Fix CN histogram x-axis

CHANGELOG

@@ -1,3 +1,7 @@
+## [1.0.2] - 2024-07-09
+
+
+
 ## [1.0.1] - 2024-07-05
 
 ### Fixed

docs/notebooks/02_cif_ensemble.ipynb

@@ -301,9 +301,6 @@
     "# Get file count per supercell atom count\n",
     "print(\"Supercell size stats:\", ensemble.supercell_size_stats)\n",
     "\n",
-    "# Get file count per min distance\n",
-    "print(\"Min distance stats:\", ensemble.min_distance_stats)\n",
-    "\n",
     "# Get file count per CN value by min dist method\n",
     "print(\"CN value using min dist method stats:\", ensemble.unique_CN_values_by_min_dist_method_stat)\n",
     "\n",
@@ -480,7 +477,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.3"
+   "version": "3.1.-1"
   }
  },
  "nbformat": 4,

src/cifkit/figures/histogram.py

@@ -4,6 +4,7 @@
 
 import os
 import matplotlib.pyplot as plt
+from matplotlib.ticker import MaxNLocator
 from cifkit.utils import folder, prompt
 
 
@@ -77,17 +78,6 @@ def plot_histogram(attribute, stats, dir_path, display, output_dir):
             },
         }
 
-    if attribute == "min_distance":
-        histogram = {
-            "data": stats,
-            "settings": {
-                "file_name": "min_distance.png",
-                "title": "Minimum Distances Distribution",
-                "xlabel": "Minimum Distance",
-                "key_data_type": "float",
-            },
-        }
-
     if attribute == "elements":
         histogram = {
             "data": stats,
@@ -161,7 +151,7 @@ def generate_histogram(data, settings, display, output_dir: str) -> None:
         data = {
             float(key): data[key] for key in sorted(data.keys(), key=float)
         }
-    elif settings.get("key_data_type") == "int":
+    if settings.get("key_data_type") == "int":
         data = {int(key): data[key] for key in sorted(data.keys(), key=int)}
 
     keys = list(data.keys())
@@ -173,14 +163,32 @@ def generate_histogram(data, settings, display, output_dir: str) -> None:
         color=settings.get("color", "blue"),
         edgecolor=settings.get("edgecolor", "black"),
     )
-    plt.title(settings["title"])
-    plt.xlabel(settings["xlabel"])
+    # Custom settings for specific histograms
+    file_name = settings.get("file_name")
+    if (
+        file_name == "CN_by_min_dist_method.png"
+        or file_name == "CN_by_best_methods.png"
+        or file_name == "composition_type.png"
+    ):
+        # Assuming keys can be cast to integers, sort and find the range for x-ticks
+        int_keys = sorted(map(int, keys))
+        plt.xticks(
+            range(min(int_keys), max(int_keys) + 1)
+        )  # Setting x-ticks to every integer within the range
+        plt.gca().xaxis.set_major_locator(MaxNLocator(integer=True))
+    else:
+        plt.xticks(rotation=settings.get("rotation", 45), ha="right")
+
+    # y-axis settings
+    plt.gca().yaxis.set_major_locator(MaxNLocator(integer=True))
     plt.ylabel("Count")
-    plt.xticks(rotation=settings.get("rotation", 45), ha="right")
+    plt.xlabel(settings["xlabel"])
+
+    plt.title(settings["title"])
     plt.grid(True, linestyle="--", linewidth=0.5)
     plt.tight_layout()
 
-    output_file_path = folder.get_file_path(output_dir, settings["file_name"])
+    output_file_path = folder.get_file_path(output_dir, file_name)
     if not os.path.exists(output_dir):
         os.makedirs(output_dir)
 

src/cifkit/models/cif_ensemble.py

@@ -189,10 +189,6 @@ def site_mixing_type_stats(self) -> dict[str, int]:
     def supercell_size_stats(self) -> dict[int, int]:
         return self._attribute_stats("supercell_points", len)
 
-    @property
-    def min_distance_stats(self) -> dict[float, int]:
-        return self._attribute_stats("shortest_distance")
-
     @property
     def unique_CN_values_by_min_dist_method_stat(
         self,
@@ -419,15 +415,6 @@ def generate_supercell_size_histogram(
             output_dir,
         )
 
-    def generate_min_distance_histogram(self, display=False, output_dir=None):
-        plot_histogram(
-            "min_distance",
-            self.min_distance_stats,
-            self.dir_path,
-            display,
-            output_dir,
-        )
-
     def generate_elements_histogram(self, display=False, output_dir=None):
         plot_histogram(
             "elements",

test.ipynb

@@ -29,7 +29,7 @@
       "# of files moved to 'error_invalid_label' folder: 0\n",
       "# of files moved to 'error_others' folder: 0\n",
       "\n",
-      "\u001b[33mInitializing 8 objects...\u001b[0m\n",
+      "\u001b[33mInitializing 8 Cif objects...\u001b[0m\n",
       "\u001b[32mFinished initialization!\u001b[0m\n"
      ]
     }
@@ -46,6 +46,13 @@
     "# print(ensemble.CN_unique_values_by_min_dist_method)\n",
     "\n"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
   }
  ],
  "metadata": {

tests/core/models/test_cif.py

@@ -596,7 +596,7 @@ def test_init_without_radius_data():
 
 
 def test_init_atomic_mixing_deficiency_without_atomic_mixing():
-    file_path = "tests/data/cif_CN_init/1956508.cif"
+    file_path = "tests/data/cif/cif_CN_init/1956508.cif"
     assert Cif(file_path).CN_unique_values_by_best_methods == {11, 14, 15}
 
 

tests/core/models/test_cif_ensemble.py

@@ -405,13 +405,6 @@ def test_supercell_size_stats(cif_ensemble_test):
     assert result == expected
 
 
-@pytest.mark.fast
-def test_min_distance_stats(cif_ensemble_test):
-    result = cif_ensemble_test.min_distance_stats
-    expected = {2.28: 1, 2.29: 1, 2.383: 1, 2.725: 2, 2.72: 1}
-    assert result == expected
-
-
 @pytest.mark.fast
 def test_unique_elements_stats(cif_ensemble_test):
     result = cif_ensemble_test.unique_elements_stats
@@ -446,8 +439,8 @@ def test_unique_CN_values_by_best_methods_stat(cif_ensemble_test):
 """
 
 
-@pytest.mark.fast
-def test_generate_histogram(cif_ensemble_test, tmp_path):
+@pytest.mark.now
+def test_generate_histogram(cif_ensemble_test):
     output_dir = "tests/data/cif/ensemble_test/histograms"
 
     # List of expected file names
@@ -458,7 +451,6 @@ def test_generate_histogram(cif_ensemble_test, tmp_path):
         "space_group_number.png",
         "space_group_name.png",
         "supercell_size.png",
-        "min_distance.png",
         "elements.png",
         "CN_by_min_dist_method.png",
         "CN_by_best_methods.png",
@@ -473,7 +465,6 @@ def test_generate_histogram(cif_ensemble_test, tmp_path):
     cif_ensemble_test.generate_space_group_number_histogram()
     cif_ensemble_test.generate_space_group_name_histogram()
     cif_ensemble_test.generate_supercell_size_histogram()
-    cif_ensemble_test.generate_min_distance_histogram()
     cif_ensemble_test.generate_elements_histogram()
     cif_ensemble_test.generate_CN_by_min_dist_method_histogram()
     cif_ensemble_test.generate_CN_by_best_methods_histogram()

tests/data/cif_CN_init/1956508.cif → tests/data/cif/cif_CN_init/1956508.cif

@@ -39,7 +39,7 @@ _chemical_formula_weight                 1071.3
 # Bibliographic data
 
 _publ_section_title
-'Nature of magnetic properties in R~3~Co~1.87~In~4~ where R= Ho, Er and Tm'   
+'Nature of magnetic properties in R~3~Co~1.87~In~4~ where R= Ho, Er and Tm'
 _journal_coden_ASTM                      PHTRDP
 _journal_name_full                       'Phase Transitions'
 _journal_year                            2018
@@ -122,4 +122,3 @@ _pd_proc_ls_proof_wR_factor              0.132
 _refine_ls_R_I_factor                    0.077
 
 # End of data set 1956508
-

tests/data/cif_CN_init/ErCo2.68In0.32.cif → tests/data/cif/cif_CN_init/ErCo2.68In0.32.cif

@@ -154,4 +154,3 @@ _refine_ls_R_factor_gt                   ?
 _refine_ls_wR_factor_gt                  ?
 
 # End of data set 1234749
-