From e3f60c1ae6092d350ffb2080ad53e23a6731f634 Mon Sep 17 00:00:00 2001
From: pubpub-zz <4083478+pubpub-zz@users.noreply.github.com>
Date: Sun, 24 Sep 2023 19:11:34 +0200
Subject: [PATCH] BUG: PDF size increases because of too high float writing
 precision (#2213)

See #1910

address regression from #2203
---
 pypdf/generic/_base.py                        |   7 +++++--
 ..._Vicksburg_Sample_OCR-crazyones-merged.pdf | Bin 217836 -> 217096 bytes
 tests/test_generic.py                         |   2 ++
 tests/test_writer.py                          |   5 ++++-
 4 files changed, 11 insertions(+), 3 deletions(-)

diff --git a/pypdf/generic/_base.py b/pypdf/generic/_base.py
index 6c3e41647..a50bb5faf 100644
--- a/pypdf/generic/_base.py
+++ b/pypdf/generic/_base.py
@@ -379,6 +379,9 @@ def readFromStream(
         return IndirectObject.read_from_stream(stream, pdf)
 
 
+FLOAT_WRITE_PRECISION = 8  # shall be min 5 digits max, allow user adj
+
+
 class FloatObject(float, PdfObject):
     def __new__(
         cls, value: Union[str, Any] = "0.0", context: Optional[Any] = None
@@ -409,8 +412,8 @@ def clone(
     def myrepr(self) -> str:
         if self == 0:
             return "0.0"
-        nb = int(log10(abs(self)))
-        s = f"{self:.{max(1,16-nb)}f}".rstrip("0").rstrip(".")
+        nb = FLOAT_WRITE_PRECISION - int(log10(abs(self)))
+        s = f"{self:.{max(1,nb)}f}".rstrip("0").rstrip(".")
         return s
 
     def __repr__(self) -> str:
diff --git a/resources/Seige_of_Vicksburg_Sample_OCR-crazyones-merged.pdf b/resources/Seige_of_Vicksburg_Sample_OCR-crazyones-merged.pdf
index 365637b94af1e4f8540ab347785d68b5f6b3574c..0e9633ac16c138eeaa90d3cf13e9f7cd6e2c006d 100644
GIT binary patch
delta 1454
zcmZuwJ&P1U5M{kXbmXGNk3*EclZ%9*tE;N33j`xCJki`h(L|0DlM75XFi=pBg$WMs
z5(AG-l@0tG`X3Da5mwLaXOA;5GyQl~@73#`zZY8{zixeh`=!}AyEzoZIyyhQW1?CN
z-@Z*rYnY)#z7Fkoi#B&o+>J$vcNEE5phAt)cPD=a1DYO=A6~t$3l6E)-D>$|buHRH
zO>(F4XVs>V*3Iv4ySCUIk}cGAPow9R5r_AdO$b6wCWC>D`K)?%^zzM%*Hw__$LH(K
zt?~YVn#1w#^!)tSmCNK<eD3bHLgKcreP{6@8Iordqp*`8b!yzj-BE_jXE%P1Nhr*M
z0ngly)z!>1?A{Dmr&cgI;#kY#TdW+x%&aixe%P9$>sHCC@L*Q9O72MHA&2AH3eX(J
zIkRoix$KxTem3#`c=F+RGWdBsx%6rB0HZjcFL&gs_5&kP>S6L9?DYA;#&Dz_?C`9&
zP*B^F0wdpRs}U@!H68)0os$9+DQ!{!%C#e7F1%Af>=6a~A}L@`BvG_2UK19tT$+{;
zftS3j?|l+ZrEn74fhDoyB~kmO^0hFnAz&XPE)oQ=Pzy5vIAX>1h7iP5N-TLS8wNGs
z1rKeyK9<_3S4;z9^sR=?Kc57U0~h-V?3If10Sw@aiGkVUN~}2{!$K?v;pl1AV9p;{
zqL#vfT%H9Gj9P372nm+4N{T&$lI(yKX({RxU3HVqD93W|0(gZ&I9Lef&lkY%)AOer
Q``4^VtoHUEJ>IYW0p_tN+yDRo

delta 2236
zcma)7&1w`u5GMY{)p!uai%Lfb>SgTeuIlO*au76lN}j-rD1t~NPcVW9b8tnNgP4<I
zBv)k*c?@4dJb98lfHnKuQ#}D^S$6rly1uWzs_xnSyz=MU%J-e&ip?Hsc81q(XMlCa
zPCp@H#raYjeD#4f7*hrL!fX$stsv#kI8_<n{2ZF~JOj*GC4Imu<crSu1eMPr7@dCq
zYw4oTxl*&acXM>%!}W77V~)ppZcM4_Svm$5C1rdGwWc($hs)FNsY{Hh;#|Pkx9Rek
zdnn6FL=Pn7MO^b`czL~))b<O(RUI?hqP;`K)cpSRT$wi7-21z{s`e7wV@kN?B9%-f
zICJH$Xp<eqRIo6RVaJ;1;OmV#?=(pv;%4PEI2lt$*Efq`iK>)sX$Ne=IUDcEFM(GI
z4An*T1}Rh_?`blnM-s(aj0ZG*93{oV!+J{8_K>7LmlKm&1kU54$Wu{0^8QZ;VO>z7
z1vN<qt;|7Yumw4#Ngv!k6=g-LiI1$@LOq4|Ks|kVJb1SK^wpCW0}suQcUPLTd-vZQ
z?Y}+RUphS6KXK=v#)xJt+-#e{_>T`1WX{?r!(xfCjWRafxC9sqOpacGg*3)wx)&rV
z*`xvyVv^2xNYGi<B<Nxx2EAq0d!waX^!RkimCOZka_#NkA!Z1L81|mDTQPd4iDSg5
z^TkB9m$TR*NiK1Ld6?L>a0|eNsGI>n?|nBPxEMecCm8djCH35Xa4BllAn8a|A{c3_
z!J|h_;z|0c#1MqoX`T2F1QAumK*X$U1(NrmCZWJl3#2HWSHyCt^eg})`dak_n2@!t
zoO~gW6Ll+*LK3pqVp8zBbq*lVly`MD{Q^d-HoLZY9}l-qZk(4aVX(ILaBE}m4<GE<
A7XSbN

diff --git a/tests/test_generic.py b/tests/test_generic.py
index 5b0dcd6a1..0e0fff677 100644
--- a/tests/test_generic.py
+++ b/tests/test_generic.py
@@ -1087,7 +1087,9 @@ def test_create_string_object_force():
         ("99900000000000000123", "99900000000000000000"),
         ("99900000000000000123.456000", "99900000000000000000"),
         ("0.00000000000000000000123", "0.00000000000000000000123"),
+        ("0.00000123", "0.00000123"),
         ("0.00000000000000000000123000", "0.00000000000000000000123"),
+        ("-4.6", "-4.6"),  # from #1910
         # (
         #    "50032481330523882508234.00000000000000000000123000",
         #    "50032481330523882508234.00000000000000000000123",
diff --git a/tests/test_writer.py b/tests/test_writer.py
index 70352833f..358fdb1cf 100644
--- a/tests/test_writer.py
+++ b/tests/test_writer.py
@@ -985,7 +985,9 @@ def test_colors_in_outline_item(pdf_file_path):
     reader2 = PdfReader(pdf_file_path)
     for outline_item in reader2.outline:
         # convert float to string because of mutability
-        assert [str(c) for c in outline_item.color] == [str(p) for p in purple_rgb]
+        assert ["%.5f" % c for c in outline_item.color] == [
+            "%.5f" % p for p in purple_rgb
+        ]
 
 
 @pytest.mark.samples()
@@ -1619,6 +1621,7 @@ def test_watermark_rendering(tmp_path):
     assert image_similarity(png_path, target_png_path) >= 0.95
 
 
+@pytest.mark.skipif(GHOSTSCRIPT_BINARY is None, reason="Requires Ghostscript")
 def test_watermarking_reportlab_rendering(tmp_path):
     """
     This test is showing a rotated+mirrored watermark in pypdf==3.15.4.