[poppler] cpp/poppler-page.cpp cpp/poppler-page.h cpp/tests

[GitLab Mirror]

 cpp/poppler-page.cpp       |   25 ++++++++++++-------------
 cpp/poppler-page.h         |    3 ++-
 cpp/tests/poppler-dump.cpp |    4 +++-
 3 files changed, 17 insertions(+), 15 deletions(-)

New commits:
commit 1e098e9b272d57478a3f23a9a6b6bb1542740aaf
Author: Jiri Jakes <freedesktop at jirijakes.eu>
Date:   Tue Mar 31 21:38:23 2020 +0000

    cpp: Add non_raw_non_physical layout for page::text()

diff --git a/cpp/poppler-page.cpp b/cpp/poppler-page.cpp
index f9cba3e5..d9533cfe 100644
--- a/cpp/poppler-page.cpp
+++ b/cpp/poppler-page.cpp
@@ -262,6 +262,10 @@ ustring page::text(const rectf &r) const
     return text(r, physical_layout);
 }
 
+static void appendToGooString(void *stream, const char *text, int len) {
+  ((GooString *) stream)->append(text, len);
+}
+
 /**
  Returns the text in the page.
 
@@ -275,22 +279,17 @@ ustring page::text(const rectf &r) const
  */
 ustring page::text(const rectf &r, text_layout_enum layout_mode) const
 {
-    std::unique_ptr<GooString> s;
-    const bool use_raw_order = (layout_mode == raw_order_layout);
-    TextOutputDev td(nullptr, false, 0, use_raw_order, false);
-    d->doc->doc->displayPage(&td, d->index + 1, 72, 72, 0, false, true, false);
+    std::unique_ptr<GooString> out(new GooString());
+    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()) {
-        PDFRectangle rect = *d->page->getCropBox();
-        const int rotate = d->page->getRotate();
-        if (rotate == 90 || rotate == 270) {
-            std::swap(rect.x1, rect.y1);
-            std::swap(rect.x2, rect.y2);
-        }
-        s.reset(td.getText(rect.x1, rect.y1, rect.x2, rect.y2));
+        d->doc->doc->displayPage(&td, d->index + 1, 72, 72, 0, false, true, false);
     } else {
-        s.reset(td.getText(r.left(), r.top(), r.right(), r.bottom()));
+        d->doc->doc->displayPageSlice(&td, d->index + 1, 72, 72, 0, false, true, false,
+				      r.left(), r.top(), r.width(), r.height());
     }
-    return ustring::from_utf8(s->c_str());
+    return ustring::from_utf8(out->c_str());
 }
 
 /*
diff --git a/cpp/poppler-page.h b/cpp/poppler-page.h
index bd7f1fcf..28cfaa95 100644
--- a/cpp/poppler-page.h
+++ b/cpp/poppler-page.h
@@ -91,7 +91,8 @@ public:
     };
     enum text_layout_enum {
         physical_layout,
-        raw_order_layout
+        raw_order_layout,
+        non_raw_non_physical_layout
     };
 
     ~page();
diff --git a/cpp/tests/poppler-dump.cpp b/cpp/tests/poppler-dump.cpp
index 857bb6cc..470327ef 100644
--- a/cpp/tests/poppler-dump.cpp
+++ b/cpp/tests/poppler-dump.cpp
@@ -81,7 +81,7 @@ static const ArgDesc the_args[] = {
     { "--show-destinations",   argFlag,  &show_destinations,   0,
       "show named destinations" },
     { "--show-text",           argString, &show_text,          sizeof(show_text),
-      "show text (physical|raw) extracted from all pages" },
+      "show text (physical|raw|none) extracted from all pages" },
     { "--show-text-list",      argFlag, &show_text_list,       0,
       "show text list (experimental)" },
     { "-h",                    argFlag,  &show_help,           0,
@@ -451,6 +451,8 @@ int main(int argc, char *argv[])
             show_text_layout = poppler::page::physical_layout;
         } else if (!memcmp(show_text, "raw", 4)) {
             show_text_layout = poppler::page::raw_order_layout;
+        } else if (!memcmp(show_text, "none", 5)) {
+            show_text_layout = poppler::page::non_raw_non_physical_layout;
         } else {
             error(std::string("unrecognized text mode: '") + show_text + "'");
         }