preliminary Diff implementation for #230 (draft)

.github/workflows/ci.yml

@@ -16,7 +16,7 @@ jobs:
     - uses: actions/checkout@v4
     - name: Checkout submodules
       # submodules are not used by the build, but needed for `make dist`:
-      run: git submodule update --init deps/json deps/pugixml
+      run: git submodule update --init deps/json deps/pugixml deps/dtl
     - name: Install dependencies
       run: |
         sudo apt-get update && \

.gitmodules

@@ -26,6 +26,9 @@
 [submodule "deps/pugixml"]
 	path = deps/pugixml
 	url = https://github.com/zeux/pugixml.git
+[submodule "deps/dtl"]
+	path = deps/dtl
+	url = https://github.com/cubicdaiya/dtl
 [submodule "deps/boost-subset"]
 	path = deps/boost
 	url = https://github.com/poedit/boost-subset.git

Makefile.am

@@ -12,6 +12,15 @@ dist_metainfo_DATA = net.poedit.Poedit.appdata.xml
 EXTRA_DIST = \
 	deps/json/LICENSE.MIT \
 	deps/json/single_include/nlohmann/json.hpp \
+	deps/dtl/COPYING \
+	deps/dtl/dtl/Diff3.hpp \
+	deps/dtl/dtl/Diff.hpp \
+	deps/dtl/dtl/dtl.hpp \
+	deps/dtl/dtl/functors.hpp \
+	deps/dtl/dtl/Lcs.hpp \
+	deps/dtl/dtl/Sequence.hpp \
+	deps/dtl/dtl/Ses.hpp \
+	deps/dtl/dtl/variables.hpp \
 	deps/pugixml/LICENSE.md \
 	deps/pugixml/src/pugiconfig.hpp \
 	deps/pugixml/src/pugixml.cpp \

configure.ac

@@ -180,7 +180,6 @@ PKG_CHECK_MODULES([ICU], [icu-uc icu-i18n >= 54],
             AC_MSG_ERROR([missing ICU library])
         ])
 
-
 dnl we need GtkSpell and GTK+ >= 2 for this, check if we're compatible
 AC_MSG_CHECKING([if wxWidgets toolkit uses GTK+ 3])
 AC_COMPILE_IFELSE(

src/sidebar.cpp

@@ -134,10 +134,11 @@ class OldMsgidSidebarBlock : public SidebarBlock
         : SidebarBlock(parent, _("Previous source text"))
     {
         m_innerSizer->AddSpacer(PX(2));
-        m_innerSizer->Add(new ExplanationLabel(parent, _("The old source text (before it changed during an update) that the now-inaccurate translation corresponds to.")),
+        m_innerSizer->Add(new ExplanationLabel(parent, _("The difference to old source text (before it changed during an update) that the now-inaccurate translation corresponds to.")),
                      wxSizerFlags().Expand());
         m_innerSizer->AddSpacer(PX(5));
-        m_text = new SelectableAutoWrappingText(parent, WinID::PreviousSourceText, "");
+        //m_text = new SelectableAutoWrappingText(parent, WinID::PreviousSourceText, "");
+        m_text = new wxStaticText(parent, wxID_ANY, "");
         m_innerSizer->Add(m_text, wxSizerFlags().Expand());
     }
 
@@ -148,11 +149,15 @@ class OldMsgidSidebarBlock : public SidebarBlock
 
     void Update(const CatalogItemPtr& item) override
     {
-        m_text->SetAndWrapLabel(item->GetOldMsgid());
+        //m_text->SetAndWrapLabel(item->GetOldMsgid());
+        m_text->SetLabelMarkup(Diff(item->GetOldMsgid(), item->GetString()).getMarkup());
+        m_text->SetMinSize(wxSize(-1, 50));
+
     }
 
 private:
-    SelectableAutoWrappingText *m_text;
+    //SelectableAutoWrappingText *m_text;
+    wxStaticText *m_text;
 };
 
 

src/utility.cpp

@@ -44,6 +44,8 @@
     #include <unistd.h>
 #endif
 
+#include "../deps/dtl/dtl/dtl.hpp"
+
 #include "str_helpers.h"
 
 wxString EscapeMarkup(const wxString& str)
@@ -117,6 +119,67 @@ wxFileName MakeFileName(const wxString& path)
     return fn;
 }
 
+/// A convertor from dtl's to our own diff action type
+Diff::Action dtl2DiffAction(dtl::edit_t lastType) {
+    switch (lastType) {
+        case dtl::SES_DELETE: return Diff::Action::Delete;
+        case dtl::SES_COMMON: return Diff::Action::Common;
+        case dtl::SES_ADD:    return Diff::Action::Add;
+    }
+    // Shouldn't happend
+    throw std::invalid_argument("Unexpected dtl::edit_t value");
+}
+
+Diff::Diff(const wxString& from, const wxString& to) {
+    // Do the diff
+    dtl::Diff<wchar_t, wxString> d(from, to);
+    d.compose();
+    auto dtlSeq = d.getSes().getSequence();
+
+    // collapes per-symbol diff into string-size chunks
+    if (!dtlSeq.empty()) {
+        dtl::edit_t lastType = dtlSeq.front().second.type;
+        wxString curStr = dtlSeq.front().first;
+
+        for ( auto it = std::next(dtlSeq.cbegin()); it != dtlSeq.cend(); ++it ) {
+
+            if ( it->second.type != lastType ) {
+                ses.emplace_back( dtl2DiffAction(lastType), std::move(curStr) );
+                curStr.Empty();
+                lastType = it->second.type;
+            }
+            curStr.Append(it->first);
+        }
+
+        ses.emplace_back( dtl2DiffAction(lastType), std::move(curStr) );
+        curStr.Empty();
+    }
+}
+
+wxString Diff::getMarkup(const wxString &addColor, const wxString &deleteColor) {
+    wxString rv;
+
+    for (const auto& el: ses) {
+        switch (el.first) {
+            case Diff::Action::Common:
+                rv.Append(EscapeMarkup(el.second));
+                break;
+            case Diff::Action::Add:
+                rv.Append("<span bgcolor=\"").Append(addColor).Append("\">");
+                rv.Append(EscapeMarkup(el.second));
+                rv.Append("</span>");
+                break;
+            case Diff::Action::Delete:
+                rv.Append("<s><span bgcolor=\"").Append(deleteColor).Append("\">");
+                rv.Append(EscapeMarkup(el.second));
+                rv.Append("</span></s>");
+                break;
+        }
+    }
+
+    return rv;
+}
+
 #if wxUSE_GUI && defined(__WXMSW__)
 bool IsRunningUnderScreenReader()
 {

src/utility.h

@@ -33,6 +33,7 @@
 #endif
 
 #include <map>
+#include <vector>
 
 #include <wx/arrstr.h>
 #include <wx/filename.h>
@@ -200,6 +201,39 @@ inline wxString MaskForType(const char *extensions, const wxString& description,
         return wxString::Format("%s|%s", description, extensions);
 }
 
+/// A helper class to calculate a display diff of strings
+class Diff
+{
+public:
+    /// Constructs a Diff object with a edit sequence from a string
+    /// @arg from to the string @arg to
+    Diff(const wxString& from, const wxString& to);
+
+    /// A type of element in the shortest edit sequence
+    enum class Action {
+        Common,  ///< symbols are the same
+        Add,     ///< symbols were added
+        Delete   ///< symbols were removed
+    };
+
+    /// An element from the shortest edit sequence
+    typedef std::pair<Action, wxString> sequenceElement;
+    /// A type to represent a shortest edit sequence i.e. the sequence of
+    /// substrings with an action (add remove, don't change) attached to them
+    typedef std::vector<sequenceElement> sequence;
+
+    /// @returns the shortest edit sequence in a form suitable for interpretation
+    const sequence& getSes() { return ses; };
+
+    /// @returns the diff ready to be displayed as a markup
+    /// @arg addColor    a background color for a string being added
+    /// @arg deleteColor a background color for a string being removed
+    wxString getMarkup(const wxString& addColor=wxString("lightgreen"),
+                       const wxString& deleteColor=wxString("pink"));
+
+private:
+    sequence ses;
+};
 
 #if wxUSE_GUI && defined(__WXMSW__)
 bool IsRunningUnderScreenReader();
@@ -267,7 +301,6 @@ class TempOutputFileFor
     wxString m_filenameFinal;
 };
 
-
 #ifdef __WXMSW__
 /// Return filename safe for passing to CLI tools (gettext).
 /// Uses 8.3 short names to avoid Unicode and codepage issues.