| 1 |
### After 1.1 |
| 2 |
- PdfTokenizer: Evaluate making sure data type is set also in case of errors, in DetermineDataType/tryReadDataType. This would help leaving the object operable with the right type even in case of parsing errors. This is done already for dictionaries/arrays, see ReadDictionary/ReadArray |
| 3 |
- PdfPredictorDecoder: Support BitsPerComponent != 8 |
| 4 |
- PdfColorSpaceFilterIndexed::GetSourceScanLineSize() handle bitsPerComponent != 8 |
| 5 |
- Cleanup PdfTreeNode |
| 6 |
- Add subsetting of PdfDifferenceEncoding |
| 7 |
- PdfDifferenceList: Validate insertion for the public Add methods (like for example enforce "Adobe Glyph List For New Fonts") |
| 8 |
- Optimize compilation SALSprep |
| 9 |
- Optimize small allocations in tryGetCodePointsFromCharNameLigatures(), |
| 10 |
tryGetCodePointsFromUnicodeHexLigatures() in PdfDifferenceEncoding.cpp |
| 11 |
- PdfEncoding: Evaluate adding move semantics |
| 12 |
- PdfVariant/PdfObject: Evaluate adding a TryGetStringLenient(string_view& str) |
| 13 |
that catches both PdfString/PdfName |
| 14 |
- Add remaining PdfNameTree(s) (also stub) |
| 15 |
- Add remaining PdfContentStreamOperators |
| 16 |
- According to 5014.CIDFont_Spec page 71-72-73 begincidchar, begincidrange, |
| 17 |
beginbfchar, beginbfrange have a limit of 100 entries. Fix it where relevant |
| 18 |
- noexcept nullable<T> methods |
| 19 |
- Check/Review doxygen doc |
| 20 |
- If the doc is updated, then should not allow to set an encryption |
| 21 |
- If the doc is updated and has an encryption, it should not allow to remove it |
| 22 |
in a following update |
| 23 |
- Evaluate make PdfObjectStream not flate filter by default in PdfMemDocument? |
| 24 |
- Evaluate move more utf8::next to utf8::unchecked::next |
| 25 |
- PdfToggleButton: Add proper IsChecked/ExportValue handling |
| 26 |
- Add version of PdfFont::TryCreateSubstituteFont for rendering |
| 27 |
(metrics/widths of loaded font override metrics found on /FontFile) |
| 28 |
- Add a fallback to search font on the system for text extraction purposes, |
| 29 |
see https://github.com/podofo/podofo/issues/123 |
| 30 |
- Check PdfWriter should really update doc trailer when saving. |
| 31 |
Now the new trailer is written but the doc still has the old one |
| 32 |
- PdfMemDocument: Check the DeviceStream is not empty before doing an incremental update/signing operation |
| 33 |
- PdfMemDocument: Prevent Save() operation after signing operation |
| 34 |
- PdfMemDocument: Evaluate release the device after all objects have been loaded (eg. after a full Save()) |
| 35 |
- PdfParserObject: Evaluate release the device after loading |
| 36 |
- PdfElement: Optimize, keep dictionary/array pointer. Evaluate Add shared_ptr PdfElement::GetObjectPtr() |
| 37 |
- Check what do with tools/restore manuals |
| 38 |
- Fix/complete handling of text extraction in rotated pages (??? Done?) |
| 39 |
- Add method to retrieve shared_ptr from PdfObject, PdfFont (and |
| 40 |
maybe others) to possibly outlive document destruction |
| 41 |
- Do more overflow checks using Chromium numerics, which is now |
| 42 |
bundled. See comments in utls::DoesMultiplicationOverflow() |
| 43 |
- PdfFontManager: Add font hash to cache descriptor |
| 44 |
- Add special SetAppearance for PdfSignature respecting |
| 45 |
"Digital Signature Appearances" document specification |
| 46 |
- Add text shaping with Harfbuzz https://github.com/harfbuzz/harfbuzz |
| 47 |
- Add fail safe sign/update mechanism, meaning the stream gets trimmed |
| 48 |
to initial length if there's a crash. Not so easy, especially since |
| 49 |
we are now using STL streams and it's not easy to trim files |
| 50 |
without access to native handle and low level I/O operations |
| 51 |
- Option to unfold Unicode ligatures to separate codepoints during encoded -> utf8 conversion |
| 52 |
- Option to convert Unicode ligatures <-> separate codepoints when drawing strings/converting to encoded |
| 53 |
- Optimize charbuff to not initialize memory, keeping std::string compatibility, |
| 54 |
see https://en.cppreference.com/w/cpp/string/basic_string/resize_and_overwrite |
| 55 |
- Add backtrace: https://github.com/boostorg/stacktrace |
| 56 |
|
| 57 |
### Ideas: |
| 58 |
- PdfFontManager: Consider also statically caching the queries and filepaths. |
| 59 |
Maybe we could also weakly (weak shared pointer) cache metrics instead of fonts |
| 60 |
- PdfName: Evaluate unescape lazily, or offer a way to debug/inspect the unescaped sequence a posteriori |
| 61 |
- Consider saving/converting to XRef stream by default |