WIP: some implementation refinement

- MAG writer output name does not need to match a cell name
  (this file is not written in this case, but the path and extension
  are taken from it). A warning is issued in this case.
- Strings for layer and cell names are normalized (special chars, UTF8)
- Cell duplicates won't be produced but existing cells are reused.
  This enables "incremental reads".

src/plugins/streamers/magic/db_plugin/dbMAGReader.cc

@@ -137,14 +137,23 @@ MAGReader::warn (const std::string &msg)
 db::cell_index_type
 MAGReader::cell_from_path (const std::string &path, db::Layout &layout)
 {
-  std::map<std::string, db::cell_index_type>::const_iterator c = m_cells_read.find (path);
+  std::string cellname = tl::filename (path);
+
+  std::map<std::string, db::cell_index_type>::const_iterator c = m_cells_read.find (cellname);
   if (c != m_cells_read.end ()) {
     return c->second;
   }
 
   //  NOTE: this can lead to cell variants if a cell is present with different library paths ... (L500_CHAR_p)
-  db::cell_index_type ci = layout.add_cell (cell_name_from_path (path).c_str ());
-  m_cells_read.insert (std::make_pair (path, ci));
+  db::cell_index_type ci;
+  if (layout.has_cell (cellname.c_str ())) {
+    //  NOTE: this reuses an existing cell and will add(!) the layout to the latter. This
+    //  enables "incremental read" like for GDS files.
+    ci = layout.cell_by_name (cellname.c_str ()).second;
+  } else {
+    ci = layout.add_cell (cell_name_from_path (path).c_str ());
+  }
+  m_cells_read.insert (std::make_pair (cellname, ci));
 
   std::string cell_file;
   if (! resolve_path (path, cell_file)) {
@@ -152,7 +161,7 @@ MAGReader::cell_from_path (const std::string &path, db::Layout &layout)
     tl::warn << tl::to_string (tr ("Unable to find a layout file for cell - skipping this cell: ")) << path;
     layout.cell (ci).set_ghost_cell (true);
   } else {
-    m_cells_to_read.insert (std::make_pair (path, std::make_pair (cell_file, ci)));
+    m_cells_to_read.insert (std::make_pair (cellname, std::make_pair (cell_file, ci)));
   }
 
   return ci;

src/plugins/streamers/magic/db_plugin/dbMAGWriter.cc

@@ -62,12 +62,8 @@ MAGWriter::write (db::Layout &layout, tl::OutputStream &stream, const db::SaveLa
   tl::URI src (stream.path ());
   std::string basename = tl::basename (src.path ());
   std::pair<bool, db::cell_index_type> ci = layout.cell_by_name (basename.c_str ());
-  if (! ci.first) {
-    throw tl::Exception (tl::to_string (tr ("Magic file names must be valid cell names. This is not a valid name: ")) + basename);
-  }
-
-  if (cell_set.find (ci.second) == cell_set.end ()) {
-    throw tl::Exception (tl::to_string (tr ("Cell name derived from file name isn't a selected cell: ")) + basename);
+  if (! ci.first || cell_set.find (ci.second) == cell_set.end ()) {
+    tl::warn << tl::to_string (tr ("Magic file names should be valid cell names - otherwise no such file is written. This is not a valid name: ")) << basename;
   }
 
   m_options = options.get_options<MAGWriterOptions> ();
@@ -135,7 +131,7 @@ MAGWriter::do_write_cell (db::cell_index_type ci, const std::vector <std::pair <
     tech = layout.meta_info_value ("technology");
   }
   if (! tech.empty ()) {
-    os << "tech " << tl::to_word_or_quoted_string (tl::to_lower_case (tech)) << "\n";
+    os << "tech " << make_string (tl::to_lower_case (tech)) << "\n";
   }
 
   os << "timestamp " << m_timestamp << "\n";
@@ -151,7 +147,7 @@ MAGWriter::do_write_cell (db::cell_index_type ci, const std::vector <std::pair <
     any = false;
     for (db::Shapes::shape_iterator s = cell.shapes (ll->first).begin (db::ShapeIterator::Boxes | db::ShapeIterator::Polygons | db::ShapeIterator::Paths); ! s.at_end (); ++s) {
       if (! any) {
-        os << "<< " << tl::to_word_or_quoted_string (tl::to_lower_case (ll->second.name)) << " >>\n";
+        os << "<< " << make_string (tl::to_lower_case (ll->second.name)) << " >>\n";
         any = true;
       }
       db::Polygon poly;
@@ -262,7 +258,7 @@ MAGWriter::write_label (const std::string &layer, const db::Text &text, const db
     s = tl::replaced (s, "\n", "\\n");
   }
 
-  os << "rlabel " << tl::to_word_or_quoted_string (layer) << " " << v.x () << " " << v.y () << " " << v.x () << " " << v.y () << " 0 " << s << "\n";
+  os << "rlabel " << make_string (layer) << " " << v.x () << " " << v.y () << " " << v.x () << " " << v.y () << " 0 " << s << "\n";
 }
 
 void
@@ -294,7 +290,7 @@ MAGWriter::write_single_instance (db::cell_index_type ci, db::ICplxTrans trans,
 
   int id = (m_cell_id [ci] += 1);
   std::string cn = layout.cell_name (ci);
-  os << "use " << tl::to_word_or_quoted_string (cn) << " " << tl::to_word_or_quoted_string (cn + "_" + tl::to_string (id)) << "\n";
+  os << "use " << make_string (cn) << " " << make_string (cn + "_" + tl::to_string (id)) << "\n";
 
   if (na > 1 || nb > 1) {
 
@@ -384,5 +380,29 @@ MAGWriter::needs_rounding (const db::Vector &v) const
   return ! db::DVector (res).equal (db::DVector (v) * m_sf);
 }
 
+static inline bool is_valid_char (uint32_t c32)
+{
+  return (c32 >= 'A' && c32 <= 'Z') ||
+         (c32 >= 'a' && c32 <= 'z') ||
+         (c32 >= '0' && c32 <= '9') ||
+         c32 == '_' || c32 == '.';
+}
+
+std::string
+MAGWriter::make_string (const std::string &s)
+{
+  std::string res;
+  for (const char *cp = s.c_str (); *cp; ) {
+    uint32_t c32 = tl::utf32_from_utf8 (cp);
+    if (! is_valid_char (c32)) {
+      res += tl::sprintf ("x%x", c32);
+    } else {
+      res += (char) c32;
+    }
+  }
+  return res;
+}
+
+
 }
 

src/plugins/streamers/magic/db_plugin/dbMAGWriter.h

@@ -104,6 +104,7 @@ private:
   void write_label (const std::string &layer, const db::Text &text, const Layout &layout, tl::OutputStream &os);
   void write_instance (const db::CellInstArray &inst, const db::Layout &layout, tl::OutputStream &os);
   void write_single_instance (db::cell_index_type ci, ICplxTrans tr, db::Vector a, db::Vector b, unsigned long na, unsigned long nb, const db::Layout &layout, tl::OutputStream &os);
+  std::string make_string (const std::string &s);
 };
 
 } // namespace db

src/tl/tl/tlGlobPattern.cc

@@ -29,49 +29,6 @@
 namespace tl
 {
 
-// ---------------------------------------------------------------------------------
-//  TODO: take from tlString.h
-
-static inline uint32_t utf32_from_utf8 (const char *&cp, const char *cpe = 0)
-{
-  uint32_t c32 = (unsigned char) *cp++;
-  if (c32 >= 0xf0 && ((cpe && cp + 2 < cpe) || (! cpe && cp [0] && cp [1] && cp [2]))) {
-    c32 = ((c32 & 0x7) << 18) | ((uint32_t (cp [0]) & 0x3f) << 12) | ((uint32_t (cp [1]) & 0x3f) << 6) | (uint32_t (cp [2]) & 0x3f);
-    cp += 3;
-  } else if (c32 >= 0xe0 && ((cpe && cp + 1 < cpe) || (! cpe && cp [0] && cp [1]))) {
-    c32 = ((c32 & 0xf) << 12) | ((uint32_t (cp [0]) & 0x3f) << 6) | (uint32_t (cp [1]) & 0x3f);
-    cp += 2;
-  } else if (c32 >= 0xc0 && ((cpe && cp < cpe) || (! cpe && cp [0]))) {
-    c32 = ((c32 & 0x1f) << 6) | (uint32_t (*cp) & 0x3f);
-    ++cp;
-  }
-
-  return c32;
-}
-
-#include "utf_casefolding.h"
-
-static inline wchar_t wdowncase (wchar_t c)
-{
-  int ch = c >> 8;
-  if (ch >= 0 && ch < int (sizeof (uc_tab) / sizeof (uc_tab[0])) && uc_tab[ch]) {
-    return uc_tab[ch][c & 0xff];
-  } else {
-    return c;
-  }
-}
-
-static inline uint32_t utf32_downcase (uint32_t c32)
-{
-  if (sizeof (wchar_t) == 2 && c32 >= 0x10000) {
-    return c32;
-  } else {
-    return uint32_t (wdowncase (wchar_t (c32)));
-  }
-}
-
-// ---------------------------------------------------------------------------------
-
 class GlobPatternOpBase
 {
 public:

src/tl/tl/tlString.cc

@@ -46,7 +46,7 @@ static std::locale c_locale ("C");
 
 #include "utf_casefolding.h"
 
-static inline wchar_t wdowncase (wchar_t c)
+wchar_t wdowncase (wchar_t c)
 {
   int ch = c >> 8;
   if (ch >= 0 && ch < int (sizeof (uc_tab) / sizeof (uc_tab[0])) && uc_tab[ch]) {
@@ -56,7 +56,7 @@ static inline wchar_t wdowncase (wchar_t c)
   }
 }
 
-static inline wchar_t wupcase (wchar_t c)
+wchar_t wupcase (wchar_t c)
 {
   int ch = c >> 8;
   if (ch >= 0 && ch < int (sizeof (lc_tab) / sizeof (lc_tab[0])) && lc_tab[ch]) {
@@ -66,7 +66,7 @@ static inline wchar_t wupcase (wchar_t c)
   }
 }
 
-static inline uint32_t utf32_downcase (uint32_t c32)
+uint32_t utf32_downcase (uint32_t c32)
 {
   if (sizeof (wchar_t) == 2 && c32 >= 0x10000) {
     return c32;
@@ -75,8 +75,7 @@ static inline uint32_t utf32_downcase (uint32_t c32)
   }
 }
 
-/* Not used yet:
-static inline uint32_t utf32_upcase (uint32_t c32)
+uint32_t utf32_upcase (uint32_t c32)
 {
   if (sizeof (wchar_t) == 2 && c32 >= 0x10000) {
     return c32;
@@ -84,12 +83,11 @@ static inline uint32_t utf32_upcase (uint32_t c32)
     return uint32_t (wupcase (wchar_t (c32)));
   }
 }
-*/
 
 // -------------------------------------------------------------------------
 //  Conversion of UTF8 to wchar_t
 
-uint32_t utf32_from_utf8 (const char *&cp, const char *cpe = 0)
+uint32_t utf32_from_utf8 (const char *&cp, const char *cpe)
 {
   uint32_t c32 = (unsigned char) *cp++;
   if (c32 >= 0xf0 && ((cpe && cp + 2 < cpe) || (! cpe && cp [0] && cp [1] && cp [2]))) {

src/tl/tl/tlString.h

@@ -878,8 +878,33 @@ TL_PUBLIC std::string trim (const std::string &s);
 TL_PUBLIC std::vector<std::string> split (const std::string &s, const std::string &sep);
 TL_PUBLIC std::string join (const std::vector<std::string> &strings, const std::string &sep);
 
+/**
+ *  @brief Returns the lower-case character for a wchar_t
+ */
+TL_PUBLIC wchar_t wdowncase (wchar_t c);
+
+/**
+ *  @brief Returns the upper-case character for a wchar_t
+ */
+TL_PUBLIC wchar_t wupcase (wchar_t c);
+
+/**
+ *  @brief Returns the lower-case UTF32 character
+ */
+TL_PUBLIC uint32_t utf32_downcase (uint32_t c32);
+
+/**
+ *  @brief Returns the upper-case UTF32 character
+ */
+TL_PUBLIC uint32_t utf32_upcase (uint32_t c32);
+
+/**
+ *  @brief Parses the next UTF32 character from an UTF-8 string
+ *  @param cp The input character's position, will be set to the next character.
+ *  @paran cpe The end of the string of 0 for "no end"
+ */
+TL_PUBLIC uint32_t utf32_from_utf8 (const char *&cp, const char *cpe = 0);
+
 } // namespace tl
 
 #endif
-
-