Table Builder — aw-table

Purpose

Read the active paper's sections/ to determine which tables are needed (dataset table, baseline comparison, ablation study, results summary, etc.), prompt the user for CSV data for each table, convert CSV to LaTeX booktabs format, write sections/tables/{name}.tex, auto-insert \input{tables/{name}} at the correct location in the parent section, and report a status summary.

Workflow

Step 0 — Verify working directory

Must be inside an active paper project under manuscripts/[paper-name]/. Abort if sections/ or main.tex is not found.

Step 1 — Determine needed tables

Read sections/methodology.tex (or the appropriate section file) to identify table references. Look for patterns:

\placeholder{tab:name}
\ref{tab:name}
Table~\ref{tab:name}

Common table names:

• tab:dataset — dataset / experimental setup
• tab:baseline — baseline comparison
• tab:ablation — ablation study
• tab:results — results summary
• tab:comparison — method comparison

Build a list of table names needed.

If no table references are found, report "No tables referenced in sections — nothing to build."

Step 2 — Collect CSV data per table

For each needed table, prompt the user:

请提供 {table-name} 的 CSV 数据，格式:
header1,header2,header3
row1col1,row1col2,row1col3
row2col1,row2col2,row2col3
...

(或直接粘贴 CSV 内容)

Accept CSV as multi-line input until the user sends an empty line or a confirmation signal (e.g., "done", "ok", "生成").

• If CSV is provided and non-empty → proceed to conversion
• If user skips or provides no data → mark as placeholder, write a \placeholder{tab:name} comment in the section, do not create a file

Step 3 — Convert CSV to LaTeX booktabs

For each table with CSV data:

1. Parse the CSV (split on commas, trim whitespace)
2. First row = column headers
3. Determine column alignment: use l for text columns, c for numeric/short, r for right-aligned
4. Generate LaTeX:

\begin{table}[htbp]
  \centering
  \caption{<Caption>}
  \begin{tabular}{<cols>}
    \toprule
    <Header1> & <Header2> & <Header3> \\
    \midrule
    <Row1Col1> & <Row1Col2> & <Row1Col3> \\
    <Row2Col1> & <Row2Col2> & <Row2Col3> \\
    \bottomrule
  \end{tabular}
  \label{tab:<name>}
\end{table}

5. Create sections/tables/ directory if it does not exist
6. Write to sections/tables/{name}.tex

Step 4 — Auto-insert \input{} in section

After generating the table file, find the appropriate section file and insert:

% --- tab:name ---
\input{tables/{name}}

near the location of \placeholder{tab:name} or the first \ref{tab:name} reference in that section. Replace \placeholder{tab:name} if found.

Step 5 — Report

# Table Builder Report

| Table | Status | File |
|-------|--------|------|
| tab:dataset | ✅ Generated | sections/tables/dataset.tex |
| tab:baseline | ✅ Generated | sections/tables/baseline.tex |
| tab:ablation | ⚠️ Placeholder | (no CSV provided) |
| tab:results | ✅ Generated | sections/tables/results.tex |

Generated: 3 | Placeholder: 1 | Skipped: 0

LaTeX booktabs Template

\begin{table}[htbp]
  \centering
  \caption{Caption text}
  \begin{tabular}{lccc}
    \toprule
    Header1 & Header2 & Header3 \\
    \midrule
    Row1 & val1 & val2 \\
    Row2 & val3 & val4 \\
    \bottomrule
  \end{tabular}
  \label{tab:name}
\end{table}

Rules

• Column separators must be &
• Use \toprule, \midrule, \bottomrule from booktabs package
• Do NOT use vertical rules (| in tabular preamble)
• caption goes ABOVE the tabular
• Always include \label{tab:name}

Error Handling

• CSV malformed → warn user, show expected format, do not generate file
• Cannot write to sections/tables/ → abort with error, do not update section
• Table reference found but no section file contains it → warn, skip auto-insert
• Empty CSV → treat as placeholder

CSV Format Reference

Method,PSNR (dB),SSIM,Time (s)
SRCNN,30.2,0.882,1.2
ESPCN,31.5,0.903,0.4
EDSR,32.1,0.912,8.6

Accepts:

• Commas as separators
• Optional spaces around values
• Empty lines (skipped)
• Lines starting with # (comment, skipped)