Refactor number of pages alias (fix #1090)

fpdf/fpdf.py

@@ -17,7 +17,7 @@
 from numbers import Number
 from os.path import splitext
 from pathlib import Path
-from typing import Callable, Iterator, NamedTuple, Optional, Union
+from typing import Callable, Dict, Iterator, NamedTuple, Optional, Union
 
 try:
     from endesive import signer
@@ -101,7 +101,7 @@ class Image:
     preload_image,
 )
 from .linearization import LinearizedOutputProducer
-from .line_break import Fragment, MultiLineBreak, TextLine
+from .line_break import Fragment, MultiLineBreak, TextLine, TotalPagesAliasFragment
 from .outline import OutlineSection
 from .output import (
     OutputProducer,
@@ -279,7 +279,9 @@ def __init__(
         but is less compatible with the PDF spec.
         """
         self.page = 0  # current page number
-        self.pages = {}  # array of PDFPage objects starting at index 1
+        self.pages: Dict[int, PDFPage] = (
+            {}
+        )  # array of PDFPage objects starting at index 1
         self.fonts = {}  # map font string keys to an instance of CoreFont or TTFFont
         # map page numbers to a set of font indices:
         self.fonts_used_per_page_number = defaultdict(set)
@@ -3129,6 +3131,8 @@ def _render_styled_text_line(
                 f"{(self.h - self.y - 0.5 * h - 0.3 * max_font_size) * k:.2f} Td"
             )
             for i, frag in enumerate(fragments):
+                if isinstance(frag, TotalPagesAliasFragment):
+                    self.pages[self.page].add_alias(frag)
                 if frag.graphics_state["text_color"] != last_used_color:
                     # allow to change color within the line of text.
                     last_used_color = frag.graphics_state["text_color"]
@@ -3381,6 +3385,22 @@ def get_fallback_font(self, char, style=""):
     def _parse_chars(self, text: str, markdown: bool) -> Iterator[Fragment]:
         "Split text into fragments"
         if not markdown and not self.is_ttf_font:
+            if self.str_alias_nb_pages:
+                for seq, fragment_text in enumerate(
+                    text.split(self.str_alias_nb_pages)
+                ):
+                    if seq > 0:
+                        yield TotalPagesAliasFragment(
+                            self.str_alias_nb_pages,
+                            self._get_current_graphics_state(),
+                            self.k,
+                        )
+                    if fragment_text:
+                        yield Fragment(
+                            fragment_text, self._get_current_graphics_state(), self.k
+                        )
+                return
+
             yield Fragment(text, self._get_current_graphics_state(), self.k)
             return
         txt_frag, in_bold, in_italics, in_underline = (
@@ -3439,6 +3459,23 @@ def frag():
                     yield frag()
                 current_text_script = text_script
 
+            if self.str_alias_nb_pages:
+                if text[: len(self.str_alias_nb_pages)] == self.str_alias_nb_pages:
+                    if txt_frag:
+                        yield frag()
+                    gstate = self._get_current_graphics_state()
+                    gstate["font_style"] = ("B" if in_bold else "") + (
+                        "I" if in_italics else ""
+                    )
+                    gstate["underline"] = in_underline
+                    yield TotalPagesAliasFragment(
+                        self.str_alias_nb_pages,
+                        gstate,
+                        self.k,
+                    )
+                    text = text[len(self.str_alias_nb_pages) :]
+                    continue
+
             # Check that previous & next characters are not identical to the marker:
             if markdown:
                 if (
@@ -4585,26 +4622,6 @@ def sign(
         )
         self.pages[self.page].annots.append(annotation)
 
-    def _substitute_page_number(self):
-        substituted = False
-        # Replace number of pages in fonts using subsets (unicode)
-        alias = self.str_alias_nb_pages.encode("utf-16-be")
-        encoded_nb = str(self.pages_count).encode("utf-16-be")
-        for page in self.pages.values():
-            substituted |= alias in page.contents
-            page.contents = page.contents.replace(alias, encoded_nb)
-        # Now repeat for no pages in non-subset fonts
-        alias = self.str_alias_nb_pages.encode("latin-1")
-        encoded_nb = str(self.pages_count).encode("latin-1")
-        for page in self.pages.values():
-            substituted |= alias in page.contents
-            page.contents = page.contents.replace(alias, encoded_nb)
-        if substituted:
-            LOGGER.debug(
-                "Substitution of '%s' was performed in the document",
-                self.str_alias_nb_pages,
-            )
-
     def _insert_table_of_contents(self):
         # Doc has been closed but we want to write to self.pages[self.page] instead of self.buffer:
         tocp = self._toc_placeholder
@@ -5158,8 +5175,14 @@ def output(
             # Generating .buffer based on .pages:
             if self._toc_placeholder:
                 self._insert_table_of_contents()
-            if self.str_alias_nb_pages:
-                self._substitute_page_number()
+            for page in self.pages.values():
+                for alias in page.get_aliases():
+                    page.contents = page.contents.replace(
+                        alias.get_alias_string().encode("latin-1"),
+                        alias.render_alias_substitution(str(self.pages_count)).encode(
+                            "latin-1"
+                        ),
+                    )
             if linearize:
                 output_producer_class = LinearizedOutputProducer
             output_producer = output_producer_class(self)

fpdf/line_break.py

@@ -9,6 +9,7 @@
 
 from typing import NamedTuple, Any, List, Optional, Union, Sequence
 from numbers import Number
+from uuid import uuid4
 
 from .enums import Align, CharVPos, TextDirection, WrapMode
 from .errors import FPDFException
@@ -332,6 +333,25 @@ def render_pdf_text_core(self, frag_ws, current_ws):
         return ret
 
 
+class TotalPagesAliasFragment(Fragment):
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.uuid = uuid4()
+
+    def get_alias_string(self):
+        return f"::alias:{self.uuid}::"
+
+    def render_pdf_text(self, *args, **kwargs):
+        self._render_args = args
+        self._render_kwargs = kwargs
+        return self.get_alias_string()
+
+    def render_alias_substitution(self, replacement_text: str):
+        self.characters = list(replacement_text)
+        return super().render_pdf_text(*self._render_args, **self._render_kwargs)
+
+
 class TextLine(NamedTuple):
     fragments: tuple
     text_width: float

fpdf/output.py

@@ -13,9 +13,11 @@
 from contextlib import contextmanager
 from io import BytesIO
 
+
 from .annotations import PDFAnnotation
 from .enums import SignatureFlag
 from .errors import FPDFException
+from .line_break import TotalPagesAliasFragment
 from .image_datastructures import RasterImageInfo
 from .outline import build_outline_objs
 from .sign import Signature, sign_content
@@ -243,6 +245,7 @@ class PDFPage(PDFObject):
         "_index",
         "_width_pt",
         "_height_pt",
+        "_aliases",
     )
 
     def __init__(
@@ -265,6 +268,7 @@ def __init__(
         self.parent = None  # must always be set before calling .serialize()
         self._index = index
         self._width_pt, self._height_pt = None, None
+        self._aliases: list[TotalPagesAliasFragment] = []
 
     def index(self):
         return self._index
@@ -277,6 +281,12 @@ def set_dimensions(self, width_pt, height_pt):
         "Accepts a pair (width, height) in the unit specified to FPDF constructor"
         self._width_pt, self._height_pt = width_pt, height_pt
 
+    def get_aliases(self):
+        return self._aliases
+
+    def add_alias(self, alias):
+        self._aliases.append(alias)
+
 
 class PDFPagesRoot(PDFObject):
     def __init__(self, count, media_box):