diff --git a/artifacts/go-benchmark/issue/xlsx/report/benchmark_coverage.json b/artifacts/go-benchmark/issue/xlsx/report/benchmark_coverage.json
index 35f4301a..c5db50bc 100644
--- a/artifacts/go-benchmark/issue/xlsx/report/benchmark_coverage.json
+++ b/artifacts/go-benchmark/issue/xlsx/report/benchmark_coverage.json
@@ -16,7 +16,7 @@
"missing_auxiliary_references": 0,
"comparison_completed": true,
"comparison_results": 24,
- "average_score": 0.49293750000000003,
+ "average_score": 0.5162083333333333,
"cases": [
{
"name": "Academic Achievement Summary Table",
@@ -25,7 +25,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/Academic Achievement Summary Table.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -37,7 +37,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/AcademicAchievement_temp.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -49,7 +49,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/Business expense budget1.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -61,7 +61,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/Business expenses budget2.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -73,7 +73,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/Business plan checklist with SWOT analysis1.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -85,7 +85,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/Event budget1.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -97,7 +97,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/Expense report basic1.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -109,7 +109,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/Grocery list1.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -121,7 +121,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/Issue202609031340.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -133,7 +133,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/payroll-calculator_f.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -145,7 +145,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/PO_anonymized.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -157,7 +157,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/Simple invoice1.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -169,7 +169,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/Small business cash flow forecast1.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -181,7 +181,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/Wedding_timeline_planner1_copy.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -193,7 +193,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/Weekly schedule planner1.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -205,7 +205,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/XlsxIssue152_ClusteredNonZeroBarChart.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -217,7 +217,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/XlsxIssue75.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -229,7 +229,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/XlsxIssue77_MergedCellAlignment.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -241,7 +241,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/XlsxIssue77_Template1.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -253,7 +253,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/XlsxIssue77_Template2_Workaround.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -265,7 +265,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/XlsxIssue81_LayoutOptions.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -277,7 +277,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/XlsxIssue82_5mb.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -289,7 +289,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/XlsxIssue82_SampleTestData5mb.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
@@ -301,7 +301,7 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/XlsxIssue82_WideTable.xlsx",
"conversion_status": "passed",
- "conversion_exit_code": null,
+ "conversion_exit_code": 0,
"candidate_exists": true,
"reference_exists": true,
"auxiliary_reference_exists": true
diff --git a/artifacts/go-benchmark/issue/xlsx/report/comparison_report.json b/artifacts/go-benchmark/issue/xlsx/report/comparison_report.json
index c5bd219d..66ec16ad 100644
--- a/artifacts/go-benchmark/issue/xlsx/report/comparison_report.json
+++ b/artifacts/go-benchmark/issue/xlsx/report/comparison_report.json
@@ -182,22 +182,22 @@
"format": "xlsx",
"source_path": "tests/Issue_Files/xlsx/Business expense budget1.xlsx",
"suite": "issue",
- "minipdf_size": 4633,
+ "minipdf_size": 85664,
"reference_size": 159864,
"pdf_valid": true,
- "minipdf_pages": 2,
+ "minipdf_pages": 4,
"reference_pages": 4,
"compared_pages": 3,
- "text_similarity": 0.4638,
- "flat_text_similarity": 0.0771,
- "word_text_similarity": 0.4638,
- "text_diff": "--- minipdf/Business expense budget1.pdf\n+++ reference/Business expense budget1.pdf\n@@ -1,42 +1,73 @@\n BUSINESS EXPENSE BUDGET\n\n Fiscal Year 2026 - Quarterly Report\n\n-CATEGORY Q1 BUDGET Q1 ACTUAL Q2 BUDGET Q2 ACTUAL VARIANCE\n\n+CATEGORY Q1 BUDGET Q1 ACTUAL Q2 BUDGET\n\n PERSONNEL\n\n-Salaries & Wages 45000 44200 46000 45800 1000\n\n-Benefits & Insurance 12000 11850 12500 12400 250\n\n-Training & Development 3000 2800 3200 3500 -100\n\n-Recruitment 2000 1500 2500 2200 800\n\n-Personnel Subtotal\n\n+Salaries & Wages 45,000.00 44,200.00 46,000.00\n\n+Benefits & Insurance 12,000.00 11,850.00 12,500.00\n\n+Training & Development 3,000.00 2,800.00 3,200.00\n\n+Recruitment 2,000.00 1,500.00 2,500.00\n\n+Personnel Subtotal 62,000.00 60,350.00 64,200.00\n\n OPERATIONS\n\n-Rent & Utilities 8000 8000 8200 8150 50\n\n-Office Supplies 1500 1650 1500 1400 -50\n\n-Equipment & Maintenance 3000 2700 3500 3800 0\n\n-IT & Software 4000 4200 4500 4300 0\n\n-Telecommunications 2000 1900 2100 2050 150\n\n-Operations Subtotal\n\n+Rent & Utilities 8,000.00 8,000.00 8,200.00\n\n+Office Supplies 1,500.00 1,650.00 1,500.00\n\n+Equipment & Maintenance 3,000.00 2,700.00 3,500.00\n\n+IT & Software 4,000.00 4,200.00 4,500.00\n\n+Telecommunications 2,000.00 1,900.00 2,100.00\n\n+Operations Subtotal 18,500.00 18,450.00 19,800.00\n\n MARKETING\n\n-Advertising 5000 4800 6000 5500 700\n\n-Digital Marketing 3000 3200 3500 3600 -300\n\n-Events & Sponsorships 2000 1800 2500 2800 -100\n\n-Printing & Design 1000 950 1200 1100 150\n\n-Marketing Subtotal\n\n+Advertising 5,000.00 4,800.00 6,000.00\n\n+Digital Marketing 3,000.00 3,200.00 3,500.00\n\n+Events & Sponsorships 2,000.00 1,800.00 2,500.00\n\n+Printing & Design 1,000.00 950.00 1,200.00\n\n+Marketing Subtotal 11,000.00 10,750.00 13,200.00\n\n TRAVEL & ENTERTAINMENT\n\n-Domestic Travel 3000 2500 3500 3200 800\n\n-International Travel 5000 4000 6000 5500 1500\n\n-Meals & Entertainment 2000 1800 2200 2100 300\n\n-Vehicle Expenses 1500 1400 1500 1600 0\n\n-Travel & Entertainment Subtotal\n\n+Domestic Travel 3,000.00 2,500.00 3,500.00\n\n+International Travel 5,000.00 4,000.00 6,000.00\n\n+Meals & Entertainment 2,000.00 1,800.00 2,200.00\n\n+Vehicle Expenses 1,500.00 1,400.00 1,500.00\n\n+Travel & Entertainment Subtotal 11,500.00 9,700.00 13,200.00\n\n PROFESSIONAL SERVICES\n\n-Legal Fees 3000 2800 3500 3200 500\n\n-Accounting & Audit 2500 2500 2500 2500 0\n\n-Consulting 4000 3500 5000 4800 700\n\n-Insurance 2000 2000 2100 2100 0\n\n-Professional Services Subtotal\n\n-GRAND TOTAL\n\n+Legal Fees 3,000.00 2,800.00 3,500.00\n\n+Accounting & Audit 2,500.00 2,500.00 2,500.00\n\n+Consulting 4,000.00 3,500.00 5,000.00\n\n+Insurance 2,000.00 2,000.00 2,100.00\n\n+Professional Services Subtotal 11,500.00 10,800.00 13,100.00\n\n+---PAGE---\n\n+GRAND TOTAL 114,500.00 110,050.00 123,500.00\n\n Category Total Budget Total Actual\n\n Personnel 126200 124250\n\n Operations 38300 38150\n\n Marketing 24200 23750\n\n Travel & Entertainment 24700 22100\n\n+Professional Services 24600 23400\n\n+Budget vs Actual by Category\n\n+Amount (\\$)\n\n+Category\n\n ---PAGE---\n\n-Professional Services 24600 23400\n+Q2 ACTUAL VARIANCE\n\n+45,800.00 1,000.00\n\n+12,400.00 250.00\n\n+3,500.00 -100.00\n\n+2,200.00 800.00\n\n+63,900.00 1,950.00\n\n+8,150.00 50.00\n\n+1,400.00 -50.00\n\n+3,800.00 0.00\n\n+4,300.00 0.00\n\n+2,050.00 150.00\n\n+19,700.00 150.00\n\n+5,500.00 700.00\n\n+3,600.00 -300.00\n\n+2,800.00 -100.00\n\n+1,100.00 150.00\n\n+13,000.00 450.00\n\n+3,200.00 800.00\n\n+5,500.00 1,500.00\n\n+2,100.00 300.00\n\n+1,600.00 0.00\n\n+12,400.00 2,600.00\n\n+3,200.00 500.00\n\n+2,500.00 0.00\n\n+4,800.00 700.00\n\n+2,100.00 0.00\n\n+12,600.00 1,200.00",
+ "text_similarity": 1.0,
+ "flat_text_similarity": 1.0,
+ "word_text_similarity": 1.0,
+ "text_diff": "(identical)",
"visual_scores": [
- 0.5166,
- 0.4059,
- 0.0
+ 0.8898,
+ 0.9049,
+ 0.9575
],
- "visual_avg": 0.3075,
+ "visual_avg": 0.9174,
"diff_images": [
{
"page": 1,
@@ -211,15 +211,15 @@
1754
],
"difference_bbox_px": [
- 113,
- 127,
+ 146,
+ 176,
996,
- 1614
+ 1579
],
- "changed_pixels": 713208,
- "changed_fraction": 0.327654,
- "mean_abs_rgb": 40.6046,
- "rmse_rgb": 86.4236,
+ "changed_pixels": 305806,
+ "changed_fraction": 0.14049,
+ "mean_abs_rgb": 21.3092,
+ "rmse_rgb": 63.9244,
"threshold": 12,
"gain": 5.0
},
@@ -237,15 +237,15 @@
1754
],
"difference_bbox_px": [
- 114,
- 96,
+ 112,
+ 178,
996,
- 1225
+ 1226
],
- "changed_pixels": 212045,
- "changed_fraction": 0.097415,
- "mean_abs_rgb": 14.3262,
- "rmse_rgb": 50.0036,
+ "changed_pixels": 207301,
+ "changed_fraction": 0.095236,
+ "mean_abs_rgb": 11.1807,
+ "rmse_rgb": 40.9815,
"threshold": 12,
"gain": 5.0
},
@@ -253,13 +253,32 @@
},
{
"page": 3,
- "minipdf_img": null,
+ "minipdf_img": "Business expense budget1_p3_minipdf.png",
"reference_img": "Business expense budget1_p3_reference.png",
"auxiliary_img": "Business expense budget1_p3_libreoffice.png",
+ "heatmap_img": "Business expense budget1_p3_heatmap.png",
+ "heatmap_metrics": {
+ "size_px": [
+ 1241,
+ 1754
+ ],
+ "difference_bbox_px": [
+ 111,
+ 296,
+ 444,
+ 1579
+ ],
+ "changed_pixels": 109374,
+ "changed_fraction": 0.050247,
+ "mean_abs_rgb": 7.5647,
+ "rmse_rgb": 37.8974,
+ "threshold": 12,
+ "gain": 5.0
+ },
"composite_img": "Business expense budget1_p3_go_minipdf_vs_microsoft_365_excel_reference_vs_libreoffice.png"
}
],
- "overall_score": 0.4085
+ "overall_score": 0.967
},
{
"name": "Business expenses budget2",
diff --git a/artifacts/go-benchmark/issue/xlsx/report/comparison_report.md b/artifacts/go-benchmark/issue/xlsx/report/comparison_report.md
index 22705a13..ccd41446 100644
--- a/artifacts/go-benchmark/issue/xlsx/report/comparison_report.md
+++ b/artifacts/go-benchmark/issue/xlsx/report/comparison_report.md
@@ -1,6 +1,6 @@
# go MiniPdf vs Microsoft 365 Excel Reference PDF Comparison Report
-Generated: 2026-09-14T20:24:21.049172
+Generated: 2026-09-14T22:49:48.552791
## Summary
@@ -8,7 +8,7 @@ Generated: 2026-09-14T20:24:21.049172
|---|-----------|-------|----------|------------|-------------|--------|
| 1 | 🔴 Academic Achievement Summary Table | ✅ | 0.3235 | 0.4218 | 9/2 | **0.3981** |
| 2 | 🔴 AcademicAchievement_temp | ✅ | 0.3235 | 0.4218 | 9/2 | **0.3981** |
-| 3 | 🔴 Business expense budget1 | ✅ | 0.4638 | 0.3075 | 2/4 | **0.4085** |
+| 3 | 🟢 Business expense budget1 | ✅ | 1.0 | 0.9174 | 4/4 | **0.967** |
| 4 | 🔴 Business expenses budget2 | ✅ | 0.119 | 0.0808 | 14/4 | **0.1799** |
| 5 | 🔴 Business plan checklist with SWOT analysis1 | ✅ | 0.8729 | 0.2639 | 1/1 | **0.6547** |
| 6 | 🔴 Event budget1 | ✅ | 0.712 | 0.5681 | 4/5 | **0.612** |
@@ -31,7 +31,7 @@ Generated: 2026-09-14T20:24:21.049172
| 23 | 🔴 XlsxIssue82_WideTable | ✅ | 0.8139 | 0.5165 | 15/13 | **0.6322** |
| 24 | 🟡 XlsxIssue152_ClusteredNonZeroBarChart | ✅ | 0.6923 | 0.5806 | 1/1 | **0.7092** |
-**Average Overall Score: 0.4929**
+**Average Overall Score: 0.5162**
## Labeled Side-by-Side Comparison
@@ -296,12 +296,17 @@ Blue areas are below the configured difference threshold; red areas have stronge
Business expense budget1
Page 1 |
 |
- changed: 713208 px (32.77%)
bbox: [113, 127, 996, 1614]
mean abs RGB: 40.6046
RMSE RGB: 86.4236
threshold: 12, gain: 5.0 |
+ changed: 305806 px (14.05%)
bbox: [146, 176, 996, 1579]
mean abs RGB: 21.3092
RMSE RGB: 63.9244
threshold: 12, gain: 5.0 |
Business expense budget1
Page 2 |
 |
- changed: 212045 px (9.74%)
bbox: [114, 96, 996, 1225]
mean abs RGB: 14.3262
RMSE RGB: 50.0036
threshold: 12, gain: 5.0 |
+ changed: 207301 px (9.52%)
bbox: [112, 178, 996, 1226]
mean abs RGB: 11.1807
RMSE RGB: 40.9815
threshold: 12, gain: 5.0 |
+
+
+ Business expense budget1
Page 3 |
+  |
+ changed: 109374 px (5.02%)
bbox: [111, 296, 444, 1579]
mean abs RGB: 7.5647
RMSE RGB: 37.8974
threshold: 12, gain: 5.0 |
Business expenses budget2
Page 1 |
@@ -601,7 +606,7 @@ Scores compare go MiniPdf against Microsoft 365 Excel Reference. LibreOffice is
Business expense budget1
format: xlsx | case: Business expense budget1 | scope: go-issue-xlsx |
- Business expense budget1 ⬤ 40.8% |
+ Business expense budget1 ⬤ 96.7% |
 |
@@ -614,7 +619,7 @@ Scores compare go MiniPdf against Microsoft 365 Excel Reference. LibreOffice is
 |
- | missing |
+  |
 |
 |
@@ -1265,171 +1270,13 @@ Scores compare go MiniPdf against Microsoft 365 Excel Reference. LibreOffice is
- **Case Metadata:** format: xlsx | case: Business expense budget1 | scope: go-issue-xlsx
- **Source:** tests/Issue_Files/xlsx/Business expense budget1.xlsx
-- **Text Similarity:** 0.4638
-- **Visual Average:** 0.3075
-- **Overall Score:** 0.4085
-- **Pages:** MiniPdf=2, Reference=4
-- **File Size:** MiniPdf=4633 bytes, Reference=159864 bytes
-
-Text Diff
-
-```diff
---- minipdf/Business expense budget1.pdf
-+++ reference/Business expense budget1.pdf
-@@ -1,42 +1,73 @@
- BUSINESS EXPENSE BUDGET
-
- Fiscal Year 2026 - Quarterly Report
-
--CATEGORY Q1 BUDGET Q1 ACTUAL Q2 BUDGET Q2 ACTUAL VARIANCE
-
-+CATEGORY Q1 BUDGET Q1 ACTUAL Q2 BUDGET
-
- PERSONNEL
-
--Salaries & Wages 45000 44200 46000 45800 1000
-
--Benefits & Insurance 12000 11850 12500 12400 250
-
--Training & Development 3000 2800 3200 3500 -100
-
--Recruitment 2000 1500 2500 2200 800
-
--Personnel Subtotal
-
-+Salaries & Wages 45,000.00 44,200.00 46,000.00
-
-+Benefits & Insurance 12,000.00 11,850.00 12,500.00
-
-+Training & Development 3,000.00 2,800.00 3,200.00
-
-+Recruitment 2,000.00 1,500.00 2,500.00
-
-+Personnel Subtotal 62,000.00 60,350.00 64,200.00
-
- OPERATIONS
-
--Rent & Utilities 8000 8000 8200 8150 50
-
--Office Supplies 1500 1650 1500 1400 -50
-
--Equipment & Maintenance 3000 2700 3500 3800 0
-
--IT & Software 4000 4200 4500 4300 0
-
--Telecommunications 2000 1900 2100 2050 150
-
--Operations Subtotal
-
-+Rent & Utilities 8,000.00 8,000.00 8,200.00
-
-+Office Supplies 1,500.00 1,650.00 1,500.00
-
-+Equipment & Maintenance 3,000.00 2,700.00 3,500.00
-
-+IT & Software 4,000.00 4,200.00 4,500.00
-
-+Telecommunications 2,000.00 1,900.00 2,100.00
-
-+Operations Subtotal 18,500.00 18,450.00 19,800.00
-
- MARKETING
-
--Advertising 5000 4800 6000 5500 700
-
--Digital Marketing 3000 3200 3500 3600 -300
-
--Events & Sponsorships 2000 1800 2500 2800 -100
-
--Printing & Design 1000 950 1200 1100 150
-
--Marketing Subtotal
-
-+Advertising 5,000.00 4,800.00 6,000.00
-
-+Digital Marketing 3,000.00 3,200.00 3,500.00
-
-+Events & Sponsorships 2,000.00 1,800.00 2,500.00
+- **Text Similarity:** 1.0
+- **Visual Average:** 0.9174
+- **Overall Score:** 0.967
+- **Pages:** MiniPdf=4, Reference=4
+- **File Size:** MiniPdf=85664 bytes, Reference=159864 bytes
-+Printing & Design 1,000.00 950.00 1,200.00
-
-+Marketing Subtotal 11,000.00 10,750.00 13,200.00
-
- TRAVEL & ENTERTAINMENT
-
--Domestic Travel 3000 2500 3500 3200 800
-
--International Travel 5000 4000 6000 5500 1500
-
--Meals & Entertainment 2000 1800 2200 2100 300
-
--Vehicle Expenses 1500 1400 1500 1600 0
-
--Travel & Entertainment Subtotal
-
-+Domestic Travel 3,000.00 2,500.00 3,500.00
-
-+International Travel 5,000.00 4,000.00 6,000.00
-
-+Meals & Entertainment 2,000.00 1,800.00 2,200.00
-
-+Vehicle Expenses 1,500.00 1,400.00 1,500.00
-
-+Travel & Entertainment Subtotal 11,500.00 9,700.00 13,200.00
-
- PROFESSIONAL SERVICES
-
--Legal Fees 3000 2800 3500 3200 500
-
--Accounting & Audit 2500 2500 2500 2500 0
-
--Consulting 4000 3500 5000 4800 700
-
--Insurance 2000 2000 2100 2100 0
-
--Professional Services Subtotal
-
--GRAND TOTAL
-
-+Legal Fees 3,000.00 2,800.00 3,500.00
-
-+Accounting & Audit 2,500.00 2,500.00 2,500.00
-
-+Consulting 4,000.00 3,500.00 5,000.00
-
-+Insurance 2,000.00 2,000.00 2,100.00
-
-+Professional Services Subtotal 11,500.00 10,800.00 13,100.00
-
-+---PAGE---
-
-+GRAND TOTAL 114,500.00 110,050.00 123,500.00
-
- Category Total Budget Total Actual
-
- Personnel 126200 124250
-
- Operations 38300 38150
-
- Marketing 24200 23750
-
- Travel & Entertainment 24700 22100
-
-+Professional Services 24600 23400
-
-+Budget vs Actual by Category
-
-+Amount (\$)
-
-+Category
-
- ---PAGE---
-
--Professional Services 24600
-... (503 more characters)
-
-```
-HEADER
` +
+ `BudgetSUM(B3:B4)
` +
+ `20000
25000
` +
+ `` +
+ ``,
+ })
+ files, err := openOfficePackage(input)
+ if err != nil {
+ t.Fatal(err)
+ }
+ worksheet, err := files.read("xl/worksheets/sheet1.xml")
+ if err != nil {
+ t.Fatal(err)
+ }
+ grid, err := readXLSXGrid(worksheet, nil)
+ if err != nil {
+ t.Fatal(err)
+ }
+ if grid.maxColumn != 2 || grid.maxRow != 4 {
+ t.Fatalf("printable grid = %dx%d, want 2x4", grid.maxColumn, grid.maxRow)
+ }
+
+ pdf, err := ConvertBytesToPDF(input)
+ assertPDFContains(t, pdf, err, "HEADER", "45,000.00", "0.12156862745098039 0.3058823529411765 0.4745098039215686 rg", " re f", " RG ")
+}
+
+func TestRenderXLSXCellMapsCalibriMetricsAndVerticalAlignment(t *testing.T) {
+ style := xlsxStyle{font: xlsxFont{name: "Calibri", size: 11, color: PDFColorBlack}}
+ page := &PDFPage{}
+ renderXLSXCell(page, xlsxCell{value: "Cell"}, 10, 20, 100, 15, style, true)
+
+ bottom := page.operations[0].(textOperation)
+ if bottom.fontSize != 11*calibriToHelveticaFontScale || bottom.horizontalScale != calibriRegularWidthScale {
+ t.Fatalf("Calibri fallback metrics = size %v, scale %v", bottom.fontSize, bottom.horizontalScale)
+ }
+ if bottom.y != 20+bottom.fontSize*helveticaBaselineRatio {
+ t.Fatalf("bottom-aligned baseline = %v", bottom.y)
+ }
+
+ style.vertical = "center"
+ style.align = "center"
+ page.operations = nil
+ renderXLSXCell(page, xlsxCell{value: "Cell"}, 10, 20, 100, 28, style, true)
+ centered := page.operations[0].(textOperation)
+ if centered.y != 20+(28-centered.fontSize)/2 {
+ t.Fatalf("centered baseline = %v", centered.y)
+ }
+ expectedX := 10 + (100-measureHelveticaText("Cell", centered.fontSize, centered.horizontalScale, false))/2
+ if centered.x != expectedX {
+ t.Fatalf("centered x = %v, want %v", centered.x, expectedX)
+ }
+
+ style.font.bold = true
+ page.operations = nil
+ renderXLSXCell(page, xlsxCell{value: "Cell"}, 10, 20, 100, 28, style, true)
+ bold := page.operations[0].(textOperation)
+ if bold.horizontalScale != calibriBoldWidthScale {
+ t.Fatalf("bold Calibri fallback scale = %v", bold.horizontalScale)
+ }
+}
+
+func TestXLSXDefaultRowHeightUsesExcelPDFScale(t *testing.T) {
+ grid := xlsxGrid{defaultRowHeight: 15, rowHeights: map[int]float64{2: 22}}
+ if height := grid.rowHeight(1); height < 14.599 || height > 14.601 {
+ t.Fatalf("default row height = %v, want 14.6", height)
+ }
+ if height := grid.rowHeight(2); height != 22 {
+ t.Fatalf("explicit row height = %v, want 22", height)
+ }
+}
+
+func TestConvertXLSXRendersClusteredColumnChart(t *testing.T) {
+ input := officePackageBytes(t, map[string]string{
+ "xl/styles.xml": ``,
+ "xl/worksheets/sheet1.xml": `` +
+ `CategoryBudgetActual
` +
+ `One108
` +
+ `Two2018
` +
+ `SUM(C2:C3)
`,
+ "xl/worksheets/_rels/sheet1.xml.rels": ``,
+ "xl/drawings/drawing1.xml": `004
0![]()
`,
+ "xl/drawings/_rels/drawing1.xml.rels": ``,
+ "xl/charts/chart1.xml": `<tx><rich><p><r><t>Budget vs Actual</t></r></p></rich></tx>` +
+ `Sheet1!B1Sheet1!A2:A3Sheet1!B2:B3` +
+ `Sheet1!C1Sheet1!A2:A3Sheet1!C2:C3` +
+ `<tx><rich><p><r><t>Category</t></r></p></rich></tx>` +
+ `<tx><rich><p><r><t>Amount</t></r></p></rich></tx>`,
+ })
+
+ pdf, err := ConvertBytesToPDF(input)
+ assertPDFContains(t, pdf, err, "Budget vs Actual", "Budget", "Actual", "Category", "Amount", "0.31 0.506 0.741 rg", "0.753 0.314 0.302 rg")
+}
+
+func TestNeedsCalculatedXLSXGridOnlyForFullyUncachedFormulas(t *testing.T) {
+ for name, test := range map[string]struct {
+ cells string
+ want bool
+ }{
+ "uncached": {cells: `SUM(A2:A3)`, want: true},
+ "cached": {cells: `SUM(A2:A3)3`},
+ "mixed": {cells: `A2B22`},
+ "plain": {cells: `1`},
+ } {
+ t.Run(name, func(t *testing.T) {
+ worksheet := []byte(`` + test.cells + `
`)
+ if got := needsCalculatedXLSXGrid(worksheet); got != test.want {
+ t.Fatalf("needsCalculatedXLSXGrid() = %v, want %v", got, test.want)
+ }
+ })
+ }
+}
+
func TestXLSXRowColumnLimitsAndOrientation(t *testing.T) {
input := officePackageBytes(t, map[string]string{
"xl/workbook.xml": ``,
diff --git a/minipdf-go/pdf.go b/minipdf-go/pdf.go
index 5d37b60d..fe5ec9d1 100644
--- a/minipdf-go/pdf.go
+++ b/minipdf-go/pdf.go
@@ -53,7 +53,13 @@ func (document *PDFDocument) AddPage(width, height float64) *PDFPage {
func (page *PDFPage) AddText(text string, x, y, fontSize float64, color PDFColor, bold bool) {
page.operations = append(page.operations, textOperation{
- text: text, x: x, y: y, fontSize: fontSize, color: color, bold: bold,
+ text: text, x: x, y: y, fontSize: fontSize, horizontalScale: 100, color: color, bold: bold,
+ })
+}
+
+func (page *PDFPage) addScaledText(text string, x, y, fontSize, horizontalScale float64, color PDFColor, bold bool) {
+ page.operations = append(page.operations, textOperation{
+ text: text, x: x, y: y, fontSize: fontSize, horizontalScale: horizontalScale, color: color, bold: bold,
})
}
@@ -70,17 +76,23 @@ func (page *PDFPage) AddLine(x1, y1, x2, y2 float64, color PDFColor, width float
}
type textOperation struct {
- text string
- x, y, fontSize float64
- color PDFColor
- bold bool
+ text string
+ x, y, fontSize float64
+ horizontalScale float64
+ color PDFColor
+ bold bool
}
func (operation textOperation) appendPDF(buffer *bytes.Buffer, embedded *embeddedFont) {
+ horizontalScale := ""
+ if operation.horizontalScale != 0 && operation.horizontalScale != 100 {
+ horizontalScale = pdfNumber(operation.horizontalScale) + " Tz "
+ }
if embedded != nil {
if encoded, ok := embedded.encode(operation.text); ok {
- fmt.Fprintf(buffer, "BT /FU1 %s Tf %s %s %s rg %s %s Td <%s> Tj ET\n",
+ fmt.Fprintf(buffer, "BT /FU1 %s Tf %s%s %s %s rg %s %s Td <%s> Tj ET\n",
pdfNumber(operation.fontSize),
+ horizontalScale,
pdfNumber(operation.color.Red),
pdfNumber(operation.color.Green),
pdfNumber(operation.color.Blue),
@@ -95,9 +107,10 @@ func (operation textOperation) appendPDF(buffer *bytes.Buffer, embedded *embedde
if operation.bold {
font = "F2"
}
- fmt.Fprintf(buffer, "BT /%s %s Tf %s %s %s rg %s %s Td (%s) Tj ET\n",
+ fmt.Fprintf(buffer, "BT /%s %s Tf %s%s %s %s rg %s %s Td (%s) Tj ET\n",
font,
pdfNumber(operation.fontSize),
+ horizontalScale,
pdfNumber(operation.color.Red),
pdfNumber(operation.color.Green),
pdfNumber(operation.color.Blue),
diff --git a/minipdf-go/xlsx.go b/minipdf-go/xlsx.go
index 21944730..7800a98d 100644
--- a/minipdf-go/xlsx.go
+++ b/minipdf-go/xlsx.go
@@ -21,6 +21,26 @@ func convertXLSX(input []byte, options ConversionOptions) ([]byte, error) {
if len(worksheets) == 0 {
return nil, errorsNewMissingWorksheets()
}
+ styles, err := readXLSXStyles(files)
+ if err != nil {
+ return nil, err
+ }
+ if len(styles) > 0 {
+ needsGridRendering := false
+ for _, name := range worksheets {
+ worksheetXML, readErr := files.read(name)
+ if readErr != nil {
+ return nil, readErr
+ }
+ if needsCalculatedXLSXGrid(worksheetXML) {
+ needsGridRendering = true
+ break
+ }
+ }
+ if needsGridRendering {
+ return renderXLSXWorksheets(files, worksheets, sharedStrings, styles, options)
+ }
+ }
pages := make([]textPage, 0, len(worksheets))
fallbackPageSize := PageSizeA4
for _, name := range worksheets {
@@ -255,24 +275,34 @@ func worksheetColumnIndex(reference string) int {
}
func decodeWorksheetCell(decoder *xml.Decoder, start xml.StartElement, sharedStrings []string) (string, error) {
+ value, _, err := decodeWorksheetCellContent(decoder, start, sharedStrings)
+ return value, err
+}
+
+func decodeWorksheetCellContent(decoder *xml.Decoder, start xml.StartElement, sharedStrings []string) (string, string, error) {
cellType := attrValue(start, "t")
var value strings.Builder
+ var formula string
depth := 1
for depth > 0 {
token, err := decoder.Token()
if err != nil {
- return "", err
+ return "", "", err
}
switch element := token.(type) {
case xml.StartElement:
depth++
- if element.Name.Local == "v" || element.Name.Local == "t" {
+ if element.Name.Local == "v" || element.Name.Local == "t" || element.Name.Local == "f" {
var text string
if err := decoder.DecodeElement(&text, &element); err != nil {
- return "", err
+ return "", "", err
}
depth--
- value.WriteString(text)
+ if element.Name.Local == "f" {
+ formula = text
+ } else {
+ value.WriteString(text)
+ }
}
case xml.EndElement:
depth--
@@ -282,16 +312,16 @@ func decodeWorksheetCell(decoder *xml.Decoder, start xml.StartElement, sharedStr
if cellType == "s" {
index, err := strconv.Atoi(strings.TrimSpace(rawValue))
if err == nil && index >= 0 && index < len(sharedStrings) {
- return sharedStrings[index], nil
+ return sharedStrings[index], formula, nil
}
}
if cellType == "b" {
switch strings.TrimSpace(rawValue) {
case "1":
- return "TRUE", nil
+ return "TRUE", formula, nil
case "0":
- return "FALSE", nil
+ return "FALSE", formula, nil
}
}
- return rawValue, nil
+ return rawValue, formula, nil
}
diff --git a/minipdf-go/xlsx_chart.go b/minipdf-go/xlsx_chart.go
new file mode 100644
index 00000000..934b7338
--- /dev/null
+++ b/minipdf-go/xlsx_chart.go
@@ -0,0 +1,457 @@
+package minipdf
+
+import (
+ "encoding/xml"
+ "fmt"
+ "math"
+ "path"
+ "strconv"
+ "strings"
+)
+
+const xlsxEmuPerPoint = 12700
+
+type xlsxChartSeries struct {
+ name string
+ categories []string
+ values []float64
+}
+
+type xlsxChart struct {
+ anchorRow, anchorColumn int
+ rowOffset, columnOffset float64
+ width, height float64
+ title, categoryTitle string
+ valueTitle, legendPosition string
+ gapWidth float64
+ series []xlsxChartSeries
+}
+
+type xlsxRelationshipsDocument struct {
+ Items []struct {
+ ID string `xml:"Id,attr"`
+ Type string `xml:"Type,attr"`
+ Target string `xml:"Target,attr"`
+ } `xml:"Relationship"`
+}
+
+type xlsxDrawingMarker struct {
+ Column int `xml:"col"`
+ ColumnOffset int64 `xml:"colOff"`
+ Row int `xml:"row"`
+ RowOffset int64 `xml:"rowOff"`
+}
+
+type xlsxChartAnchor struct {
+ From xlsxDrawingMarker `xml:"from"`
+ To xlsxDrawingMarker `xml:"to"`
+ Extension struct {
+ Width int64 `xml:"cx,attr"`
+ Height int64 `xml:"cy,attr"`
+ } `xml:"ext"`
+ Frame struct {
+ Reference struct {
+ ID string `xml:"id,attr"`
+ } `xml:"graphic>graphicData>chart"`
+ } `xml:"graphicFrame"`
+}
+
+type xlsxDrawingDocument struct {
+ OneCell []xlsxChartAnchor `xml:"oneCellAnchor"`
+ TwoCell []xlsxChartAnchor `xml:"twoCellAnchor"`
+}
+
+type xlsxChartTitleXML struct {
+ Runs []string `xml:"tx>rich>p>r>t"`
+}
+
+type xlsxChartSeriesXML struct {
+ Name string `xml:"tx>v"`
+ NameFormula string `xml:"tx>strRef>f"`
+ CategoryNum string `xml:"cat>numRef>f"`
+ CategoryStr string `xml:"cat>strRef>f"`
+ Value string `xml:"val>numRef>f"`
+}
+
+type xlsxChartDocument struct {
+ Chart struct {
+ Title xlsxChartTitleXML `xml:"title"`
+ Plot struct {
+ Bar struct {
+ Series []xlsxChartSeriesXML `xml:"ser"`
+ GapWidth struct {
+ Value float64 `xml:"val,attr"`
+ } `xml:"gapWidth"`
+ } `xml:"barChart"`
+ CategoryAxis struct {
+ Title xlsxChartTitleXML `xml:"title"`
+ } `xml:"catAx"`
+ ValueAxis struct {
+ Title xlsxChartTitleXML `xml:"title"`
+ } `xml:"valAx"`
+ } `xml:"plotArea"`
+ Legend struct {
+ Position struct {
+ Value string `xml:"val,attr"`
+ } `xml:"legendPos"`
+ } `xml:"legend"`
+ } `xml:"chart"`
+}
+
+func readXLSXCharts(files officePackage, worksheetName string, grid *xlsxGrid) ([]xlsxChart, error) {
+ relationshipsName := xlsxRelationshipPartName(worksheetName)
+ _, ok := files[relationshipsName]
+ if !ok {
+ return nil, nil
+ }
+ relationshipsData, err := files.read(relationshipsName)
+ if err != nil {
+ return nil, err
+ }
+ var relationships xlsxRelationshipsDocument
+ if err := xml.Unmarshal(relationshipsData, &relationships); err != nil {
+ return nil, fmt.Errorf("parse %s: %w", relationshipsName, err)
+ }
+ drawingName := ""
+ for _, relationship := range relationships.Items {
+ if strings.HasSuffix(relationship.Type, "/drawing") {
+ drawingName = xlsxResolveRelationshipTarget(worksheetName, relationship.Target)
+ break
+ }
+ }
+ if drawingName == "" {
+ return nil, nil
+ }
+
+ drawingData, err := files.read(drawingName)
+ if err != nil {
+ return nil, err
+ }
+ drawingRelationshipsName := xlsxRelationshipPartName(drawingName)
+ drawingRelationshipsData, err := files.read(drawingRelationshipsName)
+ if err != nil {
+ return nil, err
+ }
+ var drawingRelationships xlsxRelationshipsDocument
+ if err := xml.Unmarshal(drawingRelationshipsData, &drawingRelationships); err != nil {
+ return nil, fmt.Errorf("parse %s: %w", drawingRelationshipsName, err)
+ }
+ chartParts := make(map[string]string)
+ for _, relationship := range drawingRelationships.Items {
+ if strings.HasSuffix(relationship.Type, "/chart") {
+ chartParts[relationship.ID] = xlsxResolveRelationshipTarget(drawingName, relationship.Target)
+ }
+ }
+
+ var drawing xlsxDrawingDocument
+ if err := xml.Unmarshal(drawingData, &drawing); err != nil {
+ return nil, fmt.Errorf("parse %s: %w", drawingName, err)
+ }
+ anchors := append(drawing.OneCell, drawing.TwoCell...)
+ charts := make([]xlsxChart, 0, len(anchors))
+ for _, anchor := range anchors {
+ chartName := chartParts[anchor.Frame.Reference.ID]
+ if chartName == "" {
+ continue
+ }
+ width := float64(anchor.Extension.Width) / xlsxEmuPerPoint
+ height := float64(anchor.Extension.Height) / xlsxEmuPerPoint
+ if width <= 0 {
+ width = grid.columnOffset(anchor.To.Column) - grid.columnOffset(anchor.From.Column)
+ }
+ if height <= 0 {
+ height = grid.rowOffset(anchor.To.Row) - grid.rowOffset(anchor.From.Row)
+ }
+ if width <= 0 || height <= 0 {
+ continue
+ }
+ chartData, err := files.read(chartName)
+ if err != nil {
+ return nil, err
+ }
+ chart, ok := parseXLSXChart(chartData, *grid)
+ if !ok {
+ continue
+ }
+ chart.anchorRow = anchor.From.Row
+ chart.anchorColumn = anchor.From.Column
+ chart.rowOffset = float64(anchor.From.RowOffset) / xlsxEmuPerPoint
+ chart.columnOffset = float64(anchor.From.ColumnOffset) / xlsxEmuPerPoint
+ chart.width, chart.height = width, height
+ charts = append(charts, chart)
+ grid.extendForChart(chart)
+ }
+ return charts, nil
+}
+
+func xlsxRelationshipPartName(partName string) string {
+ return path.Join(path.Dir(partName), "_rels", path.Base(partName)+".rels")
+}
+
+func xlsxResolveRelationshipTarget(partName, target string) string {
+ if strings.HasPrefix(target, "/") {
+ return strings.TrimPrefix(path.Clean(target), "/")
+ }
+ return path.Clean(path.Join(path.Dir(partName), target))
+}
+
+func parseXLSXChart(data []byte, grid xlsxGrid) (xlsxChart, bool) {
+ var document xlsxChartDocument
+ if xml.Unmarshal(data, &document) != nil || len(document.Chart.Plot.Bar.Series) == 0 {
+ return xlsxChart{}, false
+ }
+ chart := xlsxChart{
+ title: xlsxChartTitle(document.Chart.Title),
+ categoryTitle: xlsxChartTitle(document.Chart.Plot.CategoryAxis.Title),
+ valueTitle: xlsxChartTitle(document.Chart.Plot.ValueAxis.Title),
+ legendPosition: document.Chart.Legend.Position.Value,
+ gapWidth: document.Chart.Plot.Bar.GapWidth.Value,
+ }
+ if chart.gapWidth <= 0 {
+ chart.gapWidth = 150
+ }
+ for _, source := range document.Chart.Plot.Bar.Series {
+ name := source.Name
+ if name == "" {
+ values := xlsxReferenceValues(grid, source.NameFormula)
+ if len(values) > 0 {
+ name = values[0]
+ }
+ }
+ categoryFormula := source.CategoryStr
+ if categoryFormula == "" {
+ categoryFormula = source.CategoryNum
+ }
+ valueStrings := xlsxReferenceValues(grid, source.Value)
+ values := make([]float64, len(valueStrings))
+ for index, value := range valueStrings {
+ values[index], _ = strconv.ParseFloat(strings.TrimSpace(value), 64)
+ }
+ chart.series = append(chart.series, xlsxChartSeries{
+ name: name, categories: xlsxReferenceValues(grid, categoryFormula), values: values,
+ })
+ }
+ return chart, len(chart.series) > 0
+}
+
+func xlsxChartTitle(title xlsxChartTitleXML) string {
+ return strings.Join(title.Runs, "")
+}
+
+func xlsxReferenceValues(grid xlsxGrid, formula string) []string {
+ if separator := strings.LastIndex(formula, "!"); separator >= 0 {
+ formula = formula[separator+1:]
+ }
+ formula = strings.ReplaceAll(formula, "$", "")
+ parts := strings.Split(formula, ":")
+ if len(parts) == 0 || len(parts) > 2 {
+ return nil
+ }
+ startRow, startColumn := xlsxCellPosition(parts[0])
+ endRow, endColumn := startRow, startColumn
+ if len(parts) == 2 {
+ endRow, endColumn = xlsxCellPosition(parts[1])
+ }
+ if startRow == 0 || startColumn == 0 || endRow < startRow || endColumn < startColumn {
+ return nil
+ }
+ values := make([]string, 0, (endRow-startRow+1)*(endColumn-startColumn+1))
+ for row := startRow; row <= endRow; row++ {
+ for column := startColumn; column <= endColumn; column++ {
+ values = append(values, grid.cells[[2]int{row, column}].value)
+ }
+ }
+ return values
+}
+
+func (grid xlsxGrid) columnOffset(columnCount int) float64 {
+ offset := 0.0
+ for column := 1; column <= columnCount; column++ {
+ offset += grid.columnWidth(column)
+ }
+ return offset
+}
+
+func (grid xlsxGrid) rowOffset(rowCount int) float64 {
+ offset := 0.0
+ for row := 1; row <= rowCount; row++ {
+ offset += grid.rowHeight(row)
+ }
+ return offset
+}
+
+func (grid *xlsxGrid) extendForChart(chart xlsxChart) {
+ chartRight := grid.columnOffset(chart.anchorColumn) + chart.columnOffset + chart.width
+ for column := chart.anchorColumn + 1; grid.columnOffset(column) < chartRight; column++ {
+ grid.maxColumn = max(grid.maxColumn, column+1)
+ }
+ chartBottom := grid.rowOffset(chart.anchorRow) + chart.rowOffset + chart.height
+ for row := chart.anchorRow + 1; grid.rowOffset(row) < chartBottom; row++ {
+ grid.maxRow = max(grid.maxRow, row+1)
+ }
+}
+
+type xlsxChartClip struct {
+ page *PDFPage
+ left, bottom, right, top float64
+}
+
+func (clip xlsxChartClip) line(x1, y1, x2, y2 float64, color PDFColor, width float64) {
+ if y1 == y2 {
+ if y1 < clip.bottom || y1 > clip.top {
+ return
+ }
+ x1, x2 = max(x1, clip.left), min(x2, clip.right)
+ } else if x1 == x2 {
+ if x1 < clip.left || x1 > clip.right {
+ return
+ }
+ y1, y2 = max(y1, clip.bottom), min(y2, clip.top)
+ }
+ if x2 >= x1 && y2 >= y1 {
+ clip.page.AddLine(x1, y1, x2, y2, color, width)
+ }
+}
+
+func (clip xlsxChartClip) rect(x, y, width, height float64, color PDFColor) {
+ left, right := max(x, clip.left), min(x+width, clip.right)
+ bottom, top := max(y, clip.bottom), min(y+height, clip.top)
+ if right > left && top > bottom {
+ clip.page.AddRect(left, bottom, right-left, top-bottom, color)
+ }
+}
+
+func (clip xlsxChartClip) text(text string, x, y, fontSize float64, color PDFColor, bold bool) {
+ if text == "" || x >= clip.right || y < clip.bottom || y > clip.top {
+ return
+ }
+ width := float64(len([]rune(text))) * fontSize * 0.52
+ if x+width <= clip.left {
+ return
+ }
+ clip.page.AddText(text, x, y, fontSize, color, bold)
+}
+
+func renderXLSXCharts(page *PDFPage, grid xlsxGrid, rows, columns xlsxRange) {
+ bandLeft := grid.columnOffset(columns.start - 1)
+ bandTop := grid.rowOffset(rows.start - 1)
+ bandWidth := grid.columnOffset(columns.end) - bandLeft
+ bandHeight := grid.rowOffset(rows.end) - bandTop
+ clip := xlsxChartClip{
+ page: page, left: grid.margins.Left, right: grid.margins.Left + bandWidth,
+ top: grid.pageSize.Height - grid.margins.Top,
+ bottom: grid.pageSize.Height - grid.margins.Top - bandHeight,
+ }
+ for _, chart := range grid.charts {
+ chartLeft := grid.columnOffset(chart.anchorColumn) + chart.columnOffset
+ chartTop := grid.rowOffset(chart.anchorRow) + chart.rowOffset
+ if chartLeft >= bandLeft+bandWidth || chartLeft+chart.width <= bandLeft ||
+ chartTop >= bandTop+bandHeight || chartTop+chart.height <= bandTop {
+ continue
+ }
+ x := grid.margins.Left + chartLeft - bandLeft
+ top := grid.pageSize.Height - grid.margins.Top - (chartTop - bandTop)
+ renderXLSXClusteredChart(clip, chart, x, top)
+ }
+}
+
+func renderXLSXClusteredChart(clip xlsxChartClip, chart xlsxChart, x, top float64) {
+ border := PDFColor{Red: 0.65, Green: 0.65, Blue: 0.65}
+ clip.line(x, top, x+chart.width, top, border, 0.5)
+ clip.line(x+chart.width, top-chart.height, x+chart.width, top, border, 0.5)
+ clip.line(x, top-chart.height, x+chart.width, top-chart.height, border, 0.5)
+ clip.line(x, top-chart.height, x, top, border, 0.5)
+
+ const fontSize = 8.0
+ titleWidth := float64(len([]rune(chart.title))) * fontSize * 0.52
+ clip.text(chart.title, x+(chart.width-titleWidth)/2, top-18, fontSize, PDFColorBlack, true)
+ legendWidth := 0.0
+ if chart.legendPosition == "r" || chart.legendPosition == "tr" {
+ legendWidth = 52
+ }
+ plotLeft := x + 48
+ plotRight := x + chart.width - 16 - legendWidth
+ plotTop := top - 34
+ plotBottom := top - chart.height + 36
+ plotWidth, plotHeight := plotRight-plotLeft, plotTop-plotBottom
+ if plotWidth <= 20 || plotHeight <= 20 {
+ return
+ }
+
+ maximum := 0.0
+ for _, series := range chart.series {
+ for _, value := range series.values {
+ maximum = max(maximum, value)
+ }
+ }
+ axisMaximum, step := xlsxChartAxis(maximum)
+ gridColor := PDFColor{Red: 0.82, Green: 0.82, Blue: 0.82}
+ for value := 0.0; value <= axisMaximum+step*0.01; value += step {
+ y := plotBottom + value/axisMaximum*plotHeight
+ clip.line(plotLeft, y, plotRight, y, gridColor, 0.5)
+ }
+
+ categories := 0
+ for _, series := range chart.series {
+ categories = max(categories, len(series.values))
+ }
+ if categories > 0 {
+ groupWidth := plotWidth / float64(categories)
+ barWidth := groupWidth / (float64(len(chart.series)) + chart.gapWidth/100)
+ groupPadding := (groupWidth - barWidth*float64(len(chart.series))) / 2
+ colors := []PDFColor{
+ {Red: 0.31, Green: 0.506, Blue: 0.741},
+ {Red: 0.753, Green: 0.314, Blue: 0.302},
+ }
+ for category := 0; category < categories; category++ {
+ for seriesIndex, series := range chart.series {
+ if category >= len(series.values) {
+ continue
+ }
+ height := max(0.5, series.values[category]/axisMaximum*plotHeight)
+ barX := plotLeft + float64(category)*groupWidth + groupPadding + float64(seriesIndex)*barWidth
+ clip.rect(barX, plotBottom, barWidth, height, colors[seriesIndex%len(colors)])
+ }
+ }
+ }
+ clip.line(plotLeft, plotBottom, plotLeft, plotTop, PDFColorBlack, 0.8)
+ clip.line(plotLeft, plotBottom, plotRight, plotBottom, PDFColorBlack, 0.8)
+ clip.text(chart.valueTitle, x+3, plotBottom+plotHeight*0.45, 7, PDFColorBlack, true)
+ categoryWidth := float64(len([]rune(chart.categoryTitle))) * 7 * 0.52
+ clip.text(chart.categoryTitle, plotLeft+(plotWidth-categoryWidth)/2, plotBottom-18, 7, PDFColorBlack, true)
+
+ if legendWidth > 0 {
+ legendX := plotRight + 12
+ legendY := plotBottom + plotHeight*0.5
+ colors := []PDFColor{
+ {Red: 0.31, Green: 0.506, Blue: 0.741},
+ {Red: 0.753, Green: 0.314, Blue: 0.302},
+ }
+ for index, series := range chart.series {
+ y := legendY - float64(index)*16
+ clip.rect(legendX, y, 7, 7, colors[index%len(colors)])
+ clip.text(series.name, legendX+11, y, 7, PDFColorBlack, false)
+ }
+ }
+}
+
+func xlsxChartAxis(maximum float64) (float64, float64) {
+ if maximum <= 0 {
+ return 1, 1
+ }
+ roughStep := maximum / 7
+ magnitude := math.Pow(10, math.Floor(math.Log10(roughStep)))
+ fraction := roughStep / magnitude
+ niceFraction := 1.0
+ if fraction > 1.5 {
+ niceFraction = 2
+ }
+ if fraction > 3 {
+ niceFraction = 5
+ }
+ if fraction > 7 {
+ niceFraction = 10
+ }
+ step := niceFraction * magnitude
+ return math.Ceil(maximum/step) * step, step
+}
diff --git a/minipdf-go/xlsx_render.go b/minipdf-go/xlsx_render.go
new file mode 100644
index 00000000..c963ad11
--- /dev/null
+++ b/minipdf-go/xlsx_render.go
@@ -0,0 +1,728 @@
+package minipdf
+
+import (
+ "bytes"
+ "encoding/xml"
+ "fmt"
+ "strconv"
+ "strings"
+)
+
+const (
+ xlsxColumnWidthPoints = 5.62
+ xlsxDefaultRowHeightScale = 14.6 / 15
+ calibriToHelveticaFontScale = 0.73
+ calibriRegularWidthScale = 90 / calibriToHelveticaFontScale
+ calibriBoldWidthScale = 80 / calibriToHelveticaFontScale
+ helveticaBaselineRatio = 0.45
+)
+
+type xlsxStyle struct {
+ font xlsxFont
+ fill *PDFColor
+ border xlsxBorder
+ numberFmt int
+ align string
+ vertical string
+}
+
+type xlsxFont struct {
+ name string
+ size float64
+ color PDFColor
+ bold bool
+}
+
+type xlsxBorder struct {
+ left, right, top, bottom float64
+}
+
+type xlsxCell struct {
+ row, column int
+ value string
+ formula string
+ style int
+ numeric bool
+}
+
+type xlsxMerge struct {
+ startRow, startColumn int
+ endRow, endColumn int
+}
+
+type xlsxGrid struct {
+ cells map[[2]int]xlsxCell
+ columnWidths map[int]float64
+ rowHeights map[int]float64
+ merges []xlsxMerge
+ charts []xlsxChart
+ maxRow, maxColumn int
+ dimensionRow int
+ dimensionColumn int
+ defaultRowHeight float64
+ pageSize PageSize
+ margins Margins
+}
+
+type xlsxStylesDocument struct {
+ Fonts struct {
+ Items []struct {
+ Bold *struct{} `xml:"b"`
+ Name struct {
+ Value string `xml:"val,attr"`
+ } `xml:"name"`
+ Size struct {
+ Value float64 `xml:"val,attr"`
+ } `xml:"sz"`
+ Color struct {
+ RGB string `xml:"rgb,attr"`
+ } `xml:"color"`
+ } `xml:"font"`
+ } `xml:"fonts"`
+ Fills struct {
+ Items []struct {
+ Pattern struct {
+ Type string `xml:"patternType,attr"`
+ Color struct {
+ RGB string `xml:"rgb,attr"`
+ } `xml:"fgColor"`
+ } `xml:"patternFill"`
+ } `xml:"fill"`
+ } `xml:"fills"`
+ Borders struct {
+ Items []struct {
+ Left struct {
+ Style string `xml:"style,attr"`
+ } `xml:"left"`
+ Right struct {
+ Style string `xml:"style,attr"`
+ } `xml:"right"`
+ Top struct {
+ Style string `xml:"style,attr"`
+ } `xml:"top"`
+ Bottom struct {
+ Style string `xml:"style,attr"`
+ } `xml:"bottom"`
+ } `xml:"border"`
+ } `xml:"borders"`
+ CellFormats struct {
+ Items []struct {
+ FontID int `xml:"fontId,attr"`
+ FillID int `xml:"fillId,attr"`
+ BorderID int `xml:"borderId,attr"`
+ NumberFmt int `xml:"numFmtId,attr"`
+ Alignment struct {
+ Horizontal string `xml:"horizontal,attr"`
+ Vertical string `xml:"vertical,attr"`
+ } `xml:"alignment"`
+ } `xml:"xf"`
+ } `xml:"cellXfs"`
+}
+
+func needsCalculatedXLSXGrid(data []byte) bool {
+ var worksheet struct {
+ Cells []struct {
+ Formula *string `xml:"f"`
+ Value *string `xml:"v"`
+ } `xml:"sheetData>row>c"`
+ }
+ if xml.Unmarshal(data, &worksheet) != nil {
+ return false
+ }
+ formulaCount := 0
+ for _, cell := range worksheet.Cells {
+ if cell.Formula == nil {
+ continue
+ }
+ formulaCount++
+ if cell.Value != nil && strings.TrimSpace(*cell.Value) != "" {
+ return false
+ }
+ }
+ return formulaCount > 0
+}
+
+func readXLSXStyles(files officePackage) ([]xlsxStyle, error) {
+ if _, ok := files["xl/styles.xml"]; !ok {
+ return nil, nil
+ }
+ data, err := files.read("xl/styles.xml")
+ if err != nil {
+ return nil, err
+ }
+ var document xlsxStylesDocument
+ if err := xml.Unmarshal(data, &document); err != nil {
+ return nil, fmt.Errorf("parse xl/styles.xml: %w", err)
+ }
+ styles := make([]xlsxStyle, len(document.CellFormats.Items))
+ for index, format := range document.CellFormats.Items {
+ style := xlsxStyle{
+ font: xlsxFont{size: 11, color: PDFColorBlack}, numberFmt: format.NumberFmt,
+ align: format.Alignment.Horizontal, vertical: format.Alignment.Vertical,
+ }
+ if format.FontID >= 0 && format.FontID < len(document.Fonts.Items) {
+ font := document.Fonts.Items[format.FontID]
+ style.font.name = font.Name.Value
+ if font.Size.Value > 0 {
+ style.font.size = font.Size.Value
+ }
+ style.font.bold = font.Bold != nil
+ if color, ok := parseXLSXColor(font.Color.RGB); ok {
+ style.font.color = color
+ }
+ }
+ if format.FillID >= 0 && format.FillID < len(document.Fills.Items) {
+ fill := document.Fills.Items[format.FillID].Pattern
+ if fill.Type == "solid" {
+ if color, ok := parseXLSXColor(fill.Color.RGB); ok {
+ style.fill = &color
+ }
+ }
+ }
+ if format.BorderID >= 0 && format.BorderID < len(document.Borders.Items) {
+ border := document.Borders.Items[format.BorderID]
+ style.border = xlsxBorder{
+ left: borderWidth(border.Left.Style), right: borderWidth(border.Right.Style),
+ top: borderWidth(border.Top.Style), bottom: borderWidth(border.Bottom.Style),
+ }
+ }
+ styles[index] = style
+ }
+ return styles, nil
+}
+
+func parseXLSXColor(value string) (PDFColor, bool) {
+ if len(value) < 6 {
+ return PDFColor{}, false
+ }
+ parsed, err := strconv.ParseUint(value[len(value)-6:], 16, 32)
+ if err != nil {
+ return PDFColor{}, false
+ }
+ return PDFColor{Red: float64(parsed>>16) / 255, Green: float64(parsed>>8&0xff) / 255, Blue: float64(parsed&0xff) / 255}, true
+}
+
+func borderWidth(style string) float64 {
+ switch style {
+ case "hair":
+ return 0.25
+ case "medium", "mediumDashed", "mediumDashDot", "mediumDashDotDot":
+ return 1.5
+ case "thick", "double":
+ return 2
+ case "thin", "dashed", "dotted", "dashDot", "dashDotDot":
+ return 0.75
+ default:
+ return 0
+ }
+}
+
+func renderXLSXWorksheets(files officePackage, worksheetNames []string, sharedStrings []string, styles []xlsxStyle, options ConversionOptions) ([]byte, error) {
+ document := NewPDFDocument()
+ for _, name := range worksheetNames {
+ data, err := files.read(name)
+ if err != nil {
+ return nil, err
+ }
+ grid, err := readXLSXGrid(data, sharedStrings)
+ if err != nil {
+ return nil, fmt.Errorf("parse %s: %w", name, err)
+ }
+ grid.charts, err = readXLSXCharts(files, name, &grid)
+ if err != nil {
+ return nil, err
+ }
+ grid.applyOptions(options)
+ renderXLSXGrid(document, grid, styles)
+ }
+ return document.BytesWithOptions(PDFSaveOptions{Compress: options.Compress}), nil
+}
+
+func readXLSXGrid(data []byte, sharedStrings []string) (xlsxGrid, error) {
+ grid := xlsxGrid{
+ cells: make(map[[2]int]xlsxCell), columnWidths: make(map[int]float64), rowHeights: make(map[int]float64),
+ defaultRowHeight: 15, pageSize: PageSizeA4, margins: Margins{Left: 54, Right: 54, Top: 72, Bottom: 72},
+ }
+ decoder := xml.NewDecoder(bytes.NewReader(data))
+ for {
+ token, err := decoder.Token()
+ if err != nil {
+ if err.Error() == "EOF" {
+ break
+ }
+ return xlsxGrid{}, err
+ }
+ start, ok := token.(xml.StartElement)
+ if !ok {
+ continue
+ }
+ switch start.Name.Local {
+ case "dimension":
+ reference := attrValue(start, "ref")
+ parts := strings.Split(reference, ":")
+ grid.dimensionRow, grid.dimensionColumn = xlsxCellPosition(parts[len(parts)-1])
+ case "sheetFormatPr":
+ if height, err := strconv.ParseFloat(attrValue(start, "defaultRowHeight"), 64); err == nil && height > 0 {
+ grid.defaultRowHeight = height
+ }
+ case "col":
+ minimum, _ := strconv.Atoi(attrValue(start, "min"))
+ maximum, _ := strconv.Atoi(attrValue(start, "max"))
+ width, _ := strconv.ParseFloat(attrValue(start, "width"), 64)
+ if grid.dimensionColumn > 0 {
+ maximum = min(maximum, grid.dimensionColumn)
+ }
+ for column := minimum; column <= maximum && column > 0; column++ {
+ grid.columnWidths[column] = width * xlsxColumnWidthPoints
+ }
+ case "row":
+ row, _ := strconv.Atoi(attrValue(start, "r"))
+ if height, err := strconv.ParseFloat(attrValue(start, "ht"), 64); err == nil && height > 0 {
+ grid.rowHeights[row] = height
+ }
+ grid.maxRow = max(grid.maxRow, row)
+ case "c":
+ reference := attrValue(start, "r")
+ row, column := xlsxCellPosition(reference)
+ value, formula, err := decodeWorksheetCellContent(decoder, start, sharedStrings)
+ if err != nil {
+ return xlsxGrid{}, err
+ }
+ style, _ := strconv.Atoi(attrValue(start, "s"))
+ cellType := attrValue(start, "t")
+ grid.cells[[2]int{row, column}] = xlsxCell{row: row, column: column, value: value, formula: formula, style: style, numeric: cellType == "" || cellType == "n"}
+ grid.maxRow = max(grid.maxRow, row)
+ grid.maxColumn = max(grid.maxColumn, column)
+ case "mergeCell":
+ if merge, ok := parseXLSXMerge(attrValue(start, "ref")); ok {
+ grid.merges = append(grid.merges, merge)
+ }
+ case "pageMargins":
+ grid.margins = Margins{
+ Left: inchesToPoints(attrValue(start, "left"), 54), Right: inchesToPoints(attrValue(start, "right"), 54),
+ Top: inchesToPoints(attrValue(start, "top"), 72), Bottom: inchesToPoints(attrValue(start, "bottom"), 72),
+ }
+ case "pageSetup":
+ if attrValue(start, "paperSize") == "1" {
+ grid.pageSize = PageSizeLetter
+ }
+ if attrValue(start, "orientation") == "landscape" {
+ grid.pageSize.Width, grid.pageSize.Height = grid.pageSize.Height, grid.pageSize.Width
+ }
+ }
+ }
+ if grid.dimensionRow > 0 && grid.dimensionColumn > 0 {
+ grid.maxRow = grid.dimensionRow
+ grid.maxColumn = grid.dimensionColumn
+ }
+ grid.evaluateFormulas()
+ return grid, nil
+}
+
+func (grid *xlsxGrid) evaluateFormulas() {
+ visiting := make(map[[2]int]bool)
+ for key, cell := range grid.cells {
+ if cell.value == "" && cell.formula != "" {
+ if value, ok := grid.evaluateCell(key, visiting); ok {
+ cell.value = strconv.FormatFloat(value, 'f', -1, 64)
+ grid.cells[key] = cell
+ }
+ }
+ }
+}
+
+func (grid *xlsxGrid) evaluateCell(key [2]int, visiting map[[2]int]bool) (float64, bool) {
+ cell, exists := grid.cells[key]
+ if !exists {
+ return 0, true
+ }
+ if cell.value != "" {
+ value, err := strconv.ParseFloat(strings.TrimSpace(cell.value), 64)
+ return value, err == nil
+ }
+ if cell.formula == "" || visiting[key] {
+ return 0, false
+ }
+ visiting[key] = true
+ defer delete(visiting, key)
+ formula := strings.TrimSpace(strings.TrimPrefix(cell.formula, "="))
+ if strings.HasPrefix(strings.ToUpper(formula), "SUM(") && strings.HasSuffix(formula, ")") {
+ merge, ok := parseXLSXMerge(formula[4 : len(formula)-1])
+ if !ok {
+ return 0, false
+ }
+ total := 0.0
+ for row := merge.startRow; row <= merge.endRow; row++ {
+ for column := merge.startColumn; column <= merge.endColumn; column++ {
+ value, ok := grid.evaluateCell([2]int{row, column}, visiting)
+ if !ok {
+ return 0, false
+ }
+ total += value
+ }
+ }
+ return total, true
+ }
+ total := 0.0
+ for _, reference := range strings.Split(formula, "+") {
+ row, column := xlsxCellPosition(strings.TrimSpace(reference))
+ if row == 0 || column == 0 {
+ return 0, false
+ }
+ value, ok := grid.evaluateCell([2]int{row, column}, visiting)
+ if !ok {
+ return 0, false
+ }
+ total += value
+ }
+ return total, true
+}
+
+func xlsxCellPosition(reference string) (int, int) {
+ column := worksheetColumnIndex(reference)
+ index := 0
+ for index < len(reference) && ((reference[index] >= 'A' && reference[index] <= 'Z') || (reference[index] >= 'a' && reference[index] <= 'z')) {
+ index++
+ }
+ row, _ := strconv.Atoi(reference[index:])
+ return row, column
+}
+
+func parseXLSXMerge(reference string) (xlsxMerge, bool) {
+ parts := strings.Split(reference, ":")
+ if len(parts) != 2 {
+ return xlsxMerge{}, false
+ }
+ startRow, startColumn := xlsxCellPosition(parts[0])
+ endRow, endColumn := xlsxCellPosition(parts[1])
+ return xlsxMerge{startRow: startRow, startColumn: startColumn, endRow: endRow, endColumn: endColumn}, startRow > 0 && startColumn > 0 && endRow >= startRow && endColumn >= startColumn
+}
+
+func inchesToPoints(value string, fallback float64) float64 {
+ inches, err := strconv.ParseFloat(value, 64)
+ if err != nil || inches < 0 {
+ return fallback
+ }
+ return inches * 72
+}
+
+func (grid *xlsxGrid) applyOptions(options ConversionOptions) {
+ if options.MaxRows > 0 {
+ grid.maxRow = min(grid.maxRow, options.MaxRows)
+ }
+ if options.MaxColumns > 0 {
+ grid.maxColumn = min(grid.maxColumn, options.MaxColumns)
+ }
+ if options.PageSize != nil {
+ grid.pageSize = *options.PageSize
+ }
+ if options.Margins != nil {
+ grid.margins = *options.Margins
+ }
+ if options.Landscape != nil && (*options.Landscape != (grid.pageSize.Width > grid.pageSize.Height)) {
+ grid.pageSize.Width, grid.pageSize.Height = grid.pageSize.Height, grid.pageSize.Width
+ }
+}
+
+type xlsxRange struct{ start, end int }
+
+func renderXLSXGrid(document *PDFDocument, grid xlsxGrid, styles []xlsxStyle) {
+ if grid.maxRow == 0 || grid.maxColumn == 0 {
+ document.AddPage(grid.pageSize.Width, grid.pageSize.Height)
+ return
+ }
+ columnBands := xlsxBands(grid.maxColumn, grid.pageSize.Width-grid.margins.Left-grid.margins.Right, func(index int) float64 { return grid.columnWidth(index) })
+ rowBands := xlsxBands(grid.maxRow, grid.pageSize.Height-grid.margins.Top-grid.margins.Bottom, func(index int) float64 { return grid.rowHeight(index) })
+ for _, columnBand := range columnBands {
+ for _, rowBand := range rowBands {
+ page := document.AddPage(grid.pageSize.Width, grid.pageSize.Height)
+ renderXLSXBand(page, grid, styles, rowBand, columnBand)
+ }
+ }
+}
+
+func xlsxBands(count int, available float64, size func(int) float64) []xlsxRange {
+ var bands []xlsxRange
+ for start := 1; start <= count; {
+ end := start
+ used := 0.0
+ for end <= count && (used+size(end) <= available || end == start) {
+ used += size(end)
+ end++
+ }
+ bands = append(bands, xlsxRange{start: start, end: end - 1})
+ start = end
+ }
+ return bands
+}
+
+func (grid xlsxGrid) columnWidth(column int) float64 {
+ if width := grid.columnWidths[column]; width > 0 {
+ return width
+ }
+ return 8.43 * xlsxColumnWidthPoints
+}
+
+func (grid xlsxGrid) rowHeight(row int) float64 {
+ if height := grid.rowHeights[row]; height > 0 {
+ return height
+ }
+ return grid.defaultRowHeight * xlsxDefaultRowHeightScale
+}
+
+func renderXLSXBand(page *PDFPage, grid xlsxGrid, styles []xlsxStyle, rows, columns xlsxRange) {
+ xPositions := map[int]float64{columns.start: grid.margins.Left}
+ for column := columns.start; column <= columns.end; column++ {
+ xPositions[column+1] = xPositions[column] + grid.columnWidth(column)
+ }
+ yPositions := map[int]float64{rows.start: grid.pageSize.Height - grid.margins.Top}
+ for row := rows.start; row <= rows.end; row++ {
+ yPositions[row+1] = yPositions[row] - grid.rowHeight(row)
+ }
+ merged := make(map[[2]int]bool)
+ for _, merge := range grid.merges {
+ for row := merge.startRow; row <= merge.endRow; row++ {
+ for column := merge.startColumn; column <= merge.endColumn; column++ {
+ merged[[2]int{row, column}] = true
+ }
+ }
+ if merge.endRow < rows.start || merge.startRow > rows.end || merge.endColumn < columns.start || merge.startColumn > columns.end {
+ continue
+ }
+ cell := grid.cells[[2]int{merge.startRow, merge.startColumn}]
+ startColumn := max(merge.startColumn, columns.start)
+ endColumn := min(merge.endColumn, columns.end)
+ startRow := max(merge.startRow, rows.start)
+ endRow := min(merge.endRow, rows.end)
+ renderXLSXCell(page, cell, xPositions[startColumn], yPositions[endRow+1], xPositions[endColumn+1]-xPositions[startColumn], yPositions[startRow]-yPositions[endRow+1], styleAt(styles, cell.style), merge.startColumn >= columns.start && merge.startColumn <= columns.end)
+ }
+ for row := rows.start; row <= rows.end; row++ {
+ for column := columns.start; column <= columns.end; column++ {
+ if merged[[2]int{row, column}] {
+ continue
+ }
+ cell, exists := grid.cells[[2]int{row, column}]
+ if !exists {
+ continue
+ }
+ renderXLSXCell(page, cell, xPositions[column], yPositions[row+1], grid.columnWidth(column), grid.rowHeight(row), styleAt(styles, cell.style), true)
+ }
+ }
+ renderXLSXCharts(page, grid, rows, columns)
+}
+
+func styleAt(styles []xlsxStyle, index int) xlsxStyle {
+ if index >= 0 && index < len(styles) {
+ return styles[index]
+ }
+ return xlsxStyle{font: xlsxFont{size: 11, color: PDFColorBlack}}
+}
+
+func renderXLSXCell(page *PDFPage, cell xlsxCell, x, y, width, height float64, style xlsxStyle, drawText bool) {
+ if style.fill != nil {
+ page.AddRect(x, y, width, height, *style.fill)
+ }
+ if style.border.left > 0 {
+ page.AddLine(x, y, x, y+height, PDFColorBlack, style.border.left)
+ }
+ if style.border.right > 0 {
+ page.AddLine(x+width, y, x+width, y+height, PDFColorBlack, style.border.right)
+ }
+ if style.border.top > 0 {
+ page.AddLine(x, y+height, x+width, y+height, PDFColorBlack, style.border.top)
+ }
+ if style.border.bottom > 0 {
+ page.AddLine(x, y, x+width, y, PDFColorBlack, style.border.bottom)
+ }
+ if !drawText || cell.value == "" {
+ return
+ }
+ text := formatXLSXValue(cell.value, cell.numeric, style.numberFmt)
+ fontSize := style.font.size
+ horizontalScale := 100.0
+ if strings.EqualFold(style.font.name, "Calibri") {
+ fontSize *= calibriToHelveticaFontScale
+ horizontalScale = calibriRegularWidthScale
+ if style.font.bold {
+ horizontalScale = calibriBoldWidthScale
+ }
+ }
+ textWidth := measureHelveticaText(text, fontSize, horizontalScale, style.font.bold)
+ textX := x + 2
+ switch style.align {
+ case "center":
+ textX = x + (width-textWidth)/2
+ case "right":
+ textX = x + width - textWidth - 2
+ default:
+ if cell.numeric {
+ textX = x + width - textWidth - 2
+ }
+ }
+ textY := y + fontSize*helveticaBaselineRatio
+ if style.vertical == "center" {
+ textY = y + (height-fontSize)/2
+ } else if style.vertical == "top" {
+ textY = y + height - fontSize
+ }
+ page.addScaledText(text, textX, textY, fontSize, horizontalScale, style.font.color, style.font.bold)
+}
+
+func measureHelveticaText(text string, fontSize, horizontalScale float64, bold bool) float64 {
+ total := 0
+ for _, character := range text {
+ total += helveticaCharacterWidth(character, bold)
+ }
+ return float64(total) * fontSize / 1000 * horizontalScale / 100
+}
+
+func helveticaCharacterWidth(character rune, bold bool) int {
+ if bold {
+ switch character {
+ case ' ', ',':
+ return 278
+ case '!', ':', ';', '(', ')', '-', '.', '/':
+ return 333
+ case '"':
+ return 474
+ case '#', '$', '0', '1', '2', '3', '4', '5', '6', '7', '8', '9':
+ return 556
+ case '%':
+ return 889
+ case '&', 'A', 'B', 'C', 'D', 'G', 'H', 'K', 'N', 'O', 'P', 'Q', 'R', 'U', 'V', 'X', 'Y':
+ return 722
+ case '\'', 'I':
+ return 238
+ case '*':
+ return 389
+ case '+', '<', '=', '>', '^', '~':
+ return 584
+ case '?', 'E', 'S':
+ return 611
+ case '@':
+ return 975
+ case 'F', 'L', 'T', 'Z':
+ return 611
+ case 'J':
+ return 556
+ case 'M':
+ return 833
+ case 'W':
+ return 944
+ case '[', ']':
+ return 333
+ case '\\':
+ return 278
+ case '_':
+ return 556
+ case '`':
+ return 333
+ case 'a', 'c', 'e', 's', 'x', 'v':
+ return 556
+ case 'b', 'd', 'g', 'h', 'n', 'o', 'p', 'q', 'u':
+ return 611
+ case 'f', 't':
+ return 333
+ case 'i', 'j', 'l':
+ return 278
+ case 'k':
+ return 556
+ case 'm':
+ return 889
+ case 'r':
+ return 389
+ case 'w':
+ return 778
+ case 'y':
+ return 556
+ case 'z':
+ return 500
+ case '{', '}':
+ return 389
+ case '|':
+ return 280
+ }
+ } else {
+ switch character {
+ case ' ', '!', ',', '.', '/', ':', ';', 'I', '[', '\\', ']':
+ return 278
+ case '"':
+ return 355
+ case '#', '$', '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '?', 'L', '_':
+ return 556
+ case '%':
+ return 889
+ case '&', 'A', 'B', 'E', 'K', 'P', 'S', 'V', 'X', 'Y':
+ return 667
+ case '\'':
+ return 191
+ case '(', ')', '-', '`':
+ return 333
+ case '*':
+ return 389
+ case '+', '<', '=', '>', '~':
+ return 584
+ case '@':
+ return 1015
+ case 'C', 'D', 'H', 'N', 'R', 'U':
+ return 722
+ case 'F', 'T', 'Z':
+ return 611
+ case 'G', 'O', 'Q':
+ return 778
+ case 'J':
+ return 500
+ case 'M':
+ return 833
+ case 'W':
+ return 944
+ case '^':
+ return 469
+ case 'a', 'b', 'd', 'e', 'g', 'h', 'n', 'o', 'p', 'q', 'u':
+ return 556
+ case 'c', 'k', 's', 'v', 'x', 'y', 'z':
+ return 500
+ case 'f':
+ return 278
+ case 'i', 'j', 'l':
+ return 222
+ case 'm':
+ return 833
+ case 'r':
+ return 333
+ case 't':
+ return 278
+ case 'w':
+ return 722
+ case '{', '}':
+ return 334
+ case '|':
+ return 260
+ }
+ }
+ if character >= 0x1100 && (character <= 0x115f || character >= 0x2e80 && character <= 0x9fff || character >= 0xac00 && character <= 0xd7af || character >= 0xf900 && character <= 0xfaff || character >= 0xfe30 && character <= 0xfe4f || character >= 0xff01 && character <= 0xff60 || character >= 0xffe0 && character <= 0xffe6) {
+ return 1000
+ }
+ return 556
+}
+
+func formatXLSXValue(value string, numeric bool, numberFormat int) string {
+ if !numeric || numberFormat != 4 {
+ return value
+ }
+ number, err := strconv.ParseFloat(strings.TrimSpace(value), 64)
+ if err != nil {
+ return value
+ }
+ formatted := strconv.FormatFloat(number, 'f', 2, 64)
+ parts := strings.SplitN(formatted, ".", 2)
+ digits := parts[0]
+ sign := ""
+ if strings.HasPrefix(digits, "-") {
+ sign, digits = "-", digits[1:]
+ }
+ for index := len(digits) - 3; index > 0; index -= 3 {
+ digits = digits[:index] + "," + digits[index:]
+ }
+ return sign + digits + "." + parts[1]
+}