Determining the headings.

KonnexionsGmbH · Jun 23, 2022 · 9762deb · 9762deb
1 parent a1dc9cd
commit 9762deb
Show file tree

Hide file tree

Showing 3 changed files with 56 additions and 105 deletions.
diff --git a/data/line_type_heading_rules.json b/data/line_type_heading_rules.json
@@ -18,6 +18,26 @@
         "(A)"
       ]
     },
+    {
+      "name": "a.",
+      "isFirstToken": true,
+      "regexp": "[a-z]\\.$",
+      "functionIsAsc": "lowercase_letters",
+      "startValues": [
+        "a",
+        "a."
+      ]
+    },
+    {
+      "name": "A.",
+      "isFirstToken": true,
+      "regexp": "[A-Z]\\.$",
+      "functionIsAsc": "uppercase_letters",
+      "startValues": [
+        "A",
+        "A."
+      ]
+    },
     {
       "name": "999.",
       "isFirstToken": true,
@@ -43,26 +63,6 @@
       "functionIsAsc": "string_floats",
       "startValues": []
     },
-    {
-      "name": "a.",
-      "isFirstToken": true,
-      "regexp": "[a-z]\\.$",
-      "functionIsAsc": "lowercase_letters",
-      "startValues": [
-        "a",
-        "a."
-      ]
-    },
-    {
-      "name": "A.",
-      "isFirstToken": true,
-      "regexp": "[A-Z]\\.$",
-      "functionIsAsc": "uppercase_letters",
-      "startValues": [
-        "A",
-        "A."
-      ]
-    },
     {
       "name": "(rom)",
       "isFirstToken": true,

diff --git a/data/line_type_heading_rules_test.json b/data/line_type_heading_rules_test.json
@@ -18,6 +18,26 @@
         "(A)"
       ]
     },
+    {
+      "name": "a.",
+      "isFirstToken": true,
+      "regexp": "[a-z]\\.$",
+      "functionIsAsc": "lowercase_letters",
+      "startValues": [
+        "a",
+        "a."
+      ]
+    },
+    {
+      "name": "A.",
+      "isFirstToken": true,
+      "regexp": "[A-Z]\\.$",
+      "functionIsAsc": "uppercase_letters",
+      "startValues": [
+        "A",
+        "A."
+      ]
+    },
     {
       "name": "Article 999:",
       "isFirstToken": false,
@@ -58,6 +78,13 @@
       "isFirstToken": true,
       "regexp": "Riders-\\d$",
       "functionIsAsc": "string_integers",
+      "startValues": ["Riders-1"]
+    },
+    {
+      "name": "Section 999.999.",
+      "isFirstToken": false,
+      "regexp": "Section \\d+\\.\\d+\\.",
+      "functionIsAsc": "string_floats",
       "startValues": []
     },
     {
@@ -78,40 +105,13 @@
         "(1)"
       ]
     },
-    {
-      "name": "Section 999.999.",
-      "isFirstToken": false,
-      "regexp": "Section \\d+\\.\\d+\\.",
-      "functionIsAsc": "string_floats",
-      "startValues": []
-    },
     {
       "name": "999.999",
       "isFirstToken": true,
       "regexp": "\\d+\\.\\d+\\.?$",
       "functionIsAsc": "string_floats",
       "startValues": []
     },
-    {
-      "name": "a.",
-      "isFirstToken": true,
-      "regexp": "[a-z]\\.$",
-      "functionIsAsc": "lowercase_letters",
-      "startValues": [
-        "a",
-        "a."
-      ]
-    },
-    {
-      "name": "A.",
-      "isFirstToken": true,
-      "regexp": "[A-Z]\\.$",
-      "functionIsAsc": "uppercase_letters",
-      "startValues": [
-        "A",
-        "A."
-      ]
-    },
     {
       "name": "(rom)",
       "isFirstToken": true,

diff --git a/src/dcr/nlp/cls_line_type_heading.py b/src/dcr/nlp/cls_line_type_heading.py
@@ -259,46 +259,18 @@ def _init_heading_rules(self) -> list[tuple[str, bool, str, collections.abc.Call
                 ["(A)"],
             ),
             (
-                "Article 999:",
-                False,
-                r"Article \d+:?",
-                self._is_asc_string_integers,
-                [],
-            ),
-            (
-                "ARTICLE 999 -",
-                False,
-                r"ARTICLE \d+ -",
-                self._is_asc_string_integers,
-                [],
-            ),
-            (
-                "ARTICLE XXX-YYY:",
-                False,
-                r"ARTICLE [A-Z]+(-[A-Z]+)?:",
-                self._is_asc_ignore,
-                [],
-            ),
-            (
-                "EXHIBIT A",
-                False,
-                r"EXHIBIT [A-Z]$",
-                self._is_asc_strings,
-                [],
-            ),
-            (
-                'EXHIBIT "A"        ',
-                False,
-                r"EXHIBIT \u201c[A-Z]\u201d$",
-                self._is_asc_strings,
-                [],
+                "a.",
+                True,
+                r"[a-z]\.$",
+                self._is_asc_lowercase_letters,
+                ["a", "a."],
             ),
             (
-                "Riders-9",
+                "A.",
                 True,
-                r"Riders-\d$",
-                self._is_asc_string_integers,
-                [],
+                r"[A-Z]\.$",
+                self._is_asc_uppercase_letters,
+                ["A", "A."],
             ),
             (
                 "999.",
@@ -314,34 +286,13 @@ def _init_heading_rules(self) -> list[tuple[str, bool, str, collections.abc.Call
                 self._is_asc_string_integers,
                 ["(1)"],
             ),
-            (
-                "Section 999.999.",
-                False,
-                r"Section \d+\.\d+\.",
-                self._is_asc_string_floats,
-                [],
-            ),
             (
                 "999.999",
                 True,
                 r"\d+\.\d+\.?$",
                 self._is_asc_string_floats,
                 [],
             ),
-            (
-                "a.",
-                True,
-                r"[a-z]\.$",
-                self._is_asc_lowercase_letters,
-                ["a", "a."],
-            ),
-            (
-                "A.",
-                True,
-                r"[A-Z]\.$",
-                self._is_asc_uppercase_letters,
-                ["A", "A."],
-            ),
             (
                 "(rom)",
                 True,