The code is shamelessly borrowed from kdelibs4support implementation:
https://cgit.kde.org/kdelibs4support.git/tree/src/kdecore/kmimetype.cpp
Personally, I do not know if there is a better way to implement this check without libmagic. Please help.
Details
Details
- Reviewers
broulik - Group Reviewers
KDE Applications
- Compile and run patched KFind.
- Try to find some short text in the folder with binary files. Nothing should be found.
- Try to find binary files with some metadata (e.g., "Radiohead" in the folder with Radiohead MP3s) and empty "Containing text" field. All files with "Radiohead" in metadata should be found.
Diff Detail
Diff Detail
- Repository
- R228 KFind
- Lint
Lint Skipped - Unit
Unit Tests Skipped
Comment Actions
The question was raised previously here:
https://mail.kde.org/pipermail/kde-frameworks-devel/2015-January/021254.html
I guess that this is the second user of isBinaryData...
Better ask @dfaure
("KDE Applications" is not meant to be used as tag for reviews, it's used to collect the projects; the last one who did significant changes to kfind is @broulik )
| src/kquery.cpp | ||
|---|---|---|
| 451 | Actually, that's 128 bytes nowadays. static inline bool isTextFile(const QByteArray &data)
{
// UTF16 byte order marks
static const char bigEndianBOM[] = "\xFE\xFF";
static const char littleEndianBOM[] = "\xFF\xFE";
if (data.startsWith(bigEndianBOM) || data.startsWith(littleEndianBOM))
return true;
// Check the first 128 bytes (see shared-mime spec)
const char *p = data.constData();
const char *e = p + qMin(128, data.size());
for ( ; p < e; ++p) {
if ((unsigned char)(*p) < 32 && *p != 9 && *p !=10 && *p != 13)
return false;
}
return true;
}But yeah I'm not sure it's worth its own public method in QMimeDatabase, it's just some inexact heuristic for last-chance fallback... | |