✨ Implement Scraping of use_cases (the gray outer left cells)

requirements.in

@@ -1,3 +1,3 @@
-ebdtable2graph
+ebdtable2graph>=0.1.4a
 python-docx
 more_itertools

requirements.txt

@@ -10,7 +10,7 @@ certifi==2022.12.7
     # via requests
 charset-normalizer==2.1.1
     # via requests
-ebdtable2graph==0.1.3
+ebdtable2graph==0.1.4a0
     # via -r requirements.in
 idna==3.4
     # via requests

src/ebddocx2table/docxtableconverter.py

@@ -3,20 +3,20 @@
 """
 import re
 from enum import Enum
-from itertools import cycle
+from itertools import cycle, groupby
 from typing import Generator, List, Literal, Optional, Tuple
 
 from docx.table import Table, _Cell, _Row  # type:ignore[import]
 from ebdtable2graph.models import EbdTable, EbdTableRow, EbdTableSubRow
-from ebdtable2graph.models.ebd_table import EbdCheckResult, EbdTableMetaData
-from more_itertools import first
+from ebdtable2graph.models.ebd_table import _STEP_NUMBER_REGEX, EbdCheckResult, EbdTableMetaData
+from more_itertools import first, first_true
 
 
-def _is_pruefende_rolle_cell(cell: _Cell) -> bool:
+def _is_pruefende_rolle_cell_text(text: str) -> bool:
     """ "
-    Returns true iff the cell mentions the market role that is responsible for applying this entscheidungsbaum
+    Returns true iff the given text mentions the market role that is responsible for applying this entscheidungsbaum
     """
-    return cell.text.startswith("Prüfende Rolle: ")
+    return text.startswith("Prüfende Rolle: ")
 
 
 def _sort_columns_in_row(docx_table_row: _Row) -> Generator[_Cell, None, None]:
@@ -31,6 +31,29 @@ def _sort_columns_in_row(docx_table_row: _Row) -> Generator[_Cell, None, None]:
 
 _subsequent_step_pattern = re.compile(r"^(?P<bool>(?:ja)|(?:nein))\s*(?P<subsequent_step_number>(?:\d+\*?)|ende)?")
 
+_step_number_pattern = re.compile(_STEP_NUMBER_REGEX)
+
+
+def _get_index_of_first_column_with_step_number(cells: List[_Cell]) -> int:
+    """
+    returns the index of the first cell in cells, that contains a step number
+    """
+    first_step_number_cell = first_true(cells, pred=lambda cell: _step_number_pattern.match(cell.text) is not None)
+    step_number_column_index = cells.index(first_step_number_cell)
+    return step_number_column_index
+
+
+def _get_use_cases(cells: List[_Cell]) -> List[str]:
+    """
+    Extract use cases from the given list of cells.
+    May return empty list, never returns None.
+    """
+    index_of_step_number = _get_index_of_first_column_with_step_number(cells)
+    if index_of_step_number != 0:
+        # "use_cases" are present; This means, that this step must only be applied for certain scenarios,
+        return [c.text for c in cells[0:index_of_step_number]]
+    return []  # we don't return None here because we need something that has a length in the calling code
+
 
 def _read_subsequent_step_cell(cell: _Cell) -> Tuple[bool, Optional[str]]:
     """
@@ -79,63 +102,77 @@ def __init__(self, docx_tables: List[Table], ebd_key: str, chapter: str, sub_cha
         self._docx_tables = docx_tables
         self._column_index_step_number: int
         self._column_index_description: int
-        self._column_index_check_result: int
-        self._column_index_result_code: int
+        self._column_index_check_result: int = 0
+        self._column_index_result_code: int = 0
         self._column_index_note: int
         self._row_index_last_header: Literal[0, 1]  # either 0  or 1
         for row_index in range(0, 2):  # the first two lines/rows are the header of the table.
             # In the constructor we just want to read the metadata from the table.
             # For this purpose the first two lines are enough.
-            for column_index, table_cell in enumerate(first(docx_tables).row_cells(row_index)):
-                if row_index == 0 and _is_pruefende_rolle_cell(table_cell):
-                    role = table_cell.text.split(":")[1].strip()
+            # Now it feels natural, to loop over the cells/columns of the first row, but before we do so, we have to
+            # remove duplicates. Although there are usually only 5 columns visible, technically there might be even 8.
+            # In these cases (e.g. for E_0453) columns like 'Prüfergebnis' simply occur twice in the docx table header.
+            distinct_cell_texts: List[str] = [
+                x[0] for x in groupby(first(docx_tables).row_cells(row_index), lambda cell: cell.text)
+            ]
+            for column_index, table_cell_text in enumerate(distinct_cell_texts):
+                if row_index == 0 and _is_pruefende_rolle_cell_text(table_cell_text):
+                    role = table_cell_text.split(":")[1].strip()
                     break  # because the prüfende rolle is always a full row with identical column cells
-                if table_cell.text == "Nr.":
+                if table_cell_text == "Nr.":
                     self._column_index_step_number = column_index
                     # In most of the cases this will be 1,
                     # but it can be 0 if the first row does _not_ contain the "Prüfende Rolle".
                     self._row_index_last_header = row_index  # type:ignore[assignment]
-                elif table_cell.text == "Prüfschritt":
+                elif table_cell_text == "Prüfschritt":
                     self._column_index_description = column_index
-                elif table_cell.text == "Prüfergebnis":
+                elif table_cell_text == "Prüfergebnis":
                     self._column_index_check_result = column_index
-                elif table_cell.text == "Code":
+                elif table_cell_text == "Code":
                     self._column_index_result_code = column_index
-                elif table_cell.text == "Hinweis":
+                elif table_cell_text == "Hinweis":
                     self._column_index_note = column_index
         self._metadata = EbdTableMetaData(ebd_code=ebd_key, sub_chapter=sub_chapter, chapter=chapter, role=role)
 
+    # I see that there are quite a few local variables, but honestly see no reason to break it down any further
+    # pylint:disable=too-many-locals
     def _handle_single_table(
         self, table: Table, row_offset: int, rows: List[EbdTableRow], sub_rows: List[EbdTableSubRow]
     ) -> None:
         """
         Handles a single table (out of possible multiple tables for 1 EBD).
         The results are written into rows and sub_rows. Those will be modified.
         """
+        upper_lower_iterator = cycle([_EbdSubRowPosition.UPPER, _EbdSubRowPosition.LOWER])
+        use_cases: List[str] = []
         for table_row, sub_row_position in zip(
             table.rows[row_offset:],
-            cycle([_EbdSubRowPosition.UPPER, _EbdSubRowPosition.LOWER]),
+            upper_lower_iterator,
         ):
             row_cells = list(_sort_columns_in_row(table_row))
+            if len(row_cells) <= self._column_index_description:
+                # These are the multi-column rows that span that contain stuff like
+                # "Alle festgestellten Antworten sind anzugeben, soweit im Format möglich (maximal 8 Antwortcodes)*."
+                _ = next(upper_lower_iterator)  # reset the iterator
+                continue
             if sub_row_position == _EbdSubRowPosition.UPPER:
+                use_cases = _get_use_cases(row_cells)
                 # clear list every second entry
                 sub_rows = []
-                step_number = row_cells[self._column_index_step_number].text.strip()
-                description = row_cells[self._column_index_description].text.strip()
+                step_number = row_cells[len(use_cases) + self._column_index_step_number].text.strip()
+                description = row_cells[len(use_cases) + self._column_index_description].text.strip()
             boolean_outcome, subsequent_step_number = _read_subsequent_step_cell(
-                row_cells[self._column_index_check_result]
+                row_cells[len(use_cases) + self._column_index_check_result]
             )
             sub_row = EbdTableSubRow(
                 check_result=EbdCheckResult(subsequent_step_number=subsequent_step_number, result=boolean_outcome),
-                result_code=row_cells[self._column_index_result_code].text.strip() or None,
-                note=row_cells[self._column_index_note].text.strip() or None,
+                result_code=row_cells[len(use_cases) + self._column_index_result_code].text.strip() or None,
+                note=row_cells[len(use_cases) + self._column_index_note].text.strip() or None,
             )
             sub_rows.append(sub_row)
             if sub_row_position == _EbdSubRowPosition.LOWER:
                 row = EbdTableRow(
-                    description=description,
-                    step_number=step_number,
-                    sub_rows=sub_rows,
+                    description=description, step_number=step_number, sub_rows=sub_rows, use_cases=use_cases or None
                 )
                 rows.append(row)