Support pdf parsing where not every postscript path command takes a s…

…ingle line

include/hrlib/io/pdfio.hpp

@@ -48,7 +48,7 @@ namespace hrlib { namespace pdf
 
   std::istream& read_linear_dict(std::istream& stream, linear_dict& dict);
   std::istream& read_next_binary(std::istream& stream, std::vector<char>& bin);
-  std::optional<path_cmd> parse_path_line(const std::string &line);
+  std::optional<path_cmd> parse_path_line(std::istream& stream);
 }}
 
 #endif //HRLIB_PDFIO_HPP

src/foillogic/foilio.cpp

@@ -255,16 +255,15 @@ Outline* foillogic::loadOutlinePdfStream(std::istream &stream, std::ostream */*e
   {
     std::string line;
     boost::interprocess::basic_ivectorstream<std::vector<char>> bin_stream(bin);
-    while(getline_safe(bin_stream, line))
+    while(auto pc = pdf::parse_path_line(bin_stream))
     {
-      if (auto pc = pdf::parse_path_line(line)) {
-        if (first.cmd!='m' && pc->cmd!='m')
-          // continue until first move
-          continue;
-        if (first.cmd!='m' && pc->cmd=='m')
-          // set first move
-          first=pc.value();
-        else if (first.cmd=='m' && pc->cmd=='m')
+      if (first.cmd!='m' && pc->cmd!='m')
+        // continue until first move
+        continue;
+      if (first.cmd!='m' && pc->cmd=='m')
+        // set first move
+        first=pc.value();
+      else if (first.cmd=='m' && pc->cmd=='m')
         {
           // break on a second move if path long enough
           // otherwise reset move command
@@ -274,19 +273,18 @@ Outline* foillogic::loadOutlinePdfStream(std::istream &stream, std::ostream */*e
           path_cmds.clear();
           first=pc.value();
         }
-        if (pc->cmd == 'l' && path_cmds.back().cmd == 'l') {
-          // merge line commands that extend each other
-          if (pc->vals[0][0] == path_cmds.back().vals[0][0]) {
-            path_cmds.back().vals[0][1] = pc->vals[0][1];
-            continue;
-          }
-          if (pc->vals[0][1] == path_cmds.back().vals[0][1]) {
-            path_cmds.back().vals[0][0] = pc->vals[0][0];
-            continue;
-          }
+      if (pc->cmd == 'l' && path_cmds.back().cmd == 'l') {
+        // merge line commands that extend each other
+        if (pc->vals[0][0] == path_cmds.back().vals[0][0]) {
+          path_cmds.back().vals[0][1] = pc->vals[0][1];
+          continue;
+        }
+        if (pc->vals[0][1] == path_cmds.back().vals[0][1]) {
+          path_cmds.back().vals[0][0] = pc->vals[0][0];
+          continue;
         }
-        path_cmds.push_back(std::move(pc.value()));
       }
+      path_cmds.push_back(std::move(pc.value()));
     }
     // stop reading file when a path is parsed
     if (path_cmds.size()>2)

src/hrlib/io/pdfio.cpp

@@ -44,13 +44,6 @@ namespace {
   const std::string N("/N");
 
   bool is_path_char(char c) { return std::any_of(path_chars.cbegin(), path_chars.cend(), [c](char f){ return c==f; }); }
-  bool is_path_line(const std::string &str)
-  {
-    return std::all_of(str.begin(), str.end(), [](char c){
-        if (utf8::is_float_char(c)) return true;
-        return is_path_char(c);
-      });
-  }
 
   bool try_read_property(const std::string &line, const std::string &prop, bool &out) {
     if(line.find(prop) != std::string::npos)
@@ -172,40 +165,33 @@ std::istream& hrlib::pdf::read_next_binary(std::istream &stream, std::vector<cha
   }
 }
 
-std::optional<path_cmd> hrlib::pdf::parse_path_line(const std::string &line)
+std::optional<path_cmd> hrlib::pdf::parse_path_line(std::istream& stream)
 {
-  if (!is_path_line(line))
-    return std::optional<path_cmd>();
-
   // Push coordinates & extract command
   path_cmd retval;
-  size_t prev = 0;
   size_t coord_idx = 0;
-  while (true)
-  {
-    auto pos = line.find_first_of(delimiters, prev);
-
-    std::string word = line.substr(prev, pos-prev);
-    if (hrlib::utf8::is_floats(word))
+  while (!stream.eof())
     {
-      double v;
-      std::istringstream(line.substr(prev, pos-prev)) >> v;
-      if (coord_idx % 2 == 0)
-        retval.vals.push_back({v, NAN});
-      else
-        retval.vals.back()[1] = v;
-      ++coord_idx;
+      std::string word;
+      stream >> word;
+      if (hrlib::utf8::is_floats(word)) {
+        double v;
+        std::istringstream(word) >> v;
+        if (coord_idx % 2 == 0)
+          retval.vals.push_back({v, NAN});
+        else
+          retval.vals.back()[1] = v;
+        ++coord_idx;
+      }
+      else if (word.size()==1 && is_path_char(word[0]) && retval.vals.size()>0) {
+        retval.cmd = word[0];
+        return retval;
+      }
+      else {
+        retval.vals.clear();
+        coord_idx=0;
+      }
     }
-    else if (word.size()==1 && is_path_char(word[0]) && retval.vals.size()>0)
-    {
-      retval.cmd = word[0];
-      return retval;
-    }
-
-    prev = pos+1;
-    if (pos==std::string::npos)
-      break;
-  }
 
   return std::optional<path_cmd>();
 }

src/version_autogen.hpp

@@ -11,7 +11,7 @@
 #define MINOR_VERSION 2
 #define REVISION 0
 
-#define BUILD_NUMBER 627
-#define COMMIT_HASH "ff8fce688f5d09b5160d9ebfb8a7dbc3cc23246a"
+#define BUILD_NUMBER 628
+#define COMMIT_HASH "c57dda812c04db7a0f8de5ef21695940fbd191ae"
 
 #endif // VERSION_AUTOGEN_H