-add-text now only a single caller to Cpdfembed

cpdfaddtext.ml

@@ -2,13 +2,6 @@
 open Pdfutil
 open Cpdferror
 
-(* Keep a record of charcodes used for font subsetting. *)
-let used = null_hash ()
-
-(* If we have used a TTF font, put its name and object number here. Then we
-   know we put it there, and can avoid writing it anew across ANDS. *)
-let glob_pdfobjnum = null_hash ()
-
 type color =
   Grey of float
 | RGB of float * float * float
@@ -58,14 +51,6 @@ let colour_op_stroke = function
   | CYMK (c, y, m, k) -> Pdfops.Op_K (c, y, m, k)
 
 let ops fontname longest_w x y rotate hoffset voffset outline linewidth unique_fontname unique_extgstatename colour fontsize text =
-  begin match Hashtbl.find used fontname with
-    | exception Not_found ->
-        let thisused = null_hash () in
-          String.iter (fun x -> Hashtbl.replace thisused x ()) text;
-          Hashtbl.add used fontname thisused
-    | thisused ->
-        String.iter (fun x -> Hashtbl.replace thisused x ()) text;
-  end;
   [Pdfops.Op_q;
   Pdfops.Op_BMC "/CPDFSTAMP";
    Pdfops.Op_cm
@@ -234,6 +219,16 @@ let replace_pairs pdf endpage extract_text_font_size filename bates batespad num
                  then numstring
                  else implode (many '0' (w - String.length numstring)) ^ numstring))]
 
+let expand_lines text time pdf endpage extract_text_font_size filename bates batespad num page lines =
+  let expanded_lines =
+    map
+      (function text ->
+         process_text time text (replace_pairs pdf endpage extract_text_font_size filename bates batespad num page))
+      lines
+  in
+    (* process URLs for justification too *)
+    map (fun line -> fst (get_urls_line line)) expanded_lines
+
 let addtext
   time lines linewidth outline fast colour fontname encoding bates batespad
   fontsize font fontpdfobj underneath position hoffset voffset text pages
@@ -300,16 +295,7 @@ let addtext
           let ops, urls, x, y, hoffset, voffset, text, joffset =
             let text = process_text time text (replace_pairs pdf endpage extract_text_font_size filename bates batespad num page) in
             let text, urls = get_urls_line text in
-
+            let expanded_lines = expand_lines text time pdf endpage extract_text_font_size filename bates batespad num page lines in
-                let expanded_lines =
-                  map
-                    (function text ->
-                       process_text time text (replace_pairs pdf endpage extract_text_font_size filename bates batespad num page))
-                    lines
-                in
-                let expanded_lines = (* process URLs for justification too *)
-                  map (fun line -> fst (get_urls_line line)) expanded_lines
-                in
             let textwidth = calc_textwidth text
             and allwidths = map calc_textwidth expanded_lines in
               let longest_w = last (sort compare allwidths) in
@@ -426,8 +412,19 @@ let
   justification midline topline filename extract_text_font_size shift
   ?(raw=false) pdf
 =
-  let time = Cpdfstrftime.current_time () in
   if pages = [] then error "addtexts: empty page range" else
+  let time = Cpdfstrftime.current_time () in
+  let endpage = Pdfpage.endpage pdf in
+  let ps = Pdfpage.pages_of_pagetree pdf in
+  let used = null_hash () in
+  let lines = map unescape_string (split_at_newline text) in
+  iter2
+    (fun num page ->
+       let expanded_lines = expand_lines text time pdf endpage extract_text_font_size filename bates batespad num page lines in
+         let codepoints = map Pdftext.codepoints_of_utf8 expanded_lines in
+         iter (iter (fun x -> Hashtbl.replace used x ())) codepoints)
+    pages
+    (map (fun x -> List.nth ps (x - 1)) pages);
   let realfontname = ref fontname in
   let font =
     match cpdffont with
@@ -436,7 +433,7 @@ let
         Some (hd (fst f))
     | Cpdfembed.EmbedInfo {fontfile; fontname; encoding} ->
         (*Printf.printf "Cpdfaddtext.addtexts: EmbedInfo\n";*)
-        let embedded = Cpdfembed.embed_truetype pdf ~fontfile ~fontname ~codepoints:[] ~encoding in
+        let embedded = Cpdfembed.embed_truetype pdf ~fontfile ~fontname ~codepoints:(map fst (list_of_hashtbl used)) ~encoding in
           Some (hd (fst embedded))
     | Cpdfembed.ExistingNamedFont -> None
   in
@@ -445,13 +442,7 @@ let
     | Some (Pdftext.StandardFont _ as font) ->
         Pdf.Indirect (Pdftext.write_font pdf font)
     | Some f ->
-        begin match Hashtbl.find glob_pdfobjnum fontname with
+        Pdf.Indirect (Pdftext.write_font pdf f)
-        | exception Not_found ->
-            let i = Pdftext.write_font pdf f in
-              Hashtbl.add glob_pdfobjnum fontname i; Pdf.Indirect i
-        | i ->
-            Pdf.Indirect i
-        end
     | None -> 
         let firstpage =
           List.nth (Pdfpage.pages_of_pagetree pdf) (hd pages - 1)
@@ -532,23 +523,14 @@ let
                      !pdf;
                    voffset := !voffset +. (linespacing *. fontsize))
               lines;
-              begin match cpdffont with
+              (*begin match cpdffont with
               | Cpdfembed.EmbedInfo {fontfile; fontname; encoding} ->
-                  let charcodes =
+                  let codepoints = map fst (list_of_hashtbl used) in
-                    match Hashtbl.find used fontname with
-                    | exception Not_found -> []
-                    | thisused -> map fst (list_of_hashtbl thisused)
-                  in
-                  let encoding_table = Pdftext.table_of_encoding encoding in
-                  let glyphlist_table = Pdfglyphlist.glyph_hashes () in
-                  let codepoints =
-                    map (fun c -> unicode_codepoint_of_pdfcode encoding_table glyphlist_table (int_of_char c)) charcodes
-                  in
                   let objnum = match fontpdfobj with Pdf.Indirect i -> i | _ -> failwith "bad fontpdfobj" in
                   let font = hd (fst (Cpdfembed.embed_truetype !pdf ~fontfile ~fontname ~codepoints ~encoding)) in
                     ignore (Pdftext.write_font ~objnum !pdf font)
               | _ -> ()
-              end;
+              end;*)
               !pdf
 
 

cpdftruetype.ml

@@ -2,9 +2,6 @@
 open Pdfutil
 open Pdfio
 
-(* FIXME Make sure -add-text calls subset once only *)
-(* FIXME Make it work with -draw *)
-(* FIXME Check proper use of fontpacks and maximal sharing in -toc -typeset -add-text -draw *)
 (* FIXME Proper widths for .notdef, and warn on .notdef being produced *)
 (* FIXME Add suport for composite glyphs *)
 (* FIXME Make sure -embed-std14 works for all commands *)