@createiq/htmldiff 1.1.0 → 1.2.0-beta.1

package/test/HtmlDiff.threeWay.tables.spec.ts

@@ -0,0 +1,301 @@
+import { describe, expect, it } from 'vitest'
+
+import HtmlDiff from '../src/HtmlDiff'
+
+/**
+ * Table tests under the genesis-spine model.
+ *
+ * `executeThreeWay(genesis, cpLatest, meCurrent)` walks the genesis
+ * table structure as the spine and attributes cp/me changes
+ * independently. Tables/rows/cells that both authors changed identically
+ * settle and emit plain; one-sided changes carry author attribution;
+ * disagreements show both authors' contributions.
+ */
+describe('HtmlDiff.executeThreeWay (tables, genesis-spine)', () => {
+  describe('cell-level attribution through table scaffolding', () => {
+    it('CP only changed, Me kept genesis → cp attribution', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>Rate</td></tr></table>',
+          '<table><tr><td>Interest Rate</td></tr></table>',
+          '<table><tr><td>Rate</td></tr></table>'
+        )
+      ).toBe("<table><tr><td><ins class='diffins cp' data-author='cp'>Interest </ins>Rate</td></tr></table>")
+    })
+
+    it('Me only changed, CP kept genesis → me attribution', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>Rate</td></tr></table>',
+          '<table><tr><td>Rate</td></tr></table>',
+          '<table><tr><td>Rate per annum</td></tr></table>'
+        )
+      ).toBe("<table><tr><td>Rate<ins class='diffins me' data-author='me'>&nbsp;per annum</ins></td></tr></table>")
+    })
+
+    it('Settled — both made the same change → no markup', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>Rate</td></tr></table>',
+          '<table><tr><td>Interest Rate</td></tr></table>',
+          '<table><tr><td>Interest Rate</td></tr></table>'
+        )
+      ).toBe('<table><tr><td>Interest Rate</td></tr></table>')
+    })
+
+    it('Disagreement — different changes at the same place', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>five</td></tr></table>',
+          '<table><tr><td>five and a half</td></tr></table>',
+          '<table><tr><td>seven</td></tr></table>'
+        )
+      ).toBe(
+        "<table><tr><td><del class='diffdel me' data-author='me'>five</del><ins class='diffins cp' data-author='cp'>&nbsp;and a half</ins><ins class='diffins me' data-author='me'>seven</ins></td></tr></table>"
+      )
+    })
+
+    it('CP deletes content, Me kept', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>five and a half</td></tr></table>',
+          '<table><tr><td>five</td></tr></table>',
+          '<table><tr><td>five and a half</td></tr></table>'
+        )
+      ).toBe("<table><tr><td>five<del class='diffdel cp' data-author='cp'>&nbsp;and a half</del></td></tr></table>")
+    })
+
+    it('Me deletes content, CP kept', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>five and a half</td></tr></table>',
+          '<table><tr><td>five and a half</td></tr></table>',
+          '<table><tr><td>five</td></tr></table>'
+        )
+      ).toBe("<table><tr><td>five<del class='diffdel me' data-author='me'>&nbsp;and a half</del></td></tr></table>")
+    })
+  })
+
+  describe('scaffolding scanner edge cases', () => {
+    it('<thead>/<tbody> wrappers', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><thead><tr><th>Term</th></tr></thead><tbody><tr><td>Old</td></tr></tbody></table>',
+          '<table><thead><tr><th>Term</th></tr></thead><tbody><tr><td>New</td></tr></tbody></table>',
+          '<table><thead><tr><th>Term</th></tr></thead><tbody><tr><td>Old</td></tr></tbody></table>'
+        )
+      ).toBe(
+        "<table><thead><tr><th>Term</th></tr></thead><tbody><tr><td><del class='diffdel cp' data-author='cp'>Old</del><ins class='diffins cp' data-author='cp'>New</ins></td></tr></tbody></table>"
+      )
+    })
+
+    it('<th> header cells', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><th>Label</th><th>Value</th></tr></table>',
+          '<table><tr><th>Description</th><th>Value</th></tr></table>',
+          '<table><tr><th>Label</th><th>Value</th></tr></table>'
+        )
+      ).toBe(
+        "<table><tr><th><del class='diffdel cp' data-author='cp'>Label</del><ins class='diffins cp' data-author='cp'>Description</ins></th><th>Value</th></tr></table>"
+      )
+    })
+
+    it('rowspan attribute preserved with author attribution on cell content', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td rowspan="2">Foo</td></tr></table>',
+          '<table><tr><td rowspan="2">Bar</td></tr></table>',
+          '<table><tr><td rowspan="2">Foo</td></tr></table>'
+        )
+      ).toBe(
+        "<table><tr><td rowspan=\"2\"><del class='diffdel cp' data-author='cp'>Foo</del><ins class='diffins cp' data-author='cp'>Bar</ins></td></tr></table>"
+      )
+    })
+
+    it('multi-row table with edit in one row only', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>a</td></tr><tr><td>b</td></tr><tr><td>c</td></tr></table>',
+          '<table><tr><td>a</td></tr><tr><td>B</td></tr><tr><td>c</td></tr></table>',
+          '<table><tr><td>a</td></tr><tr><td>b</td></tr><tr><td>c</td></tr></table>'
+        )
+      ).toBe(
+        "<table><tr><td>a</td></tr><tr><td><del class='diffdel cp' data-author='cp'>b</del><ins class='diffins cp' data-author='cp'>B</ins></td></tr><tr><td>c</td></tr></table>"
+      )
+    })
+
+    it('identity input passes through unchanged', () => {
+      const text = '<table><tr><td>a</td><td>b</td></tr><tr><td>c</td><td>d</td></tr></table>'
+      expect(HtmlDiff.executeThreeWay(text, text, text)).toBe(text)
+    })
+  })
+
+  describe('structural row changes (rows added or removed)', () => {
+    it('CP inserted a row, Me kept the absence → cp-attributed row insertion', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>a</td></tr></table>',
+          '<table><tr><td>a</td></tr><tr><td>b</td></tr></table>',
+          '<table><tr><td>a</td></tr></table>'
+        )
+      ).toBe(
+        "<table><tr><td>a</td></tr><tr class='diffins cp' data-author='cp'><td class='diffins cp' data-author='cp'><ins class='diffins cp' data-author='cp'>b</ins></td></tr></table>"
+      )
+    })
+
+    it('Me inserted a row, CP kept the absence → me-attributed row insertion', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>a</td></tr></table>',
+          '<table><tr><td>a</td></tr></table>',
+          '<table><tr><td>a</td></tr><tr><td>new</td></tr></table>'
+        )
+      ).toBe(
+        "<table><tr><td>a</td></tr><tr class='diffins me' data-author='me'><td class='diffins me' data-author='me'><ins class='diffins me' data-author='me'>new</ins></td></tr></table>"
+      )
+    })
+
+    it('CP deleted a row, Me kept it → cp-attributed row deletion (content from Me)', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>a</td></tr><tr><td>b</td></tr></table>',
+          '<table><tr><td>a</td></tr></table>',
+          '<table><tr><td>a</td></tr><tr><td>b</td></tr></table>'
+        )
+      ).toBe(
+        "<table><tr><td>a</td></tr><tr class='diffdel cp' data-author='cp'><td class='diffdel cp' data-author='cp'><del class='diffdel cp' data-author='cp'>b</del></td></tr></table>"
+      )
+    })
+
+    it('Me deleted a row, CP kept it → me-attributed row deletion', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>a</td></tr><tr><td>b</td></tr></table>',
+          '<table><tr><td>a</td></tr><tr><td>b</td></tr></table>',
+          '<table><tr><td>a</td></tr></table>'
+        )
+      ).toBe(
+        "<table><tr><td>a</td></tr><tr class='diffdel me' data-author='me'><td class='diffdel me' data-author='me'><del class='diffdel me' data-author='me'>b</del></td></tr></table>"
+      )
+    })
+
+    it('Both inserted same row → settled, no markup', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>a</td></tr></table>',
+          '<table><tr><td>a</td></tr><tr><td>b</td></tr></table>',
+          '<table><tr><td>a</td></tr><tr><td>b</td></tr></table>'
+        )
+      ).toBe('<table><tr><td>a</td></tr><tr><td>b</td></tr></table>')
+    })
+  })
+
+  describe('multi-table (table count diverges)', () => {
+    it('CP added a whole new table, Me kept absence → cp insertion of whole table', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>a</td></tr></table>',
+          '<table><tr><td>a</td></tr></table><table><tr><td>b</td></tr></table>',
+          '<table><tr><td>a</td></tr></table>'
+        )
+      ).toBe(
+        "<table><tr><td>a</td></tr></table><ins class='diffins cp' data-author='cp'><table><tr><td>b</td></tr></table></ins>"
+      )
+    })
+
+    it('Me added a whole new table, CP kept absence → me insertion', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>a</td></tr></table>',
+          '<table><tr><td>a</td></tr></table>',
+          '<table><tr><td>a</td></tr></table><table><tr><td>new</td></tr></table>'
+        )
+      ).toBe(
+        "<table><tr><td>a</td></tr></table><ins class='diffins me' data-author='me'><table><tr><td>new</td></tr></table></ins>"
+      )
+    })
+
+    it('CP deleted a table from genesis, Me kept it → cp deletion', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>a</td></tr></table><table><tr><td>b</td></tr></table>',
+          '<table><tr><td>a</td></tr></table>',
+          '<table><tr><td>a</td></tr></table><table><tr><td>b</td></tr></table>'
+        )
+      ).toBe(
+        "<table><tr><td>a</td></tr></table><del class='diffdel cp' data-author='cp'><table><tr><td>b</td></tr></table></del>"
+      )
+    })
+
+    it('Me deleted a table from genesis, CP kept it → me deletion', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>a</td></tr></table><table><tr><td>b</td></tr></table>',
+          '<table><tr><td>a</td></tr></table><table><tr><td>b</td></tr></table>',
+          '<table><tr><td>a</td></tr></table>'
+        )
+      ).toBe(
+        "<table><tr><td>a</td></tr></table><del class='diffdel me' data-author='me'><table><tr><td>b</td></tr></table></del>"
+      )
+    })
+
+    it('Both added the same new table → settled, plain output', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<table><tr><td>orig</td></tr></table>',
+          '<table><tr><td>orig</td></tr></table><table><tr><td>added</td></tr></table>',
+          '<table><tr><td>orig</td></tr></table><table><tr><td>added</td></tr></table>'
+        )
+      ).toBe('<table><tr><td>orig</td></tr></table><table><tr><td>added</td></tr></table>')
+    })
+  })
+
+  describe('cell-level edge cases', () => {
+    it('exceedsSizeLimit bail-out: oversized table falls through to word-level path', () => {
+      // 1501-row table: preprocessTablesThreeWay returns null on size cap,
+      // so executeThreeWay treats the table as raw HTML.
+      const rows = Array.from({ length: 1501 }, (_, i) => `<tr><td>row ${i}</td></tr>`).join('')
+      const html = `<table>${rows}</table>`
+      expect(HtmlDiff.executeThreeWay(html, html, html)).toBe(html)
+    })
+  })
+
+  describe('nested tables', () => {
+    it('handles a table nested inside a cell — both attributed correctly', () => {
+      const out = HtmlDiff.executeThreeWay(
+        '<table><tr><td><table><tr><td>inner</td></tr></table></td></tr></table>',
+        '<table><tr><td><table><tr><td>INNER</td></tr></table></td></tr></table>',
+        '<table><tr><td><table><tr><td>inner</td></tr></table></td></tr></table>'
+      )
+      expect(out).toMatch(/<del[^>]*data-author='cp'[^>]*>inner<\/del>/)
+      expect(out).toMatch(/<ins[^>]*data-author='cp'[^>]*>INNER<\/ins>/)
+      expect(out.startsWith('<table><tr><td><table>')).toBe(true)
+      expect(out.endsWith('</table></td></tr></table>')).toBe(true)
+    })
+
+    it('deeply-nested identity tables pass through unchanged (depth cap exercised)', () => {
+      let v = 'leaf'
+      for (let i = 0; i < 10; i++) v = `<table><tr><td>${v}</td></tr></table>`
+      expect(HtmlDiff.executeThreeWay(v, v, v)).toBe(v)
+    })
+  })
+
+  describe('integration with surrounding prose', () => {
+    it('prose and table both attributed correctly', () => {
+      expect(
+        HtmlDiff.executeThreeWay(
+          '<p>Rates as follows:</p><table><tr><td>five</td></tr></table>',
+          '<p>Interest rates as follows:</p><table><tr><td>five and a half</td></tr></table>',
+          '<p>Rates as follows:</p><table><tr><td>five</td></tr></table>'
+        )
+      ).toBe(
+        "<p><del class='diffdel cp' data-author='cp'>Rates</del><ins class='diffins cp' data-author='cp'>Interest rates</ins> as follows:</p><table><tr><td>five<ins class='diffins cp' data-author='cp'>&nbsp;and a half</ins></td></tr></table>"
+      )
+    })
+
+    it('no-tables path is unaffected', () => {
+      expect(HtmlDiff.executeThreeWay('<p>a</p>', '<p>a</p>', '<p>a</p>')).toBe('<p>a</p>')
+    })
+  })
+})

package/test/TableDiff.bench.ts

@@ -110,6 +110,45 @@ describe('TableDiff — column added', () => {
   )
 })
 
+describe('TableDiff — column-position DP worst case', () => {
+  // The pre-DP combinatorial path was O(C(N, K)) per row. At N=40 and
+  // K=6 (the caps), that's 3.8M combos × ~34 string ops, multiplied by
+  // every preserved row. This case exercises that path: 50 preserved
+  // rows × 40-cell wide new (34-cell old + 6 inserted columns).
+  const oldHtml = buildTable(50, 34, defaultCell)
+  const newHtml = buildTable(50, 40, (r, c) => {
+    if (c < 3) return `NEW_PREFIX_r${r}_c${c}`
+    if (c >= 37) return `NEW_SUFFIX_r${r}_c${c}`
+    return defaultCell(r, c - 3)
+  })
+
+  bench(
+    '50×40 — 6 columns inserted (was C(40,6) ≈ 3.8M combos/row pre-DP)',
+    () => {
+      expect(HtmlDiff.execute(oldHtml, newHtml)).toBeTruthy()
+    },
+    benchOptions
+  )
+})
+
+describe('TableDiff — column-position DP common case', () => {
+  // Companion to the worst-case bench above: the common case is a wide
+  // row with a single column added. This should stay sub-ms and not
+  // hide behind the worst-case headline if anything regresses.
+  const oldHtml = buildTable(50, 40, defaultCell)
+  const newHtml = buildTable(50, 41, (r, c) =>
+    c < 20 ? defaultCell(r, c) : c === 20 ? `Inserted col r${r}` : defaultCell(r, c - 1)
+  )
+
+  bench(
+    '50×40 — 1 column inserted (common case)',
+    () => {
+      expect(HtmlDiff.execute(oldHtml, newHtml)).toBeTruthy()
+    },
+    benchOptions
+  )
+})
+
 describe('TableDiff — horizontal merge (colspan)', () => {
   // Old: <tr><td>a</td><td>b</td><td>c</td></tr>
   // New: <tr><td colspan=2>a b</td><td>c</td></tr>

package/test/Utils.spec.ts

@@ -117,4 +117,52 @@ describe('Utils', () => {
       expect(Utils.isWord('b')).toBe(true)
     })
   })
+
+  describe('wrapText()', () => {
+    it('produces the historical shape when no metadata is passed', () => {
+      expect(Utils.wrapText('hello', 'ins', 'diffins')).toBe("<ins class='diffins'>hello</ins>")
+    })
+
+    it('appends extraClasses after the base class', () => {
+      expect(Utils.wrapText('hello', 'ins', 'diffins', { extraClasses: 'cp' })).toBe(
+        "<ins class='diffins cp'>hello</ins>"
+      )
+    })
+
+    it('emits data-* attributes for each entry in dataAttrs', () => {
+      expect(Utils.wrapText('hello', 'ins', 'diffins', { dataAttrs: { author: 'cp' } })).toBe(
+        "<ins class='diffins' data-author='cp'>hello</ins>"
+      )
+    })
+
+    it('combines extraClasses and dataAttrs in one call', () => {
+      expect(
+        Utils.wrapText('hello', 'del', 'diffdel', {
+          extraClasses: 'me rejects-cp',
+          dataAttrs: { author: 'me', rejects: 'cp' },
+        })
+      ).toBe("<del class='diffdel me rejects-cp' data-author='me' data-rejects='cp'>hello</del>")
+    })
+
+    it('skips the metadata path entirely when neither extraClasses nor dataAttrs is set', () => {
+      // Empty metadata should still pass through the chosen branch deterministically.
+      expect(Utils.wrapText('hello', 'ins', 'diffins', {})).toBe("<ins class='diffins'>hello</ins>")
+    })
+  })
+
+  describe('composeTagAttributes()', () => {
+    it('returns just the class attribute when metadata is empty', () => {
+      expect(Utils.composeTagAttributes('diffins', {})).toBe(" class='diffins'")
+    })
+
+    it('appends extraClasses inside the class attribute', () => {
+      expect(Utils.composeTagAttributes('mod strong', { extraClasses: 'cp' })).toBe(" class='mod strong cp'")
+    })
+
+    it('includes all data-* attrs after the class attr', () => {
+      expect(Utils.composeTagAttributes('mod', { dataAttrs: { author: 'me', flag: 'on' } })).toBe(
+        " class='mod' data-author='me' data-flag='on'"
+      )
+    })
+  })
 })