first commit

milestone_attribution_word_export_module/generate_milestones_report.py

@@ -0,0 +1,366 @@
+import os
+from pathlib import Path
+from collections import defaultdict
+import argparse
+from functools import partial
+
+from docx import Document
+from docx.shared import Pt, Inches
+from docx.enum.text import WD_ALIGN_PARAGRAPH
+
+from pasbdc_word_library import WordDocumentBuilder, PageConfig #pyright:ignore
+
+def parse_and_organize_images(source_folder):
+    """
+    Parses filenames and groups them by Center and Fiscal Year, 
+    keyed by the specific chart tag.
+    """
+    source_path = Path(source_folder)
+    
+    # Structure: organization[center][fiscal_year][chart_tag] = full_path
+    # We change the inner default from 'list' to 'dict'
+    organization = defaultdict(lambda: defaultdict(dict))
+
+    for image_path in source_path.glob('*.png'):
+        filename_stem = image_path.stem 
+        parts = filename_stem.split('_')
+        
+        if len(parts) < 3:
+            continue
+
+        center = parts[0]
+        fiscal_year = parts[-1]
+        
+        # The key you will use to access this specific file later
+        chart_tag = "_".join(parts[1:-1])
+
+        # Map the tag directly to the path
+        organization[center][fiscal_year][chart_tag] = str(image_path)
+
+    return organization
+
+def single_year_nbs_page(
+    builder: WordDocumentBuilder, 
+    pie_chart_path:str, 
+    grouped_bar_path:str, 
+    center_name:str,
+    fiscal_year_tag:str,
+    add_documentation_note:bool
+):
+    # 1. Setup Margins
+    for section in builder.doc.sections:
+        section.top_margin = Inches(0.5)
+        section.bottom_margin = Inches(0.5)
+        section.left_margin = Inches(0.5)
+        section.right_margin = Inches(0.5)
+
+    # 2. Add Title
+    title = builder.doc.add_heading(f'{builder.current_section}. {center_name} New Business Starts Attribution Analysis {fiscal_year_tag}', level=1)
+    title.alignment = WD_ALIGN_PARAGRAPH.CENTER #pyright:ignore
+    
+    # 3. Create a 2x2 Table for Side-by-Side Layout
+    # Row 0 = Images, Row 1 = Captions
+    table = builder.doc.add_table(rows=2, cols=2)
+    table.autofit = False 
+    
+    # --- LEFT COLUMN (Pie Chart) ---
+    
+    # Image (Row 0, Col 0)
+    cell_pie_img = table.cell(0, 0)
+    p_pie = cell_pie_img.paragraphs[0]
+    p_pie.alignment = WD_ALIGN_PARAGRAPH.CENTER
+    # Width set to 3.2 inches to ensure two fit on a 7.5 inch wide printable area
+    p_pie.add_run().add_picture(pie_chart_path, width=Inches(3.2))
+
+    # Caption (Row 1, Col 0)
+    cell_pie_cap = table.cell(1, 0)
+    p_pie_cap = cell_pie_cap.paragraphs[0]
+    p_pie_cap.alignment = WD_ALIGN_PARAGRAPH.CENTER
+    builder.figure_number += 1
+    p_pie_cap.add_run(f"Figure {builder.current_section}.{builder.figure_number} shows the proportion of new business start milestones that fell into each documentation level").bold = True
+    # --- RIGHT COLUMN (Bar Chart) ---
+
+    # Image (Row 0, Col 1)
+    cell_bar_img = table.cell(0, 1)
+    p_bar = cell_bar_img.paragraphs[0]
+    p_bar.alignment = WD_ALIGN_PARAGRAPH.CENTER
+    p_bar.add_run().add_picture(grouped_bar_path, width=Inches(3.2))
+
+    # Caption (Row 1, Col 1)
+    cell_bar_cap = table.cell(1, 1)
+    p_bar_cap = cell_bar_cap.paragraphs[0]
+    p_bar_cap.alignment = WD_ALIGN_PARAGRAPH.CENTER
+    builder.figure_number += 1
+    p_bar_cap.add_run(f"Figure {builder.current_section}.{builder.figure_number} shows how many of each documentation level were a part of each attribution source").bold = True
+
+    # 4. Add Spacing after table
+    builder.doc.add_paragraph()
+
+    if add_documentation_note:
+        # 6. Add the "NOTE" section (Definitions)
+        builder.doc.add_heading("NOTE: Data is based on business start impact date and new business start milestone records on the client file, extracted as of 1/27/26.", level=4)
+        builder.doc.add_heading('In situations where multiple milestone records existed for a client, a de-duplication / merger script was utilized for data analysis (no changes in NS).The script created some unique combination types (e.g., "Email from Client,Requested on Ecenter"); these combined types were left as separate categories to more easily identify potential data entry training opportunities.', level=4)
+        builder.doc.add_heading("NOTE 2: Documentation levels were determined as follows:", level=4)
+        
+        note_paragraph = builder.doc.add_paragraph()
+        note_paragraph.add_run("No documentation (red) - Will NOT be submitted to Nexus\n").bold = True
+        note_paragraph.add_run("\t1. Attribution Source was blank, OR \n\t2. Attribution source was 'Requested on eCenter'")
+        note_paragraph.add_run("\n\n")
+
+        note_paragraph.add_run("Affirmation Missing (yellow) - Will NOT be submitted to Nexus\n").bold = True
+        note_paragraph.add_run("\t1. Attribution source field is a non-blank, non-'Requested on eCenter' attribution source, BUT\n\t2. Affirmation Statement was blank")
+        note_paragraph.add_run("\n\n")
+        
+        note_paragraph.add_run("Documented (green) - Will be submitted to Nexus as long as 'Director Verified' is checked\n").bold = True
+        note_paragraph.add_run("\t1. There is a non-blank, non-'Requested on eCenter' attribution source AND\n\t2. Affirmation Statement was non-blank")
+        note_paragraph.add_run("\n")
+        note_paragraph.add_run("\tNOTE: If the attribution source is eCenter, then no value is required in the affirmation column.")
+        note_paragraph.add_run("\n\n")
+
+def single_year_funding_page(
+    builder: WordDocumentBuilder, 
+    pie_chart_path:str, 
+    grouped_bar_path:str, 
+    center_name:str,
+    fiscal_year_tag:str,
+    add_documentation_note:bool
+):
+    # 1. Setup Margins
+    for section in builder.doc.sections:
+        section.top_margin = Inches(0.5)
+        section.bottom_margin = Inches(0.5)
+        section.left_margin = Inches(0.5)
+        section.right_margin = Inches(0.5)
+
+    # 2. Add Title
+    title = builder.doc.add_heading(f'{builder.current_section}. {center_name} Capital Transaction Attribution Analysis {fiscal_year_tag}', level=1)
+    title.alignment = WD_ALIGN_PARAGRAPH.CENTER #pyright:ignore
+    
+    # 3. Create a 2x2 Table for Side-by-Side Layout
+    # Row 0 = Images, Row 1 = Captions
+    table = builder.doc.add_table(rows=2, cols=2)
+    table.autofit = False 
+    
+    # --- LEFT COLUMN (Pie Chart) ---
+    
+    # Image (Row 0, Col 0)
+    cell_pie_img = table.cell(0, 0)
+    p_pie = cell_pie_img.paragraphs[0]
+    p_pie.alignment = WD_ALIGN_PARAGRAPH.CENTER
+    # Width set to 3.2 inches to ensure two fit on a 7.5 inch wide printable area
+    p_pie.add_run().add_picture(pie_chart_path, width=Inches(3.2))
+
+    # Caption (Row 1, Col 0)
+    cell_pie_cap = table.cell(1, 0)
+    p_pie_cap = cell_pie_cap.paragraphs[0]
+    p_pie_cap.alignment = WD_ALIGN_PARAGRAPH.CENTER
+    builder.figure_number += 1
+    p_pie_cap.add_run(f"Figure {builder.current_section}.{builder.figure_number} shows the proportion of funding milestones that fell into each documentation level").bold = True
+    # --- RIGHT COLUMN (Bar Chart) ---
+
+    # Image (Row 0, Col 1)
+    cell_bar_img = table.cell(0, 1)
+    p_bar = cell_bar_img.paragraphs[0]
+    p_bar.alignment = WD_ALIGN_PARAGRAPH.CENTER
+    p_bar.add_run().add_picture(grouped_bar_path, width=Inches(3.2))
+
+    # Caption (Row 1, Col 1)
+    cell_bar_cap = table.cell(1, 1)
+    p_bar_cap = cell_bar_cap.paragraphs[0]
+    p_bar_cap.alignment = WD_ALIGN_PARAGRAPH.CENTER
+    builder.figure_number += 1
+    p_bar_cap.add_run(f"Figure {builder.current_section}.{builder.figure_number} shows how many of each documentation level were a part of each attribution source").bold = True
+
+    # 4. Add Spacing after table
+    builder.doc.add_paragraph()
+
+    if add_documentation_note:
+        builder.doc.add_heading("NOTE: Data is based on those impacts showing up in the q1 scorecard as of 1/27/26", level=4)
+        builder.doc.add_heading("NOTE 2: Documentation levels were determined as follows:", level=4)
+        
+        note_paragraph = builder.doc.add_paragraph()
+        note_paragraph.add_run("No documentation (red) - Will NOT be submitted to Nexus\n").bold = True
+        note_paragraph.add_run("\t1. Attribution Source was blank, OR \n\t2. Attribution source was 'Requested on eCenter'")
+        note_paragraph.add_run("\n\n")
+        
+        note_paragraph.add_run("Affirmation Missing (yellow) - Will NOT be submitted to Nexus\n").bold = True
+        note_paragraph.add_run("\t1. Attribution source field is a non-blank, non-'Requested on eCenter' attribution source, BUT\n\t2. Affirmation Statement was blank")
+        note_paragraph.add_run("\n\n")
+        
+        note_paragraph.add_run("Documented (green) - Will be submitted to Nexus as long as 'Director Verified' is checked\n").bold = True
+        note_paragraph.add_run("\t1. There is a non-blank, non-'Requested on eCenter' attribution source AND\n\t2. Affirmation Statement was non-blank")
+        note_paragraph.add_run("\n")
+        note_paragraph.add_run("\tNOTE: If the attribution source is eCenter, then no value is required in the affirmation column.")
+        note_paragraph.add_run("\n\n")
+
+def process_documents(organization_data, out_path):
+    network_wide_nbs = []
+    network_wide_funding = []
+    is_first_network_funding_page = True
+    is_first_network_nbs_page = True
+    for fiscal_year, chart_map in sorted(organization_data["network"].items()):
+        path_funding_grouped = chart_map.get('funding_attribution_grouped')
+        if not path_funding_grouped:
+            print(f"   WARNING: Missing 'funding_attribution_grouped' for network wide fy {fiscal_year}")
+        else:
+            print(path_funding_grouped)
+
+        path_funding_pie = chart_map.get('funding_attribution_pie')
+        if not path_funding_pie:
+            print(f"   WARNING: Missing 'funding_attribution_pie' for network wide fy {fiscal_year}")
+        else:
+            print(path_funding_pie)
+
+        path_nbs_grouped = chart_map.get('nbs_attribution_grouped')
+        if not path_nbs_grouped:
+            print(f"   WARNING: Missing 'nbs_attribution_grouped' for network wide fy {fiscal_year}")
+        else:
+            print(path_nbs_grouped)
+
+        path_nbs_pie = chart_map.get('nbs_attribution_pie')
+        if not path_nbs_pie:
+            print(f"   WARNING: Missing 'nbs_attribution_pie' for network wide fy {fiscal_year}")
+        else:
+            print(path_nbs_pie)
+
+        if path_nbs_pie and path_nbs_grouped:
+            nbs_page = partial(single_year_nbs_page, 
+                                pie_chart_path=path_nbs_pie, 
+                                grouped_bar_path=path_nbs_grouped,
+                                center_name="Network Wide",
+                                fiscal_year_tag=fiscal_year,
+                                add_documentation_note=True if is_first_network_nbs_page else False
+                                )
+            network_wide_nbs.append(PageConfig(page_function=nbs_page, add_page_break=True if is_first_network_nbs_page else False))
+            if is_first_network_nbs_page:
+                is_first_network_nbs_page = False
+        else:
+            print(f"   SKIPPING: Network Wide NBS page for {fiscal_year} due to missing graphs.")
+
+        if path_funding_pie and path_funding_grouped:
+            funding_page = partial(single_year_funding_page, 
+                                pie_chart_path=path_funding_pie, 
+                                grouped_bar_path=path_funding_grouped,
+                                center_name="Network Wide",
+                                fiscal_year_tag=fiscal_year,
+                                add_documentation_note=True if is_first_network_funding_page else False
+                                )
+            network_wide_funding.append(PageConfig(page_function=funding_page, add_page_break=True if is_first_network_funding_page else False))
+            if is_first_network_funding_page:
+                is_first_network_funding_page = False
+        else:
+            print(f"   SKIPPING: Network Wide Funding page for {fiscal_year} due to missing graphs.")
+
+    nbs_pages = []
+    funding_pages = [] 
+    for center, fy_groups in sorted(organization_data.items()):
+        if center == "network":
+            continue
+        # Used to store completed PageConfig objects
+        center_nbs_pages = []
+        # Used to tell the script whether to place the documentation level notes or not
+        is_first_nbs_page = True
+
+        center_funding_pages = []
+        is_first_funding_page = True
+
+        for fy, chart_map in sorted(fy_groups.items()):
+            
+            print(f"--- Creating Doc for {center} {fy} ---")
+
+            # Create the funding pages
+            path_funding_grouped = chart_map.get('funding_attribution_grouped')
+            if not path_funding_grouped:
+                print(f"   WARNING: Missing 'funding_attribution_grouped' for {center}")
+            else:
+                print(path_funding_grouped)
+
+            # Example: Getting the 'pie' chart
+            path_funding_pie = chart_map.get('funding_attribution_pie')
+            if not path_funding_pie:
+                print(f"   WARNING: Missing 'funding_attribution_pie' for {center}")
+            else:
+                print(path_funding_pie)
+
+            # Create the new busiuness start pages
+            path_nbs_grouped = chart_map.get('nbs_attribution_grouped')
+            if not path_nbs_grouped:
+                print(f"   WARNING: Missing 'nbs_attribution_grouped' for {center}")
+            else:
+                print(path_nbs_grouped)
+
+            # Example: Getting the 'pie' chart
+            path_nbs_pie = chart_map.get('nbs_attribution_pie')
+            if not path_nbs_pie:
+                print(f"   WARNING: Missing 'nbs_attribution_pie' for {center}")
+            else:
+                print(path_nbs_pie)
+
+            if path_nbs_pie and path_nbs_grouped:
+                nbs_page = partial(single_year_nbs_page, 
+                                pie_chart_path=path_nbs_pie, 
+                                grouped_bar_path=path_nbs_grouped,
+                                center_name=center,
+                                fiscal_year_tag=fy,
+                                add_documentation_note=True if is_first_nbs_page else False
+                                )
+
+                nbs_pages.append(PageConfig(page_function=nbs_page, add_page_break=True if is_first_nbs_page else False))
+                center_nbs_pages.append(PageConfig(page_function=nbs_page, add_page_break=True if is_first_nbs_page else False))
+                if is_first_nbs_page:
+                    is_first_nbs_page = False
+            else:
+                print(f"   SKIPPING: NBS page for {center} {fy} due to missing graphs.")
+
+            if path_funding_pie and path_funding_grouped:
+                funding_page = partial(single_year_funding_page, 
+                                pie_chart_path=path_funding_pie, 
+                                grouped_bar_path=path_funding_grouped,
+                                center_name=center,
+                                fiscal_year_tag=fy,
+                                add_documentation_note=True if is_first_funding_page else False
+                                )
+
+                funding_pages.append(PageConfig(page_function=funding_page, add_page_break=True if is_first_funding_page else False))
+                center_funding_pages.append(PageConfig(page_function=funding_page, add_page_break=True if is_first_funding_page else False))
+                if is_first_funding_page:
+                    is_first_funding_page = False
+            else:
+                print(f"   SKIPPING: Funding page for {center} {fy} due to missing graphs.")
+
+        
+        center_builder = WordDocumentBuilder()
+        center_builder.create_document(
+            center_nbs_pages + network_wide_nbs + center_funding_pages + network_wide_funding,
+            os.path.join(out_path, f"{center}_MilestoneAttributionStudy.docx"),
+        )
+
+    master_builder = WordDocumentBuilder()
+    master_builder.create_document(
+        nbs_pages + network_wide_nbs + funding_pages + network_wide_funding, 
+        os.path.join(out_path, f"AllCenters_MilestoneAttributionStudy.docx")
+    )
+    
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--images",
+                        type=str,
+                        required=True,
+                        help="The folder containing all of the report images for the milestones report")
+
+    parser.add_argument("--out",
+                        type=str,
+                        required=True,
+                        help="The folder to place the generated reports into")
+
+    args = parser.parse_args()
+
+    if not os.path.exists(args.out):
+        os.makedirs(args.out)
+
+    # Run the parsing logic
+    grouped_image_data = parse_and_organize_images(args.images)
+    # print(grouped_image_data)
+    # Run the document generation loops
+    process_documents(grouped_image_data, args.out)