initial implementation of ResourceCatalog

fpdf/enums.py

@@ -1068,3 +1068,14 @@ def coerce(cls, value):
         if isinstance(value, str):
             value = value.upper()
         return super(cls, cls).coerce(value)
+
+
+class PDFResourceType(Enum):
+    EXT_G_STATE = intern("ExtGState")
+    COLOR_SPACE = intern("ColorSpece")
+    PATTERN = intern("Pattern")
+    SHADDING = intern("Shading")
+    X_OBJECT = intern("XObject")
+    FONT = intern("Font")
+    PROC_SET = intern("ProcSet")
+    PROPERTIES = intern("Properties")

fpdf/fpdf.py

@@ -85,6 +85,7 @@ class Image:
     PageMode,
     PageOrientation,
     PathPaintRule,
+    PDFResourceType,
     RenderStyle,
     TextDirection,
     TextEmphasis,
@@ -123,6 +124,7 @@ class Image:
     OutputProducer,
     PDFPage,
     PDFPageLabel,
+    ResourceCatalog,
     stream_content_for_raster_image,
 )
 from .recorder import FPDFRecorder
@@ -366,12 +368,8 @@ def __init__(
         self._drawing_graphics_state_registry = GraphicsStateDictRegistry()
         # map page numbers to a set of GraphicsState names:
         self.graphics_style_names_per_page_number = defaultdict(set)
-        self.shadings_per_page_number = defaultdict(set)
-        self.patterns_per_page_number = defaultdict(set)
-        self._shading_registry = {}
-        self._pattern_registry = {}
-
         self._record_text_quad_points = False
+        self._resource_catalog = ResourceCatalog()
 
         # page number -> array of 8 × n numbers:
         self._text_quad_points = defaultdict(list)
@@ -1275,14 +1273,12 @@ def use_pattern(self, shading):
         """
         Create a context for using a shading pattern on the current page.
         """
-        if shading not in self._shading_registry:
-            self._shading_registry[shading] = f"Sh{len(self._shading_registry) + 1}"
+        self._resource_catalog.add(PDFResourceType.SHADDING, shading, self.page)
         pattern = shading.get_pattern()
-        if pattern not in self._pattern_registry:
-            self._pattern_registry[pattern] = "P" + str(len(self._pattern_registry) + 1)
-        self._out(f"/Pattern cs /{self._pattern_registry[pattern]} scn")
-        self.shadings_per_page_number[self.page].add(self._shading_registry[shading])
-        self.patterns_per_page_number[self.page].add(self._pattern_registry[pattern])
+        pattern_name = self._resource_catalog.add(
+            PDFResourceType.PATTERN, pattern, self.page
+        )
+        self._out(f"/Pattern cs /{pattern_name} scn")
         try:
             yield
         finally:
@@ -4875,7 +4871,6 @@ def _default_file_id(self, buffer):
         # > When a file is first written, both identifiers shall be set to the same value.
         id_hash = hashlib.new("md5", usedforsecurity=False)  # nosec B324
         id_hash.update(buffer)
-        print(str(buffer))
         if self.creation_date:
             id_hash.update(self.creation_date.strftime("%Y%m%d%H%M%S").encode("utf8"))
         hash_hex = id_hash.hexdigest().upper()

fpdf/output.py

@@ -16,7 +16,7 @@
 from fontTools import subset as ftsubset
 
 from .annotations import PDFAnnotation
-from .enums import PageLabelStyle, SignatureFlag
+from .enums import PDFResourceType, PageLabelStyle, SignatureFlag
 from .errors import FPDFException
 from .line_break import TotalPagesSubstitutionFragment
 from .image_datastructures import RasterImageInfo
@@ -432,6 +432,41 @@ def serialize(self, _security_handler=None):
         return "\n".join(out)
 
 
+class ResourceCatalog:
+    "Manage the indexing of resources and association to the pages they are used"
+
+    def __init__(self):
+        self.resources = defaultdict(dict)
+        self.resources_per_page = defaultdict(set)
+
+    def add(self, resource_type: PDFResourceType, resource, page_number: int):
+        registry = self.resources[resource_type]
+        if resource not in registry:
+            registry[resource] = f"{self._get_prefix(resource_type)}{len(registry) + 1}"
+        self.resources_per_page[(page_number, resource_type)].add(registry[resource])
+        return registry[resource]
+
+    def get_items(self, resource_type: PDFResourceType):
+        return self.resources[resource_type].items()
+
+    @classmethod
+    def _get_prefix(cls, resource_type: PDFResourceType):
+        match resource_type:
+            case PDFResourceType.EXT_G_STATE:
+                return "GS"
+            # COLOR_SPACE = intern("ColorSpece")
+            case PDFResourceType.PATTERN:
+                return "P"
+            case PDFResourceType.SHADDING:
+                return "Sh"
+            case PDFResourceType.X_OBJECT:
+                return "I"
+            case PDFResourceType.FONT:
+                return "F"
+            case _:
+                raise ValueError(f"No prefix for resource type {resource_type}")
+
+
 class OutputProducer:
     "Generates the final bytearray representing the PDF document, based on a FPDF instance."
 
@@ -893,7 +928,9 @@ def _add_gfxstates(self):
 
     def _add_shadings(self):
         shading_objs_per_name = OrderedDict()
-        for shading, name in self.fpdf._shading_registry.items():
+        for shading, name in self.fpdf._resource_catalog.get_items(
+            PDFResourceType.SHADDING
+        ):
             for function in shading.functions:
                 self._add_pdf_obj(function, "function")
             shading_obj = shading.get_shading_object()
@@ -903,7 +940,9 @@ def _add_shadings(self):
 
     def _add_patterns(self):
         pattern_objs_per_name = OrderedDict()
-        for pattern, name in self.fpdf._pattern_registry.items():
+        for pattern, name in self.fpdf._resource_catalog.get_items(
+            PDFResourceType.PATTERN
+        ):
             self._add_pdf_obj(pattern, "pattern")
             pattern_objs_per_name[name] = pattern
         return pattern_objs_per_name
@@ -943,11 +982,15 @@ def _insert_resources(self, page_objs):
                 }
                 page_shading_objs_per_name = {
                     shading_name: shading_objs_per_name[shading_name]
-                    for shading_name in self.fpdf.shadings_per_page_number[page_number]
+                    for shading_name in self.fpdf._resource_catalog.resources_per_page[
+                        (page_number, PDFResourceType.SHADDING)
+                    ]
                 }
                 page_pattern_objs_per_name = {
                     pattern_name: pattern_objs_per_name[pattern_name]
-                    for pattern_name in self.fpdf.patterns_per_page_number[page_number]
+                    for pattern_name in self.fpdf._resource_catalog.resources_per_page[
+                        (page_number, PDFResourceType.PATTERN)
+                    ]
                 }
                 page_obj.resources = self._add_resources_dict(
                     page_font_objs_per_index,

fpdf/pattern.py

@@ -79,7 +79,9 @@ def __init__(
         super().__init__()
         self.shading_type = shading_type
         self.background = (
-            f'[{" ".join(f"{c:.2f}" for c in background.colors)}]' if background else None
+            f'[{" ".join(f"{c:.2f}" for c in background.colors)}]'
+            if background
+            else None
         )
         self.color_space = Name(color_space)
         self.coords = coords