polymer.protein polymer.nucleic multifilesave ...

* new selection keywords: polymer.protein polymer.nucleic
* new command: multifilesave
* MMTF export (requires simplemmtf python module)
* select: present, state -1 -> current object state
* create: source_state = -1 -> current
  (was: source_state = -1/0 -> all)
* cif: support quoted '?' and '.' as values
  cif_get_array -> None instead of empty string for ? and .
* fix copy discrete object w/o coords
* fix loading pdbqr AutoDock 4 atom types
* fix get_model('none') and get_str('mol', 'none')
* h_add refactored: 5x faster, support discrete objects

ChangeLog

@@ -2,6 +2,30 @@
 CHANGE LOG
 =========================================================================
 
+2018-02-07 Thomas Holder <[email protected]>
+
+	* new selection keywords: polymer.protein polymer.nucleic
+
+	* new command: multifilesave
+
+	* MMTF export (requires simplemmtf python module)
+
+	* select: present, state -1 -> current object state
+
+	* create: source_state = -1 -> current
+	  (was: source_state = -1/0 -> all)
+
+	* cif: support quoted '?' and '.' as values
+	  cif_get_array -> None instead of empty string for ? and .
+
+	* fix copy discrete object w/o coords
+
+	* fix loading pdbqr AutoDock 4 atom types
+
+	* fix get_model('none') and get_str('mol', 'none')
+
+	* h_add refactored: 5x faster, support discrete objects
+
 2017-12-21 Thomas Holder, Piotr Rotkiewicz
 
 	* Python 3.7 support (https://sf.net/p/pymol/bugs/197/)

contrib/mmtf-c/mmtf_parser.cpp

@@ -304,6 +304,10 @@ void MMTF_container_destroy(MMTF_container* thing) {
     free(thing->sequenceIndexList);
     free(thing->groupsPerChain);
     free(thing->chainsPerModel);
+
+    // PyMOL
+    free(thing->pymolRepsList);
+    free(thing->pymolColorList);
 }
 void MMTF_BioAssembly_destroy(MMTF_BioAssembly* bio_assembly) {
     IF_NULL_PTRERROR_RETURN(bio_assembly, );
@@ -962,6 +966,11 @@ bool MMTF_unpack_from_msgpack_object(const msgpack_object* object, MMTF_containe
     FETCH_AND_ASSIGN_DUMMYCOUNT(thing, int32_array, groupsPerChain);
     FETCH_AND_ASSIGN_WITHCOUNT(thing, string_array, experimentalMethods);
     FETCH_AND_ASSIGN_ARRAY(thing, float, unitCell);
+
+    // PyMOL
+    FETCH_AND_ASSIGN_DUMMYCOUNT(thing, int32_array, pymolRepsList);
+    FETCH_AND_ASSIGN_DUMMYCOUNT(thing, int32_array, pymolColorList);
+
     MAP_ITERATE_END();
 
     return true;

contrib/mmtf-c/mmtf_parser.h

@@ -137,6 +137,10 @@ typedef struct {
     char** WITHCOUNT(chainNameList);
     int32_t* groupsPerChain;
     int32_t* chainsPerModel;
+
+    // PyMOL
+    int32_t* pymolRepsList;
+    int32_t* pymolColorList;
 } MMTF_container;
 
 /*

layer0/strcasecmp.h

@@ -0,0 +1,13 @@
+/*
+ * strcasecmp, strncasecmp
+ */
+
+#pragma once
+
+#ifdef _WIN32
+  #include <string.h>
+  #define strcasecmp(s1, s2) _stricmp(s1, s2)
+  #define strncasecmp(s1, s2, n) _strnicmp(s1, s2, n)
+#else
+  #include <strings.h>
+#endif

layer1/PConv.h

@@ -213,6 +213,11 @@ inline PyObject * PConvToPyObject(const std::string &v) {
 }
 
 inline PyObject * PConvToPyObject(const char * v) {
+#ifndef _PYMOL_NOPY
+  if (!v) {
+    Py_RETURN_NONE;
+  }
+#endif
   return PyString_FromString(v);
 }
 

layer1/PyMOLObject.cpp

@@ -671,7 +671,16 @@ void ObjectMakeValidName(PyMOLGlobals * G, char *name)
   ObjectMakeValidName(name);
 
   if (SelectorNameIsKeyword(G, name)) {
+    PRINTFB(G, FB_Executive, FB_Warnings)
+      " Warning: '%s' is a reserved keyword, appending underscore\n", name
+      ENDFB(G);
     strcat(name, "_");
+  } else if (strcmp(name, "protein") == 0 || strcmp(name, "nucleic") == 0) {
+    // Warn the user if "protein" or "nucleic" are used as names, but
+    // don't modify the name (yet).
+    PRINTFB(G, FB_Executive, FB_Warnings)
+      " Warning: '%s' may become a reserved selection keyword in the future\n", name
+      ENDFB(G);
   }
 }
 

layer1/Scene.cpp

@@ -1245,10 +1245,10 @@ void SceneSetStereo(PyMOLGlobals * G, int flag)
     I->StereoMode = 0;
   }
 
-  if((cur_stereo != I->StereoMode) && ((cur_stereo == 4) || (I->StereoMode == 4))) {
+  if((cur_stereo != I->StereoMode) && (cur_stereo == cStereo_geowall || I->StereoMode == cStereo_geowall)) {
     OrthoReshape(G, G->Option->winX, G->Option->winY, true);
 #ifndef _PYMOL_NOPY
-    if((cur_stereo == 4) && I->StereoMode) {
+    if(cur_stereo == cStereo_geowall) {
       PParse(G, "viewport");
     }
 #endif

layer1/Setting.cpp

@@ -2891,7 +2891,7 @@ void SettingInitGlobal(PyMOLGlobals * G, int alloc, int reset_gui, int use_defau
     if(G->Option->stereo_mode) {
       set_i(I, cSetting_stereo_mode, G->Option->stereo_mode);
     } else if(G->StereoCapable || G->Option->blue_line) {
-      set_i(I, cSetting_stereo_mode, 1);      /* quadbuffer if we can */
+      set_i(I, cSetting_stereo_mode, cStereo_quadbuffer);      /* quadbuffer if we can */
     }
 
     /* In order to get electrostatic potentials in kT from the Coulomb equation... 

layer2/AtomInfo.cpp

@@ -1260,7 +1260,6 @@ int AtomInfoUniquefyNames(PyMOLGlobals * G, const AtomInfoType * atInfo0, int n0
       } else {
         sprintf(name, "%1d%1s%02d", c / 100, ai1->elem, c % 100);
       }
-      name[4] = 0;              /* just is case we go over */
       LexAssign(G, ai1->name, name);
       result++;
       c = c + 1;

layer2/CifFile.cpp

@@ -12,11 +12,11 @@
 
 #include <vector>
 #include <iostream>
-#include <stdexcept>
 
 #include "CifFile.h"
 #include "File.h"
 #include "MemoryDebug.h"
+#include "strcasecmp.h"
 
 // basic IO and string handling
 
@@ -90,24 +90,38 @@ static void tolowerinplace(char *p) {
 static const char * EMPTY_STRING = "";
 static cif_array EMPTY_ARRAY(NULL);
 
+/*
+ * Class to store CIF loops. Only for parsing, do not use in any higher level
+ * reading functions.
+ */
+class cif_loop {
+public:
+  int ncols;
+  int nrows;
+  const char **values;
+
+  // methods
+  const char * get_value_raw(int row, int col) const;
+};
+
 // get table value, return NULL if indices out of bounds
 const char * cif_loop::get_value_raw(int row, int col) const {
   if (row >= nrows)
     return NULL;
   return values[row * ncols + col];
 }
 
+// get the number of elements in this array
+int cif_array::get_nrows() const {
+  return (col < 0) ? 1 : pointer.loop->nrows;
+}
+
 // get array value, return NULL if row-index out of bounds
-const char * cif_array::get_value_raw(int row) const {
+// or value in ['.', '?']
+const char * cif_array::get_value(int row) const {
   if (col < 0)
     return (row > 0) ? NULL : pointer.value;
   return pointer.loop->get_value_raw(row, col);
-};
-
-// get array value, return NULL if value in ['.', '?']
-const char * cif_array::get_value(int row) const {
-  const char * s = get_value_raw(row);
-  return (s && (s[0] == '?' || s[0] == '.') && !s[1]) ? NULL : s;
 }
 
 // get array value, return an empty string if missing
@@ -141,7 +155,7 @@ bool cif_array::is_missing_all() const {
 }
 
 // templated getters
-template <> const char* cif_array::as<const char* >(int row) const { return as_s(row); }
+template <> const char* cif_array::as<const char* >(int row) const { return get_value(row); }
 template <> std::string cif_array::as<std::string >(int row) const { return as_s(row); }
 template <> int         cif_array::as<int         >(int row) const { return as_i(row); }
 template <> double      cif_array::as<double      >(int row) const { return as_d(row); }
@@ -228,7 +242,7 @@ bool cif_file::parse() {
   char quote;
   char prev = '\0';
 
-  std::vector<char> codes;
+  std::vector<bool> keypossible;
 
   // tokenize
   while (true) {
@@ -243,14 +257,14 @@ bool cif_file::parse() {
       prev = *p;
     } else if (isquote(*p)) { // will NULL the closing quote
       quote = *p;
-      codes.push_back('Q');
+      keypossible.push_back(false);
       tokens.push_back(p + 1);
       while (*++p && !(*p == quote && iswhitespace0(p[1])));
       if (*p)
         *(p++) = 0;
       prev = *p;
     } else if (*p == ';' && islinefeed(prev)) { // will NULL the line feed before the closing semicolon
-      codes.push_back('Q');
+      keypossible.push_back(false);
       tokens.push_back(p + 1);
       while (*++p && !(islinefeed(*p) && p[1] == ';'));
       if (*p) {
@@ -259,23 +273,35 @@ bool cif_file::parse() {
       }
       prev = ';';
     } else { // will null the whitespace
-      codes.push_back('R');
-      tokens.push_back(p);
+      char * q = p++;
       while (!iswhitespace0(*p)) ++p;
       prev = *p;
-      if (*p)
-        *(p++) = 0;
+      if (p - q == 1 && (*q == '?' || *q == '.')) {
+        // store values '.' (inapplicable) and '?' (unknown) as null-pointers
+        q = NULL;
+        keypossible.push_back(false);
+      } else {
+        if (*p)
+          *(p++) = 0;
+        keypossible.push_back(true);
+      }
+      tokens.push_back(q);
     }
   }
 
   cif_data *current_data = NULL, *current_frame = NULL, *global_block = NULL;
 
   // parse into dictionary
   for (unsigned int i = 0, n = tokens.size(); i < n; i++) {
-    if (codes[i] == 'Q') {
+    if (!keypossible[i]) {
       std::cout << "ERROR" << std::endl;
       break;
     } else if (tokens[i][0] == '_') {
+      if (i + 1 == n) {
+        std::cout << "ERROR truncated" << std::endl;
+        break;
+      }
+
       if (current_frame) {
         tolowerinplace(tokens[i]);
         current_frame->dict[tokens[i]].set_value(tokens[i + 1]);
@@ -296,7 +322,7 @@ bool cif_file::parse() {
       }
 
       // columns
-      while (++i < n && codes[i] != 'Q' && tokens[i][0] == '_') {
+      while (++i < n && keypossible[i] && tokens[i][0] == '_') {
         tolowerinplace(tokens[i]);
 
         if (current_frame) {
@@ -313,8 +339,14 @@ bool cif_file::parse() {
       }
 
       // rows
-      while (i < n && (codes[i] == 'Q' || !isspecial(tokens[i]))) {
+      while (i < n && !(keypossible[i] && isspecial(tokens[i]))) {
         i += ncols;
+
+        if (i > n) {
+          std::cout << "ERROR truncated loop" << std::endl;
+          break;
+        }
+
         nrows++;
       }
 
@@ -334,7 +366,7 @@ bool cif_file::parse() {
       global_block = current_data = current_frame = new cif_data;
 
     } else if (strncasecmp("save_", tokens[i], 5) == 0) {
-      if (tokens[i][5]) {
+      if (tokens[i][5] && current_data) {
         // begin
         const char * key(tokens[i] + 5);
         current_data->saveframes[key] = current_frame = new cif_data;

layer2/CifFile.h

@@ -9,15 +9,9 @@
 
 #include <vector>
 #include <map>
-#include <stdexcept>
 
 #include <string.h>
 
-#ifdef WIN32
-  #define strcasecmp(s1, s2) _stricmp(s1, s2)
-  #define strncasecmp(s1, s2, n) _strnicmp(s1, s2, n)
-#endif
-
 /*
  * C string comparison class
  */
@@ -59,24 +53,12 @@ class cif_file {
   bool parse();
 };
 
-/*
- * Class to store CIF loops. Only for parsing, do not use in any higher level
- * reading functions.
- */
-class cif_loop {
-public:
-  int ncols;
-  int nrows;
-  const char **values;
-
-  // methods
-  const char * get_value_raw(int row, int col) const;
-};
-
 /*
  * High-level access to CIF arrays
  */
 class cif_array {
+  friend class cif_file;
+
 private:
   // column index, -1 if not in loop
   short col;
@@ -88,10 +70,8 @@ class cif_array {
   } pointer;
 
   // methods
-  const char * get_value_raw(int row = 0) const;
   const char * get_value(int row = 0) const;
 
-public:
   // point this array to a loop (only for parsing)
   void set_loop(const cif_loop * loop, short col_) {
     col = col_;
@@ -104,6 +84,7 @@ class cif_array {
     pointer.value = value;
   };
 
+public:
   // constructor
   cif_array() {
   };
@@ -114,9 +95,7 @@ class cif_array {
   };
 
   // get the number of elements in this array
-  int get_nrows() const {
-    return (col < 0) ? 1 : pointer.loop->nrows;
-  };
+  int get_nrows() const;
 
   // get element as string, integer or double. If index is out of bounds,
   // then return a default value.