[cpp] separate the font info in text_box to another struct.

* add new API, page::text_list(int opt_flag). The old one
taking no argument is kept for ABI compatibility.
The opt_flag is a bitmask-multiple of the new enum,
page::text_list_option_enum.

* text_box.m_data->text_box_font is an unique pointer to
the storage (if text_list() requests the font info), or
just a null pointer (if text_list() does not request the
font info).

* new option "--show-text-list-with-font" showing font
info, to tests/poppler-dump.cpp. "--show-text-list"
does not load the font info at all.

Co-authored-by: Adam Reichold <adam.reichold@t-online.de>
Co-authored-by: Albert Astals Cid <aacid@kde.org>

4 files changed, 121 insertions, 50 deletions

diff --git a/cpp/poppler-page.cpp b/cpp/poppler-page.cpp
index f274ca5b..01b0409d 100644
--- a/cpp/poppler-page.cpp
+++ b/cpp/poppler-page.cpp
@@ -299,7 +299,7 @@ static void appendToGooString(void *stream, const char *text, int len) {
 ustring page::text(const rectf &r, text_layout_enum layout_mode) const
 {
     std::unique_ptr<GooString> out(new GooString());
-    const bool use_raw_order = (layout_mode == raw_order_layout); 
+    const bool use_raw_order = (layout_mode == raw_order_layout);
     const bool use_physical_layout = (layout_mode == physical_layout);
     TextOutputDev td(&appendToGooString, out.get(), use_physical_layout, 0, use_raw_order, false);
     if (r.is_empty()) {
@@ -312,6 +312,11 @@ ustring page::text(const rectf &r, text_layout_enum layout_mode) const
 }
 
 /*
+ * text_box_font_info object for text_box
+ */
+text_box_font_info_data::~text_box_font_info_data() = default;
+
+/*
  * text_box object for page::text_list()
  */
 text_box_data::~text_box_data() = default;
@@ -352,30 +357,41 @@ bool text_box::has_space_after() const
     return m_data->has_space_after;
 }
 
+bool text_box::has_font_info() const
+{
+    return (m_data->text_box_font != nullptr);
+}
+
 text_box::writing_mode_enum text_box::get_wmode(int i) const
 {
-    return m_data->wmodes[i];
+    if (this->has_font_info())
+        return m_data->text_box_font->wmodes[i];
+    else
+        return text_box::invalid_wmode;
 }
 
 double text_box::get_font_size() const
 {
-    return m_data->font_size;
+    if (this->has_font_info())
+        return m_data->text_box_font->font_size;
+    else
+        return -1;
 }
 
 std::string text_box::get_font_name(int i) const
 {
-    int j = m_data->glyph_to_cache_index[i];
+    if (!this->has_font_info())
+        return std::string("*ignored*");
+
+    int j = m_data->text_box_font->glyph_to_cache_index[i];
     if (j < 0) {
         return std::string("");
     }
-    return m_data->font_info_cache[j].name();
+    return m_data->text_box_font->font_info_cache[j].name();
 }
 
-
-std::vector<text_box> page::text_list() const
+std::vector<text_box> page::text_list(int opt_flag) const
 {
-    d->init_font_info_cache();
-
     std::vector<text_box>  output_list;
 
     /* config values are same with Qt5 Page::TextList() */
@@ -419,41 +435,55 @@ std::vector<text_box> page::text_list() const
                 word->getRotation(),
                 {},
                 word->hasSpaceAfter() == true,
-                {},
-                word->getFontSize(),
-                d->font_info_cache,
-                {}
+                nullptr
             }};
 
+            std::unique_ptr<text_box_font_info_data> tb_font_info = nullptr;
+            if (opt_flag & page::text_list_include_font) {
+                d->init_font_info_cache();
+
+                std::unique_ptr<text_box_font_info_data> tb_font{new text_box_font_info_data{
+                    word->getFontSize(), // double font_size
+                    {},                  // std::vector<text_box::writing_mode> wmodes;
+                    d->font_info_cache,  // std::vector<font_info> font_info_cache;
+                    {}                   // std::vector<int> glyph_to_cache_index;
+                }};
+
+                tb_font_info = std::move(tb_font);
+            };
+
             tb.m_data->char_bboxes.reserve(word->getLength());
             for (int j = 0; j < word->getLength(); j ++) {
                 word->getCharBBox(j, &xMin, &yMin, &xMax, &yMax);
                 tb.m_data->char_bboxes.emplace_back(xMin, yMin, xMax-xMin, yMax-yMin);
             }
 
-            tb.m_data->glyph_to_cache_index.reserve(word->getLength());
-            for (int j = 0; j < word->getLength(); j++) {
-                const TextFontInfo* cur_text_font_info = word->getFontInfo(j);
-
-                // filter-out the invalid WMode value here.
-                switch (cur_text_font_info->getWMode()) {
-                case 0:
-                    tb.m_data->wmodes.push_back(text_box::horizontal_wmode);
-                    break;
-                case 1:
-                    tb.m_data->wmodes.push_back(text_box::vertical_wmode);
-                    break;
-                default:
-                    tb.m_data->wmodes.push_back(text_box::invalid_wmode);
-                };
-
-                tb.m_data->glyph_to_cache_index[j] = -1;
-                for (size_t k = 0; k < d->font_info_cache.size(); k++) {
-                    if (cur_text_font_info->matches(&(d->font_info_cache[k].d->ref))) {
-                        tb.m_data->glyph_to_cache_index[j] = k;
+            if (tb_font_info && d->font_info_cache_initialized) {
+                tb_font_info->glyph_to_cache_index.reserve(word->getLength());
+                for (int j = 0; j < word->getLength(); j++) {
+                    const TextFontInfo* cur_text_font_info = word->getFontInfo(j);
+
+                    // filter-out the invalid WMode value here.
+                    switch (cur_text_font_info->getWMode()) {
+                    case 0:
+                        tb_font_info->wmodes.push_back(text_box::horizontal_wmode);
                         break;
+                    case 1:
+                        tb_font_info->wmodes.push_back(text_box::vertical_wmode);
+                        break;
+                    default:
+                        tb_font_info->wmodes.push_back(text_box::invalid_wmode);
+                    };
+
+                    tb_font_info->glyph_to_cache_index[j] = -1;
+                    for (size_t k = 0; k < tb_font_info->font_info_cache.size(); k++) {
+                        if (cur_text_font_info->matches(&(tb_font_info->font_info_cache[k].d->ref))) {
+                            tb_font_info->glyph_to_cache_index[j] = k;
+                            break;
+                        }
                     }
                 }
+                tb.m_data->text_box_font = std::move(tb_font_info);
             }
 
             output_list.push_back(std::move(tb));
@@ -462,3 +492,8 @@ std::vector<text_box> page::text_list() const
 
     return output_list;
 }
+
+std::vector<text_box> page::text_list() const
+{
+    return text_list(0);
+}
diff --git a/cpp/poppler-page.h b/cpp/poppler-page.h
index ca5be2fd..dd6ebf2c 100644
--- a/cpp/poppler-page.h
+++ b/cpp/poppler-page.h
@@ -66,6 +66,12 @@ public:
     rectf     char_bbox(size_t i) const;
     bool      has_space_after() const;
 
+
+    /**
+      \since 0.89
+     */
+    bool      has_font_info() const;
+
     /**
        Get a writing mode for the i-th glyph
 
@@ -186,6 +192,22 @@ public:
     */
     std::vector<text_box> text_list() const;
 
+    /*
+     * text_list_option_enum is a bitmask-style flags for text_list(),
+     * 0 means the default & simplest behaviour.
+     */
+    enum text_list_option_enum {
+        text_list_include_font = 1 // \since 0.89
+    };
+
+    /**
+       Extended version of text_list() taking an option flag.
+       The option flag should be the multiple of text_list_option_enum.
+
+       \since 0.89
+    */
+    std::vector<text_box> text_list(int opt_flag) const;
+
 private:
     page(document_private *doc, int index);
 
diff --git a/cpp/poppler-private.h b/cpp/poppler-private.h
index 83e46319..4ec159a8 100644
--- a/cpp/poppler-private.h
+++ b/cpp/poppler-private.h
@@ -73,23 +73,17 @@ void delete_all(const Collection &c)
 }
 
 class font_info;
-struct text_box_data
+struct text_box_font_info_data
 {
-    ~text_box_data();
-
-    ustring text;
-    rectf bbox;
-    int rotation;
-    std::vector<rectf> char_bboxes;
-    bool has_space_after;
+    ~text_box_font_info_data();
 
-    std::vector<text_box::writing_mode_enum> wmodes; 
     double font_size;
+    std::vector<text_box::writing_mode_enum> wmodes; 
 
     /*
      * a duplication of the font_info_cache created by the
      * poppler::font_iterator and owned by the poppler::page
-     *  object. Its lifetime might differ from that of text_box
+     * object. Its lifetime might differ from that of text_box
      * object (think about collecting all text_box objects
      * from all pages), so we have to duplicate it into all
      * text_box instances.
@@ -97,7 +91,7 @@ struct text_box_data
     std::vector<font_info> font_info_cache;
 
     /*
-     * a std::vector from the glyph index in the current
+     * a std::vector from the glyph index in the owner
      * text_box to the font_info index in font_info_cache. 
      * The "-1" means no corresponding fonts found in the
      * cache.
@@ -105,6 +99,20 @@ struct text_box_data
     std::vector<int> glyph_to_cache_index;
 };
 
+class font_info;
+struct text_box_data
+{
+    ~text_box_data();
+
+    ustring text;
+    rectf bbox;
+    int rotation;
+    std::vector<rectf> char_bboxes;
+    bool has_space_after;
+
+    std::unique_ptr<text_box_font_info_data> text_box_font;
+};
+
 }
 
 #endif
diff --git a/cpp/tests/poppler-dump.cpp b/cpp/tests/poppler-dump.cpp
index 7864979e..ef391d78 100644
--- a/cpp/tests/poppler-dump.cpp
+++ b/cpp/tests/poppler-dump.cpp
@@ -60,6 +60,7 @@ bool show_help = false;
 bool show_version = false;
 char show_text[32];
 bool show_text_list = false;
+bool show_text_list_with_font = false;
 poppler::page::text_layout_enum show_text_layout = poppler::page::physical_layout;
 
 static const ArgDesc the_args[] = {
@@ -85,6 +86,8 @@ static const ArgDesc the_args[] = {
       "show text (physical|raw|none) extracted from all pages" },
     { "--show-text-list",      argFlag, &show_text_list,       0,
       "show text list (experimental)" },
+    { "--show-text-list-with-font",  argFlag, &show_text_list_with_font, 0,
+      "show text list with font info (experimental)" },
     { "-h",                    argFlag,  &show_help,           0,
       "print usage information" },
     { "--help",                argFlag,  &show_help,           0,
@@ -417,14 +420,14 @@ static void print_page_text(poppler::page *p)
     std::cout << std::endl;
 }
 
-static void print_page_text_list(poppler::page *p)
+static void print_page_text_list(poppler::page *p, int opt_flag = 0)
 {
     if (!p) {
         std::cout << std::setw(out_width) << "Broken Page. Could not be parsed" << std::endl;
         std::cout << std::endl;
         return;
     }
-    auto text_list = p->text_list();
+    auto text_list = p->text_list(opt_flag);
 
     std::cout << "---" << std::endl;
     for (const poppler::text_box &text : text_list) {
@@ -435,9 +438,9 @@ static void print_page_text_list(poppler::page *p)
         std::string font_name = text.get_font_name();
         std::cout << "[" << ustr << "] @ ";
         std::cout << "( x=" << bbox.x() << " y=" << bbox.y() << " w=" << bbox.width() << " h=" << bbox.height() << " )";
-        std::cout << "( fontname=" << font_name << " fontsize=" << font_size << " wmode=" << wmode << " )";
+        if (text.has_font_info())
+            std::cout << "( fontname=" << font_name << " fontsize=" << font_size << " wmode=" << wmode << " )";
         std::cout << std::endl;
-
     }
     std::cout << "---" << std::endl;
 }
@@ -538,12 +541,15 @@ int main(int argc, char *argv[])
             print_page_text(p.get());
         }
     }
-    if (show_text_list) {
+    if (show_text_list || show_text_list_with_font) {
         const int pages = doc->pages();
         for (int i = 0; i < pages; ++i) {
             std::cout << "Page " << (i + 1) << "/" << pages << ":" << std::endl;
             std::unique_ptr<poppler::page> p(doc->create_page(i));
-            print_page_text_list(p.get());
+            if (show_text_list_with_font)
+                print_page_text_list(p.get(), poppler::page::text_list_include_font);
+            else
+                print_page_text_list(p.get(), 0);
         }
     }