cdx export #941 (#1012)

* rgroups (#946) Co-authored-by: Roman Porozhnetov <roman_porozhnetov@epam.com> * CDX import: Reaction arrows disappear when opening a file #943 (#948) * marvin cdx fix * clang fix Co-authored-by: Roman Porozhnetov <roman_porozhnetov@epam.com> * bond order fix * clang fix * test fix * cdx export * clang fixes * cdx python binding * clang fixes * api fixes * cdx export * indigo py fix * cdx * cdx * cdx fixes * clang fix * PR comments * clang fix * pr fix * pr fix * b64cdx java&dotnet fix * b64cdx java&dotnet fix * b64cdx java&dotnet fix * diff fix * test fix * clang fix * mol_to_cdx.out deleted * clang fix * tests fix * tests fix * tests fix * tests fix * tests fix * tests fix * tests fix * tests fix * fix dir * fix dir * fix dir --------- Co-authored-by: Roman Porozhnetov <roman_porozhnetov@epam.com>
epam · Jan 30, 2023 · a7d0a74 · a7d0a74
1 parent 9760b99
commit a7d0a74
Show file tree

Hide file tree

Showing 105 changed files with 58,392 additions and 1,148 deletions.
diff --git a/.github/workflows/indigo-ci.yaml b/.github/workflows/indigo-ci.yaml
@@ -25,7 +25,9 @@ jobs:
           lfs: false
           fetch-depth: 500
       - name: Git fetch tags
-        run: git fetch --tags -f
+        run: |
+         git config --global --add safe.directory /__w/Indigo/Indigo
+         git fetch --tags -f
       - name: Check
         run: |
           python3 -m pip install -r api/python/requirements_dev.txt

diff --git a/api/c/indigo/indigo.h b/api/c/indigo/indigo.h
@@ -203,6 +203,7 @@ CEXPORT int indigoSaveJson(int item, int output);
 CEXPORT int indigoSaveCml(int object, int output);
 CEXPORT int indigoSaveCmlToFile(int object, const char* filename);
 CEXPORT const char* indigoCml(int object);
+CEXPORT const char* indigoCdxBase64(int object);
 
 // accepts molecules and reactions
 CEXPORT int indigoSaveCdxml(int object, int output);
@@ -1021,6 +1022,7 @@ CEXPORT int indigoExpandAbbreviations(int molecule);
 /* Other */
 
 CEXPORT const char* indigoToString(int handle);
+CEXPORT const char* indigoToBase64String(int handle);
 CEXPORT int indigoToBuffer(int handle, char** buf, int* size);
 
 /* Reaction products enumeration */

diff --git a/api/c/indigo/src/indigo_internal.h b/api/c/indigo/src/indigo_internal.h
@@ -200,6 +200,8 @@ class DLLEXPORT IndigoObject
     virtual MonomersProperties& getMonomersProperties();
     virtual void copyProperties(PropertiesMap&);
 
+    void toBase64String(Array<char>& str);
+
 private:
     IndigoObject(const IndigoObject&);
 };

diff --git a/api/c/indigo/src/indigo_io.cpp b/api/c/indigo/src/indigo_io.cpp
@@ -187,6 +187,20 @@ CEXPORT const char* indigoToString(int handle)
     INDIGO_END(0);
 }
 
+CEXPORT const char* indigoToBase64String(int handle)
+{
+    INDIGO_BEGIN
+    {
+        IndigoObject& obj = self.getObject(handle);
+
+        auto& tmp = self.getThreadTmpData();
+        obj.toBase64String(tmp.string);
+        tmp.string.push(0);
+        return tmp.string.ptr();
+    }
+    INDIGO_END(0);
+}
+
 CEXPORT int indigoToBuffer(int handle, char** buf, int* size)
 {
     INDIGO_BEGIN

diff --git a/api/c/indigo/src/indigo_macros.c b/api/c/indigo/src/indigo_macros.c
@@ -142,6 +142,22 @@ CEXPORT const char* indigoMolfile(int molecule)
     return res;
 }
 
+CEXPORT const char* indigoCdxBase64(int object)
+{
+    int b = indigoWriteBuffer();
+    const char* res;
+
+    if (b == -1)
+        return 0;
+
+    if (indigoSaveCdx(object, b) == -1)
+        return 0;
+
+    res = indigoToBase64String(b);
+    indigoFree(b);
+    return res;
+}
+
 CEXPORT const char* indigoCml(int molecule)
 {
     int b = indigoWriteBuffer();

diff --git a/api/c/indigo/src/indigo_object.cpp b/api/c/indigo/src/indigo_object.cpp
@@ -16,14 +16,14 @@
  * limitations under the License.
  ***************************************************************************/
 
-#include "indigo_internal.h"
-
-#include <map>
-
 #include "base_cpp/output.h"
 #include "base_cpp/properties_map.h"
+#include "indigo_internal.h"
 #include "reaction/reaction.h"
 
+#include <../cppcodec/cppcodec/base64_default_rfc4648.hpp>
+#include <map>
+
 using IndigoObjectTypesMap = std::map<int, const char* const>;
 class IndigoObjectTypes : public IndigoObjectTypesMap, public NonCopyable
 {
@@ -161,6 +161,14 @@ const char* IndigoObject::debugInfo() const
     return getTypeName();
 }
 
+void IndigoObject::toBase64String(Array<char>& str)
+{
+    Array<char> temp;
+    toString(temp);
+    auto encoded = base64::encode(temp.ptr(), temp.size());
+    str.copy(encoded.c_str(), encoded.size());
+}
+
 void IndigoObject::toString(Array<char>& str)
 {
     throw IndigoError("can not convert %s to string", debugInfo());

diff --git a/api/dotnet/src/IndigoLib.cs b/api/dotnet/src/IndigoLib.cs
@@ -164,6 +164,9 @@ public unsafe class IndigoLib
         [DllImport("indigo"), SuppressUnmanagedCodeSecurity]
         public static extern byte* indigoCdxml(int molecule);
 
+        [DllImport("indigo"), SuppressUnmanagedCodeSecurity]
+        public static extern byte* indigoCdxBase64(int molecule);
+
         [DllImport("indigo"), SuppressUnmanagedCodeSecurity]
         public static extern int indigoSaveMDLCT(int item, int output);
 

diff --git a/api/dotnet/src/IndigoObject.cs b/api/dotnet/src/IndigoObject.cs
@@ -112,6 +112,12 @@ public string cdxml()
             return dispatcher.checkResult(IndigoLib.indigoCdxml(self));
         }
 
+        public string b64cdx()
+        {
+            dispatcher.setSessionID();
+            return dispatcher.checkResult(IndigoLib.indigoCdxBase64(self));
+        }
+
         public void saveCdxml(string filename)
         {
             dispatcher.setSessionID();

diff --git a/api/java/indigo/src/main/java/com/epam/indigo/IndigoLib.java b/api/java/indigo/src/main/java/com/epam/indigo/IndigoLib.java
@@ -140,6 +140,8 @@ public interface IndigoLib extends Library {
 
     Pointer indigoCdxml(int object);
 
+    Pointer indigoCdxBase64(int object);
+
     int indigoSaveMDLCT(int item, int output);
 
     int indigoLoadReaction(int source);

diff --git a/api/java/indigo/src/main/java/com/epam/indigo/IndigoObject.java b/api/java/indigo/src/main/java/com/epam/indigo/IndigoObject.java
@@ -102,6 +102,11 @@ public String cdxml() {
         return Indigo.checkResultString(this, lib.indigoCdxml(self));
     }
 
+    public String b64cdx() {
+        dispatcher.setSessionID();
+        return Indigo.checkResultString(this, lib.indigoCdxBase64(self));
+    }
+
     public void saveCdxml(String filename) {
         dispatcher.setSessionID();
         Indigo.checkResult(this, lib.indigoSaveCdxmlToFile(self, filename));

diff --git a/api/python/indigo/indigo/indigo_lib.py b/api/python/indigo/indigo/indigo_lib.py
@@ -254,8 +254,12 @@ def __init__(self) -> None:
         IndigoLib.lib.indigoCml.argtypes = [c_int]
         IndigoLib.lib.indigoSaveCdxmlToFile.restype = c_int
         IndigoLib.lib.indigoSaveCdxmlToFile.argtypes = [c_int, c_char_p]
+        IndigoLib.lib.indigoSaveCdxToFile.restype = c_int
+        IndigoLib.lib.indigoSaveCdxToFile.argtypes = [c_int, c_char_p]
         IndigoLib.lib.indigoCdxml.restype = c_char_p
         IndigoLib.lib.indigoCdxml.argtypes = [c_int]
+        IndigoLib.lib.indigoCdxBase64.restype = c_char_p
+        IndigoLib.lib.indigoCdxBase64.argtypes = [c_int]
         IndigoLib.lib.indigoJson.restype = c_char_p
         IndigoLib.lib.indigoJson.argtypes = [c_int]
         IndigoLib.lib.indigoSaveMDLCT.restype = c_int

diff --git a/api/python/indigo/indigo/indigo_object.py b/api/python/indigo/indigo/indigo_object.py
@@ -296,6 +296,20 @@ def saveCdxml(self, filename):
             self._lib().indigoSaveCdxmlToFile(self.id, filename.encode())
         )
 
+    def saveCdx(self, filename):
+        """Molecule method saves the structure into a CDX file
+
+        Args:
+            filename (str): full path to the output file
+
+        Returns:
+            int: 1 if the file is saved successfully
+        """
+
+        return IndigoLib.checkResult(
+            self._lib().indigoSaveCdxToFile(self.id, filename.encode())
+        )
+
     def cdxml(self):
         """Molecule method returns the structure as a string in CDXML format
 
@@ -305,6 +319,17 @@ def cdxml(self):
 
         return IndigoLib.checkResultString(self._lib().indigoCdxml(self.id))
 
+    def b64cdx(self):
+        """Molecule method returns the structure as a string in CDX base64 encoded format
+
+        Returns:
+            str: base64 encoded CDX string
+        """
+
+        return IndigoLib.checkResultString(
+            self._lib().indigoCdxBase64(self.id)
+        )
+
     def json(self):
         """Structure method returns the structure as a string in KET format
 

diff --git a/api/tests/integration/ref/formats/cdx_export.py.out b/api/tests/integration/ref/formats/cdx_export.py.out
@@ -0,0 +1,5 @@
+*** Mol to CDX ***
+enhanced_stereo1:success
+enhanced_stereo2:success
+enhanced_stereo3:success
+stereo_either-0020:success
diff --git a/api/tests/integration/tests/formats/cdx_export.py b/api/tests/integration/tests/formats/cdx_export.py
@@ -0,0 +1,54 @@
+import difflib
+import os
+import sys
+
+
+def find_diff(a, b):
+    return "\n".join(difflib.unified_diff(a.splitlines(), b.splitlines()))
+
+
+sys.path.append(
+    os.path.normpath(
+        os.path.join(os.path.abspath(__file__), "..", "..", "..", "common")
+    )
+)
+from env_indigo import *  # noqa
+
+indigo = Indigo()
+
+print("*** Mol to CDX ***")
+
+root = joinPathPy("molecules/", __file__)
+files = [
+    "stereo_either-0020",
+    "enhanced_stereo1",
+    "enhanced_stereo2",
+    "enhanced_stereo3",
+]
+
+ref_path = joinPathPy("ref/", __file__)
+files.sort()
+
+for filename in files:
+    try:
+        mol = indigo.loadMoleculeFromFile(
+            os.path.join(root, filename + ".mol")
+        )
+        resb64 = mol.b64cdx()
+        # with open(os.path.join(ref_path, filename + ".cdxml"), 'w') as file:
+        #    data = file.write(mol.cdxml())
+        with open(os.path.join(ref_path, filename + ".b64cdx"), "r") as file:
+            refb64 = file.read()
+        print(filename + (":success" if refb64 == resb64 else ":failed"))
+    except IndigoException as e:
+        print(getIndigoExceptionText(e))
+        print("*** Try as Query ***")
+        mol = indigo.loadQueryMoleculeFromFile(
+            os.path.join(root, filename + ".mol")
+        )
+        resb64 = mol.b64cdx()
+        # with open(os.path.join(ref_path, filename + ".cdxml"), 'w') as file:
+        #   data = file.write(mol.cdxml())
+        with open(os.path.join(ref_path, filename + ".b64cdx"), "r") as file:
+            refb64 = file.read()
+        print(filename + (":success" if refb64 == resb64 else ":failed"))