sn-da-excel-workflow

1# Excel Data Analysis Workflow
2 
3End-to-end workflow for structured Excel analysis. Each step maps to a
4capability sub-skill that can be loaded for detailed patterns.
5 
6## Workflow
7 
8### Step 1 — Count rows across all sheets (lightweight, no full load)
9 
10Count rows per sheet **without loading data into memory**. Use openpyxl
11`read_only` mode — this works for any file size.
12 
13```python
14import openpyxl, gc
15 
16wb = openpyxl.load_workbook(file_path, read_only=True, data_only=True)
17total_rows = 0
18sheet_info = {}
19for name in wb.sheetnames:
20    ws = wb[name]
21    row_count = sum(1 for _ in ws.iter_rows(min_row=2, values_only=True))
22    total_rows += row_count
23    sheet_info[name] = row_count
24    print(f"Sheet '{name}': {row_count} rows")
25wb.close()
26print(f"总行数={total_rows}")
27```
28 
29⚠️ **Do NOT use `pd.read_excel()` to count rows** — it loads all data into
30memory, which will OOM on large files.
31 
32→ capability: `excel-reading/multi-sheet-reading`
33 
34### Step 2 — Large file gate (CRITICAL — choose strategy by row count)
35 
36| total_rows | Strategy | What to do |
37|-----------|----------|------------|
38| < 10k | Direct read | `df = pd.read_excel(file_path, sheet_name=target_sheet)` |
39| 10k – 100k | Parquet cache | `pd.read_excel()` once → `df.to_parquet()` → all later reads from Parquet |
40| **>= 100k** | **STOP. Load `sn-da-large-file-analysis` skill** | Read its SKILL.md, then follow its streaming read + Parquet pattern. **Do NOT use `pd.read_excel()` at all** — it will OOM or timeout on 100k+ rows. |
41…