🎉 v0.4.0 released — web_table: HTML tables as structured rows
The poll asked what v0.4.0 should add — web_table won the top slot and is now shipped.
What's new
web_table — extract the HTML tables of a page as structured rows (arrays of cell strings): the tables counterpart of web_fetch, for spec sheets, pricing pages and comparison charts.
- Header detection: the first row is the header when it sits in
<thead> or every cell is a <th>
colspan / rowspan expand into a rectangular grid; the spanning text repeats in every covered slot so rows stay self-contained- Defensive parsing: implicit cell/row closes, unclosed tables at EOF, script/style content never reaches a cell, empty tables ignored, nested tables flatten into the containing cell
- Caps:
maxTables (default 5 — or one table by 1-based position), maxRows (default 50), 200 chars per cell
- Semantics are anchored to an independent Python oracle (25 fixtures,
test/anchors-table.py) — expectation values are generated, not hand-inferred.
- Tests: 97 → 125; coverage ratchet 86.08% → 88.40%.
dsh plugin --profile web add github:TYEclipse/dsh-webfetch#v0.4.0
Still on the bench
The other poll candidates remain open for the next round — web_meta (page metadata pack), web_fetch_many (batch fetch), web_selector (CSS-scoped extraction), web_sitemap (robots/sitemap discovery). If your pick isn't listed, say what you'd add and why.
中文摘要
投票选出的 v0.4.0 头号功能 web_table 已发布：把页面上的 HTML 表格提取为结构化行——表头识别（<thead> 或全 <th>）、colspan/rowspan 网格展开（跨格文本填充到每个覆盖格）、防御式解析（隐式闭合/未闭合/script 内容一律安全）、行列与单元格上限；语义由独立 Python oracle（25 fixture）锚定。测试 97→125、覆盖率 86.08%→88.40%。
安装：dsh plugin --profile web add github:TYEclipse/dsh-webfetch#v0.4.0
下一个版本继续从候选池挑（web_meta / web_fetch_many / web_selector / web_sitemap）——有更好的想法直接说。
🎉 v0.4.0 released —
web_table: HTML tables as structured rowsThe poll asked what v0.4.0 should add —
web_tablewon the top slot and is now shipped.What's new
web_table— extract the HTML tables of a page as structured rows (arrays of cell strings): the tables counterpart ofweb_fetch, for spec sheets, pricing pages and comparison charts.<thead>or every cell is a<th>colspan/rowspanexpand into a rectangular grid; the spanning text repeats in every covered slot so rows stay self-containedmaxTables(default 5 — or one table by 1-based position),maxRows(default 50), 200 chars per celltest/anchors-table.py) — expectation values are generated, not hand-inferred.Still on the bench
The other poll candidates remain open for the next round —
web_meta(page metadata pack),web_fetch_many(batch fetch),web_selector(CSS-scoped extraction),web_sitemap(robots/sitemap discovery). If your pick isn't listed, say what you'd add and why.中文摘要
投票选出的 v0.4.0 头号功能
web_table已发布：把页面上的 HTML 表格提取为结构化行——表头识别（<thead>或全<th>）、colspan/rowspan 网格展开（跨格文本填充到每个覆盖格）、防御式解析（隐式闭合/未闭合/script 内容一律安全）、行列与单元格上限；语义由独立 Python oracle（25 fixture）锚定。测试 97→125、覆盖率 86.08%→88.40%。安装：
dsh plugin --profile web add github:TYEclipse/dsh-webfetch#v0.4.0下一个版本继续从候选池挑（web_meta / web_fetch_many / web_selector / web_sitemap）——有更好的想法直接说。