Fix some typos found by codespell

ccstruct/rect.cpp

@@ -29,7 +29,7 @@
  *
  **********************************************************************/
 
-TBOX::TBOX(                   //construtor
+TBOX::TBOX(                   //constructor
          const ICOORD pt1,  //one corner
          const ICOORD pt2   //the other corner
         ) {

ccstruct/rect.h

@@ -307,7 +307,7 @@ class DLLSYM TBOX  {  // bounding box
  *
  **********************************************************************/
 
-inline TBOX::TBOX(               // construtor
+inline TBOX::TBOX(               // constructor
                 const FCOORD pt  // floating centre
                ) {
   bot_left = ICOORD ((inT16) floor (pt.x ()), (inT16) floor (pt.y ()));

ccutil/genericvector.h

@@ -331,7 +331,7 @@ class GenericVector {
   void init(int size);
 
   // We are assuming that the object generally placed in thie
-  // vector are small enough that for efficiency it makes sence
+  // vector are small enough that for efficiency it makes sense
   // to start with a larger initial size.
   static const int kDefaultVectorSize = 4;
   inT32   size_used_;

ccutil/unicharset.h

@@ -219,7 +219,7 @@ class UNICHARSET {
 
   // Return the UTF8 representation corresponding to the given UNICHAR_ID after
   // resolving any private encodings internal to Tesseract. This method is
-  // preferrable to id_to_unichar for outputting text that will be visible to
+  // preferable to id_to_unichar for outputting text that will be visible to
   // external applications.
   const char* const id_to_unichar_ext(UNICHAR_ID id) const;
 

cube/char_samp.h

@@ -81,7 +81,7 @@ class CharSamp : public Bmp8 {
       label32_ = NULL;
     }
     if (label32 != NULL) {
-      // remove any byte order markes if any
+      // remove any byte order marks if any
       if (label32[0] == 0xfeff) {
         label32++;
       }

opencl/oclkernels.h

@@ -807,7 +807,7 @@ KERNEL(
 // HistogramRect Kernel: Accumulate
 // assumes 4 channels, i.e., bytes_per_pixel = 4
 // assumes number of pixels is multiple of 8
-// data is layed out as
+// data is laid out as
 // ch0                                           ch1 ...
 // bin0          bin1            bin2...         bin0...
 // rpt0,1,2...256  rpt0,1,2...

opencl/openclwrapper.cpp

@@ -495,7 +495,7 @@ int OpenclDevice::GeneratBinFromKernelSource( cl_program program, const char * c
                 printf("[OD] write binary[%s] failed\n", fileName);
                 return 0;
             } //else
-            printf("[OD] write binary[%s] succesfully\n", fileName);
+            printf("[OD] write binary[%s] successfully\n", fileName);
         }
     }
 
@@ -1630,7 +1630,7 @@ pixDilateCL(l_int32  hsize, l_int32  vsize, l_int32  wpl, l_int32  h)
     }
     else if (xp > 0 || xn > 0 )
     {
-        //Specfic Horizontal pass kernel for half width < 32
+        //Specific Horizontal pass kernel for half width < 32
         rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "morphoDilateHor_32word", &status );
         isEven = (xp != xn);
 
@@ -2371,7 +2371,7 @@ OpenclDevice::pixGetLinesCL(PIX  *pixd,
 /*************************************************************************
  *  HistogramRect
  *  Otsu Thresholding Operations
- *  histogramAllChannels is layed out as all channel 0, then all channel 1...
+ *  histogramAllChannels is laid out as all channel 0, then all channel 1...
  *  only supports 1 or 4 channels (bytes_per_pixel)
  ************************************************************************/
 int OpenclDevice::HistogramRectOCL(

testing/runtestset.sh

@@ -56,7 +56,7 @@ rm -f testing/reports/$setname.times
 while read page dir
 do
   # A pages file may be a list of files with subdirs or maybe just
-  # a plain list of files so accomodate both.
+  # a plain list of files so accommodate both.
   if [ "$dir" ]
   then
      srcdir="$imdir/$dir"

textord/alignedblob.cpp

@@ -402,7 +402,7 @@ BLOBNBOX* AlignedBlob::FindAlignedBlob(const AlignedBlobParams& p,
   // Compute skew tolerance.
   int skew_tolerance = p.max_v_gap / kMaxSkewFactor;
   // Calculate xmin and xmax of the search box so that it contains
-  // all possibly relevant boxes upto p.max_v_gap above or below accoording
+  // all possibly relevant boxes up to p.max_v_gap above or below accoording
   // to top_to_bottom.
   // Start with a notion of vertical with the current estimate.
   int x2 = (p.max_v_gap * p.vertical.x() + p.vertical.y()/2) / p.vertical.y();
@@ -502,7 +502,7 @@ BLOBNBOX* AlignedBlob::FindAlignedBlob(const AlignedBlobParams& p,
     }
     if ((p.right_tab && neighbour->leader_on_right()) ||
         (!p.right_tab && neighbour->leader_on_left()))
-      continue;  // Neigbours of leaders are not allowed to be used.
+      continue;  // Neighbours of leaders are not allowed to be used.
     if (n_x <= x_at_n_y + p.r_align_tolerance &&
         n_x >= x_at_n_y - p.l_align_tolerance) {
       // Aligned so keep it. If it is a marked tab save it as result,

textord/bbgrid.cpp

@@ -154,7 +154,7 @@ IntGrid* IntGrid::NeighbourhoodSum() const {
 }
 
 // Returns true if more than half the area of the rect is covered by grid
-// cells that are over the theshold.
+// cells that are over the threshold.
 bool IntGrid::RectMostlyOverThreshold(const TBOX& rect, int threshold) const {
   int min_x, min_y, max_x, max_y;
   GridCoords(rect.left(), rect.bottom(), &min_x, &min_y);

textord/bbgrid.h

@@ -127,7 +127,7 @@ class IntGrid : public GridBase {
     grid_[grid_y * gridwidth_ + grid_x] = value;
   }
   // Returns true if more than half the area of the rect is covered by grid
-  // cells that are over the theshold.
+  // cells that are over the threshold.
   bool RectMostlyOverThreshold(const TBOX& rect, int threshold) const;
 
   // Returns true if any cell value in the given rectangle is zero.
@@ -292,7 +292,7 @@ template<class BBC, class BBC_CLIST, class BBC_C_IT> class GridSearch {
   // Return the next bbox in the search or NULL if done.
   BBC* NextFullSearch();
 
-  // Start a new radius search. Will search in a spiral upto a
+  // Start a new radius search. Will search in a spiral up to a
   // given maximum radius in grid cells from the given center in pixels.
   void StartRadSearch(int x, int y, int max_radius);
   // Return the next bbox in the radius search or NULL if the
@@ -750,7 +750,7 @@ void GridSearch<BBC, BBC_CLIST, BBC_C_IT>::StartSideSearch(int x,
                                                            int ymin, int ymax) {
   // Right search records the x in x_origin_, the ymax in y_origin_
   // and the size of the vertical strip to search in radius_.
-  // To guarantee finding overlapping objects of upto twice the
+  // To guarantee finding overlapping objects of up to twice the
   // given size, double the height.
   radius_ = ((ymax - ymin) * 2 + grid_->gridsize_ - 1) / grid_->gridsize_;
   rad_index_ = 0;

textord/blkocc.h

@@ -39,7 +39,7 @@ CLASS REGION_OCC
   built in sorted order of min x. Overlapping REGION_OCCs are not permitted on
   a single list. An overlapping region to be added causes the existing region
   to be extended. This extension may result in the following REGION_OCC on the
-  list overlapping the ammended one. In this case the ammended REGION_OCC is
+  list overlapping the amended one. In this case the amended REGION_OCC is
   further extended to include the range of the following one, so that the
   following one can be deleted.
 

textord/ccnontextdetect.cpp

@@ -67,7 +67,7 @@ CCNonTextDetect::CCNonTextDetect(int gridsize,
     noise_density_(NULL) {
   // TODO(rays) break max_noise_count_ out into an area-proportional
   // value, as now plus an additive constant for the number of text blobs
-  // in the 3x3 neigbourhood - maybe 9.
+  // in the 3x3 neighbourhood - maybe 9.
 }
 
 CCNonTextDetect::~CCNonTextDetect() {

textord/colfind.cpp

@@ -340,7 +340,7 @@ int ColumnFinder::FindBlocks(PageSegMode pageseg_mode, Pix* scaled_color,
   // into thinking they are dealing with left-to-right text.
   // To do this, we reflect the needed data in the y-axis and then reflect
   // the blocks back after they have been created. This is a temporary
-  // arrangment that is confined to this function only, so the reflection
+  // arrangement that is confined to this function only, so the reflection
   // is completely invisible in the output blocks.
   // The only objects reflected are:
   // The vertical separator lines that have already been found;
@@ -869,7 +869,7 @@ void ColumnFinder::ShrinkRangeToLongestRun(int** column_set_costs,
   }
 }
 
-// Moves start in the direction of step, upto, but not including end while
+// Moves start in the direction of step, up to, but not including end while
 // the only incompatible regions are no more than kMaxIncompatibleColumnCount
 // in size, and the compatible regions beyond are bigger.
 void ColumnFinder::ExtendRangePastSmallGaps(int** column_set_costs,

textord/colfind.h

@@ -212,7 +212,7 @@ class ColumnFinder : public TabFind {
                                const bool* any_columns_possible,
                                int column_set_id,
                                int* best_start, int* best_end);
-  // Moves start in the direction of step, upto, but not including end while
+  // Moves start in the direction of step, up to, but not including end while
   // the only incompatible regions are no more than kMaxIncompatibleColumnCount
   // in size, and the compatible regions beyond are bigger.
   void ExtendRangePastSmallGaps(int** column_set_costs,

textord/colpartition.cpp

@@ -981,7 +981,7 @@ int ColPartition::CountOverlappingBoxes(const TBOX& box) {
   return overlap_count;
 }
 
-// Computes and sets the type_ and first_colum_, last_column_ and column_set_.
+// Computes and sets the type_ and first_column_, last_column_ and column_set_.
 // resolution refers to the ppi resolution of the image.
 void ColPartition::SetPartitionType(int resolution, ColPartitionSet* columns) {
   int first_spanned_col = -1;
@@ -2194,7 +2194,7 @@ bool ColPartition::IsInSameColumnAs(const ColPartition& part) const {
 void ColPartition::SmoothSpacings(int resolution, int page_height,
                                   ColPartition_LIST* parts) {
   // The task would be trivial if we didn't have to allow for blips -
-  // occasional offsets in spacing caused by anomolous text, such as all
+  // occasional offsets in spacing caused by anomalous text, such as all
   // caps, groups of descenders, joined words, Arabic etc.
   // The neighbourhood stores a consecutive group of partitions so that
   // blips can be detected correctly, yet conservatively enough to not

textord/colpartition.h

@@ -695,7 +695,7 @@ class ColPartition : public ELIST2_LINK {
   // one partner. This makes block creation simpler.
   // If get_desperate is true, goes to more desperate merge methods
   // to merge flowing text before breaking partnerships.
-  void RefinePartners(PolyBlockType type, bool get_desparate,
+  void RefinePartners(PolyBlockType type, bool get_desperate,
                       ColPartitionGrid* grid);
 
   // Returns true if this column partition is in the same column as
@@ -713,7 +713,7 @@ class ColPartition : public ELIST2_LINK {
   }
 
  private:
-  // enum to refer to the entries in a neigbourhood of lines.
+  // enum to refer to the entries in a neighbourhood of lines.
   // Used by SmoothSpacings to test for blips with OKSpacingBlip.
   enum SpacingNeighbourhood {
     PN_ABOVE2,

textord/colpartitiongrid.cpp

@@ -106,7 +106,7 @@ void ColPartitionGrid::HandleClick(int x, int y) {
 // Merges ColPartitions in the grid that look like they belong in the same
 // textline.
 // For all partitions in the grid, calls the box_cb permanent callback
-// to compute the search box, seaches the box, and if a candidate is found,
+// to compute the search box, searches the box, and if a candidate is found,
 // calls the confirm_cb to check any more rules. If the confirm_cb returns
 // true, then the partitions are merged.
 // Both callbacks are deleted before returning.
@@ -1438,7 +1438,7 @@ void ColPartitionGrid::FindMergeCandidates(const ColPartition* part,
 }
 
 // Smoothes the region type/flow type of the given part by looking at local
-// neigbours and the given image mask. Searches a padded rectangle with the
+// neighbours and the given image mask. Searches a padded rectangle with the
 // padding truncated on one size of the part's box in turn for each side,
 // using the result (if any) that has the least distance to all neighbours
 // that contribute to the decision. This biases in favor of rectangular
@@ -1759,7 +1759,7 @@ void ColPartitionGrid::FindPartitionMargins(ColPartitionSet* columns,
   part->set_right_margin(right_margin);
 }
 
-// Starting at x, and going in the specified direction, upto x_limit, finds
+// Starting at x, and going in the specified direction, up to x_limit, finds
 // the margin for the given y range by searching sideways,
 // and ignoring not_this.
 int ColPartitionGrid::FindMargin(int x, bool right_to_left, int x_limit,

textord/colpartitiongrid.h

@@ -45,7 +45,7 @@ class ColPartitionGrid : public BBGrid<ColPartition,
   // Merges ColPartitions in the grid that look like they belong in the same
   // textline.
   // For all partitions in the grid, calls the box_cb permanent callback
-  // to compute the search box, seaches the box, and if a candidate is found,
+  // to compute the search box, searches the box, and if a candidate is found,
   // calls the confirm_cb to check any more rules. If the confirm_cb returns
   // true, then the partitions are merged.
   // Both callbacks are deleted before returning.
@@ -200,7 +200,7 @@ class ColPartitionGrid : public BBGrid<ColPartition,
                            bool debug, ColPartition_CLIST* candidates);
 
   // Smoothes the region type/flow type of the given part by looking at local
-  // neigbours and the given image mask. Searches a padded rectangle with the
+  // neighbours and the given image mask. Searches a padded rectangle with the
   // padding truncated on one size of the part's box in turn for each side,
   // using the result (if any) that has the least distance to all neighbours
   // that contribute to the decision. This biases in favor of rectangular
@@ -246,7 +246,7 @@ class ColPartitionGrid : public BBGrid<ColPartition,
   // neighbours that vertically overlap significantly.
   void FindPartitionMargins(ColPartitionSet* columns, ColPartition* part);
 
-  // Starting at x, and going in the specified direction, upto x_limit, finds
+  // Starting at x, and going in the specified direction, up to x_limit, finds
   // the margin for the given y range by searching sideways,
   // and ignoring not_this.
   int FindMargin(int x, bool right_to_left, int x_limit,

textord/fpchop.cpp

@@ -735,7 +735,7 @@ C_OUTLINE *join_chopped_fragments(                         //join pieces
  * join_segments
  *
  * Join the two edgestep fragments such that the second comes after
- * the first and the gap beween them is closed.
+ * the first and the gap between them is closed.
  **********************************************************************/
 
 void join_segments(                         //join pieces

textord/imagefind.cpp

@@ -648,7 +648,7 @@ static void CutChunkFromParts(const TBOX& box, const TBOX& im_box,
     TBOX part_box = part->bounding_box();
     if (part_box.overlap(box)) {
       // This part must be cut and replaced with the remains. There are
-      // upto 4 pieces to be made. Start with the first one and use
+      // up to 4 pieces to be made. Start with the first one and use
       // add_before_stay_put. For each piece if it has no black pixels
       // left, just don't make the box.
       // Above box.

textord/makerow.cpp

@@ -671,7 +671,7 @@ BOOL8 find_best_dropout_row(                    //find neighbours
                             TO_ROW_IT *row_it,  //current position
                             BOOL8 testing_on    //correct orientation
                            ) {
-  inT32 next_index;              //of neigbouring row
+  inT32 next_index;              //of neighbouring row
   inT32 row_offset;              //from current row
   inT32 abs_dist;                //absolute distance
   inT8 row_inc;                  //increment to row_index

textord/pithsync.h

@@ -66,7 +66,7 @@ class FPCUTPT
       inT16 pitch,               //proposed pitch
       inT16 pitch_error);        //allowed tolerance
 
-    inT32 position() {  //acces func
+    inT32 position() {  //access func
       return xpos;
     }
     double cost_function() {

textord/pitsync1.h

@@ -46,7 +46,7 @@ class FPSEGPT:public ELIST_LINK
             FPSEGPT_LIST *prev_list);  //previous segment
     FPSEGPT(FPCUTPT *cutpt);  //build from new type
 
-    inT32 position() {  //acces func
+    inT32 position() {  //access func
       return xpos;
     }
     double cost_function() {

textord/tabfind.cpp

@@ -75,7 +75,7 @@ const int kMaxTextLineBlobRatio = 5;
 const int kMinTextLineBlobRatio = 3;
 // Fraction of box area covered by image to make a blob image.
 const double kMinImageArea = 0.5;
-// Upto 30 degrees is allowed for rotations of diacritic blobs.
+// Up to 30 degrees is allowed for rotations of diacritic blobs.
 // Keep this value slightly larger than kCosSmallAngle in blobbox.cpp
 // so that the assert there never fails.
 const double kCosMaxSkewAngle = 0.866025;

textord/tabfind.h

@@ -371,7 +371,7 @@ class TabFind : public AlignedBlob {
   TabVector_LIST vectors_;        //< List of rule line and tabstops.
   TabVector_IT v_it_;             //< Iterator for searching vectors_.
   TabVector_LIST dead_vectors_;   //< Separators and unpartnered tab vectors.
-  // List of commonly occuring width ranges with x=min and y=max.
+  // List of commonly occurring width ranges with x=min and y=max.
   ICOORDELT_LIST column_widths_;  //< List of commonly occurring width ranges.
   /** Callback to test an int for being a common width. */
   WidthCallback* width_cb_;

textord/tablefind.cpp

@@ -943,7 +943,7 @@ bool TableFinder::HasWideOrNoInterWordGap(ColPartition* part) const {
     return true;
 
   // return true if the maximum gap found is smaller than the minimum allowed
-  // max_gap in a text partition. This indicates that there is no signficant
+  // max_gap in a text partition. This indicates that there is no significant
   // space in the partition, hence it is likely a single word.
   return largest_partition_gap_found < min_gap;
 }
@@ -954,7 +954,7 @@ bool TableFinder::HasWideOrNoInterWordGap(ColPartition* part) const {
 // Note that this includes overlapping leaders. However, it does not
 // include leaders in different columns on the page.
 // Possible false-positive will include lists, such as a table of contents.
-// As these arise, the agressive nature of this search may need to be
+// As these arise, the aggressive nature of this search may need to be
 // trimmed down.
 bool TableFinder::HasLeaderAdjacent(const ColPartition& part) {
   if (part.flow() == BTFT_LEADER)

textord/tablerecog.cpp

@@ -685,7 +685,7 @@ int StructuredTable::CountHorizontalIntersections(int y) {
 
 // Counts how many text partitions are in this box.
 // This is used to count partitons in cells, as that can indicate
-// how "strong" a potential table row/colum (or even full table) actually is.
+// how "strong" a potential table row/column (or even full table) actually is.
 int StructuredTable::CountPartitions(const TBOX& box) {
   ColPartitionGridSearch gsearch(text_grid_);
   gsearch.SetUniqueMode(true);
@@ -740,7 +740,7 @@ StructuredTable* TableRecognizer::RecognizeTable(const TBOX& guess) {
   table->set_line_grid(line_grid_);
   table->set_max_text_height(max_text_height_);
 
-  // Try to solve ths simple case, a table with *both*
+  // Try to solve this simple case, a table with *both*
   // vertical and horizontal lines.
   if (RecognizeLinedTable(guess, table))
     return table;