lupdate: remove number heuristics

Adapt the troll print example

[ChangeLog][lupdate][Important Behavior Changes] Removed the
number heuristics feature. Users are advised to avoid hard-coded
numbers in translatable strings and use QString::arg instead.

Task-number: QTBUG-115962
Change-Id: I64e920e7b6496924f2c1ee6d07bea6e73964177a
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
(cherry picked from commit cda38d3d08e689deab506a11ed8c54a04d3f40d4)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
(cherry picked from commit de619b5eec88a677e51d206fb32b5eb47a018108)

12 files changed, 20 insertions, 285 deletions

diff --git a/examples/linguist/doc/src/trollprint.qdoc b/examples/linguist/doc/src/trollprint.qdoc
index fc664594a..868bf69b1 100644
--- a/examples/linguist/doc/src/trollprint.qdoc
+++ b/examples/linguist/doc/src/trollprint.qdoc
@@ -195,20 +195,6 @@
     Linguist}. The following items are of special interest:
 
     \list
-    \li \c MainWindow
-         \list
-         \li Troll Print 1.0 - marked "(obs.)", obsolete
-         \li About Troll Print 1.0 - marked "(obs.)", obsolete
-         \li Troll Print 1.0. Copyright 1999 Software, Inc. -
-       marked obsolete
-         \li Troll Print 1.1 - automatically translated as
-       "Troll Imprimir 1.1"
-         \li About Troll Print 1.1 - automatically translated as
-       "Troll Imprimir 1.1"
-         \li Troll Print 1.1. Copyright 1999-2000 Software,
-       Inc. - automatically translated as "Troll Imprimir 1.1.
-       Copyright 1999-2000 Software, Inc."
-         \endlist
     \li \c PrintPanel
          \list
          \li 2-sided - marked "(obs.)", obsolete
@@ -217,8 +203,9 @@
          \endlist
     \endlist
 
-    Notice that \c lupdate works hard behind the scenes to make revisions
-    easier, and it's pretty smart with numbers.
+    In order to make revisions easier, instead of tr("Troll Print 1.0")
+    you can use tr("Troll Print %1").arg("1.0") or such to avoid having to update
+    the string on every release.
 
     Go over the translations in \c MainWindow and mark these as "done".
     Translate "\<b\>TROLL PRINT\</b\>" as "\<b\>TROLL IMPRIMIR\</b\>".
diff --git a/examples/linguist/trollprint/mainwindow.cpp b/examples/linguist/trollprint/mainwindow.cpp
index eb8fef9a4..b099af15b 100644
--- a/examples/linguist/trollprint/mainwindow.cpp
+++ b/examples/linguist/trollprint/mainwindow.cpp
@@ -15,15 +15,15 @@ MainWindow::MainWindow()
     createMenus();
 
 //! [0]
-    setWindowTitle(tr("Troll Print 1.0"));
+    setWindowTitle(tr("Troll Print %1").arg("1.0"));
 //! [0]
 }
 
 void MainWindow::about()
 {
-    QMessageBox::information(this, tr("About Troll Print 1.0"),
-                      tr("Troll Print 1.0.\n\n"
-                      "Copyright 1999 Software, Inc."));
+    QMessageBox::information(this, tr("About Troll Print %1").arg("1.0"),
+                      tr("Troll Print %1.\n\n"
+                      "Copyright 1999 Software, Inc.").arg("1.0"));
 }
 
 //! [1]
diff --git a/examples/linguist/trollprint/trollprint_pt.ts b/examples/linguist/trollprint/trollprint_pt.ts
index e5871bd08..82165ffd3 100644
--- a/examples/linguist/trollprint/trollprint_pt.ts
+++ b/examples/linguist/trollprint/trollprint_pt.ts
@@ -2,8 +2,8 @@
 <context>
     <name>MainWindow</name>
     <message>
-        <source>Troll Print 1.0</source>
-        <translation>Troll Imprimir 1.0</translation>
+        <source>Troll Print %1</source>
+        <translation>Troll Imprimir %1</translation>
     </message>
     <message>
         <source>E&amp;xit</source>
@@ -26,14 +26,14 @@
         <translation>A&amp;juda</translation>
     </message>
     <message>
-        <source>About Troll Print 1.0</source>
-        <translation>Sobre Troll Imprimir 1.0</translation>
+        <source>About Troll Print %1</source>
+        <translation>Sobre Troll Imprimir %1</translation>
     </message>
     <message>
-        <source>Troll Print 1.0.
+        <source>Troll Print %1.
 
 Copyright 1999 Software, Inc.</source>
-        <translation>Troll Imprimir 1.0
+        <translation>Troll Imprimir %1
 
 Copyright 1999 Software, Inc.</translation>
     </message>
diff --git a/src/linguist/linguist/doc/src/linguist-manual.qdoc b/src/linguist/linguist/doc/src/linguist-manual.qdoc
index cd026b40b..fe557048e 100644
--- a/src/linguist/linguist/doc/src/linguist-manual.qdoc
+++ b/src/linguist/linguist/doc/src/linguist-manual.qdoc
@@ -279,7 +279,7 @@
         \li \c {-no-ui-lines}
         \li Do not record line numbers in references to UI files.
     \row
-        \li \c {-disable-heuristic {sametext|similartext|number}}
+        \li \c {-disable-heuristic {sametext|similartext}}
         \li Disable the named merge heuristic. Can be specified multiple times.
     \row
         \li \c {-project <filename>}
diff --git a/src/linguist/lupdate/lupdate.1 b/src/linguist/lupdate/lupdate.1
index 861e5bfee..a866965ca 100644
--- a/src/linguist/lupdate/lupdate.1
+++ b/src/linguist/lupdate/lupdate.1
@@ -28,7 +28,7 @@ used with version control systems if required.
 .PP
 .SH OPTIONS
 .TP
-.I "-disable-heuristic {sametext|similartext|number}"
+.I "-disable-heuristic {sametext|similartext}"
 Disable the named merge heuristic. Can be specified multiple times.
 .TP
 .I "-extensions <ext>[,<ext>...]"
diff --git a/src/linguist/lupdate/lupdate.h b/src/linguist/lupdate/lupdate.h
index 6b7b25e02..5cc0f8c2b 100644
--- a/src/linguist/lupdate/lupdate.h
+++ b/src/linguist/lupdate/lupdate.h
@@ -27,7 +27,6 @@ enum UpdateOption {
     NoSort = 8,
     HeuristicSameText = 16,
     HeuristicSimilarText = 32,
-    HeuristicNumber = 64,
     AbsoluteLocations = 256,
     RelativeLocations = 512,
     NoLocations = 1024,
diff --git a/src/linguist/lupdate/main.cpp b/src/linguist/lupdate/main.cpp
index e3d83b7d2..0ebe43214 100644
--- a/src/linguist/lupdate/main.cpp
+++ b/src/linguist/lupdate/main.cpp
@@ -254,7 +254,7 @@ static void printUsage()
         "           Default is absolute for new files.\n"
         "    -no-ui-lines\n"
         "           Do not record line numbers in references to UI files.\n"
-        "    -disable-heuristic {sametext|similartext|number}\n"
+        "    -disable-heuristic {sametext|similartext}\n"
         "           Disable the named merge heuristic. Can be specified multiple times.\n"
         "    -project <filename>\n"
         "           Name of a file containing the project's description in JSON format.\n"
@@ -728,7 +728,7 @@ int main(int argc, char **argv)
 
     UpdateOptions options =
         Verbose | // verbose is on by default starting with Qt 4.2
-        HeuristicSameText | HeuristicSimilarText | HeuristicNumber;
+        HeuristicSameText | HeuristicSimilarText;
     int numFiles = 0;
     bool metTsFlag = false;
     bool metXTsFlag = false;
@@ -799,8 +799,6 @@ int main(int argc, char **argv)
                 options &= ~HeuristicSameText;
             } else if (arg == QLatin1String("similartext")) {
                 options &= ~HeuristicSimilarText;
-            } else if (arg == QLatin1String("number")) {
-                options &= ~HeuristicNumber;
             } else {
                 printErr(u"Invalid heuristic name passed to -disable-heuristic.\n"_s);
                 return 1;
diff --git a/src/linguist/lupdate/merge.cpp b/src/linguist/lupdate/merge.cpp
index 273ace6ad..f204fc73d 100644
--- a/src/linguist/lupdate/merge.cpp
+++ b/src/linguist/lupdate/merge.cpp
@@ -14,212 +14,6 @@
 
 QT_BEGIN_NAMESPACE
 
-static bool isDigitFriendly(QChar c)
-{
-    return c.isPunct() || c.isSpace();
-}
-
-static int numberLength(const QString &s, int i)
-{
-    if (i >= s.size() || !s.at(i).isDigit())
-        return 0;
-
-    int pos = i;
-    do {
-        ++i;
-    } while (i < s.size()
-             && (s.at(i).isDigit()
-                 || (isDigitFriendly(s[i])
-                     && i + 1 < s.size()
-                     && (s[i + 1].isDigit()
-                         || (isDigitFriendly(s[i + 1])
-                             && i + 2 < s.size()
-                             && s[i + 2].isDigit())))));
-    return i - pos;
-}
-
-
-/*
-  Returns a version of 'key' where all numbers have been replaced by zeroes.  If
-  there were none, returns "".
-*/
-static QString zeroKey(const QString &key)
-{
-    QString zeroed;
-    bool metSomething = false;
-
-    for (int i = 0; i < key.size(); ++i) {
-        int len = numberLength(key, i);
-        if (len > 0) {
-            i += len;
-            zeroed.append(QLatin1Char('0'));
-            metSomething = true;
-        } else {
-            zeroed.append(key.at(i));
-        }
-    }
-    return metSomething ? zeroed : QString();
-}
-
-static QString translationAttempt(const QString &oldTranslation,
-    const QString &oldSource, const QString &newSource)
-{
-    int p = zeroKey(oldSource).count(QLatin1Char('0'));
-    QString attempt;
-    QStringList oldNumbers;
-    QStringList newNumbers;
-    QList<bool> met(p);
-    QList<int> matchedYet(p);
-    int i, j;
-    int k = 0, ell, best;
-    int m, n;
-    int pass;
-
-    /*
-      This algorithm is hard to follow, so we'll consider an example
-      all along: oldTranslation is "XeT 3.0", oldSource is "TeX 3.0"
-      and newSource is "XeT 3.1".
-
-      First, we set up two tables: oldNumbers and newNumbers. In our
-      example, oldNumber[0] is "3.0" and newNumber[0] is "3.1".
-    */
-    for (i = 0, j = 0; i < oldSource.size(); i++, j++) {
-        m = numberLength(oldSource, i);
-        n = numberLength(newSource, j);
-        if (m > 0) {
-            oldNumbers.append(oldSource.mid(i, m + 1));
-            newNumbers.append(newSource.mid(j, n + 1));
-            i += m;
-            j += n;
-            met[k] = false;
-            matchedYet[k] = 0;
-            k++;
-        }
-    }
-
-    /*
-      We now go over the old translation, "XeT 3.0", one letter at a
-      time, looking for numbers found in oldNumbers. Whenever such a
-      number is met, it is replaced with its newNumber equivalent. In
-      our example, the "3.0" of "XeT 3.0" becomes "3.1".
-    */
-    for (i = 0; i < oldTranslation.size(); i++) {
-        attempt += oldTranslation[i];
-        for (k = 0; k < p; k++) {
-            if (matchedYet[k] < oldNumbers[k].size() &&
-                oldTranslation[i] == oldNumbers[k][matchedYet[k]]) {
-                matchedYet[k]++;
-            } else {
-                matchedYet[k] = 0;
-            }
-        }
-
-        /*
-          Let's find out if the last character ended a match. We make
-          two passes over the data. In the first pass, we try to
-          match only numbers that weren't matched yet; if that fails,
-          the second pass does the trick. This is useful in some
-          suspicious cases, flagged below.
-        */
-        for (pass = 0; pass < 2; pass++) {
-            best = p; // an impossible value
-            for (k = 0; k < p; k++) {
-                if ((!met[k] || pass > 0) &&
-                     matchedYet[k] == oldNumbers[k].size() &&
-                     numberLength(oldTranslation, i + 1 - matchedYet[k]) == matchedYet[k]) {
-                    // the longer the better
-                    if (best == p || matchedYet[k] > matchedYet[best])
-                        best = k;
-                }
-            }
-            if (best != p) {
-                attempt.truncate(attempt.size() - matchedYet[best]);
-                attempt += newNumbers[best];
-                met[best] = true;
-                for (k = 0; k < p; k++)
-                    matchedYet[k] = 0;
-                break;
-            }
-        }
-    }
-
-    /*
-      We flag two kinds of suspicious cases. They are identified as
-      such with comments such as "{2000?}" at the end.
-
-      Example of the first kind: old source text "TeX 3.0" translated
-      as "XeT 2.0" is flagged "TeX 2.0 {3.0?}", no matter what the
-      new text is.
-    */
-    for (k = 0; k < p; k++) {
-        if (!met[k])
-            attempt += QLatin1String(" {") + newNumbers[k] + QLatin1String("?}");
-    }
-
-    /*
-      Example of the second kind: "1 of 1" translated as "1 af 1",
-      with new source text "1 of 2", generates "1 af 2 {1 or 2?}"
-      because it's not clear which of "1 af 2" and "2 af 1" is right.
-    */
-    for (k = 0; k < p; k++) {
-        for (ell = 0; ell < p; ell++) {
-            if (k != ell && oldNumbers[k] == oldNumbers[ell] &&
-                    newNumbers[k] < newNumbers[ell])
-                attempt += QLatin1String(" {") + newNumbers[k] + QLatin1String(" or ") +
-                           newNumbers[ell] + QLatin1String("?}");
-        }
-    }
-    return attempt;
-}
-
-
-/*
-  Augments a Translator with translations easily derived from
-  similar existing (probably obsolete) translations.
-
-  For example, if "TeX 3.0" is translated as "XeT 3.0" and "TeX 3.1"
-  has no translation, "XeT 3.1" is added to the translator and is
-  marked Unfinished.
-
-  Returns the number of additional messages that this heuristic translated.
-*/
-int applyNumberHeuristic(Translator &tor)
-{
-    QMap<QString, QPair<QString, QString> > translated;
-    QList<bool> untranslated(tor.messageCount());
-    int inserted = 0;
-
-    for (int i = 0; i < tor.messageCount(); ++i) {
-        const TranslatorMessage &msg = tor.message(i);
-        bool hasTranslation = msg.isTranslated();
-        if (msg.type() == TranslatorMessage::Unfinished) {
-            if (!hasTranslation)
-                untranslated[i] = true;
-        } else if (hasTranslation && msg.translations().size() == 1) {
-            const QString &key = zeroKey(msg.sourceText());
-            if (!key.isEmpty())
-                translated.insert(key, qMakePair(msg.sourceText(), msg.translation()));
-        }
-    }
-
-    for (int i = 0; i < tor.messageCount(); ++i) {
-        if (untranslated[i]) {
-            TranslatorMessage &msg = tor.message(i);
-            const QString &key = zeroKey(msg.sourceText());
-            if (!key.isEmpty()) {
-                const auto t = translated.constFind(key);
-                if (t != translated.constEnd() && t->first != msg.sourceText()) {
-                    msg.setTranslation(translationAttempt(t->second, t->first,
-                                                          msg.sourceText()));
-                    inserted++;
-                }
-            }
-        }
-    }
-    return inserted;
-}
-
-
 /*
   Augments a Translator with trivially derived translations.
 
@@ -493,13 +287,6 @@ Translator merge(
     */
     int sameTextHeuristicCount = (options & HeuristicSameText) ? applySameTextHeuristic(outTor) : 0;
 
-    /*
-      The number heuristic handles cases where a message has an
-      obsolete counterpart with mostly numbers differing in the
-      source text.
-    */
-    int sameNumberHeuristicCount = (options & HeuristicNumber) ? applyNumberHeuristic(outTor) : 0;
-
     if (options & Verbose) {
         int totalFound = neww + known;
         err += QStringLiteral("    Found %1 source text(s) (%2 new and %3 already existing)\n")
@@ -513,9 +300,6 @@ Translator merge(
             }
         }
 
-        if (sameNumberHeuristicCount)
-            err += QStringLiteral("    Number heuristic provided %1 translation(s)\n")
-                .arg(sameNumberHeuristicCount);
         if (sameTextHeuristicCount)
             err += QStringLiteral("    Same-text heuristic provided %1 translation(s)\n")
                 .arg(sameTextHeuristicCount);
diff --git a/tests/auto/linguist/lupdate/testdata/good/heuristics/expectedoutput.txt b/tests/auto/linguist/lupdate/testdata/good/heuristics/expectedoutput.txt
index e363cc8cb..f8c994f34 100644
--- a/tests/auto/linguist/lupdate/testdata/good/heuristics/expectedoutput.txt
+++ b/tests/auto/linguist/lupdate/testdata/good/heuristics/expectedoutput.txt
@@ -1,5 +1,4 @@
 Updating 'project\.ts'\.\.\.
-    Found 4 source text\(s\) \(4 new and 0 already existing\)
-    Removed 6 obsolete entries
-    Number heuristic provided 2 translation\(s\)
+    Found 2 source text\(s\) \(2 new and 0 already existing\)
+    Removed 4 obsolete entries
     Same-text heuristic provided 1 translation\(s\)
diff --git a/tests/auto/linguist/lupdate/testdata/good/heuristics/main.cpp b/tests/auto/linguist/lupdate/testdata/good/heuristics/main.cpp
index 0915b556c..ef37410ec 100644
--- a/tests/auto/linguist/lupdate/testdata/good/heuristics/main.cpp
+++ b/tests/auto/linguist/lupdate/testdata/good/heuristics/main.cpp
@@ -36,17 +36,11 @@ class A: public QObject {
     Q_OBJECT
     void foo()
     {
-        // number Heuristics
-        tr("version 2.0 now");
-
         // same text match
         tr("this is the matched same text");
 
         // failed same text
         tr("this is the non-matched same text");
-
-        // number heuristics QTBUG-111775
-        tr("1xb");
     }
 };
 
diff --git a/tests/auto/linguist/lupdate/testdata/good/heuristics/project.ts.before b/tests/auto/linguist/lupdate/testdata/good/heuristics/project.ts.before
index ca731d961..2d1b201a9 100644
--- a/tests/auto/linguist/lupdate/testdata/good/heuristics/project.ts.before
+++ b/tests/auto/linguist/lupdate/testdata/good/heuristics/project.ts.before
@@ -2,14 +2,6 @@
 <!DOCTYPE TS>
 <TS version="2.0" language="de_DE">
 <context>
-    <name>A</name>
-    <message>
-        <location filename="main.cpp" line="53"/>
-        <source>version 1.0 now</source>
-        <translation>teraz wersja 1.0</translation>
-    </message>
-</context>
-<context>
     <name>B</name>
     <message>
         <location filename="main.cpp" line="56"/>
@@ -35,12 +27,4 @@
         <translation>völlig andere variante des reinfalls mit same-text</translation>
     </message>
 </context>
-<context>
-    <name>D</name>
-    <message>
-        <location filename="main.cpp" line="62"/>
-        <source>1ab</source>
-        <translation>1ab</translation>
-    </message>
-</context>
 </TS>
diff --git a/tests/auto/linguist/lupdate/testdata/good/heuristics/project.ts.result b/tests/auto/linguist/lupdate/testdata/good/heuristics/project.ts.result
index 546d1f1f4..3c150a30c 100644
--- a/tests/auto/linguist/lupdate/testdata/good/heuristics/project.ts.result
+++ b/tests/auto/linguist/lupdate/testdata/good/heuristics/project.ts.result
@@ -5,23 +5,13 @@
     <name>A</name>
     <message>
         <location filename="main.cpp" line="40"/>
-        <source>version 2.0 now</source>
-        <translation type="unfinished">teraz wersja 1.0 {2.0 ?}</translation>
-    </message>
-    <message>
-        <location filename="main.cpp" line="43"/>
         <source>this is the matched same text</source>
         <translation type="unfinished">der same-text-treffer</translation>
     </message>
     <message>
-        <location filename="main.cpp" line="46"/>
+        <location filename="main.cpp" line="43"/>
         <source>this is the non-matched same text</source>
         <translation type="unfinished"></translation>
     </message>
-    <message>
-        <location filename="main.cpp" line="49"/>
-        <source>1xb</source>
-        <translation type="unfinished">1ab {1x?}</translation>
-    </message>
 </context>
 </TS>