Mercurial > octave
changeset 32023:bce1850f8104
gui: Use iconv/gnulib to load editor content from files in any encoding.
* bootstrap.conf: Add module "uniconv/u16-conv-from-enc".
* liboctave/wrappers/uniconv-wrappers.c, liboctave/wrappers/uniconv-wrappers.h
(octave_u16_conv_from_encoding, octave_u16_conv_from_encoding_strict): Add
wrappers for gnulib function.
* liboctave/util/oct-string.cc, liboctave/util/oct-string.h
(octave::string::u16_to_encoding): Add new function.
* libgui/src/m-editor/file-editor-tab.h: Remove header that is no longer needed.
* libgui/src/m-editor/file-editor-tab.cc
(octave::file_editor_tab::load_file): Use iconv/gnulib functions to load file
editor content from file in any encoding. The interpreter needs to be able to
read these files using the iconv/gnulib functions anyway. And Qt6 can only
convert from UTF-8, UTF-16, UTF-32 or the system locale encoding.
| author | Markus Mützel <markus.muetzel@gmx.de> |
|---|---|
| date | Sat, 15 Apr 2023 20:51:50 +0200 |
| parents | 5945e1bd73ea |
| children | 6db223403fdb |
| files | bootstrap.conf libgui/src/m-editor/file-editor-tab.cc libgui/src/m-editor/file-editor-tab.h liboctave/wrappers/uniconv-wrappers.c liboctave/wrappers/uniconv-wrappers.h |
| diffstat | 5 files changed, 72 insertions(+), 14 deletions(-) [+] |
line wrap: on
line diff
--- a/bootstrap.conf Sat Apr 15 07:17:15 2023 -0700 +++ b/bootstrap.conf Sat Apr 15 20:51:50 2023 +0200 @@ -105,6 +105,7 @@ uname unicase/u8-tolower unicase/u8-toupper + uniconv/u16-conv-from-enc uniconv/u16-conv-to-enc uniconv/u32-conv-to-enc uniconv/u8-conv-from-enc
--- a/libgui/src/m-editor/file-editor-tab.cc Sat Apr 15 07:17:15 2023 -0700 +++ b/libgui/src/m-editor/file-editor-tab.cc Sat Apr 15 20:51:50 2023 +0200 @@ -1881,21 +1881,50 @@ text_data.chop (1); } - // decode - QTextCodec::ConverterState st; - QTextCodec *codec = QTextCodec::codecForName (m_encoding.toLatin1 ()); - if (codec == nullptr) - codec = QTextCodec::codecForLocale (); - - const QString text = codec->toUnicode(text_data.constData(), - text_data.size(), &st); - - // Decoding with invalid characters? - if (st.invalidChars > 0) + // expected file encoding + std::string encoding = m_encoding.toStdString (); + if (encoding.compare (0, 6, "SYSTEM") == 0) + encoding = octave_locale_charset_wrapper (); + + // check if the selected encoding can be used to decode the file + + const char *src = text_data.constData (); + std::size_t srclen = text_data.length (); + + std::size_t length; + uint16_t *u16_str; + + // try to convert encoding in strict mode + u16_str = octave_u16_conv_from_encoding_strict (encoding.c_str (), + src, srclen, &length); + + // check for invalid characters in input file + if (! u16_str) { // Set read only m_edit_area->setReadOnly (true); + // convert encoding allowing replacements + u16_str = octave_u16_conv_from_encoding (encoding.c_str (), + src, srclen, &length); + + if (! u16_str) + { + // FIXME: Can this ever happen? + + // non-modal error message box + QMessageBox *msgBox + = new QMessageBox (QMessageBox::Critical, + tr ("Octave Editor"), + tr ("Unable to read file '%1'\n" + "with selected encoding '%2': %3") + .arg (file_to_load).arg (m_encoding) + .arg (std::strerror (errno)), + QMessageBox::Ok, nullptr); + show_dialog (msgBox, false); + return QString (); + } + // Message box for user decision QString msg = tr ("There were problems reading the file\n" "%1\n" @@ -1919,13 +1948,18 @@ msg_box->show (); } + unwind_action free_u16_str ([=] () { ::free (u16_str); }); + + QString text + = QString::fromUtf16 (reinterpret_cast<char16_t *> (u16_str), length); + m_edit_area->setText (text); m_edit_area->setEolMode (detect_eol_mode ()); QApplication::restoreOverrideCursor (); - m_copy_available = false; // no selection yet available - m_edit_area->setModified (false); // loaded file is not modified yet + m_copy_available = false; // no selection yet available + m_edit_area->setModified (false); // loaded file is not modified yet set_file_name (file_to_load); update_eol_indicator ();
--- a/libgui/src/m-editor/file-editor-tab.h Sat Apr 15 07:17:15 2023 -0700 +++ b/libgui/src/m-editor/file-editor-tab.h Sat Apr 15 20:51:50 2023 +0200 @@ -33,7 +33,6 @@ #include <QFileSystemWatcher> #include <QLabel> #include <QStatusBar> -#include <QTextCodec> #include <QWidget> #include <Qsci/qsciapis.h>
--- a/liboctave/wrappers/uniconv-wrappers.c Sat Apr 15 07:17:15 2023 -0700 +++ b/liboctave/wrappers/uniconv-wrappers.c Sat Apr 15 20:51:50 2023 +0200 @@ -92,6 +92,22 @@ src, srclen, NULL, lengthp); } +uint16_t * +octave_u16_conv_from_encoding (const char *fromcode, const char *src, + size_t srclen, size_t *lengthp) +{ + return u16_conv_from_encoding (fromcode, iconveh_question_mark, + src, srclen, NULL, NULL, lengthp); +} + +uint16_t * +octave_u16_conv_from_encoding_strict (const char *fromcode, const char *src, + size_t srclen, size_t *lengthp) +{ + return u16_conv_from_encoding (fromcode, iconveh_error, + src, srclen, NULL, NULL, lengthp); +} + char * octave_u16_conv_to_encoding (const char *tocode, const uint16_t *src, size_t srclen, size_t *lengthp)
--- a/liboctave/wrappers/uniconv-wrappers.h Sat Apr 15 07:17:15 2023 -0700 +++ b/liboctave/wrappers/uniconv-wrappers.h Sat Apr 15 20:51:50 2023 +0200 @@ -53,6 +53,14 @@ octave_u8_conv_to_encoding_strict (const char *tocode, const uint8_t *src, size_t srclen, size_t *lengthp); +extern OCTAVE_API uint16_t * +octave_u16_conv_from_encoding (const char *fromcode, const char *src, + size_t srclen, size_t *lengthp); + +extern OCTAVE_API uint16_t * +octave_u16_conv_from_encoding_strict (const char *fromcode, const char *src, + size_t srclen, size_t *lengthp); + extern OCTAVE_API char * octave_u16_conv_to_encoding (const char *tocode, const uint16_t *src, size_t srclen, size_t *lengthp);