Merge branch 'database' of github.com:CalebBell/chemicals into database

chemicals/acentric.py

@@ -52,6 +52,7 @@
            'omega_methods', 'omega_all_methods', 'omega_definition']
 
 from chemicals import critical
+from chemicals import data_reader as dr
 from chemicals.data_reader import (list_available_methods_from_df_dict,
                                    database_constant_lookup,
                                    retrieve_any_from_df_dict,
@@ -150,8 +151,9 @@ def omega(CASRN, method=None):
        Hydrocarbons, Second Edition. Amsterdam Boston: Gulf Professional
        Publishing, 2014.
     '''
-    val, found = database_constant_lookup(CASRN, 'omega') if method is None else (None, False)
-    if found: return val
+    if dr.USE_CONSTANTS_DATABASE and method is None:
+        val, found = database_constant_lookup(CASRN, 'omega')
+        if found: return val
     if method:
         return retrieve_from_df_dict(critical.omega_sources, CASRN, 'omega', method)
     else:

chemicals/combustion.py

@@ -75,7 +75,7 @@
 """
 
 from fluids.numerics import normalize
-
+from chemicals import data_reader as dr
 from chemicals.data_reader import (data_source, database_constant_lookup,
                                    list_available_methods_from_df_dict,
                                    register_df_source,
@@ -260,8 +260,9 @@ def RON(CASRN, method=None):
        Learning Projects for Fuel Property Prediction." Journal of Open Source 
        Software 2, no. 17 (2017): 401.
     '''
-    val, found = database_constant_lookup(CASRN, 'RON') if method is None else (None, False)
-    if found: return val
+    if dr.USE_CONSTANTS_DATABASE and method is None:
+        val, found = database_constant_lookup(CASRN, 'RON')
+        if found: return val
     if not _combustion_data_loaded: _load_combustion_data()
     if method:
         value = retrieve_from_df_dict(RON_sources, CASRN, 'RON', method)
@@ -354,8 +355,9 @@ def MON(CASRN, method=None):
        Learning Projects for Fuel Property Prediction." Journal of Open Source 
        Software 2, no. 17 (2017): 401.
     '''
-    val, found = database_constant_lookup(CASRN, 'MON') if method is None else (None, False)
-    if found: return val
+    if dr.USE_CONSTANTS_DATABASE and method is None:
+        val, found = database_constant_lookup(CASRN, 'MON')
+        if found: return val
     if not _combustion_data_loaded: _load_combustion_data()
     if method:
         value = retrieve_from_df_dict(MON_sources, CASRN, 'MON', method)
@@ -387,8 +389,9 @@ def ignition_delay_methods(CASRN):
     --------
     ignition_delay
     """
-    val, found = database_constant_lookup(CASRN, 'IGNITION_DELAY') if method is None else (None, False)
-    if found: return val
+    if dr.USE_CONSTANTS_DATABASE:
+        val, found = database_constant_lookup(CASRN, 'IGNITION_DELAY')
+        if found: return val
     if not _combustion_data_loaded: _load_combustion_data()
     return list_available_methods_from_df_dict(ignition_delay_sources, CASRN, 'IGNITION_DELAY')
 

chemicals/critical.py

@@ -100,7 +100,7 @@
 import os
 
 from fluids.constants import N_A, R, R_inv
-
+from chemicals import data_reader as dr
 from chemicals import miscdata
 from chemicals.data_reader import (data_source,
                                    list_available_methods_from_df_dict,
@@ -405,8 +405,9 @@ def Tc(CASRN, method=None):
        Estimation Based on Zero, First and Second Order Methods." In 
        Proceedings of the AIChE Spring Meeting, 21, 1996.
     '''
-    val, found = database_constant_lookup(CASRN, 'Tc') if method is None else (None, False)
-    if found: return val
+    if dr.USE_CONSTANTS_DATABASE and method is None:
+        val, found = database_constant_lookup(CASRN, 'Tc')
+        if found: return val
     if not _critical_data_loaded: _load_critical_data()
     if method:
         return retrieve_from_df_dict(Tc_sources, CASRN, 'Tc', method)
@@ -583,8 +584,9 @@ def Pc(CASRN, method=None):
        Estimation Based on Zero, First and Second Order Methods." In 
        Proceedings of the AIChE Spring Meeting, 21, 1996.
     '''
-    val, found = database_constant_lookup(CASRN, 'Pc') if method is None else (None, False)
-    if found: return val
+    if dr.USE_CONSTANTS_DATABASE and method is None:
+        val, found = database_constant_lookup(CASRN, 'Pc')
+        if found: return val
     if not _critical_data_loaded: _load_critical_data()
     if method:
         return retrieve_from_df_dict(Pc_sources, CASRN, 'Pc', method)
@@ -755,8 +757,9 @@ def Vc(CASRN, method=None):
     .. [20] Fedors, R. F. "A Method to Estimate Critical Volumes." AIChE 
        Journal 25, no. 1 (1979): 202-202. https://doi.org/10.1002/aic.690250129.
     '''
-    val, found = database_constant_lookup(CASRN, 'Vc') if method is None else (None, False)
-    if found: return val
+    if dr.USE_CONSTANTS_DATABASE and method is None:
+        val, found = database_constant_lookup(CASRN, 'Vc')
+        if found: return val
     if not _critical_data_loaded: _load_critical_data()
     if method:
         return retrieve_from_df_dict(Vc_sources, CASRN, 'Vc', method)
@@ -924,8 +927,9 @@ def Zc(CASRN, method=None):
        Tc-PR." AIChE Journal n/a, no. n/a (n.d.): e17518. 
        https://doi.org/10.1002/aic.17518.
     '''
-    val, found = database_constant_lookup(CASRN, 'Zc') if method is None else (None, False)
-    if found: return val
+    if dr.USE_CONSTANTS_DATABASE and method is None:
+        val, found = database_constant_lookup(CASRN, 'Zc')
+        if found: return val
     if not _critical_data_loaded: _load_critical_data()
     if method:
         return retrieve_from_df_dict(Zc_sources, CASRN, 'Zc', method)

chemicals/data_reader.py

@@ -78,8 +78,7 @@ def make_df_sparse(df, non_sparse_columns=[]):
 
 def register_df_source(folder, name, sep='\t', index_col=0, csv_kwargs=None,
                        postload=None, sparsify=False, int_CAS=False):
-    if csv_kwargs is None:
-        csv_kwargs = {}
+    if csv_kwargs is None: csv_kwargs = {}
     load_cmds[name] = (folder, name, sep, index_col, csv_kwargs, postload, sparsify, int_CAS)
 
 '''The following flags will strip out the excess memory usage of redundant 
@@ -166,7 +165,7 @@ def retrieve_from_df(df, index, key):
     df_index = df.index
     if df_index.dtype is int64_dtype and isinstance(index, str):
         try: index = CAS_to_int(index)
-        except: pass
+        except: return None
     if index in df_index:
         if isinstance(key, (int, str)):
             return get_value_from_df(df, index, key)
@@ -177,7 +176,7 @@ def retrieve_any_from_df(df, index, keys):
     df_index = df.index
     if df_index.dtype is int64_dtype and isinstance(index, str):
         try: index = CAS_to_int(index)
-        except: pass
+        except: return None
     if index not in df.index: return None
     for key in keys:
         value = df.at[index, key]
@@ -191,24 +190,23 @@ def get_value_from_df(df, index, key):
     value = df.at[index, key]
     try:
         return None if isnan(value) else float(value)
-    except TypeError:
-        # Not a number
+    except TypeError: # Not a number
         return value
 
 def list_available_methods_from_df_dict(df_dict, index, key):
     methods = []
-    int_index = None
+    int_index = None if isinstance(index, str) else index # Assume must be string or int
     for method, df in df_dict.items():
         df_index = df.index
-        if df_index.dtype is int64_dtype and isinstance(index, str):
+        if df_index.dtype is int64_dtype:
             if int_index is None:
                 try:
                     int_index = CAS_to_int(index)
                 except:
                     int_index = 'skip'
-            elif int_index == 'skip':
-                continue
-            if (int_index in df_index) and not isnan(df.at[int_index, key]):
+                else:
+                    (int_index in df_index) and not isnan(df.at[int_index, key]) and methods.append(method)
+            elif int_index != 'skip' and (int_index in df_index) and not isnan(df.at[int_index, key]):
                 methods.append(method)
         elif (index in df_index) and not isnan(df.at[index, key]):
             methods.append(method)
@@ -239,41 +237,40 @@ def list_available_methods_from_df(df, index, keys_by_method):
 
 DATABASE_CONSTANTS_CACHE = {}
 def cached_constant_lookup(CASi, prop):
-    if CONSTANTS_CURSOR is None:
-        init_constants_db()
-    prop_idx = CONSTANT_DATABASE_NAME_TO_IDX[prop]
-    try:
-        return DATABASE_CONSTANTS_CACHE[CASi][prop_idx], True
-    except KeyError:
-        pass
-    # Fetch and store the whole row
-    CONSTANTS_CURSOR.execute("SELECT * FROM constants WHERE `index`=?", (str(CASi),))
-    result = CONSTANTS_CURSOR.fetchone()
-    DATABASE_CONSTANTS_CACHE[CASi] = result
-    if result is not None:
+    if CONSTANTS_CURSOR is None: init_constants_db()
+    if CASi in DATABASE_CONSTANTS_CACHE:
+        result = DATABASE_CONSTANTS_CACHE[CASi]
+    else:
+        # Fetch and store the whole row
+        CONSTANTS_CURSOR.execute("SELECT * FROM constants WHERE `index`=?", (str(CASi),))
+        result = CONSTANTS_CURSOR.fetchone()
+        DATABASE_CONSTANTS_CACHE[CASi] = result
+    if result is None:
+        # Result the value, and whether the compound was in the index
+        return result, False
+    else:
+        prop_idx = CONSTANT_DATABASE_NAME_TO_IDX[prop]
         return result[prop_idx], True
-
-    # Result the value, and whether the compound was in the index
-    return result, False
 
 def init_constants_db():
     global CONSTANTS_CURSOR
     import sqlite3
     conn = sqlite3.connect(path_join(source_path, 'Misc', 'default.sqlite'))
     CONSTANTS_CURSOR = conn.cursor()
 
-def database_constant_lookup(CAS, prop):
-    if not USE_CONSTANTS_DATABASE:
-        return None, False
-    try:
-        CASi = CAS_to_int(CAS)
-    except:
-        if type(CAS) is not int:
+def database_constant_lookup(CASi, prop):
+    if isinstance(CASi, str): # Assume it must be either an int or string
+        try:
+            CASi = CAS_to_int(CASi)
+        except:
             return None, False
-        else:
-            # Was already an int
-            CASi = CAS
     try:
         return cached_constant_lookup(CASi, prop)
-    except:
+    except (TypeError, KeyError) as e:
+        raise e from None
+    except Exception:
+        # Prevent database lookup after first failure considering it should work everytime.
+        # It will possibly fail for users every time if database has not been created.
+        global USE_CONSTANTS_DATABASE
+        USE_CONSTANTS_DATABASE = False
         return None, False

chemicals/dipole.py

@@ -39,10 +39,11 @@
            'dipole_moment_methods',
            'dipole_moment_all_methods']
 
+from chemicals import data_reader as dr
 from chemicals.data_reader import (data_source, database_constant_lookup,
+                                   retrieve_any_from_df_dict,
                                    list_available_methods_from_df_dict,
                                    register_df_source,
-                                   retrieve_any_from_df_dict,
                                    retrieve_from_df_dict)
 from chemicals.miscdata import PSI4_2022A
 from chemicals.utils import (PY37, can_load_data, mark_numba_incompatible,
@@ -186,8 +187,9 @@ def dipole_moment(CASRN, method=None):
        Program." WIREs Computational Molecular Science 2, no. 4 (2012): 556-65. 
        https://doi.org/10.1002/wcms.93.
     '''
-    val, found = database_constant_lookup(CASRN, 'dipole_moment') if method is None else (None, False)
-    if found: return val
+    if dr.USE_CONSTANTS_DATABASE and method is None:
+        val, found = database_constant_lookup(CASRN, 'dipole_moment')
+        if found: return val
     if not _dipole_data_loaded: _load_dipole_data()
     if method:
         return retrieve_from_df_dict(dipole_sources, CASRN, 'dipole_moment', method)

chemicals/environment.py

@@ -54,6 +54,7 @@
 __all__ = ['GWP', 'ODP', 'logP',
            'GWP_all_methods', 'ODP_all_methods', 'logP_all_methods',
            'GWP_methods', 'ODP_methods', 'logP_methods']
+from chemicals import data_reader as dr
 from chemicals import miscdata
 from chemicals.data_reader import (
     data_source, list_available_methods_from_df, database_constant_lookup,
@@ -248,8 +249,9 @@ def GWP(CASRN, method=None):
        Other Weak Atmospheric Absorbers." Reviews of Geophysics 58, no. 3 
        (2020): e2019RG000691. https://doi.org/10.1029/2019RG000691.
     '''
-    val, found = database_constant_lookup(CASRN, 'GWP') if method is None else (None, False)
-    if found: return val
+    if dr.USE_CONSTANTS_DATABASE and method is None:
+        val, found = database_constant_lookup(CASRN, 'GWP')
+        if found: return val
     if not _GWP_ODP_data_loaded: _load_GWP_ODP_data()
     if method:
         if method in _IPCC_2014_GWP_keys_by_method:
@@ -364,8 +366,9 @@ def ODP(CASRN, method=None):
        Project-Report No. 52, Geneva, Switzerland, 516 p.
        https://www.wmo.int/pages/prog/arep/gaw/ozone_2010/documents/Ozone-Assessment-2010-complete.pdf
     '''
-    val, found = database_constant_lookup(CASRN, 'ODP') if method is None else (None, False)
-    if found: return val
+    if dr.USE_CONSTANTS_DATABASE and method is None:
+        val, found = database_constant_lookup(CASRN, 'ODP')
+        if found: return val
     if not _GWP_ODP_data_loaded: _load_GWP_ODP_data()
     if method:
         key = _ODP_keys_by_method[method]
@@ -448,8 +451,9 @@ def logP(CASRN, method=None):
     .. [2] Haynes, W.M., Thomas J. Bruno, and David R. Lide. CRC Handbook of
        Chemistry and Physics, 95E. Boca Raton, FL: CRC press, 2014.
     '''
-    val, found = database_constant_lookup(CASRN, 'logP') if method is None else (None, False)
-    if found: return val
+    if dr.USE_CONSTANTS_DATABASE and method is None:
+        val, found = database_constant_lookup(CASRN, 'logP')
+        if found: return val
     if not _logP_data_loaded: _load_logP_data()
     if method:
         return retrieve_from_df_dict(logP_sources, CASRN, 'logP', method)

chemicals/lennard_jones.py

@@ -95,7 +95,7 @@
 import os
 
 from fluids.constants import k
-
+from chemicals import data_reader as dr
 from chemicals.data_reader import (data_source, database_constant_lookup,
                                    list_available_methods_from_df_dict,
                                    register_df_source,
@@ -252,8 +252,9 @@ def Stockmayer(CASRN='', Tm=None, Tb=None, Tc=None, Zc=None, omega=None,
        Ranges of Temperature and Density." The Journal of Supercritical Fluids
        76 (April 2013): 94-114. doi:10.1016/j.supflu.2013.02.002.
     '''
-    val, found = database_constant_lookup(CASRN, 'Stockmayer') if method is None else (None, False)
-    if found: return val
+    if dr.USE_CONSTANTS_DATABASE and method is None:
+        val, found = database_constant_lookup(CASRN, 'Stockmayer')
+        if found: return val
     if not _LJ_data_loaded: _load_LJ_data()
     if method is not None:
         if method == FLYNN:
@@ -411,8 +412,9 @@ def molecular_diameter(CASRN=None, Tc=None, Pc=None, Vc=None, Zc=None, omega=Non
        Ranges of Temperature and Density." The Journal of Supercritical Fluids
        76 (April 2013): 94-114. doi:10.1016/j.supflu.2013.02.002.
     '''
-    val, found = database_constant_lookup(CASRN, 'molecular_diameter') if method is None else (None, False)
-    if found: return val
+    if dr.USE_CONSTANTS_DATABASE and method is None:
+        val, found = database_constant_lookup(CASRN, 'molecular_diameter')
+        if found: return val
     if not _LJ_data_loaded: _load_LJ_data()
     if method is not None:
         if method == FLYNN:

chemicals/molecular_geometry.py

@@ -43,7 +43,7 @@
            'linear', 'linear_methods', 'linear_all_methods']
 
 from fluids.numerics import horner, interp
-
+from chemicals import data_reader as dr
 from chemicals.data_reader import (data_source, database_constant_lookup,
                                    list_available_methods_from_df_dict,
                                    register_df_source,
@@ -171,8 +171,9 @@ def RG(CASRN, method=None):
     .. [2] Kooijman, Harry A., and Ross Taylor. The ChemSep Book. Books on 
        Demand Norderstedt, Germany, 2000.
     '''
-    val, found = database_constant_lookup(CASRN, 'RG') if method is None else (None, False)
-    if found: return val
+    if dr.USE_CONSTANTS_DATABASE and method is None:
+        val, found = database_constant_lookup(CASRN, 'RG')
+        if found: return val
     if not _RG_data_loaded: _load_RG_data()
     if method:
         value = retrieve_from_df_dict(RG_sources, CASRN, 'RG', method)
@@ -256,8 +257,9 @@ def linear(CASRN, method=None):
        Program." WIREs Computational Molecular Science 2, no. 4 (2012): 556-65. 
        https://doi.org/10.1002/wcms.93.
     '''
-    val, found = database_constant_lookup(CASRN, 'linear') if method is None else (None, False)
-    if found: return bool(val)
+    if dr.USE_CONSTANTS_DATABASE and method is None:
+        val, found = database_constant_lookup(CASRN, 'linear')
+        if found: return bool(val)
     if not _RG_data_loaded: _load_RG_data()
     if method:
         value = retrieve_from_df_dict(linear_sources, CASRN, 'linear', method)