TextOutputDev: memory cleanups

Return some GooStrings by value, and convert a std::vector** to a vector
of vectors

6 files changed, 74 insertions, 134 deletions

diff --git a/glib/poppler-page.cc b/glib/poppler-page.cc
index 38e5e680..29871887 100644
--- a/glib/poppler-page.cc
+++ b/glib/poppler-page.cc
@@ -726,7 +726,6 @@ cairo_region_t *poppler_page_get_selected_region(PopplerPage *page, gdouble scal
  **/
 char *poppler_page_get_selected_text(PopplerPage *page, PopplerSelectionStyle style, PopplerRectangle *selection)
 {
-    GooString *sel_text;
     char *result;
     TextPage *text;
     SelectionStyle selection_style = selectionStyleGlyph;
@@ -753,9 +752,8 @@ char *poppler_page_get_selected_text(PopplerPage *page, PopplerSelectionStyle st
     }
 
     text = poppler_page_get_text_page(page);
-    sel_text = text->getSelectionText(&pdf_selection, selection_style);
-    result = g_strdup(sel_text->c_str());
-    delete sel_text;
+    GooString sel_text = text->getSelectionText(&pdf_selection, selection_style);
+    result = g_strdup(sel_text.c_str());
 
     return result;
 }
@@ -2327,12 +2325,10 @@ gboolean poppler_page_get_text_layout_for_area(PopplerPage *page, PopplerRectang
     TextPage *text;
     PopplerRectangle *rect;
     PDFRectangle selection;
-    int i, k;
     guint offset = 0;
     guint n_rects = 0;
     gdouble x1, y1, x2, y2;
     gdouble x3, y3, x4, y4;
-    int n_lines;
 
     g_return_val_if_fail(POPPLER_IS_PAGE(page), FALSE);
     g_return_val_if_fail(area != nullptr, FALSE);
@@ -2345,19 +2341,18 @@ gboolean poppler_page_get_text_layout_for_area(PopplerPage *page, PopplerRectang
     selection.y2 = area->y2;
 
     text = poppler_page_get_text_page(page);
-    std::vector<TextWordSelection *> **word_list = text->getSelectionWords(&selection, selectionStyleGlyph, &n_lines);
-    if (!word_list) {
+    std::vector<std::vector<std::unique_ptr<TextWordSelection>>> word_list = text->getSelectionWords(&selection, selectionStyleGlyph);
+    if (word_list.empty()) {
         return FALSE;
     }
 
-    n_rects += n_lines - 1;
-    for (i = 0; i < n_lines; i++) {
-        std::vector<TextWordSelection *> *line_words = word_list[i];
-        n_rects += line_words->size() - 1;
-        for (std::size_t j = 0; j < line_words->size(); j++) {
-            const TextWordSelection *word_sel = (*line_words)[j];
+    n_rects += word_list.size() - 1;
+    for (const std::vector<std::unique_ptr<TextWordSelection>> &line_words : word_list) {
+        n_rects += line_words.size() - 1;
+        for (std::size_t j = 0; j < line_words.size(); j++) {
+            const TextWordSelection *word_sel = line_words[j].get();
             n_rects += word_sel->getEnd() - word_sel->getBegin();
-            if (!word_sel->getWord()->hasSpaceAfter() && j < line_words->size() - 1) {
+            if (!word_sel->getWord()->hasSpaceAfter() && j < line_words.size() - 1) {
                 n_rects--;
             }
         }
@@ -2366,14 +2361,14 @@ gboolean poppler_page_get_text_layout_for_area(PopplerPage *page, PopplerRectang
     *rectangles = g_new(PopplerRectangle, n_rects);
     *n_rectangles = n_rects;
 
-    for (i = 0; i < n_lines; i++) {
-        std::vector<TextWordSelection *> *line_words = word_list[i];
-        for (std::size_t j = 0; j < line_words->size(); j++) {
-            TextWordSelection *word_sel = (*line_words)[j];
+    for (size_t i = 0; i < word_list.size(); i++) {
+        std::vector<std::unique_ptr<TextWordSelection>> &line_words = word_list[i];
+        for (std::size_t j = 0; j < line_words.size(); j++) {
+            TextWordSelection *word_sel = line_words[j].get();
             const TextWord *word = word_sel->getWord();
             int end = word_sel->getEnd();
 
-            for (k = word_sel->getBegin(); k < end; k++) {
+            for (int k = word_sel->getBegin(); k < end; k++) {
                 rect = *rectangles + offset;
                 word->getCharBBox(k, &(rect->x1), &(rect->y1), &(rect->x2), &(rect->y2));
                 offset++;
@@ -2382,8 +2377,8 @@ gboolean poppler_page_get_text_layout_for_area(PopplerPage *page, PopplerRectang
             rect = *rectangles + offset;
             word->getBBox(&x1, &y1, &x2, &y2);
 
-            if (word->hasSpaceAfter() && j < line_words->size() - 1) {
-                TextWordSelection *next_word_sel = (*line_words)[j + 1];
+            if (word->hasSpaceAfter() && j < line_words.size() - 1) {
+                TextWordSelection *next_word_sel = line_words[j + 1].get();
 
                 next_word_sel->getWord()->getBBox(&x3, &y3, &x4, &y4);
                 // space is from one word to other and with the same height as
@@ -2394,11 +2389,9 @@ gboolean poppler_page_get_text_layout_for_area(PopplerPage *page, PopplerRectang
                 rect->y2 = y2;
                 offset++;
             }
-
-            delete word_sel;
         }
 
-        if (i < n_lines - 1 && offset > 0) {
+        if (i < word_list.size() - 1 && offset > 0) {
             // end of line
             rect->x1 = x2;
             rect->y1 = y2;
@@ -2406,12 +2399,7 @@ gboolean poppler_page_get_text_layout_for_area(PopplerPage *page, PopplerRectang
             rect->y2 = y2;
             offset++;
         }
-
-        delete line_words;
     }
-
-    gfree(static_cast<void *>(word_list));
-
     return TRUE;
 }
 
@@ -2503,11 +2491,9 @@ GList *poppler_page_get_text_attributes_for_area(PopplerPage *page, PopplerRecta
 {
     TextPage *text;
     PDFRectangle selection;
-    int n_lines;
     PopplerTextAttributes *attrs = nullptr;
     const TextWord *word, *prev_word = nullptr;
     gint word_i, prev_word_i;
-    gint i;
     gint offset = 0;
     GList *attributes = nullptr;
 
@@ -2520,15 +2506,15 @@ GList *poppler_page_get_text_attributes_for_area(PopplerPage *page, PopplerRecta
     selection.y2 = area->y2;
 
     text = poppler_page_get_text_page(page);
-    std::vector<TextWordSelection *> **word_list = text->getSelectionWords(&selection, selectionStyleGlyph, &n_lines);
-    if (!word_list) {
+    std::vector<std::vector<std::unique_ptr<TextWordSelection>>> word_list = text->getSelectionWords(&selection, selectionStyleGlyph);
+    if (word_list.empty()) {
         return nullptr;
     }
 
-    for (i = 0; i < n_lines; i++) {
-        std::vector<TextWordSelection *> *line_words = word_list[i];
-        for (std::size_t j = 0; j < line_words->size(); j++) {
-            TextWordSelection *word_sel = (*line_words)[j];
+    for (size_t i = 0; i < word_list.size(); i++) {
+        std::vector<std::unique_ptr<TextWordSelection>> &line_words = word_list[i];
+        for (std::size_t j = 0; j < line_words.size(); j++) {
+            TextWordSelection *word_sel = line_words[j].get();
             int end = word_sel->getEnd();
 
             word = word_sel->getWord();
@@ -2545,23 +2531,16 @@ GList *poppler_page_get_text_attributes_for_area(PopplerPage *page, PopplerRecta
                 prev_word_i = word_i;
             }
 
-            if (word->hasSpaceAfter() && j < line_words->size() - 1) {
+            if (word->hasSpaceAfter() && j < line_words.size() - 1) {
                 attrs->end_index = offset;
                 offset++;
             }
-
-            delete word_sel;
         }
 
-        if (i < n_lines - 1) {
+        if (i < word_list.size() - 1) {
             attrs->end_index = offset;
             offset++;
         }
-
-        delete line_words;
     }
-
-    gfree(static_cast<void *>(word_list));
-
     return g_list_reverse(attributes);
 }
diff --git a/poppler/TextOutputDev.cc b/poppler/TextOutputDev.cc
index 7a817ee8..1a53b5e2 100644
--- a/poppler/TextOutputDev.cc
+++ b/poppler/TextOutputDev.cc
@@ -4284,9 +4284,8 @@ bool TextPage::findText(const Unicode *s, int len, bool startAtTop, bool stopAtB
     return false;
 }
 
-GooString *TextPage::getText(double xMin, double yMin, double xMax, double yMax, EndOfLineKind textEOL) const
+GooString TextPage::getText(double xMin, double yMin, double xMax, double yMax, EndOfLineKind textEOL) const
 {
-    GooString *s;
     const UnicodeMap *uMap;
     TextBlock *blk;
     TextLine *line;
@@ -4300,7 +4299,7 @@ GooString *TextPage::getText(double xMin, double yMin, double xMax, double yMax,
     int col, idx0, idx1, i, j;
     bool multiLine, oneRot;
 
-    s = new GooString();
+    GooString s;
 
     // get the output encoding
     if (!(uMap = globalParams->getTextEncoding())) {
@@ -4318,7 +4317,7 @@ GooString *TextPage::getText(double xMin, double yMin, double xMax, double yMax,
                 word->getCharBBox(j, &gXMin, &gYMin, &gXMax, &gYMax);
                 if (xMin <= gXMin && gXMax <= xMax && yMin <= gYMin && gYMax <= yMax) {
                     mbc_len = uMap->mapUnicode(*(word->getChar(j)), mbc, sizeof(mbc));
-                    s->append(mbc, mbc_len);
+                    s.append(mbc, mbc_len);
                 }
             }
         }
@@ -4489,22 +4488,22 @@ GooString *TextPage::getText(double xMin, double yMin, double xMax, double yMax,
 
             // insert a return
             if (frag->col < col || (i > 0 && fabs(frag->base - frags[i - 1].base) > maxIntraLineDelta * frags[i - 1].line->words->fontSize)) {
-                s->append(eol, eolLen);
+                s.append(eol, eolLen);
                 col = 0;
                 multiLine = true;
             }
 
             // column alignment
             for (; col < frag->col; ++col) {
-                s->append(space, spaceLen);
+                s.append(space, spaceLen);
             }
 
             // get the fragment text
-            col += dumpFragment(frag->line->text + frag->start, frag->len, uMap, s);
+            col += dumpFragment(frag->line->text + frag->start, frag->len, uMap, &s);
         }
 
         if (multiLine) {
-            s->append(eol, eolLen);
+            s.append(eol, eolLen);
         }
     }
 
@@ -4543,66 +4542,44 @@ public:
     void visitWord(TextWord *word, int begin, int end, const PDFRectangle *selection) override;
     void endPage();
 
-    GooString *getText();
-    std::vector<TextWordSelection *> **takeWordList(int *nLines);
+    GooString getText();
+    std::vector<std::vector<std::unique_ptr<TextWordSelection>>> takeWordList();
 
 private:
     void startLine();
     void finishLine();
 
-    std::vector<TextWordSelection *> **lines;
-    int nLines, linesSize;
-    std::vector<TextWordSelection *> *words;
+    std::vector<std::vector<std::unique_ptr<TextWordSelection>>> lines;
+    std::vector<std::unique_ptr<TextWordSelection>> words;
     int tableId;
     TextBlock *currentBlock;
 };
 
 TextSelectionDumper::TextSelectionDumper(TextPage *p) : TextSelectionVisitor(p)
 {
-    linesSize = 256;
-    lines = (std::vector<TextWordSelection *> **)gmallocn(linesSize, sizeof(std::vector<TextWordSelection *> *));
-    nLines = 0;
-
     tableId = -1;
     currentBlock = nullptr;
-    words = nullptr;
 }
 
-TextSelectionDumper::~TextSelectionDumper()
-{
-    for (int i = 0; i < nLines; i++) {
-        for (auto entry : *(lines[i])) {
-            delete entry;
-        }
-        delete lines[i];
-    }
-    gfree(static_cast<void *>(lines));
-}
+TextSelectionDumper::~TextSelectionDumper() = default;
 
 void TextSelectionDumper::startLine()
 {
     finishLine();
-    words = new std::vector<TextWordSelection *>();
+    words.clear();
 }
 
 void TextSelectionDumper::finishLine()
 {
-    if (nLines == linesSize) {
-        linesSize *= 2;
-        lines = (std::vector<TextWordSelection *> **)grealloc(static_cast<void *>(lines), linesSize * sizeof(std::vector<TextWordSelection *> *));
-    }
-
-    if (words && !words->empty()) {
+    if (!words.empty()) {
         // Reverse word order for RTL text. Fixes #53 for glib backend (Evince)
         if (!page->primaryLR) {
-            std::ranges::reverse(*words);
+            std::ranges::reverse(words);
         }
 
-        lines[nLines++] = words;
-    } else {
-        delete words;
+        lines.push_back(std::move(words));
     }
-    words = nullptr;
+    words.clear();
 }
 
 void TextSelectionDumper::visitLine(TextLine *line, TextWord *begin, TextWord *end, int edge_begin, int edge_end, const PDFRectangle *selection)
@@ -4639,7 +4616,7 @@ void TextSelectionDumper::visitLine(TextLine *line, TextWord *begin, TextWord *e
 
 void TextSelectionDumper::visitWord(TextWord *word, int begin, int end, const PDFRectangle *selection)
 {
-    words->push_back(new TextWordSelection(word, begin, end));
+    words.push_back(std::make_unique<TextWordSelection>(word, begin, end));
 }
 
 void TextSelectionDumper::endPage()
@@ -4647,16 +4624,13 @@ void TextSelectionDumper::endPage()
     finishLine();
 }
 
-GooString *TextSelectionDumper::getText()
+GooString TextSelectionDumper::getText()
 {
-    GooString *text;
-    int i;
+    GooString text;
     const UnicodeMap *uMap;
     char space[8], eol[16];
     int spaceLen, eolLen;
 
-    text = new GooString();
-
     if (!(uMap = globalParams->getTextEncoding())) {
         return text;
     }
@@ -4665,39 +4639,31 @@ GooString *TextSelectionDumper::getText()
     eolLen = uMap->mapUnicode(0x0a, eol, sizeof(eol));
 
     std::vector<Unicode> uText;
-    for (i = 0; i < nLines; i++) {
-        std::vector<TextWordSelection *> *lineWords = lines[i];
-        for (std::size_t j = 0; j < lineWords->size(); j++) {
-            TextWordSelection *sel = (*lineWords)[j];
+    for (size_t i = 0; i < lines.size(); i++) {
+        const auto &line = lines[i];
+        for (size_t j = 0; j < line.size(); j++) {
+            const auto &sel = line[j];
 
             uText.resize(sel->end - sel->begin);
             std::transform(sel->word->chars.begin() + sel->begin, sel->word->chars.begin() + sel->end, uText.begin(), [](auto &c) { return c.text; });
-            page->dumpFragment(uText.data(), uText.size(), uMap, text);
+            page->dumpFragment(uText.data(), uText.size(), uMap, &text);
 
-            if (j < lineWords->size() - 1 && sel->word->spaceAfter) {
-                text->append(space, spaceLen);
+            if (j < line.size() - 1 && sel->word->spaceAfter) {
+                text.append(space, spaceLen);
             }
         }
-        if (i < nLines - 1) {
-            text->append(eol, eolLen);
+        if (i < lines.size() - 1) {
+            text.append(eol, eolLen);
         }
     }
 
     return text;
 }
 
-std::vector<TextWordSelection *> **TextSelectionDumper::takeWordList(int *nLinesOut)
+std::vector<std::vector<std::unique_ptr<TextWordSelection>>> TextSelectionDumper::takeWordList()
 {
-    std::vector<TextWordSelection *> **returnValue = lines;
-
-    *nLinesOut = nLines;
-    if (nLines == 0) {
-        return nullptr;
-    }
-
-    nLines = 0;
-    lines = nullptr;
-
+    std::vector<std::vector<std::unique_ptr<TextWordSelection>>> returnValue;
+    std::swap(lines, returnValue);
     return returnValue;
 }
 
@@ -5338,7 +5304,7 @@ std::vector<PDFRectangle *> *TextPage::getSelectionRegion(const PDFRectangle *se
     return sizer.takeRegion();
 }
 
-GooString *TextPage::getSelectionText(const PDFRectangle *selection, SelectionStyle style)
+GooString TextPage::getSelectionText(const PDFRectangle *selection, SelectionStyle style)
 {
     TextSelectionDumper dumper(this);
 
@@ -5348,14 +5314,14 @@ GooString *TextPage::getSelectionText(const PDFRectangle *selection, SelectionSt
     return dumper.getText();
 }
 
-std::vector<TextWordSelection *> **TextPage::getSelectionWords(const PDFRectangle *selection, SelectionStyle style, int *nLines)
+std::vector<std::vector<std::unique_ptr<TextWordSelection>>> TextPage::getSelectionWords(const PDFRectangle *selection, SelectionStyle style)
 {
     TextSelectionDumper dumper(this);
 
     visitSelection(&dumper, selection, style);
     dumper.endPage();
 
-    return dumper.takeWordList(nLines);
+    return dumper.takeWordList();
 }
 
 bool TextPage::findCharRange(int pos, int length, double *xMin, double *yMin, double *xMax, double *yMax) const
@@ -6060,7 +6026,7 @@ bool TextOutputDev::findText(const Unicode *s, int len, bool startAtTop, bool st
     return text->findText(s, len, startAtTop, stopAtBottom, startAtLast, stopAtLast, caseSensitive, backward, wholeWord, xMin, yMin, xMax, yMax);
 }
 
-GooString *TextOutputDev::getText(double xMin, double yMin, double xMax, double yMax) const
+GooString TextOutputDev::getText(double xMin, double yMin, double xMax, double yMax) const
 {
     return text->getText(xMin, yMin, xMax, yMax, textEOL);
 }
@@ -6075,7 +6041,7 @@ std::vector<PDFRectangle *> *TextOutputDev::getSelectionRegion(const PDFRectangl
     return text->getSelectionRegion(selection, style, scale);
 }
 
-GooString *TextOutputDev::getSelectionText(const PDFRectangle *selection, SelectionStyle style)
+GooString TextOutputDev::getSelectionText(const PDFRectangle *selection, SelectionStyle style)
 {
     return text->getSelectionText(selection, style);
 }
diff --git a/poppler/TextOutputDev.h b/poppler/TextOutputDev.h
index 20a71ae8..4ca5630e 100644
--- a/poppler/TextOutputDev.h
+++ b/poppler/TextOutputDev.h
@@ -637,7 +637,7 @@ public:
                   double *xMax, double *yMax, PDFRectangle *continueMatch, bool *ignoredHyphen);
 
     // Get the text which is inside the specified rectangle.
-    GooString *getText(double xMin, double yMin, double xMax, double yMax, EndOfLineKind textEOL) const;
+    GooString getText(double xMin, double yMin, double xMax, double yMax, EndOfLineKind textEOL) const;
 
     void visitSelection(TextSelectionVisitor *visitor, const PDFRectangle *selection, SelectionStyle style);
 
@@ -645,9 +645,9 @@ public:
 
     std::vector<PDFRectangle *> *getSelectionRegion(const PDFRectangle *selection, SelectionStyle style, double scale);
 
-    GooString *getSelectionText(const PDFRectangle *selection, SelectionStyle style);
+    GooString getSelectionText(const PDFRectangle *selection, SelectionStyle style);
 
-    std::vector<TextWordSelection *> **getSelectionWords(const PDFRectangle *selection, SelectionStyle style, int *nLines);
+    [[nodiscard]] std::vector<std::vector<std::unique_ptr<TextWordSelection>>> getSelectionWords(const PDFRectangle *selection, SelectionStyle style);
 
     // Find a string by character position and length.  If found, sets
     // the text bounding rectangle and returns true; otherwise returns
@@ -850,7 +850,7 @@ public:
     bool findText(const Unicode *s, int len, bool startAtTop, bool stopAtBottom, bool startAtLast, bool stopAtLast, bool caseSensitive, bool backward, bool wholeWord, double *xMin, double *yMin, double *xMax, double *yMax) const;
 
     // Get the text which is inside the specified rectangle.
-    GooString *getText(double xMin, double yMin, double xMax, double yMax) const;
+    GooString getText(double xMin, double yMin, double xMax, double yMax) const;
 
     // Find a string by character position and length.  If found, sets
     // the text bounding rectangle and returns true; otherwise returns
@@ -861,7 +861,7 @@ public:
 
     std::vector<PDFRectangle *> *getSelectionRegion(const PDFRectangle *selection, SelectionStyle style, double scale);
 
-    GooString *getSelectionText(const PDFRectangle *selection, SelectionStyle style);
+    GooString getSelectionText(const PDFRectangle *selection, SelectionStyle style);
 
     // If true, will combine characters when a base and combining
     // character are drawn on eachother.
diff --git a/qt5/src/poppler-page.cc b/qt5/src/poppler-page.cc
index 96ff2875..08eb7f27 100644
--- a/qt5/src/poppler-page.cc
+++ b/qt5/src/poppler-page.cc
@@ -678,7 +678,7 @@ QImage Page::thumbnail() const
 QString Page::text(const QRectF &r, TextLayout textLayout) const
 {
     TextOutputDev *output_dev;
-    GooString *s;
+    GooString s;
     QString result;
 
     const bool rawOrder = textLayout == RawOrderLayout;
@@ -695,10 +695,9 @@ QString Page::text(const QRectF &r, TextLayout textLayout) const
         s = output_dev->getText(r.left(), r.top(), r.right(), r.bottom());
     }
 
-    result = QString::fromStdString(s->toStr());
+    result = QString::fromStdString(s.toStr());
 
     delete output_dev;
-    delete s;
     return result;
 }
 
diff --git a/qt6/src/poppler-page.cc b/qt6/src/poppler-page.cc
index 8b99894b..8c88142b 100644
--- a/qt6/src/poppler-page.cc
+++ b/qt6/src/poppler-page.cc
@@ -678,7 +678,7 @@ QImage Page::thumbnail() const
 QString Page::text(const QRectF &r, TextLayout textLayout) const
 {
     TextOutputDev *output_dev;
-    GooString *s;
+    GooString s;
     QString result;
 
     const bool rawOrder = textLayout == RawOrderLayout;
@@ -695,10 +695,9 @@ QString Page::text(const QRectF &r, TextLayout textLayout) const
         s = output_dev->getText(r.left(), r.top(), r.right(), r.bottom());
     }
 
-    result = QString::fromStdString(s->toStr());
+    result = QString::fromStdString(s.toStr());
 
     delete output_dev;
-    delete s;
     return result;
 }
 
diff --git a/test/perf-test.cc b/test/perf-test.cc
index 6526a04b..5f4b71b9 100644
--- a/test/perf-test.cc
+++ b/test/perf-test.cc
@@ -608,7 +608,6 @@ static bool ShowPreview()
 static void RenderPdfAsText(const char *fileName)
 {
     PDFDoc *pdfDoc = nullptr;
-    GooString *txt = nullptr;
     int pageCount;
     double timeInMs;
 
@@ -650,15 +649,13 @@ static void RenderPdfAsText(const char *fileName)
         bool crop = true;
         bool doLinks = false;
         pdfDoc->displayPage(textOut, curPage, 72, 72, rotate, useMediaBox, crop, doLinks);
-        txt = textOut->getText(0.0, 0.0, 10000.0, 10000.0);
+        GooString txt = textOut->getText(0.0, 0.0, 10000.0, 10000.0);
         msTimer.stop();
         timeInMs = msTimer.getElapsed();
         if (gfTimings) {
             LogInfo("page %d: %.2f ms\n", curPage, timeInMs);
         }
-        printf("%s\n", txt->c_str());
-        delete txt;
-        txt = nullptr;
+        printf("%s\n", txt.c_str());
     }
 
 Exit: