_child(child)
             return child
 
-        def add_link(self, text=None):
+        def add_link(self, href, text=None):
             """Add a hyperlink ("a" tag)"""
             child = self.create_element("a")
-            if type(text) is str:
-               child.append_child(self.create_text_node(text)) 
+            if not isinstance(text, str):
+                text = href
+            child.set_attribute("href", href)
+            child.append_child(self.create_text_node(text)) 
             prev = self.span_bottom()
             if prev == None:
                 prev = self

fitz/helper-annot.i

@@ -310,19 +310,25 @@ PyObject *JM_get_annot_id_list(fz_context *ctx, pdf_page *page)
 PyObject *JM_get_annot_xref_list(fz_context *ctx, pdf_obj *page_obj)
 {
     PyObject *names = PyList_New(0);
-    pdf_obj *id, *annot_obj = NULL;
+    pdf_obj *id, *subtype, *annot_obj = NULL;
     pdf_obj *annots = pdf_dict_get(ctx, page_obj, PDF_NAME(Annots));
+    const char *name=NULL;
+    int type;
     if (!annots) return names;
     fz_try(ctx) {
         int i, n = pdf_array_len(ctx, annots);
         for (i = 0; i < n; i++) {
             annot_obj = pdf_array_get(ctx, annots, i);
             int xref = pdf_to_num(ctx, annot_obj);
-            pdf_obj *subtype = pdf_dict_get(ctx, annot_obj, PDF_NAME(Subtype));
-            int type = PDF_ANNOT_UNKNOWN;
+            subtype = pdf_dict_get(ctx, annot_obj, PDF_NAME(Subtype));
             if (subtype) {
-                const char *name = pdf_to_name(ctx, subtype);
+                name = pdf_to_name(ctx, subtype);
                 type = pdf_annot_type_from_string(ctx, name);
+                if (type == PDF_ANNOT_UNKNOWN) {
+                    continue;  // only accept valid annot types
+                }
+            } else {
+                continue;
             }
             id = pdf_dict_gets(ctx, annot_obj, "NM");
             LIST_APPEND_DROP(names, Py_BuildValue("iis", xref, type, pdf_to_text_string(ctx, id)));

fitz/helper-devices.i

@@ -101,6 +101,7 @@ jm_checkrect()
 	dev_linecount = 0; // reset line count
 	long orientation = 0;
 	fz_point ll, lr, ur, ul;
+	fz_rect r;
 	PyObject *rect;
 	PyObject *line0, *line2;
 	PyObject *items = PyDict_GetItem(dev_pathdict, dictkey_items);
@@ -109,50 +110,35 @@ jm_checkrect()
 	line0 = PyList_GET_ITEM(items, len - 3);
 	ll = JM_point_from_py(PyTuple_GET_ITEM(line0, 1));
 	lr = JM_point_from_py(PyTuple_GET_ITEM(line0, 2));
-
+	// no need to extract "line1"!
 	line2 = PyList_GET_ITEM(items, len - 1);
 	ur = JM_point_from_py(PyTuple_GET_ITEM(line2, 1));
 	ul = JM_point_from_py(PyTuple_GET_ITEM(line2, 2));
 
 	/*
 	---------------------------------------------------------------------
-	Three connected lines: at least a quad! Check whether even a rect.
-	For this, the lines must be parallel to the axes.
 	Assumption:
 	For decomposing rects, MuPDF always starts with a horizontal line,
 	followed by a vertical line, followed by a horizontal line.
 	We will also check orientation of the enclosed area and add this info
 	as '+1' for anti-clockwise, '-1' for clockwise orientation.
 	---------------------------------------------------------------------
 	*/
-	if (ll.y != lr.y) {  // not horizontal
-		goto drop_out;
-	}
-	if (lr.x != ur.x) {  // not vertical
-		goto drop_out;
-	}
-	if (ur.y != ul.y) {  // not horizontal
-		goto drop_out;
+	if (ll.y != lr.y ||
+		ll.x != ul.x ||
+		ur.y != ul.y ||
+		ur.x != lr.x) {
+		goto drop_out;  // not a rectangle
 	}
-	// we have a rect, determine orientation
-	if (ll.x < lr.x) {  // move left to right
-		if (lr.y > ur.y) {  // move upwards
-			orientation = 1;
-		} else {
-			orientation = -1;
-		}
-	} else {  // move right to left
-		if (lr.y < ur.y) {  // move downwards
-			orientation = 1;
-		} else {
-			orientation = -1;
-		}
+
+	// we have a rect, replace last 3 "l" items by one "re" item.
+	if (ul.y < lr.y) {
+		r = fz_make_rect(ul.x, ul.y, lr.x, lr.y);
+		orientation = 1;
+	} else {
+		r = fz_make_rect(ll.x, ll.y, ur.x, ur.y);
+		orientation = -1;
 	}
-	// Replace last 3 "l" items by one "re" item.
-	fz_rect r = fz_make_rect(ul.x, ul.y, ul.x, ul.y);
-	r = fz_include_point_in_rect(r, ur);
-	r = fz_include_point_in_rect(r, ll);
-	r = fz_include_point_in_rect(r, lr);
 	rect = PyTuple_New(3);
 	PyTuple_SET_ITEM(rect, 0, PyUnicode_FromString("re"));
 	PyTuple_SET_ITEM(rect, 1, JM_py_from_rect(r));
@@ -493,7 +479,7 @@ jm_trace_text_span(fz_context *ctx, PyObject *out, fz_text_span *span, int type,
 	// make the span dictionary
 	PyObject *span_dict = PyDict_New();
 	DICT_SETITEMSTR_DROP(span_dict, "dir", JM_py_from_point(dir));
-	DICT_SETITEM_DROP(span_dict, dictkey_font, Py_BuildValue("s",fontname));
+	DICT_SETITEM_DROP(span_dict, dictkey_font, JM_EscapeStrFromStr(fontname));
 	DICT_SETITEM_DROP(span_dict, dictkey_wmode, PyLong_FromLong((long) span->wmode));
 	DICT_SETITEM_DROP(span_dict, dictkey_flags, PyLong_FromLong((long) fflags));
 	DICT_SETITEMSTR_DROP(span_dict, "bidi_lvl", PyLong_FromLong((long) span->bidi_level));

fitz/helper-fields.i

@@ -1030,8 +1030,7 @@ class Widget(object):
         return "'%s' widget on %s" % (self.field_type_string, str(self.parent))
 
     def __del__(self):
-        annot = getattr(self, "_annot")
-        if annot:
+        if hasattr(self, "_annot"):
             del self._annot
 
     @property

fitz/helper-python.i

@@ -844,6 +844,7 @@ class linkDest(object):
             self.page = -1
             self.kind = LINK_NONE
         if isInt and self.uri:
+            self.uri = self.uri.replace("&zoom=nan", "&zoom=0")
             if self.uri.startswith("#"):
                 self.named = ""
                 self.kind = LINK_GOTO

fitz/helper-stext.i

@@ -16,7 +16,11 @@ JM_font_ascender(fz_context *ctx, fz_font *font)
     if (skip_quad_corrections) {
         return 0.8f;
     }
-    return fz_font_ascender(ctx, font);
+    float asc = fz_font_ascender(ctx, font);
+    if (asc > 10) {
+        return 1.0f;
+    }
+    return asc;
 }
 
 static const float
@@ -25,7 +29,11 @@ JM_font_descender(fz_context *ctx, fz_font *font)
     if (skip_quad_corrections) {
         return -0.2f;
     }
-    return fz_font_descender(ctx, font);
+    float dsc =fz_font_descender(ctx, font);
+    if (dsc < -10) {
+        return -0.2f;
+    }
+    return dsc;
 }
 
 
@@ -97,16 +105,18 @@ JM_char_quad(fz_context *ctx, fz_stext_line *line, fz_stext_char *ch)
         return ch->quad;
     }
     fz_font *font = ch->font;
+    fz_matrix trm1, trm2, xlate1, xlate2;
+    fz_quad quad;
+    float c, s, fsize = ch->size;
     float asc = JM_font_ascender(ctx, font);
     float dsc = JM_font_descender(ctx, font);
-    float c, s, fsize = ch->size;
     float asc_dsc = asc - dsc + FLT_EPSILON;
     if (asc_dsc >= 1 && small_glyph_heights == 0) {  // no problem
        return ch->quad;
     }
     if (asc < 1e-3) {  // probably Tesseract glyphless font
-        dsc = -0.1f;
-        asc = 0.9f;
+        dsc = -0.2f;
+        asc = 0.8f;
         asc_dsc = 1.0f;
     }
 
@@ -123,8 +133,6 @@ JM_char_quad(fz_context *ctx, fz_stext_line *line, fz_stext_char *ch)
     Move ch->origin to (0,0) and de-rotate quad, then adjust the corners,
     re-rotate and move back to ch->origin location.
     ------------------------------ */
-    fz_matrix trm1, trm2, xlate1, xlate2;
-    fz_quad quad;
     c = line->dir.x;  // cosine
     s = line->dir.y;  // sine
     trm1 = fz_make_matrix(c, -s, s, c, 0, 0);  // derotate
@@ -139,35 +147,44 @@ JM_char_quad(fz_context *ctx, fz_stext_line *line, fz_stext_char *ch)
     quad = fz_transform_quad(ch->quad, xlate1);  // move origin to (0,0)
     quad = fz_transform_quad(quad, trm1);  // de-rotate corners
 
+    quad.ul.y = -asc;
+    quad.ur.y = -asc;
+    quad.ll.y = -dsc;
+    quad.lr.y = -dsc;
+
     // adjust vertical coordinates
     if (c == 1 && quad.ul.y > 0) {  // up-down flip
         quad.ul.y = asc;
         quad.ur.y = asc;
         quad.ll.y = dsc;
         quad.lr.y = dsc;
-    } else {
+    } else if (abs(c) == 1) {
         quad.ul.y = -asc;
         quad.ur.y = -asc;
         quad.ll.y = -dsc;
         quad.lr.y = -dsc;
     }
 
     // adjust horizontal coordinates that are too crazy:
-    // (1) left x must be >= 0
+    // (1) left and right x must be >= 0
     // (2) if bbox width is 0, lookup char advance in font.
     if (quad.ll.x < 0) {
-        quad.ll.x = 0;
-        quad.ul.x = 0;
+        quad.ll.x = quad.ul.x = 0;
+    }
+    if (quad.lr.x <= 0) {
+        quad.lr.x = quad.ur.x = 0.5 * fsize;
     }
     float cwidth = quad.lr.x - quad.ll.x;
     if (cwidth < FLT_EPSILON) {
         int glyph = fz_encode_character(ctx, font, ch->c);
-        if (glyph) {
+        if (glyph>=0) {
             float fwidth = fz_advance_glyph(ctx, font, glyph, line->wmode);
-            quad.lr.x = quad.ll.x + fwidth * fsize;
-            quad.ur.x = quad.lr.x;
-        }
-    }
+            if (fwidth > cwidth) {
+                quad.lr.x = quad.ll.x + fwidth;
+            	quad.ur.x = quad.lr.x;
+        	}
+    	}
+	}
 
     quad = fz_transform_quad(quad, trm2);  // rotate back
     quad = fz_transform_quad(quad, xlate2);  // translate back
@@ -567,8 +584,8 @@ JM_make_spanlist(fz_context *ctx, PyObject *line_dict,
             span = PyDict_New();
             float asc = style.asc, desc = style.desc;
             if (style.asc < 1e-3) {
-                asc = 0.9f;
-                desc = -0.1f;
+                asc = 0.8f;
+                desc = -0.2f;
             }
 
             DICT_SETITEM_DROP(span, dictkey_size, Py_BuildValue("f", style.size));