Text copying produces broken/spaced characters + inaccurate text selection (Turkish PDF)

[96273]

I'm experiencing two related issues when working with a Turkish-language PDF in SumatraPDF.

1. Copied text has excessive spaces between characters
   When I select and copy text, it is pasted with spaces inserted between almost every character (and sometimes between syllables).

Example:
Original text (should be):

HP VE PSÖRİAZİS
Kronik, immün aracılı inflamatuar bir hastalık olan psöriazisin Hp enfeksiyonu ile tetiklenebileceği düşünülmüş...

Copied text (actual):

HP VE PSÖ Rİ A ZİS
Kro nik, im mün ara cı lı inflamatuar bir has ta lık olan
psö ri a zi sin Hp en fek si yo nu ile te tik le ne bi le ce ği dü-
şü nül müş ol mak la be ra ber ça lış ma lar çe liş ki li so-
nuç lar gös ter miş tir.

---

2. Text selection is inaccurate / overshoots

When trying to select a specific paragraph or line, SumatraPDF often selects extra text from the line above and/or below. It is very difficult to select only the desired text block (see attached screenshots).
PDF used:
https://www.turkiyeklinikleri.com/pdf/?pdf=829d05d6c3a204658233bc43fd943779

---

Screenshots:

Image

---

Environment:

SumatraPDF version: [3.6.1]
Windows version: [Windows 10 22H2]

Could you please investigate this? Accurate text selection and copying is very important for academic work, especially when dealing with non-English documents.
Thank you!

[GitHubRulesOK]

PDF does not use words like a word processor, so each character (or even part of one) is in effect a token that is then drawn in any order on the chart (page).
With luck the PDF writer placed them close together without using justification (kerning) so when reading that text it looks like a word but is not really the case. and also with LUCK it even wrote the whole page in top to bottom (sometimes) or for Arabic right to left but it's common to see that fail miserably in American writers.

SumatraPDF 3.7.### should do better as it has been adjusted for more oddballs but there is never a guarantee that any reader can allow for all writer fails to place text in order so here we see selection is now better for this case but with clipboard adding the kerning it will seem wrong. Even on extract whole page in block order.

HP VE PSÖ Rİ A ZİS
Kro nik, im mün ara cı lı inflamatuar bir has ta lık olan
psö ri a zi sin Hp en fek si yo nu ile te tik le ne bi le ce ği dü-
şü nül müş ol mak la be ra ber ça lış ma lar çe liş ki li so-
nuç lar gös ter miş tir.47-51 Psö ri a zis te Hp se ro po zi tif-
li ği nin art tı ğı nı dü şün dü ren ça lış ma la rın ya nı sı ra
era di kas yon te da vi siy le psö ri a zis lez yon la rın da ge-
ri le me gö rü len olgu sunumlarından olu şan li te ra -
tür ler de mev cut tur.52,53 An cak psöri a zis li has ta lar-
da Hp en fek si yo nu sık lı ğı art mış ol sa bi le era di kas -
yon te da vi si nin psö ri a zis lez yon la rın da ge ri le me ye
yol aç ma dı ğı nı gös te ren bir ça lış ma da mev cut tur.50

The best tool is a very old one called Poppler PDFtoText using -layout switch to place each character it finds in the best top down layout location. Or simply use the Acrobat in Edge. but guess what it reads it the same as SumatraPDF