Determining the numbered lists.

docs/running_configuration.md

@@ -160,6 +160,7 @@ The customisable entries are:
     lt_list_bullet_min_entries = 2
     lt_list_bullet_rule_file = none
     lt_list_bullet_tolerance_llx = 5
+    lt_list_number_file_incl_regexp = false
     lt_list_number_min_entries = 2
     lt_list_number_rule_file = none
     lt_list_number_tolerance_llx = 5
@@ -220,6 +221,7 @@ The customisable entries are:
 | lt_list_bullet_min_entries       | **`2`**                                     | Minimum number of entries to determine a bulleted list.                                                                |
 | lt_list_bullet_rule_file         | **`data/line_type_list_bullet_rules.json`** | File with rules to determine the bulleted lists.                                                                       |
 | lt_list_bullet_tolerance_llx     | **`5`**                                     | Tolerance of vertical indentation in percent.                                                                          |
+| lt_list_number_file_incl_regexp  | **`false`**                                 | If it is set to **`true`**, the regular expression for the numbered list is included in the **`JSON`** file.           |
 | lt_list_number_min_entries       | **`2`**                                     | Minimum number of entries to determine a numbered list.                                                                |
 | lt_list_number_rule_file         | **`data/line_type_list_number_rules.json`** | File with rules to determine the numbered lists.                                                                       |
 | lt_list_number_tolerance_llx     | **`5`**                                     | Tolerance of vertical indentation in percent.                                                                          |

setup.cfg

@@ -40,6 +40,7 @@ lt_heading_tolerance_llx = 5
 lt_list_bullet_min_entries = 2
 lt_list_bullet_rule_file = none
 lt_list_bullet_tolerance_llx = 5
+lt_list_number_file_incl_regexp = false
 lt_list_number_min_entries = 2
 lt_list_number_rule_file = none
 lt_list_number_tolerance_llx = 5
@@ -66,10 +67,12 @@ db_connection_port = 5433
 db_database = dcr_db_dev
 db_database_admin = dcr_db_dev_admin
 db_initial_data_file = data/db_initial_data_file_dev.json
+delete_auxiliary_files = false
 directory_inbox = data/inbox_dev
 directory_inbox_accepted = data/inbox_dev_accepted
 directory_inbox_rejected = data/inbox_dev_rejected
 lt_heading_file_incl_regexp = true
+lt_list_number_file_incl_regexp = true
 
 [dcr.env.test]
 create_extra_file_list_bullet = true
@@ -110,6 +113,7 @@ lt_heading_tolerance_llx = 5
 lt_list_bullet_min_entries = 2
 lt_list_bullet_rule_file = data/line_type_list_bullet_rules_test.json
 lt_list_bullet_tolerance_llx = 5
+lt_list_number_file_incl_regexp = false
 lt_list_number_min_entries = 2
 lt_list_number_rule_file = data/line_type_list_number_rules_test.json
 lt_list_number_tolerance_llx = 5

src/dcr/cfg/cls_setup.py

@@ -20,7 +20,7 @@ class Setup:
     # -----------------------------------------------------------------------------
     # Class variables.
     # -----------------------------------------------------------------------------
-    _CONFIG_PARAM_NO = 125
+    _CONFIG_PARAM_NO = 126
 
     _DCR_CFG_CREATE_EXTRA_FILE_LIST_BULLET: ClassVar[str] = "create_extra_file_list_bullet"
     _DCR_CFG_CREATE_EXTRA_FILE_LIST_NUMBER: ClassVar[str] = "create_extra_file_list_number"
@@ -61,6 +61,7 @@ class Setup:
     _DCR_CFG_LT_LIST_BULLET_MIN_ENTRIES: ClassVar[str] = "lt_list_bullet_min_entries"
     _DCR_CFG_LT_LIST_BULLET_RULE_FILE: ClassVar[str] = "lt_list_bullet_rule_file"
     _DCR_CFG_LT_LIST_BULLET_TOLERANCE_LLX: ClassVar[str] = "lt_list_bullet_tolerance_llx"
+    _DCR_CFG_LT_LIST_NUMBER_FILE_INCL_REGEXP: ClassVar[str] = "lt_list_number_file_incl_regexp"
     _DCR_CFG_LT_LIST_NUMBER_MIN_ENTRIES: ClassVar[str] = "lt_list_number_min_entries"
     _DCR_CFG_LT_LIST_NUMBER_RULE_FILE: ClassVar[str] = "lt_list_number_rule_file"
     _DCR_CFG_LT_LIST_NUMBER_TOLERANCE_LLX: ClassVar[str] = "lt_list_number_tolerance_llx"
@@ -228,6 +229,9 @@ def __init__(self) -> None:
         self.lt_list_bullet_min_entries = 2
         self.lt_list_bullet_rule_file = "none"
         self.lt_list_bullet_tolerance_llx = 5
+
+        self.is_lt_list_number_file_incl_regexp = False
+
         self.lt_list_number_min_entries = 2
         self.lt_list_number_rule_file = "none"
         self.lt_list_number_tolerance_llx = 5
@@ -412,6 +416,9 @@ def _check_config(self) -> None:
         self.lt_list_bullet_tolerance_llx = self._determine_config_param_integer(
             Setup._DCR_CFG_LT_LIST_BULLET_TOLERANCE_LLX, self.lt_list_bullet_tolerance_llx
         )
+        self.is_lt_list_number_file_incl_regexp = self._determine_config_param_boolean(
+            Setup._DCR_CFG_LT_LIST_NUMBER_FILE_INCL_REGEXP, self.is_lt_list_number_file_incl_regexp
+        )
         self.lt_list_number_min_entries = self._determine_config_param_integer(
             Setup._DCR_CFG_LT_LIST_NUMBER_MIN_ENTRIES, self.lt_list_number_min_entries
         )
@@ -898,6 +905,7 @@ def _load_config(self) -> None:
                     | Setup._DCR_CFG_LT_HEADING_TOLERANCE_LLX
                     | Setup._DCR_CFG_LT_LIST_BULLET_MIN_ENTRIES
                     | Setup._DCR_CFG_LT_LIST_BULLET_TOLERANCE_LLX
+                    | Setup._DCR_CFG_LT_LIST_NUMBER_FILE_INCL_REGEXP
                     | Setup._DCR_CFG_LT_LIST_NUMBER_MIN_ENTRIES
                     | Setup._DCR_CFG_LT_LIST_NUMBER_TOLERANCE_LLX
                     | Setup._DCR_CFG_LT_TABLE_FILE_INCL_EMPTY_COLUMNS

src/dcr/nlp/cls_line_type_list_number.py

@@ -19,7 +19,7 @@
 Entry = dict[str, int | str]
 Entries = list[Entry]
 
-List = dict[str, Entries | float | int | str]
+List = dict[str, Entries | float | int | object | str]
 Lists = list[List]
 
 RuleExtern = tuple[str, str, collections.abc.Callable[[str, str], bool], list[str]]
@@ -182,18 +182,23 @@ def _finish_list(self) -> None:
             #     "noEntries": 99,
             #     "pageNoFrom": 99,
             #     "pageNoTill": 99,
+            #     "regexp": "xxx",
             #     "entries": []
-            # },
-            self._lists.append(
-                {
-                    nlp.cls_nlp_core.NLPCore.JSON_NAME_NUMBER: self._rule[0].rstrip(),
-                    nlp.cls_nlp_core.NLPCore.JSON_NAME_LIST_NO: self.no_lists,
-                    nlp.cls_nlp_core.NLPCore.JSON_NAME_NO_ENTRIES: len(entries),
-                    nlp.cls_nlp_core.NLPCore.JSON_NAME_PAGE_NO_FROM: int(self._entries[0][0]) + 1,
-                    nlp.cls_nlp_core.NLPCore.JSON_NAME_PAGE_NO_TILL: int(self._entries[-1][0]) + 1,
-                    nlp.cls_nlp_core.NLPCore.JSON_NAME_ENTRIES: entries,
-                }
-            )
+            # }
+            entry = {
+                nlp.cls_nlp_core.NLPCore.JSON_NAME_NUMBER: self._rule[0].rstrip(),
+                nlp.cls_nlp_core.NLPCore.JSON_NAME_LIST_NO: self.no_lists,
+                nlp.cls_nlp_core.NLPCore.JSON_NAME_NO_ENTRIES: len(entries),
+                nlp.cls_nlp_core.NLPCore.JSON_NAME_PAGE_NO_FROM: int(self._entries[0][0]) + 1,
+                nlp.cls_nlp_core.NLPCore.JSON_NAME_PAGE_NO_TILL: int(self._entries[-1][0]) + 1,
+            }
+
+            if cfg.glob.setup.is_lt_list_number_file_incl_regexp:
+                entry[nlp.cls_nlp_core.NLPCore.JSON_NAME_REGEXP] = self._rule[-1]
+
+            entry[nlp.cls_nlp_core.NLPCore.JSON_NAME_ENTRIES] = entries
+
+            self._lists.append(entry)
 
         self._reset_list()
 

tests/conftest.py

@@ -735,6 +735,7 @@ def fxtr_before_any_test():
         (cfg.cls_setup.Setup._DCR_CFG_LT_LIST_BULLET_MIN_ENTRIES, "2"),
         (cfg.cls_setup.Setup._DCR_CFG_LT_LIST_BULLET_RULE_FILE, "data/line_type_list_bullet_rules_test.json"),
         (cfg.cls_setup.Setup._DCR_CFG_LT_LIST_BULLET_TOLERANCE_LLX, "5"),
+        (cfg.cls_setup.Setup._DCR_CFG_LT_LIST_NUMBER_FILE_INCL_REGEXP, "false"),
         (cfg.cls_setup.Setup._DCR_CFG_LT_LIST_NUMBER_MIN_ENTRIES, "2"),
         (cfg.cls_setup.Setup._DCR_CFG_LT_LIST_NUMBER_RULE_FILE, "data/line_type_list_number_rules_test.json"),
         (cfg.cls_setup.Setup._DCR_CFG_LT_LIST_NUMBER_TOLERANCE_LLX, "5"),

tests/test_nlp_cls_line_type_list_number.py

@@ -50,6 +50,7 @@ def test_line_type_list_number_1(fxtr_rmdir_opt, fxtr_setup_empty_db_and_inbox):
             (cfg.cls_setup.Setup._DCR_CFG_LT_HEADING_FILE_INCL_NO_CTX, "3"),
             (cfg.cls_setup.Setup._DCR_CFG_LT_HEADING_FILE_INCL_REGEXP, "true"),
             (cfg.cls_setup.Setup._DCR_CFG_LT_HEADING_RULE_FILE, "none"),
+            (cfg.cls_setup.Setup._DCR_CFG_LT_LIST_NUMBER_FILE_INCL_REGEXP, "true"),
             (cfg.cls_setup.Setup._DCR_CFG_TETML_PAGE, "true"),
             (cfg.cls_setup.Setup._DCR_CFG_TETML_WORD, "true"),
             (cfg.cls_setup.Setup._DCR_CFG_VERBOSE_LT_HEADING, "true"),