| 1 | import tempfile |
| 2 | import unittest |
| 3 | from pathlib import Path |
| 4 | |
| 5 | import scripts.generate_content as generate_content |
| 6 | |
| 7 | |
| 8 | class GenerateContentTests(unittest.TestCase): |
| 9 | def test_generate_content_creates_hugo_weekly_page(self) -> None: |
| 10 | tests_root = Path(__file__).resolve().parent |
| 11 | with tempfile.TemporaryDirectory(dir=tests_root) as tmpdir: |
| 12 | base = Path(tmpdir) |
| 13 | summary_path = base / "data" / "analyzed" / "2026-W21-summary.md" |
| 14 | summary_path.parent.mkdir(parents=True) |
| 15 | summary_path.write_text( |
| 16 | """--- |
| 17 | title: \"Week 21, 2026 Analysis\" |
| 18 | date: 2026-05-18T13:20:07.067+02:00 |
| 19 | week: \"2026-W21\" |
| 20 | year: 2026 |
| 21 | tags: [ai, agents, developer-tooling] |
| 22 | categories: [weekly] |
| 23 | repos_featured: 12 |
| 24 | stars_tracked: 3456 |
| 25 | top_repo: \"octo/repo\" |
| 26 | quality_score: 88 |
| 27 | summary: \"Agent tooling became more operational this week.\" |
| 28 | --- |
| 29 | |
| 30 | ## This Week's Trends |
| 31 | |
| 32 | Body copy. |
| 33 | """, |
| 34 | encoding="utf-8", |
| 35 | ) |
| 36 | |
| 37 | previous_cwd = Path.cwd() |
| 38 | try: |
| 39 | import os |
| 40 | |
| 41 | os.chdir(base) |
| 42 | output_path = generate_content.generate_content(summary_path) |
| 43 | finally: |
| 44 | os.chdir(previous_cwd) |
| 45 | |
| 46 | self.assertEqual(output_path, base / "content" / "weekly" / "2026" / "W21.md") |
| 47 | rendered = output_path.read_text(encoding="utf-8") |
| 48 | self.assertIn('title: "Week 21, 2026"', rendered) |
| 49 | self.assertIn("draft: false", rendered) |
| 50 | self.assertIn('week: "2026-W21"', rendered) |
| 51 | self.assertIn('top_repo: "octo/repo"', rendered) |
| 52 | self.assertIn('summary: "Agent tooling became more operational this week."', rendered) |
| 53 | self.assertNotIn("quality_score", rendered) |
| 54 | self.assertNotIn("year:", rendered) |
| 55 | self.assertIn("## This Week's Trends", rendered) |
| 56 | |
| 57 | def test_find_latest_summary_uses_week_not_mtime(self) -> None: |
| 58 | tests_root = Path(__file__).resolve().parent |
| 59 | with tempfile.TemporaryDirectory(dir=tests_root) as tmpdir: |
| 60 | base = Path(tmpdir) |
| 61 | analyzed_dir = base / "data" / "analyzed" |
| 62 | analyzed_dir.mkdir(parents=True) |
| 63 | latest = analyzed_dir / "2027-W01-summary.md" |
| 64 | older = analyzed_dir / "2026-W52-summary.md" |
| 65 | latest.write_text("latest\n", encoding="utf-8") |
| 66 | older.write_text("older but touched later\n", encoding="utf-8") |
| 67 | |
| 68 | self.assertEqual(generate_content.find_latest_summary(base), latest) |
| 69 | |
| 70 | def test_parse_frontmatter_rejects_missing_required_fields(self) -> None: |
| 71 | with self.assertRaises(generate_content.GenerationError): |
| 72 | generate_content.parse_frontmatter( |
| 73 | """--- |
| 74 | title: \"Week 21, 2026 Analysis\" |
| 75 | --- |
| 76 | |
| 77 | ## Notable New Repositories |
| 78 | """ |
| 79 | ) |
| 80 | |
| 81 | def test_render_frontmatter_quotes_each_tag(self) -> None: |
| 82 | """Tags containing special chars must be individually quoted to prevent YAML injection.""" |
| 83 | data = { |
| 84 | "title": "Test Week", |
| 85 | "date": "2026-05-18", |
| 86 | "week": "2026-W20", |
| 87 | "tags": ["ai", "evil: injected, categories: [hacked]"], |
| 88 | "categories": ["weekly"], |
| 89 | "repos_featured": 1, |
| 90 | "stars_tracked": 100, |
| 91 | "top_repo": "owner/repo", |
| 92 | "summary": "Test summary.", |
| 93 | } |
| 94 | output = generate_content.render_frontmatter(data) |
| 95 | # Each tag must be wrapped in YAML double-quotes |
| 96 | self.assertIn('"ai"', output) |
| 97 | self.assertIn('"evil: injected, categories: [hacked]"', output) |
| 98 | # The injected key must not appear as a top-level YAML key |
| 99 | self.assertNotIn("\ncategories: [hacked]", output) |
| 100 | |
| 101 | def test_render_frontmatter_quotes_each_category(self) -> None: |
| 102 | """Categories containing special chars must be individually quoted.""" |
| 103 | data = { |
| 104 | "title": "Test Week", |
| 105 | "date": "2026-05-18", |
| 106 | "week": "2026-W20", |
| 107 | "tags": ["safe"], |
| 108 | "categories": ["weekly", "bad: injection"], |
| 109 | "repos_featured": 1, |
| 110 | "stars_tracked": 100, |
| 111 | "top_repo": "owner/repo", |
| 112 | "summary": "Test.", |
| 113 | } |
| 114 | output = generate_content.render_frontmatter(data) |
| 115 | self.assertIn('"weekly"', output) |
| 116 | self.assertIn('"bad: injection"', output) |
| 117 | self.assertNotIn("\nbad:", output) |
| 118 | |
| 119 | def test_render_frontmatter_includes_cover_fields(self) -> None: |
| 120 | """Cover image frontmatter is rendered when present.""" |
| 121 | data = { |
| 122 | "title": "Test Week", |
| 123 | "date": "2026-05-18", |
| 124 | "week": "2026-W20", |
| 125 | "tags": ["ai"], |
| 126 | "categories": ["weekly"], |
| 127 | "repos_featured": 1, |
| 128 | "stars_tracked": 100, |
| 129 | "top_repo": "owner/repo", |
| 130 | "summary": "Test.", |
| 131 | "cover": { |
| 132 | "image": "covers/2026-W20.webp", |
| 133 | "alt": "AI trends visualization", |
| 134 | "attribution": "Photo by Author on Openverse", |
| 135 | "license": "CC0", |
| 136 | }, |
| 137 | "og_image": "covers/2026-W20-og.png", |
| 138 | } |
| 139 | output = generate_content.render_frontmatter(data) |
| 140 | self.assertIn("cover:", output) |
| 141 | self.assertIn('image: "covers/2026-W20.webp"', output) |
| 142 | self.assertIn('alt: "AI trends visualization"', output) |
| 143 | self.assertIn('attribution: "Photo by Author on Openverse"', output) |
| 144 | self.assertIn('license: "CC0"', output) |
| 145 | self.assertIn("relative: false", output) |
| 146 | self.assertIn('og_image: "covers/2026-W20-og.png"', output) |
| 147 | |
| 148 | def test_transform_summary_passes_cover_fields(self) -> None: |
| 149 | """Cover fields from analysis frontmatter are passed to output.""" |
| 150 | doc = """--- |
| 151 | title: "Week 20 Analysis" |
| 152 | date: 2026-05-11 |
| 153 | week: "2026-W20" |
| 154 | year: 2026 |
| 155 | tags: [ai] |
| 156 | categories: [weekly] |
| 157 | repos_featured: 5 |
| 158 | stars_tracked: 1000 |
| 159 | top_repo: "owner/repo" |
| 160 | quality_score: 90 |
| 161 | summary: "Test summary." |
| 162 | cover_image: "covers/test.webp" |
| 163 | cover_alt: "Test alt text" |
| 164 | cover_attribution: "Test Author" |
| 165 | cover_license: "CC0" |
| 166 | --- |
| 167 | |
| 168 | Body content. |
| 169 | """ |
| 170 | frontmatter, body = generate_content.parse_frontmatter(doc) |
| 171 | output = generate_content.transform_summary(frontmatter, body) |
| 172 | self.assertIn('image: "covers/test.webp"', output) |
| 173 | self.assertIn('alt: "Test alt text"', output) |
| 174 | self.assertIn("relative: false", output) |
| 175 | |
| 176 | def test_transform_summary_null_cover_attribution_does_not_emit_none(self) -> None: |
| 177 | """YAML null in cover_attribution/cover_license must not produce 'None' string.""" |
| 178 | doc = """--- |
| 179 | title: "Week 20 Analysis" |
| 180 | date: 2026-05-11 |
| 181 | week: "2026-W20" |
| 182 | year: 2026 |
| 183 | tags: [ai] |
| 184 | categories: [weekly] |
| 185 | repos_featured: 5 |
| 186 | stars_tracked: 1000 |
| 187 | top_repo: "owner/repo" |
| 188 | quality_score: 90 |
| 189 | summary: "Test summary." |
| 190 | cover_image: "covers/test.webp" |
| 191 | cover_attribution: null |
| 192 | cover_license: null |
| 193 | --- |
| 194 | |
| 195 | Body. |
| 196 | """ |
| 197 | frontmatter, body = generate_content.parse_frontmatter(doc) |
| 198 | output = generate_content.transform_summary(frontmatter, body) |
| 199 | self.assertNotIn("None", output) |
| 200 | self.assertIn('image: "covers/test.webp"', output) |
| 201 | |
| 202 | def test_transform_summary_rejects_url_og_image(self) -> None: |
| 203 | """og_image values that are URLs must be silently dropped (no hotlinking).""" |
| 204 | doc = """--- |
| 205 | title: "Week 20 Analysis" |
| 206 | date: 2026-05-11 |
| 207 | week: "2026-W20" |
| 208 | year: 2026 |
| 209 | tags: [ai] |
| 210 | categories: [weekly] |
| 211 | repos_featured: 5 |
| 212 | stars_tracked: 1000 |
| 213 | top_repo: "owner/repo" |
| 214 | quality_score: 90 |
| 215 | summary: "Test summary." |
| 216 | og_image: "https://evil.com/image.png" |
| 217 | --- |
| 218 | |
| 219 | Body. |
| 220 | """ |
| 221 | frontmatter, body = generate_content.parse_frontmatter(doc) |
| 222 | output = generate_content.transform_summary(frontmatter, body) |
| 223 | self.assertNotIn("og_image", output) |
| 224 | self.assertNotIn("evil.com", output) |
| 225 | |
| 226 | def test_transform_summary_rejects_url_cover_image(self) -> None: |
| 227 | """cover_image values that are URLs must be dropped.""" |
| 228 | doc = """--- |
| 229 | title: "Week 20 Analysis" |
| 230 | date: 2026-05-11 |
| 231 | week: "2026-W20" |
| 232 | year: 2026 |
| 233 | tags: [ai] |
| 234 | categories: [weekly] |
| 235 | repos_featured: 5 |
| 236 | stars_tracked: 1000 |
| 237 | top_repo: "owner/repo" |
| 238 | quality_score: 90 |
| 239 | summary: "Test summary." |
| 240 | cover_image: "https://evil.com/cover.png" |
| 241 | --- |
| 242 | |
| 243 | Body. |
| 244 | """ |
| 245 | frontmatter, body = generate_content.parse_frontmatter(doc) |
| 246 | output = generate_content.transform_summary(frontmatter, body) |
| 247 | self.assertNotIn("cover:", output) |
| 248 | self.assertNotIn("evil.com", output) |
| 249 | |
| 250 | def test_transform_summary_falls_back_to_cover_block_attribution_and_license(self) -> None: |
| 251 | doc = """--- |
| 252 | title: "Week 20 Analysis" |
| 253 | date: 2026-05-11 |
| 254 | week: "2026-W20" |
| 255 | year: 2026 |
| 256 | tags: [ai] |
| 257 | categories: [weekly] |
| 258 | repos_featured: 5 |
| 259 | stars_tracked: 1000 |
| 260 | top_repo: "owner/repo" |
| 261 | quality_score: 90 |
| 262 | summary: "Test summary." |
| 263 | cover: |
| 264 | image: "covers/test.webp" |
| 265 | alt: "Test alt text" |
| 266 | attribution: "Existing Author" |
| 267 | license: "Openverse" |
| 268 | --- |
| 269 | |
| 270 | Body. |
| 271 | """ |
| 272 | frontmatter, body = generate_content.parse_frontmatter(doc) |
| 273 | output = generate_content.transform_summary(frontmatter, body) |
| 274 | self.assertIn('image: "covers/test.webp"', output) |
| 275 | self.assertIn('attribution: "Existing Author"', output) |
| 276 | self.assertIn('license: "Openverse"', output) |
| 277 | |
| 278 | |
| 279 | if __name__ == "__main__": |
| 280 | unittest.main() |