Fix memory leak (OSS-Fuzz issue #32246)

Signed-off-by: Stefan Weil <sw@weilnetz.de>

src/ccstruct/blobbox.cpp

@@ -378,7 +378,7 @@ void BLOBNBOX::DeleteNoiseBlobs(BLOBNBOX_LIST *blobs) {
   for (blob_it.mark_cycle_pt(); !blob_it.cycled_list(); blob_it.forward()) {
     BLOBNBOX *blob = blob_it.data();
     if (blob->DeletableNoise()) {
-      delete blob->cblob();
+      delete blob->remove_cblob();
       delete blob_it.extract();
     }
   }
@@ -926,16 +926,6 @@ TO_BLOCK::TO_BLOCK(  // make a block
   block = src_block;
 }
 
-static void clear_blobnboxes(BLOBNBOX_LIST *boxes) {
-  BLOBNBOX_IT it = boxes;
-  // A BLOBNBOX generally doesn't own its blobs, so if they do, you
-  // have to delete them explicitly.
-  for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) {
-    BLOBNBOX *box = it.data();
-    delete box->cblob();
-  }
-}
-
 /**********************************************************************
  * TO_BLOCK::clear
  *
@@ -963,11 +953,11 @@ void TO_BLOCK::clear() {
 
 TO_BLOCK::~TO_BLOCK() {
   // Any residual BLOBNBOXes at this stage own their blobs, so delete them.
-  clear_blobnboxes(&blobs);
+  BLOBNBOX::clear_blobnboxes(&blobs);
-  clear_blobnboxes(&underlines);
+  BLOBNBOX::clear_blobnboxes(&underlines);
-  clear_blobnboxes(&noise_blobs);
+  BLOBNBOX::clear_blobnboxes(&noise_blobs);
-  clear_blobnboxes(&small_blobs);
+  BLOBNBOX::clear_blobnboxes(&small_blobs);
-  clear_blobnboxes(&large_blobs);
+  BLOBNBOX::clear_blobnboxes(&large_blobs);
 }
 
 // Helper function to divide the input blobs over noise, small, medium

src/ccstruct/blobbox.h

@@ -154,6 +154,18 @@ public:
       delete cblob_ptr;
     }
   }
+
+  static void clear_blobnboxes(BLOBNBOX_LIST *boxes) {
+    BLOBNBOX_IT it = boxes;
+    // A BLOBNBOX generally doesn't own its blobs, so if they do, you
+    // have to delete them explicitly.
+    for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) {
+      BLOBNBOX *box = it.data();
+      // TODO: remove next line, currently still needed for resultiterator_test.
+      delete box->remove_cblob();
+    }
+  }
+
   static BLOBNBOX *RealBlob(C_OUTLINE *outline) {
     auto *blob = new C_BLOB(outline);
     return new BLOBNBOX(blob);
@@ -265,6 +277,12 @@ public:
   C_BLOB *cblob() const {
     return cblob_ptr;
   }
+  C_BLOB *remove_cblob() {
+    auto blob = cblob_ptr;
+    cblob_ptr = nullptr;
+    owns_cblob_ = false;
+    return blob;
+  }
   TabType left_tab_type() const {
     return left_tab_type_;
   }

src/ccstruct/werd.cpp

@@ -360,14 +360,9 @@ WERD &WERD::operator=(const WERD &source) {
   flags = source.flags;
   script_id_ = source.script_id_;
   correct = source.correct;
-  if (!cblobs.empty()) {
   cblobs.clear();
-  }
   cblobs.deep_copy(&source.cblobs, &C_BLOB::deep_copy);
-
-  if (!rej_cblobs.empty()) {
   rej_cblobs.clear();
-  }
   rej_cblobs.deep_copy(&source.rej_cblobs, &C_BLOB::deep_copy);
   return *this;
 }
@@ -537,7 +532,7 @@ bool WERD::AddSelectedOutlines(const std::vector<bool> &wanted,
     *make_next_word_fuzzy = false;
   }
   C_BLOB_IT rej_it(&rej_cblobs);
-  for (int i = 0; i < outlines.size(); ++i) {
+  for (unsigned i = 0; i < outlines.size(); ++i) {
     C_OUTLINE *outline = outlines[i];
     if (outline == nullptr) {
       continue; // Already used it.

src/textord/ccnontextdetect.cpp

@@ -278,9 +278,8 @@ void CCNonTextDetect::MarkAndDeleteNonTextBlobs(BLOBNBOX_LIST *blobs, int max_bl
       // It is safe to delete the cblob now, as it isn't used by the grid
       // or BlobOverlapsTooMuch, and the BLOBNBOXes will go away with the
       // dead_blobs list.
-       // TODO(rays) delete the delete when the BLOBNBOX destructor deletes
+      // TODO: remove next line, currently still needed for resultiterator_test.
-       // the cblob.
+      delete blob->remove_cblob();
-      delete blob->cblob();
       dead_it.add_to_end(blob_it.extract());
     }
   }

src/textord/colpartition.cpp

@@ -322,7 +322,8 @@ void ColPartition::DeleteBoxes() {
   // and the BLOBNBOXes own the underlying C_BLOBs.
   for (BLOBNBOX_C_IT bb_it(&boxes_); !bb_it.empty(); bb_it.forward()) {
     BLOBNBOX *bblob = bb_it.extract();
-    delete bblob->cblob();
+    // TODO: remove next line, currently still needed for resultiterator_test.
+    delete bblob->remove_cblob();
     delete bblob;
   }
 }

src/textord/fpchop.cpp

@@ -267,7 +267,7 @@ void split_to_blob(                 // split the blob
   C_BLOB *real_cblob; // cblob to chop
 
   if (blob != nullptr) {
-    real_cblob = blob->cblob();
+    real_cblob = blob->remove_cblob();
   } else {
     real_cblob = nullptr;
   }

src/textord/tordmain.cpp

@@ -181,6 +181,7 @@ void assign_blobs_to_blocks2(Image pix,
     for (blob_it.mark_cycle_pt(); !blob_it.cycled_list(); blob_it.forward()) {
       blob = blob_it.extract();
       newblob = new BLOBNBOX(blob); // Convert blob to BLOBNBOX.
+      newblob->set_owns_cblob(true);
       SetBlobStrokeWidth(pix, newblob);
       port_box_it.add_after_then_move(newblob);
     }

src/textord/tospace.cpp

@@ -864,8 +864,6 @@ the gap between the word being built and the next one. */
   float repetition_spacing; // gap between repetitions
   int32_t xstarts[2];       // row ends
   int32_t prev_x;           // end of prev blob
-  BLOBNBOX *bblob;          // current blob
-  TBOX blob_box;            // bounding box
   BLOBNBOX_IT box_it;       // iterator
   TBOX prev_blob_box;
   TBOX next_blob_box;
@@ -935,18 +933,21 @@ the gap between the word being built and the next one. */
 
     peek_at_next_gap(row, box_it, next_blob_box, next_gap, next_within_xht_gap);
     do {
-      bblob = box_it.data();
+      auto bblob = box_it.data();
-      blob_box = bblob->bounding_box();
+      auto blob_box = bblob->bounding_box();
       if (bblob->joined_to_prev()) {
-        if (bblob->cblob() != nullptr) {
+        auto cblob = bblob->remove_cblob();
+        if (cblob != nullptr) {
           cout_it.set_to_list(cblob_it.data()->out_list());
           cout_it.move_to_last();
-          cout_it.add_list_after(bblob->cblob()->out_list());
+          cout_it.add_list_after(cblob->out_list());
-          delete bblob->cblob();
+          delete cblob;
         }
       } else {
-        if (bblob->cblob() != nullptr) {
+        auto cblob = bblob->cblob();
-          cblob_it.add_after_then_move(bblob->cblob());
+        if (cblob != nullptr) {
+          bblob->set_owns_cblob(false);
+          cblob_it.add_after_then_move(cblob);
         }
         prev_x = blob_box.right();
       }

src/textord/underlin.cpp

@@ -84,11 +84,8 @@ void restore_underlined_blobs( // get chop points
         ru_it.add_after_then_move(new BLOBNBOX(new C_BLOB(&left_coutlines)));
       }
     }
-    if (u_line != nullptr) {
-      delete u_line->cblob();
     delete u_line;
   }
-  }
   if (!ru_it.empty()) {
     ru_it.move_to_first();
     for (ru_it.mark_cycle_pt(); !ru_it.cycled_list(); ru_it.forward()) {

src/textord/wordseg.cpp

@@ -62,15 +62,17 @@ void make_single_word(bool one_blob, TO_ROW_LIST *rows, ROW_LIST *real_rows) {
     for (; !box_it.empty(); box_it.forward()) {
       BLOBNBOX *bblob = box_it.extract();
       if (bblob->joined_to_prev() || (one_blob && !cblob_it.empty())) {
-        if (bblob->cblob() != nullptr) {
+        auto cblob = bblob->remove_cblob();
+        if (cblob != nullptr) {
           C_OUTLINE_IT cout_it(cblob_it.data()->out_list());
           cout_it.move_to_last();
-          cout_it.add_list_after(bblob->cblob()->out_list());
+          cout_it.add_list_after(cblob->out_list());
-          delete bblob->cblob();
+          delete cblob;
         }
       } else {
-        if (bblob->cblob() != nullptr) {
+        auto cblob = bblob->remove_cblob();
-          cblob_it.add_after_then_move(bblob->cblob());
+        if (cblob != nullptr) {
+          cblob_it.add_after_then_move(cblob);
         }
       }
       delete bblob;