@aroman22/codegraph-vba-darwin-arm64 1.5.1 → 1.6.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (335) hide show
  1. package/bin/codegraph +5 -0
  2. package/lib/dist/bin/codegraph.d.ts +1 -1
  3. package/lib/dist/bin/codegraph.d.ts.map +1 -1
  4. package/lib/dist/bin/codegraph.js +307 -81
  5. package/lib/dist/bin/codegraph.js.map +1 -1
  6. package/lib/dist/bin/command-supervision.d.ts +38 -1
  7. package/lib/dist/bin/command-supervision.d.ts.map +1 -1
  8. package/lib/dist/bin/command-supervision.js +22 -3
  9. package/lib/dist/bin/command-supervision.js.map +1 -1
  10. package/lib/dist/db/index.d.ts +84 -5
  11. package/lib/dist/db/index.d.ts.map +1 -1
  12. package/lib/dist/db/index.js +209 -12
  13. package/lib/dist/db/index.js.map +1 -1
  14. package/lib/dist/db/migrations.d.ts +1 -1
  15. package/lib/dist/db/migrations.d.ts.map +1 -1
  16. package/lib/dist/db/migrations.js +80 -3
  17. package/lib/dist/db/migrations.js.map +1 -1
  18. package/lib/dist/db/queries.d.ts +127 -4
  19. package/lib/dist/db/queries.d.ts.map +1 -1
  20. package/lib/dist/db/queries.js +317 -7
  21. package/lib/dist/db/queries.js.map +1 -1
  22. package/lib/dist/db/schema.sql +43 -1
  23. package/lib/dist/db/wal-valve.d.ts +106 -0
  24. package/lib/dist/db/wal-valve.d.ts.map +1 -0
  25. package/lib/dist/db/wal-valve.js +208 -0
  26. package/lib/dist/db/wal-valve.js.map +1 -0
  27. package/lib/dist/directory.d.ts +9 -5
  28. package/lib/dist/directory.d.ts.map +1 -1
  29. package/lib/dist/directory.js +210 -19
  30. package/lib/dist/directory.js.map +1 -1
  31. package/lib/dist/extraction/cfml-extractor.d.ts +107 -0
  32. package/lib/dist/extraction/cfml-extractor.d.ts.map +1 -0
  33. package/lib/dist/extraction/cfml-extractor.js +494 -0
  34. package/lib/dist/extraction/cfml-extractor.js.map +1 -0
  35. package/lib/dist/extraction/grammars.d.ts +25 -1
  36. package/lib/dist/extraction/grammars.d.ts.map +1 -1
  37. package/lib/dist/extraction/grammars.js +177 -23
  38. package/lib/dist/extraction/grammars.js.map +1 -1
  39. package/lib/dist/extraction/index.d.ts +47 -2
  40. package/lib/dist/extraction/index.d.ts.map +1 -1
  41. package/lib/dist/extraction/index.js +568 -51
  42. package/lib/dist/extraction/index.js.map +1 -1
  43. package/lib/dist/extraction/languages/arkts.d.ts +3 -0
  44. package/lib/dist/extraction/languages/arkts.d.ts.map +1 -0
  45. package/lib/dist/extraction/languages/arkts.js +127 -0
  46. package/lib/dist/extraction/languages/arkts.js.map +1 -0
  47. package/lib/dist/extraction/languages/c-cpp.d.ts +98 -0
  48. package/lib/dist/extraction/languages/c-cpp.d.ts.map +1 -1
  49. package/lib/dist/extraction/languages/c-cpp.js +563 -1
  50. package/lib/dist/extraction/languages/c-cpp.js.map +1 -1
  51. package/lib/dist/extraction/languages/cfquery.d.ts +12 -0
  52. package/lib/dist/extraction/languages/cfquery.d.ts.map +1 -0
  53. package/lib/dist/extraction/languages/cfquery.js +28 -0
  54. package/lib/dist/extraction/languages/cfquery.js.map +1 -0
  55. package/lib/dist/extraction/languages/cfscript.d.ts +3 -0
  56. package/lib/dist/extraction/languages/cfscript.d.ts.map +1 -0
  57. package/lib/dist/extraction/languages/cfscript.js +73 -0
  58. package/lib/dist/extraction/languages/cfscript.js.map +1 -0
  59. package/lib/dist/extraction/languages/cobol.d.ts +33 -0
  60. package/lib/dist/extraction/languages/cobol.d.ts.map +1 -0
  61. package/lib/dist/extraction/languages/cobol.js +499 -0
  62. package/lib/dist/extraction/languages/cobol.js.map +1 -0
  63. package/lib/dist/extraction/languages/erlang.d.ts +3 -0
  64. package/lib/dist/extraction/languages/erlang.d.ts.map +1 -0
  65. package/lib/dist/extraction/languages/erlang.js +350 -0
  66. package/lib/dist/extraction/languages/erlang.js.map +1 -0
  67. package/lib/dist/extraction/languages/index.d.ts.map +1 -1
  68. package/lib/dist/extraction/languages/index.js +18 -0
  69. package/lib/dist/extraction/languages/index.js.map +1 -1
  70. package/lib/dist/extraction/languages/nix.d.ts +3 -0
  71. package/lib/dist/extraction/languages/nix.d.ts.map +1 -0
  72. package/lib/dist/extraction/languages/nix.js +294 -0
  73. package/lib/dist/extraction/languages/nix.js.map +1 -0
  74. package/lib/dist/extraction/languages/solidity.d.ts +3 -0
  75. package/lib/dist/extraction/languages/solidity.d.ts.map +1 -0
  76. package/lib/dist/extraction/languages/solidity.js +293 -0
  77. package/lib/dist/extraction/languages/solidity.js.map +1 -0
  78. package/lib/dist/extraction/languages/terraform.d.ts +3 -0
  79. package/lib/dist/extraction/languages/terraform.d.ts.map +1 -0
  80. package/lib/dist/extraction/languages/terraform.js +641 -0
  81. package/lib/dist/extraction/languages/terraform.js.map +1 -0
  82. package/lib/dist/extraction/languages/vbnet.d.ts +11 -0
  83. package/lib/dist/extraction/languages/vbnet.d.ts.map +1 -0
  84. package/lib/dist/extraction/languages/vbnet.js +141 -0
  85. package/lib/dist/extraction/languages/vbnet.js.map +1 -0
  86. package/lib/dist/extraction/mybatis-extractor.d.ts +30 -10
  87. package/lib/dist/extraction/mybatis-extractor.d.ts.map +1 -1
  88. package/lib/dist/extraction/mybatis-extractor.js +140 -38
  89. package/lib/dist/extraction/mybatis-extractor.js.map +1 -1
  90. package/lib/dist/extraction/parse-pool.d.ts +28 -1
  91. package/lib/dist/extraction/parse-pool.d.ts.map +1 -1
  92. package/lib/dist/extraction/parse-pool.js +76 -7
  93. package/lib/dist/extraction/parse-pool.js.map +1 -1
  94. package/lib/dist/extraction/parse-worker.js +11 -4
  95. package/lib/dist/extraction/parse-worker.js.map +1 -1
  96. package/lib/dist/extraction/tree-sitter-helpers.d.ts.map +1 -1
  97. package/lib/dist/extraction/tree-sitter-helpers.js +1 -0
  98. package/lib/dist/extraction/tree-sitter-helpers.js.map +1 -1
  99. package/lib/dist/extraction/tree-sitter-types.d.ts +20 -1
  100. package/lib/dist/extraction/tree-sitter-types.d.ts.map +1 -1
  101. package/lib/dist/extraction/tree-sitter.d.ts +60 -1
  102. package/lib/dist/extraction/tree-sitter.d.ts.map +1 -1
  103. package/lib/dist/extraction/tree-sitter.js +905 -20
  104. package/lib/dist/extraction/tree-sitter.js.map +1 -1
  105. package/lib/dist/extraction/vba/call-sweep.d.ts +3 -0
  106. package/lib/dist/extraction/vba/call-sweep.d.ts.map +1 -0
  107. package/lib/dist/extraction/vba/call-sweep.js +206 -0
  108. package/lib/dist/extraction/vba/call-sweep.js.map +1 -0
  109. package/lib/dist/extraction/vba/calls.d.ts +55 -0
  110. package/lib/dist/extraction/vba/calls.d.ts.map +1 -0
  111. package/lib/dist/extraction/vba/calls.js +439 -0
  112. package/lib/dist/extraction/vba/calls.js.map +1 -0
  113. package/lib/dist/extraction/vba/constants.d.ts +42 -0
  114. package/lib/dist/extraction/vba/constants.d.ts.map +1 -0
  115. package/lib/dist/extraction/vba/constants.js +112 -0
  116. package/lib/dist/extraction/vba/constants.js.map +1 -0
  117. package/lib/dist/extraction/vba/context.d.ts +175 -0
  118. package/lib/dist/extraction/vba/context.d.ts.map +1 -0
  119. package/lib/dist/extraction/vba/context.js +268 -0
  120. package/lib/dist/extraction/vba/context.js.map +1 -0
  121. package/lib/dist/extraction/vba/controls.d.ts +24 -0
  122. package/lib/dist/extraction/vba/controls.d.ts.map +1 -0
  123. package/lib/dist/extraction/vba/controls.js +89 -0
  124. package/lib/dist/extraction/vba/controls.js.map +1 -0
  125. package/lib/dist/extraction/vba/declarations.d.ts +3 -0
  126. package/lib/dist/extraction/vba/declarations.d.ts.map +1 -0
  127. package/lib/dist/extraction/vba/declarations.js +154 -0
  128. package/lib/dist/extraction/vba/declarations.js.map +1 -0
  129. package/lib/dist/extraction/vba/dims.d.ts +3 -0
  130. package/lib/dist/extraction/vba/dims.d.ts.map +1 -0
  131. package/lib/dist/extraction/vba/dims.js +186 -0
  132. package/lib/dist/extraction/vba/dims.js.map +1 -0
  133. package/lib/dist/extraction/vba/docmd.d.ts +16 -0
  134. package/lib/dist/extraction/vba/docmd.d.ts.map +1 -0
  135. package/lib/dist/extraction/vba/docmd.js +179 -0
  136. package/lib/dist/extraction/vba/docmd.js.map +1 -0
  137. package/lib/dist/extraction/vba/enums-consts.d.ts +15 -0
  138. package/lib/dist/extraction/vba/enums-consts.d.ts.map +1 -0
  139. package/lib/dist/extraction/vba/enums-consts.js +178 -0
  140. package/lib/dist/extraction/vba/enums-consts.js.map +1 -0
  141. package/lib/dist/extraction/vba/implements.d.ts +3 -0
  142. package/lib/dist/extraction/vba/implements.d.ts.map +1 -0
  143. package/lib/dist/extraction/vba/implements.js +52 -0
  144. package/lib/dist/extraction/vba/implements.js.map +1 -0
  145. package/lib/dist/extraction/vba/procedures.d.ts +3 -0
  146. package/lib/dist/extraction/vba/procedures.d.ts.map +1 -0
  147. package/lib/dist/extraction/vba/procedures.js +239 -0
  148. package/lib/dist/extraction/vba/procedures.js.map +1 -0
  149. package/lib/dist/extraction/vba/sql-wrapper.d.ts +17 -0
  150. package/lib/dist/extraction/vba/sql-wrapper.d.ts.map +1 -0
  151. package/lib/dist/extraction/vba/sql-wrapper.js +212 -0
  152. package/lib/dist/extraction/vba/sql-wrapper.js.map +1 -0
  153. package/lib/dist/extraction/vba/tempvars.d.ts +10 -0
  154. package/lib/dist/extraction/vba/tempvars.d.ts.map +1 -0
  155. package/lib/dist/extraction/vba/tempvars.js +138 -0
  156. package/lib/dist/extraction/vba/tempvars.js.map +1 -0
  157. package/lib/dist/extraction/vba/text-utils.d.ts +73 -0
  158. package/lib/dist/extraction/vba/text-utils.d.ts.map +1 -0
  159. package/lib/dist/extraction/vba/text-utils.js +195 -0
  160. package/lib/dist/extraction/vba/text-utils.js.map +1 -0
  161. package/lib/dist/extraction/vba-extractor.d.ts +3 -951
  162. package/lib/dist/extraction/vba-extractor.d.ts.map +1 -1
  163. package/lib/dist/extraction/vba-extractor.js +46 -2934
  164. package/lib/dist/extraction/vba-extractor.js.map +1 -1
  165. package/lib/dist/extraction/vba-preprocess.d.ts +1 -1
  166. package/lib/dist/extraction/vba-preprocess.d.ts.map +1 -1
  167. package/lib/dist/extraction/vba-preprocess.js +300 -88
  168. package/lib/dist/extraction/vba-preprocess.js.map +1 -1
  169. package/lib/dist/extraction/wasm/tree-sitter-arkts.wasm +0 -0
  170. package/lib/dist/extraction/wasm/tree-sitter-cfml.wasm +0 -0
  171. package/lib/dist/extraction/wasm/tree-sitter-cfquery.wasm +0 -0
  172. package/lib/dist/extraction/wasm/tree-sitter-cfscript.wasm +0 -0
  173. package/lib/dist/extraction/wasm/tree-sitter-cobol.wasm +0 -0
  174. package/lib/dist/extraction/wasm/tree-sitter-erlang.wasm +0 -0
  175. package/lib/dist/extraction/wasm/tree-sitter-nix.wasm +0 -0
  176. package/lib/dist/extraction/wasm/tree-sitter-terraform.wasm +0 -0
  177. package/lib/dist/extraction/wasm/tree-sitter-vbnet.wasm +0 -0
  178. package/lib/dist/graph/traversal.d.ts.map +1 -1
  179. package/lib/dist/graph/traversal.js +76 -17
  180. package/lib/dist/graph/traversal.js.map +1 -1
  181. package/lib/dist/index.d.ts +81 -1
  182. package/lib/dist/index.d.ts.map +1 -1
  183. package/lib/dist/index.js +467 -10
  184. package/lib/dist/index.js.map +1 -1
  185. package/lib/dist/installer/index.d.ts +42 -0
  186. package/lib/dist/installer/index.d.ts.map +1 -1
  187. package/lib/dist/installer/index.js +97 -2
  188. package/lib/dist/installer/index.js.map +1 -1
  189. package/lib/dist/mcp/daemon.d.ts +35 -3
  190. package/lib/dist/mcp/daemon.d.ts.map +1 -1
  191. package/lib/dist/mcp/daemon.js +101 -9
  192. package/lib/dist/mcp/daemon.js.map +1 -1
  193. package/lib/dist/mcp/dynamic-boundaries.d.ts.map +1 -1
  194. package/lib/dist/mcp/dynamic-boundaries.js +2 -1
  195. package/lib/dist/mcp/dynamic-boundaries.js.map +1 -1
  196. package/lib/dist/mcp/early-ppid.d.ts +26 -0
  197. package/lib/dist/mcp/early-ppid.d.ts.map +1 -0
  198. package/lib/dist/mcp/early-ppid.js +29 -0
  199. package/lib/dist/mcp/early-ppid.js.map +1 -0
  200. package/lib/dist/mcp/index.d.ts.map +1 -1
  201. package/lib/dist/mcp/index.js +29 -4
  202. package/lib/dist/mcp/index.js.map +1 -1
  203. package/lib/dist/mcp/liveness-watchdog.d.ts +18 -1
  204. package/lib/dist/mcp/liveness-watchdog.d.ts.map +1 -1
  205. package/lib/dist/mcp/liveness-watchdog.js +73 -13
  206. package/lib/dist/mcp/liveness-watchdog.js.map +1 -1
  207. package/lib/dist/mcp/proxy.d.ts.map +1 -1
  208. package/lib/dist/mcp/proxy.js +30 -4
  209. package/lib/dist/mcp/proxy.js.map +1 -1
  210. package/lib/dist/mcp/query-pool.d.ts +14 -0
  211. package/lib/dist/mcp/query-pool.d.ts.map +1 -1
  212. package/lib/dist/mcp/query-pool.js +18 -0
  213. package/lib/dist/mcp/query-pool.js.map +1 -1
  214. package/lib/dist/mcp/session.d.ts +14 -0
  215. package/lib/dist/mcp/session.d.ts.map +1 -1
  216. package/lib/dist/mcp/session.js +28 -1
  217. package/lib/dist/mcp/session.js.map +1 -1
  218. package/lib/dist/mcp/startup-handshake.d.ts +44 -0
  219. package/lib/dist/mcp/startup-handshake.d.ts.map +1 -0
  220. package/lib/dist/mcp/startup-handshake.js +73 -0
  221. package/lib/dist/mcp/startup-handshake.js.map +1 -0
  222. package/lib/dist/mcp/tools.d.ts +22 -0
  223. package/lib/dist/mcp/tools.d.ts.map +1 -1
  224. package/lib/dist/mcp/tools.js +289 -23
  225. package/lib/dist/mcp/tools.js.map +1 -1
  226. package/lib/dist/mcp/transport.d.ts.map +1 -1
  227. package/lib/dist/mcp/transport.js +18 -0
  228. package/lib/dist/mcp/transport.js.map +1 -1
  229. package/lib/dist/project-config.d.ts +44 -0
  230. package/lib/dist/project-config.d.ts.map +1 -1
  231. package/lib/dist/project-config.js +151 -2
  232. package/lib/dist/project-config.js.map +1 -1
  233. package/lib/dist/resolution/c-fnptr-synthesizer.d.ts +2 -1
  234. package/lib/dist/resolution/c-fnptr-synthesizer.d.ts.map +1 -1
  235. package/lib/dist/resolution/c-fnptr-synthesizer.js +170 -146
  236. package/lib/dist/resolution/c-fnptr-synthesizer.js.map +1 -1
  237. package/lib/dist/resolution/callback-synthesizer.d.ts +1 -1
  238. package/lib/dist/resolution/callback-synthesizer.d.ts.map +1 -1
  239. package/lib/dist/resolution/callback-synthesizer.js +1109 -97
  240. package/lib/dist/resolution/callback-synthesizer.js.map +1 -1
  241. package/lib/dist/resolution/cooperative-yield.d.ts +32 -0
  242. package/lib/dist/resolution/cooperative-yield.d.ts.map +1 -0
  243. package/lib/dist/resolution/cooperative-yield.js +42 -0
  244. package/lib/dist/resolution/cooperative-yield.js.map +1 -0
  245. package/lib/dist/resolution/frameworks/cics.d.ts +20 -0
  246. package/lib/dist/resolution/frameworks/cics.d.ts.map +1 -0
  247. package/lib/dist/resolution/frameworks/cics.js +90 -0
  248. package/lib/dist/resolution/frameworks/cics.js.map +1 -0
  249. package/lib/dist/resolution/frameworks/index.d.ts.map +1 -1
  250. package/lib/dist/resolution/frameworks/index.js +6 -0
  251. package/lib/dist/resolution/frameworks/index.js.map +1 -1
  252. package/lib/dist/resolution/frameworks/java.d.ts.map +1 -1
  253. package/lib/dist/resolution/frameworks/java.js +14 -6
  254. package/lib/dist/resolution/frameworks/java.js.map +1 -1
  255. package/lib/dist/resolution/frameworks/terraform.d.ts +38 -0
  256. package/lib/dist/resolution/frameworks/terraform.d.ts.map +1 -0
  257. package/lib/dist/resolution/frameworks/terraform.js +277 -0
  258. package/lib/dist/resolution/frameworks/terraform.js.map +1 -0
  259. package/lib/dist/resolution/goframe-synthesizer.d.ts +2 -1
  260. package/lib/dist/resolution/goframe-synthesizer.d.ts.map +1 -1
  261. package/lib/dist/resolution/goframe-synthesizer.js +8 -3
  262. package/lib/dist/resolution/goframe-synthesizer.js.map +1 -1
  263. package/lib/dist/resolution/import-resolver.d.ts +7 -0
  264. package/lib/dist/resolution/import-resolver.d.ts.map +1 -1
  265. package/lib/dist/resolution/import-resolver.js +129 -4
  266. package/lib/dist/resolution/import-resolver.js.map +1 -1
  267. package/lib/dist/resolution/index.d.ts +85 -2
  268. package/lib/dist/resolution/index.d.ts.map +1 -1
  269. package/lib/dist/resolution/index.js +535 -58
  270. package/lib/dist/resolution/index.js.map +1 -1
  271. package/lib/dist/resolution/name-matcher.d.ts +22 -3
  272. package/lib/dist/resolution/name-matcher.d.ts.map +1 -1
  273. package/lib/dist/resolution/name-matcher.js +613 -39
  274. package/lib/dist/resolution/name-matcher.js.map +1 -1
  275. package/lib/dist/resolution/strip-comments.d.ts +1 -1
  276. package/lib/dist/resolution/strip-comments.d.ts.map +1 -1
  277. package/lib/dist/resolution/strip-comments.js +49 -0
  278. package/lib/dist/resolution/strip-comments.js.map +1 -1
  279. package/lib/dist/resolution/types.d.ts +29 -0
  280. package/lib/dist/resolution/types.d.ts.map +1 -1
  281. package/lib/dist/resolution/workspace-packages.d.ts +10 -0
  282. package/lib/dist/resolution/workspace-packages.d.ts.map +1 -1
  283. package/lib/dist/resolution/workspace-packages.js +142 -4
  284. package/lib/dist/resolution/workspace-packages.js.map +1 -1
  285. package/lib/dist/search/identifier-segments.d.ts +60 -0
  286. package/lib/dist/search/identifier-segments.d.ts.map +1 -0
  287. package/lib/dist/search/identifier-segments.js +176 -0
  288. package/lib/dist/search/identifier-segments.js.map +1 -0
  289. package/lib/dist/sync/git-hooks.d.ts.map +1 -1
  290. package/lib/dist/sync/git-hooks.js +2 -0
  291. package/lib/dist/sync/git-hooks.js.map +1 -1
  292. package/lib/dist/sync/watcher.d.ts +10 -5
  293. package/lib/dist/sync/watcher.d.ts.map +1 -1
  294. package/lib/dist/sync/watcher.js +51 -14
  295. package/lib/dist/sync/watcher.js.map +1 -1
  296. package/lib/dist/sync/worktree.d.ts +9 -0
  297. package/lib/dist/sync/worktree.d.ts.map +1 -1
  298. package/lib/dist/sync/worktree.js +45 -0
  299. package/lib/dist/sync/worktree.js.map +1 -1
  300. package/lib/dist/types.d.ts +25 -2
  301. package/lib/dist/types.d.ts.map +1 -1
  302. package/lib/dist/types.js +10 -0
  303. package/lib/dist/types.js.map +1 -1
  304. package/lib/dist/upgrade/index.d.ts +32 -0
  305. package/lib/dist/upgrade/index.d.ts.map +1 -1
  306. package/lib/dist/upgrade/index.js +163 -14
  307. package/lib/dist/upgrade/index.js.map +1 -1
  308. package/lib/dist/upgrade/remove-binary.d.ts +87 -0
  309. package/lib/dist/upgrade/remove-binary.d.ts.map +1 -0
  310. package/lib/dist/upgrade/remove-binary.js +289 -0
  311. package/lib/dist/upgrade/remove-binary.js.map +1 -0
  312. package/lib/dist/upgrade/update-check.d.ts +92 -0
  313. package/lib/dist/upgrade/update-check.d.ts.map +1 -0
  314. package/lib/dist/upgrade/update-check.js +258 -0
  315. package/lib/dist/upgrade/update-check.js.map +1 -0
  316. package/lib/node_modules/.modules.yaml +1 -1
  317. package/lib/node_modules/.pnpm-workspace-state-v1.json +1 -1
  318. package/lib/package.json +1 -1
  319. package/package.json +1 -1
  320. package/lib/dist/reasoning/config.d.ts +0 -45
  321. package/lib/dist/reasoning/config.d.ts.map +0 -1
  322. package/lib/dist/reasoning/config.js +0 -171
  323. package/lib/dist/reasoning/config.js.map +0 -1
  324. package/lib/dist/reasoning/credentials.d.ts +0 -5
  325. package/lib/dist/reasoning/credentials.d.ts.map +0 -1
  326. package/lib/dist/reasoning/credentials.js +0 -83
  327. package/lib/dist/reasoning/credentials.js.map +0 -1
  328. package/lib/dist/reasoning/login.d.ts +0 -21
  329. package/lib/dist/reasoning/login.d.ts.map +0 -1
  330. package/lib/dist/reasoning/login.js +0 -85
  331. package/lib/dist/reasoning/login.js.map +0 -1
  332. package/lib/dist/reasoning/reasoner.d.ts +0 -43
  333. package/lib/dist/reasoning/reasoner.d.ts.map +0 -1
  334. package/lib/dist/reasoning/reasoner.js +0 -308
  335. package/lib/dist/reasoning/reasoner.js.map +0 -1
@@ -61,62 +61,33 @@ exports.VbaExtractor = void 0;
61
61
  * Rejects `.form.txt` / `.report.txt` (REQ-CODE-9 — the form UI extractor
62
62
  * owns those). `Option ...` directives are inert — `stripVbaComments()`
63
63
  * drops them.
64
+ *
65
+ * This class is a THIN ORCHESTRATOR: it owns the input, drives the file-level
66
+ * node creation, and sequences the per-concern sweeps. All shared extraction
67
+ * state (node/edge accumulators, the `functionNodeByName` / `localVarTypeMap`
68
+ * maps, `moduleOrClassNode`) and the sweeps themselves live on
69
+ * `VbaExtractorContext` under `src/extraction/vba/`.
64
70
  */
65
71
  const path = __importStar(require("path"));
66
72
  const tree_sitter_helpers_1 = require("./tree-sitter-helpers");
67
73
  const vba_preprocess_1 = require("./vba-preprocess");
74
+ const context_1 = require("./vba/context");
75
+ const procedures_1 = require("./vba/procedures");
76
+ const declarations_1 = require("./vba/declarations");
77
+ const dims_1 = require("./vba/dims");
78
+ const enums_consts_1 = require("./vba/enums-consts");
79
+ const implements_1 = require("./vba/implements");
80
+ const call_sweep_1 = require("./vba/call-sweep");
68
81
  class VbaExtractor {
69
82
  filePath;
70
83
  source;
71
- nodes = [];
72
- edges = [];
73
- errors = [];
74
- unresolvedReferences = [];
75
- moduleOrClassNode = null;
76
- /**
77
- * Class-name prefix for `qualifiedName` composition.
78
- *
79
- * B3 (hueco 5): for `.cls` files every function node's `qualifiedName`
80
- * is composed as `${className}.${procName}` (e.g. `Form_TestForm.Form_Load`)
81
- * so cross-class callers can disambiguate which form each `Form_Load`
82
- * belongs to. For `.bas` files this is `null` — module-scoped procs
83
- * keep their bare-name `qualifiedName` (e.g. `DoThing`), preserving
84
- * the pre-hueco-5 behavior and every existing qualifiedName-based
85
- * assertion in `extraction-vba.test.ts`.
86
- *
87
- * Resolved once per `extract()` from `isCls` + `vbName` (both already
88
- * computed for `createModuleOrClassNode`); `sweepProcedures` reads it
89
- * to set `ProcInfo.qualifiedName` and the function node's qualifiedName.
90
- */
91
- classNamePrefix = null;
92
- /**
93
- * Map of procedure name (within this file) → list of ProcInfo. The list
94
- * (rather than a single value) is needed because VBA allows multiple
95
- * Property accessors with the same name: `Property Get Foo`, `Property
96
- * Let Foo`, and `Property Set Foo` all share the name `Foo`. Audit W6
97
- * (June 2026): the previous Map<string, ProcInfo> kept only the last
98
- * accessor, breaking same-file call resolution and emitting the wrong
99
- * `findFunctionNodeByName` match.
100
- */
101
- localProcs = new Map();
102
- /**
103
- * Cache: procedure name → first matching function node emitted for that
104
- * name. Audit S2 (June 2026): the previous implementation did an O(n)
105
- * linear scan of all nodes for every call site; with W6's multimap
106
- * the same-name collision can return multiple matches but bare-name
107
- * call resolution only needs the first.
108
- */
109
- functionNodeByName = new Map();
110
- /**
111
- * Cache: startLine → function node emitted for that line. Used by Fix 1
112
- * (June 2026): when Property Get/Let/Set share a name, we need to find
113
- * the SPECIFIC accessor's node by its declaration line rather than relying
114
- * on `functionNodeByName` which only stores the first accessor.
115
- */
116
- functionNodeByStartLine = new Map();
117
- constructor(filePath, source) {
84
+ ctx;
85
+ vbaTargets;
86
+ constructor(filePath, source, vbaTargets) {
118
87
  this.filePath = filePath;
119
88
  this.source = source;
89
+ this.vbaTargets = vbaTargets;
90
+ this.ctx = new context_1.VbaExtractorContext(filePath);
120
91
  }
121
92
  extract() {
122
93
  const startTime = Date.now();
@@ -124,18 +95,18 @@ class VbaExtractor {
124
95
  // REQ-CODE-9: hand form/report files off to VbaFormExtractor. If
125
96
  // somehow routed here, emit just a file node and return.
126
97
  if (this.isFormOrReportFile()) {
127
- this.nodes.push(this.createFileNode());
98
+ this.ctx.nodes.push(this.createFileNode());
128
99
  return this.result(startTime);
129
100
  }
130
- // Pre-process pipeline (per design): join continuations, blank inactive
131
- // conditional-compilation branches, strip comments, then sweep the
132
- // joined-but-uncommented source. The conditional preprocessor preserves
133
- // line count by replacing directives/inactive lines with empty strings.
134
- const joined = (0, vba_preprocess_1.joinLineContinuations)(this.source);
135
- const preprocessed = (0, vba_preprocess_1.preprocessConditionalCompilation)(joined);
136
- const uncommented = (0, vba_preprocess_1.stripVbaComments)(preprocessed);
101
+ // Pre-process pipeline (per design): strip comments first to prevent
102
+ // comments from blocking or causing false line continuations, then join
103
+ // continuations, and blank inactive conditional-compilation branches.
104
+ // Line count is preserved by all stages.
105
+ const cleanComments = (0, vba_preprocess_1.stripVbaComments)(this.source);
106
+ const joined = (0, vba_preprocess_1.joinLineContinuations)(cleanComments);
107
+ const uncommented = (0, vba_preprocess_1.preprocessConditionalCompilation)(joined, this.vbaTargets);
137
108
  // Create the file node.
138
- this.nodes.push(this.createFileNode());
109
+ this.ctx.nodes.push(this.createFileNode());
139
110
  // Resolve the file's "kind" — .cls → class, else module (including .bas
140
111
  // and the legacy .frm/.dsr stubs).
141
112
  const isCls = this.filePath.toLowerCase().endsWith('.cls');
@@ -152,68 +123,68 @@ class VbaExtractor {
152
123
  .basename(this.filePath)
153
124
  .replace(/\.[^.]+$/, '');
154
125
  const className = isCls ? (vbName ?? fallbackName) : null;
155
- this.classNamePrefix = className;
126
+ this.ctx.classNamePrefix = className;
156
127
  // Module/class node — created lazily so a file containing only
157
128
  // Option directives (REQ-CODE-10: "emits nothing") doesn't carry a
158
129
  // module node with no symbols attached.
159
130
  let hasAnySymbols = false;
160
131
  // Sweep: procedures → function nodes + contains edges.
161
- const procs = this.sweepProcedures(uncommented);
132
+ const procs = (0, procedures_1.sweepProcedures)(this.ctx, uncommented);
162
133
  if (procs.length > 0)
163
134
  hasAnySymbols = true;
164
135
  // Sweep: first-class Event / Type / Declare declarations (roadmap #26).
165
136
  // This runs before call-site scanning so `RaiseEvent Foo` and calls to a
166
137
  // `Declare` can resolve to real local nodes rather than fall through.
167
- const declarationCount = this.sweepEventsTypesAndDeclares(uncommented);
138
+ const declarationCount = (0, declarations_1.sweepEventsTypesAndDeclares)(this.ctx, uncommented);
168
139
  if (declarationCount > 0)
169
140
  hasAnySymbols = true;
170
141
  // Sweep: Implements (REQ-CODE-5).
171
- const implCount = this.sweepImplements(uncommented);
142
+ const implCount = (0, implements_1.sweepImplements)(this.ctx, uncommented);
172
143
  if (implCount > 0)
173
144
  hasAnySymbols = true;
174
145
  // Sweep: qualified Dim (REQ-CODE-6) and WithEvents (REQ-CODE-7).
175
- const dimCount = this.sweepDimsAndWithEvents(uncommented);
146
+ const dimCount = (0, dims_1.sweepDimsAndWithEvents)(this.ctx, uncommented);
176
147
  if (dimCount > 0)
177
148
  hasAnySymbols = true;
178
149
  // Sweep: Enum/Const declarations (REQ-CODE-12, REQ-CODE-13). Dysflow
179
150
  // exports the full module text, so a constants module carries its
180
151
  // `Const` lines and `Enum ... End Enum` blocks verbatim — these are the
181
152
  // project's domain dictionary and must be in the graph.
182
- const enumConstCount = this.sweepEnumsAndConsts(uncommented);
153
+ const enumConstCount = (0, enums_consts_1.sweepEnumsAndConsts)(this.ctx, uncommented);
183
154
  if (enumConstCount > 0)
184
155
  hasAnySymbols = true;
185
156
  // Sweep: call sites (REQ-CODE-4) and SQL-in-strings (REQ-CODE-8).
186
157
  // Both walk the same per-line view; combining them in one pass is
187
158
  // simpler and keeps line tracking consistent.
188
- this.sweepCallsAndSql(uncommented);
159
+ (0, call_sweep_1.sweepCallsAndSql)(this.ctx, uncommented);
189
160
  // Create the module/class node ONLY when the file has actual symbols
190
161
  // (REQ-CODE-10 — a file with ONLY Option directives emits zero symbol
191
162
  // nodes. A file with Enum/Const but no procedures DOES emit a module
192
163
  // node, since those are real symbols — see REQ-CODE-12/13).
193
164
  if (hasAnySymbols) {
194
- this.moduleOrClassNode = this.createModuleOrClassNode(isCls, vbName);
195
- this.nodes.push(this.moduleOrClassNode);
165
+ this.ctx.moduleOrClassNode = this.createModuleOrClassNode(isCls, vbName);
166
+ this.ctx.nodes.push(this.ctx.moduleOrClassNode);
196
167
  // Re-attribute contains/implements/reference edges whose source was
197
168
  // held in `pendingModuleOrClassSource` — see below. Since we now
198
169
  // know the module id, redirect those edges to it.
199
- for (const edge of this.pendingModuleOrClassSource) {
200
- edge.source = this.moduleOrClassNode.id;
170
+ for (const edge of this.ctx.pendingModuleOrClassSource) {
171
+ edge.source = this.ctx.moduleOrClassNode.id;
201
172
  }
202
- this.pendingModuleOrClassSource.length = 0;
173
+ this.ctx.pendingModuleOrClassSource.length = 0;
203
174
  // Sub New marker on the class node (REQ-CODE-3).
204
175
  if (isCls) {
205
176
  const hasNew = procs.some((p) => p.name === 'New');
206
177
  if (hasNew) {
207
- const md = (this.moduleOrClassNode.metadata ?? {});
178
+ const md = (this.ctx.moduleOrClassNode.metadata ?? {});
208
179
  md.hasClassInitializer = true;
209
180
  md.initializerName = 'New';
210
- this.moduleOrClassNode.metadata = md;
181
+ this.ctx.moduleOrClassNode.metadata = md;
211
182
  }
212
183
  }
213
184
  }
214
185
  }
215
186
  catch (error) {
216
- this.errors.push({
187
+ this.ctx.errors.push({
217
188
  message: `VBA extraction error: ${error instanceof Error ? error.message : String(error)}`,
218
189
  filePath: this.filePath,
219
190
  severity: 'error',
@@ -227,10 +198,10 @@ class VbaExtractor {
227
198
  // ---------------------------------------------------------------------------
228
199
  result(startTime) {
229
200
  return {
230
- nodes: this.nodes,
231
- edges: this.edges,
232
- unresolvedReferences: this.unresolvedReferences,
233
- errors: this.errors,
201
+ nodes: this.ctx.nodes,
202
+ edges: this.ctx.edges,
203
+ unresolvedReferences: this.ctx.unresolvedReferences,
204
+ errors: this.ctx.errors,
234
205
  durationMs: Date.now() - startTime,
235
206
  };
236
207
  }
@@ -300,2865 +271,6 @@ class VbaExtractor {
300
271
  updatedAt: Date.now(),
301
272
  };
302
273
  }
303
- /** Sub/Function/Property regex — captures visibility prefix, kind, and name. */
304
- static PROC_RE = /^\s*((?:Public|Private|Friend|Static)\s+)?(?:Static\s+)?(Sub|Function|Property(?:\s+(?:Get|Let|Set))?)\s+(\p{L}[\p{L}\p{N}_]*)/iu;
305
- /**
306
- * Walk the (uncommented, line-joined) source and emit one `function` node
307
- * per `Sub` / `Function` / `Property` declaration. Also records the proc
308
- * in `localProcs` so call-site resolution can distinguish same-file from
309
- * cross-module calls.
310
- */
311
- sweepProcedures(src) {
312
- const procs = [];
313
- const lines = src.split('\n');
314
- for (let i = 0; i < lines.length; i++) {
315
- const line = lines[i] ?? '';
316
- const m = VbaExtractor.PROC_RE.exec(line);
317
- if (!m)
318
- continue;
319
- const visibilityRaw = (m[1] ?? '').trim();
320
- const kindRaw = (m[2] ?? '').trim().toLowerCase();
321
- const name = m[3] ?? '';
322
- if (!name)
323
- continue;
324
- const lineNum = i + 1;
325
- // Normalize the visibility. VBA's "Static" keyword is a storage
326
- // specifier, not visibility, so we treat bare-static declarations as
327
- // 'Public' (the default) per spec R1. "Friend" is not in the Node
328
- // visibility enum; treat it as 'Public' since it's the closest
329
- // broader-than-private modifier.
330
- let visibility;
331
- switch (visibilityRaw.toLowerCase()) {
332
- case 'private':
333
- visibility = 'private';
334
- break;
335
- case 'public':
336
- case 'static':
337
- case 'friend':
338
- case '':
339
- default:
340
- visibility = 'public';
341
- break;
342
- }
343
- const kind = kindRaw.startsWith('sub')
344
- ? 'sub'
345
- : kindRaw.startsWith('function')
346
- ? 'function'
347
- : 'property';
348
- const proc = {
349
- name,
350
- // B3 (hueco 5): when this file is a `.cls`, prefix the
351
- // qualifiedName with the resolved class name so cross-class
352
- // queries (e.g. `Form_Load`) match only the owning class.
353
- // `.bas` files leave `qualifiedName === name`.
354
- qualifiedName: this.classNamePrefix
355
- ? `${this.classNamePrefix}.${name}`
356
- : name,
357
- kind,
358
- visibility,
359
- startLine: lineNum,
360
- };
361
- procs.push(proc);
362
- const bucket = this.localProcs.get(name);
363
- if (bucket)
364
- bucket.push(proc);
365
- else
366
- this.localProcs.set(name, [proc]);
367
- const nodeId = (0, tree_sitter_helpers_1.generateNodeId)(this.filePath, 'function', name, lineNum);
368
- const fnNode = {
369
- id: nodeId,
370
- kind: 'function',
371
- name,
372
- // B3 (hueco 5): same prefix rule as the ProcInfo above — class
373
- // methods get `${className}.${name}`, module-level Subs keep
374
- // their bare name.
375
- qualifiedName: this.classNamePrefix
376
- ? `${this.classNamePrefix}.${name}`
377
- : name,
378
- filePath: this.filePath,
379
- language: 'vba',
380
- startLine: lineNum,
381
- endLine: lineNum,
382
- startColumn: 0,
383
- endColumn: line.length,
384
- visibility,
385
- updatedAt: Date.now(),
386
- };
387
- this.nodes.push(fnNode);
388
- // Cache the first node emitted for this name — `findFunctionNodeByName`
389
- // (audit S2) becomes O(1) instead of O(n) per call site.
390
- if (!this.functionNodeByName.has(name)) {
391
- this.functionNodeByName.set(name, fnNode);
392
- }
393
- // Fix 1: also index by startLine so Property Get/Let/Set with the same
394
- // name can each be found by their exact declaration line.
395
- this.functionNodeByStartLine.set(lineNum, fnNode);
396
- // Hueco 3: synthesize the event-handler edge for Access naming
397
- // convention `<ControlName>_<EventName>` (e.g. `ComandoAltaPM_Click`,
398
- // `MotivoBorrado_AfterUpdate`). The `<X>_<Y>` shape is split on the
399
- // LAST underscore so multi-word events like `BeforeDelConfirm` parse
400
- // correctly. Form-level events (`Form_Load`, `Form_Open`,
401
- // `Form_Unload`, …) are NOT control handlers — they fire on the
402
- // form object itself, not on a control — so they are skipped here
403
- // (the form module node is the conceptual source for those).
404
- //
405
- // Scope guard: only emit when this .cls looks like a form code-
406
- // behind (`Form_*.cls`). Without this guard, regular service classes
407
- // whose methods happen to have underscores in their names
408
- // (e.g. `InformeRiesgoPDFServicio.cls` declares
409
- // `GenerarHTML_Principal`, `GetEstilosCSS_PDF`,
410
- // `Class_Initialize`) would synthesize ~550 spurious
411
- // `form-instance-control` stubs in real Dysflow projects. The
412
- // `Form_` prefix is the canonical Access code-behind naming
413
- // convention and matches the .form.txt siblings' basename.
414
- //
415
- // Cross-file synthesis caveat: the edge's source is a
416
- // `form-instance-control` node that lives in the sibling .form.txt,
417
- // not this .cls file. Two consequences:
418
- // 1. We also emit a STUB form-instance-control node locally with
419
- // the deterministic id so the per-file edge filter
420
- // (`insertedIds.has(source)`) accepts the edge. When the
421
- // sibling .form.txt is later indexed, VbaFormExtractor emits
422
- // the real form-instance-control node with the same id; the
423
- // `INSERT OR REPLACE` semantics in queries.ts:insertNode
424
- // overwrite the stub with the real one (preserving the
425
- // metadata.controlType, filePath, line range, etc.).
426
- // 2. When the .form.txt is processed FIRST (alphabetically
427
- // unlikely but possible), the real node exists in the DB
428
- // before this edge is committed; insertEdges's DB-level
429
- // endpoint check passes the edge naturally without the stub.
430
- // Either order converges on the same final state.
431
- // See vba-form-extractor.ts:findControlName for the matching real
432
- // form-instance-control node emission.
433
- const handler = parseEventHandlerName(name);
434
- // Prefix-driven sibling binding (issue #41). Both `Form_*.cls` and
435
- // `Report_*.cls` Dysflow code-behind files share the same code path;
436
- // only the sibling extension differs (`.form.txt` vs `.report.txt`).
437
- // The check is on the BASENAME prefix so a class called
438
- // `FormularioVentas.cls` or `ReportingHelper.cls` (no trailing
439
- // underscore) does not match — the trailing `_` is the discriminator.
440
- // Any other `.cls` (e.g. `InformeRiesgoPDFServicio.cls` with methods
441
- // like `GenerarHTML_Principal`) gets `codeBehindExt === null` and is
442
- // skipped, preserving the original Form_-only guard's behaviour for
443
- // non-form classes.
444
- const basename = path.basename(this.filePath).toLowerCase();
445
- const codeBehindExt = basename.startsWith('report_')
446
- ? '.report.txt'
447
- : basename.startsWith('form_')
448
- ? '.form.txt'
449
- : null;
450
- const isFormCodeBehind = codeBehindExt !== null;
451
- if (handler && isFormCodeBehind) {
452
- const siblingPath = this.filePath.replace(/\.cls$/i, codeBehindExt);
453
- const controlNodeId = (0, tree_sitter_helpers_1.generateNodeId)(siblingPath, 'form-instance-control', handler.controlName, 0);
454
- // Stub form-instance-control: local so the per-file edge filter
455
- // passes the event-handler edge. Overwritten by the real node
456
- // emitted from the sibling .form.txt (or .report.txt) at index time
457
- // (same id, same schema, INSERT OR REPLACE). No metadata.controlType
458
- // here — the sibling side carries the real control type.
459
- this.nodes.push({
460
- id: controlNodeId,
461
- kind: 'form-instance-control',
462
- name: handler.controlName,
463
- qualifiedName: `${siblingPath}::${handler.controlName}`,
464
- filePath: siblingPath,
465
- language: 'vba',
466
- startLine: 0,
467
- endLine: 0,
468
- startColumn: 0,
469
- endColumn: 0,
470
- updatedAt: Date.now(),
471
- });
472
- this.edges.push({
473
- source: nodeId,
474
- target: controlNodeId,
475
- kind: 'event-handler',
476
- provenance: 'heuristic',
477
- metadata: { eventName: handler.eventName },
478
- line: lineNum,
479
- column: 0,
480
- });
481
- }
482
- if (this.moduleOrClassNode) {
483
- this.edges.push({
484
- source: this.moduleOrClassNode.id,
485
- target: nodeId,
486
- kind: 'contains',
487
- });
488
- }
489
- else {
490
- // Module/class node is created lazily (see extract). Hold the edge in
491
- // pending so its source can be rewritten once the module exists.
492
- const edge = {
493
- source: '',
494
- target: nodeId,
495
- kind: 'contains',
496
- };
497
- this.edges.push(edge);
498
- this.pendingModuleOrClassSource.push(edge);
499
- }
500
- }
501
- return procs;
502
- }
503
- /** `[visibility] Event <Name>(...)` custom event declaration. */
504
- static EVENT_DECL_RE = /^\s*((?:Public|Private|Friend)\s+)?Event\s+(\p{L}[\p{L}\p{N}_]*)\b/iu;
505
- /** `[visibility] Type <Name>` user-defined type block start. */
506
- static TYPE_START_RE = /^\s*((?:Public|Private|Friend)\s+)?Type\s+(\p{L}[\p{L}\p{N}_]*)\b/iu;
507
- /** `End Type` user-defined type block end. */
508
- static TYPE_END_RE = /^\s*End\s+Type\b/iu;
509
- /** `<MemberName> As <Type>` inside a user-defined type block. */
510
- static TYPE_MEMBER_RE = /^\s*(\p{L}[\p{L}\p{N}_]*)\s*(?:\([^)]*\))?\s+As\s+(.+?)\s*$/iu;
511
- /** `[visibility] Declare [PtrSafe] Sub|Function <Name> Lib "dll" [Alias "x"] ...` */
512
- static DLL_DECLARE_RE = /^\s*((?:Public|Private)\s+)?Declare\s+(PtrSafe\s+)?(Sub|Function)\s+(\p{L}[\p{L}\p{N}_]*)\s+Lib\s+"([^"]+)"(?:\s+Alias\s+"([^"]+)")?/iu;
513
- /**
514
- * Roadmap #26 declaration sweep:
515
- * - Event declarations become `event` nodes and `RaiseEvent` can point to them.
516
- * - Type...End Type blocks become `type` + `type_member` nodes.
517
- * - Win32 API Declare statements become `declare` nodes, while still being
518
- * cached by name so normal call-site scanning can emit `calls` edges.
519
- */
520
- sweepEventsTypesAndDeclares(src) {
521
- const lines = src.split('\n');
522
- let count = 0;
523
- let currentType = null;
524
- for (let i = 0; i < lines.length; i++) {
525
- const line = lines[i] ?? '';
526
- const lineNum = i + 1;
527
- if (currentType) {
528
- if (VbaExtractor.TYPE_END_RE.test(line)) {
529
- currentType = null;
530
- continue;
531
- }
532
- const member = VbaExtractor.TYPE_MEMBER_RE.exec(line);
533
- if (member) {
534
- const memberName = member[1] ?? '';
535
- const memberType = (member[2] ?? '').trim();
536
- if (!memberName)
537
- continue;
538
- const memberId = (0, tree_sitter_helpers_1.generateNodeId)(this.filePath, 'type_member', memberName, lineNum);
539
- this.nodes.push({
540
- id: memberId,
541
- kind: 'type_member',
542
- name: memberName,
543
- qualifiedName: `${currentType.name}.${memberName}`,
544
- filePath: this.filePath,
545
- language: 'vba',
546
- startLine: lineNum,
547
- endLine: lineNum,
548
- startColumn: 0,
549
- endColumn: line.length,
550
- metadata: { memberType },
551
- updatedAt: Date.now(),
552
- });
553
- this.edges.push({
554
- source: currentType.id,
555
- target: memberId,
556
- kind: 'type-member',
557
- provenance: 'parser',
558
- });
559
- }
560
- continue;
561
- }
562
- const eventDecl = VbaExtractor.EVENT_DECL_RE.exec(line);
563
- if (eventDecl) {
564
- const visibility = VbaExtractor.foldVisibility(eventDecl[1] ?? '');
565
- const name = eventDecl[2] ?? '';
566
- if (!name)
567
- continue;
568
- const eventId = (0, tree_sitter_helpers_1.generateNodeId)(this.filePath, 'event', name, lineNum);
569
- const eventNode = {
570
- id: eventId,
571
- kind: 'event',
572
- name,
573
- signature: line.trim(),
574
- qualifiedName: this.classNamePrefix ? `${this.classNamePrefix}.${name}` : name,
575
- filePath: this.filePath,
576
- language: 'vba',
577
- startLine: lineNum,
578
- endLine: lineNum,
579
- startColumn: 0,
580
- endColumn: line.length,
581
- visibility,
582
- updatedAt: Date.now(),
583
- };
584
- this.nodes.push(eventNode);
585
- this.localEvents.set(name.toLowerCase(), eventNode);
586
- this.pushContainsFromModule(eventId);
587
- count++;
588
- continue;
589
- }
590
- const typeStart = VbaExtractor.TYPE_START_RE.exec(line);
591
- if (typeStart) {
592
- const visibility = VbaExtractor.foldVisibility(typeStart[1] ?? '');
593
- const name = typeStart[2] ?? '';
594
- if (!name)
595
- continue;
596
- const typeId = (0, tree_sitter_helpers_1.generateNodeId)(this.filePath, 'type', name, lineNum);
597
- this.nodes.push({
598
- id: typeId,
599
- kind: 'type',
600
- name,
601
- qualifiedName: this.classNamePrefix ? `${this.classNamePrefix}.${name}` : name,
602
- filePath: this.filePath,
603
- language: 'vba',
604
- startLine: lineNum,
605
- endLine: lineNum,
606
- startColumn: 0,
607
- endColumn: line.length,
608
- visibility,
609
- updatedAt: Date.now(),
610
- });
611
- this.pushContainsFromModule(typeId);
612
- currentType = { id: typeId, name };
613
- count++;
614
- continue;
615
- }
616
- const declaration = VbaExtractor.DLL_DECLARE_RE.exec(line);
617
- if (declaration) {
618
- const visibility = VbaExtractor.foldVisibility(declaration[1] ?? '');
619
- const ptrSafe = !!declaration[2];
620
- const declareKind = (declaration[3] ?? '').toLowerCase();
621
- const name = declaration[4] ?? '';
622
- const dll = declaration[5] ?? '';
623
- const aliasName = declaration[6] ?? undefined;
624
- if (!name)
625
- continue;
626
- const declareId = (0, tree_sitter_helpers_1.generateNodeId)(this.filePath, 'declare', name, lineNum);
627
- const declareNode = {
628
- id: declareId,
629
- kind: 'declare',
630
- name,
631
- qualifiedName: this.classNamePrefix ? `${this.classNamePrefix}.${name}` : name,
632
- filePath: this.filePath,
633
- language: 'vba',
634
- startLine: lineNum,
635
- endLine: lineNum,
636
- startColumn: 0,
637
- endColumn: line.length,
638
- visibility,
639
- metadata: {
640
- isDeclare: true,
641
- dll,
642
- declareKind,
643
- ptrSafe,
644
- ...(aliasName ? { aliasName } : {}),
645
- },
646
- updatedAt: Date.now(),
647
- };
648
- this.nodes.push(declareNode);
649
- if (!this.functionNodeByName.has(name)) {
650
- this.functionNodeByName.set(name, declareNode);
651
- }
652
- this.pushContainsFromModule(declareId);
653
- count++;
654
- }
655
- }
656
- return count;
657
- }
658
- /** Implements regex. */
659
- static IMPLEMENTS_RE = /^\s*Implements\s+(\p{L}[\p{L}\p{N}_]*)/iu;
660
- /** Edges whose source needs to be set to the module/class id once it exists. */
661
- pendingModuleOrClassSource = [];
662
- sweepImplements(src) {
663
- const lines = src.split('\n');
664
- let count = 0;
665
- for (let i = 0; i < lines.length; i++) {
666
- const line = lines[i] ?? '';
667
- const m = VbaExtractor.IMPLEMENTS_RE.exec(line);
668
- if (!m)
669
- continue;
670
- const name = m[1] ?? '';
671
- if (!name)
672
- continue;
673
- const lineNum = i + 1;
674
- const targetId = (0, tree_sitter_helpers_1.generateNodeId)(this.filePath, 'interface', name, lineNum);
675
- this.nodes.push({
676
- id: targetId,
677
- kind: 'interface',
678
- name,
679
- qualifiedName: name,
680
- filePath: this.filePath,
681
- language: 'vba',
682
- startLine: lineNum,
683
- endLine: lineNum,
684
- startColumn: 0,
685
- endColumn: line.length,
686
- updatedAt: Date.now(),
687
- });
688
- const edge = {
689
- source: '', // placeholder; rewritten after module node exists
690
- target: targetId,
691
- kind: 'implements',
692
- // S4 fix: `Implements IFoo` is a static, source-declared fact —
693
- // not a guess. Use the `parser` provenance (generalizes
694
- // `tree-sitter` for non-tree-sitter extractors like our regex
695
- // sweepers) instead of `heuristic`, which is reserved for
696
- // guessed/inferred edges.
697
- provenance: 'parser',
698
- line: lineNum,
699
- column: 0,
700
- };
701
- this.edges.push(edge);
702
- this.pendingModuleOrClassSource.push(edge);
703
- count++;
704
- }
705
- return count;
706
- }
707
- /**
708
- * Check that a line is a variable declaration and NOT a Sub/Function/
709
- * Property/Const/WithEvents header (those have their own sweeps).
710
- * Fix 1+3 (Issues #1,#3): replaced the old single-match DIM_QUAL_RE /
711
- * DIM_UNQUAL_RE pair with a prefix-check + global scan that handles
712
- * `As New <Type>`, multi-variable `Dim a As Foo, b As Bar`, and all
713
- * visibility keywords in one pass.
714
- * Issue #47: now also accepts `Global` (module-level typed instance) and
715
- * `Static` (procedure-local retention modifier) so they emit the same
716
- * `references` edge and `localVarTypeMap` registration as their `Dim`
717
- * siblings today. The negative lookahead is unchanged: `Const` is still
718
- * routed to `sweepEnumsAndConsts`.
719
- */
720
- static DIM_DECL_PREFIX_RE = /^\s*(?:Dim|Private|Public|Global|Static)\s+(?!(?:Function|Sub|Property|Const|WithEvents)\b)/i;
721
- /**
722
- * Globally scan all `identifier As [New] TypePart1[.TypePart2]` on a
723
- * variable declaration line. Run with /g after confirming DIM_DECL_PREFIX_RE.
724
- *
725
- * Groups: (1) variable name, (2) bracketed outer type, (3) unbracketed
726
- * outer type, (4) bracketed inner type (if qualified), (5) unbracketed
727
- * inner type. The variable name is always bare (`Dim` cannot declare a
728
- * bracketed variable). The TYPE position accepts BOTH bracketed names
729
- * with spaces (e.g. `[Clase Con Espacios]`) and bare identifiers — the
730
- * bracketed capture wins when present. Only one of (2)/(3) and one of
731
- * (4)/(5) is ever populated per match.
732
- * `(?:New\s+)?` consumes the VBA auto-instantiation keyword so it is
733
- * never captured as the type name (Fix 1).
734
- *
735
- * Issue #54: extends the type alternative to accept `[Name With Spaces]`
736
- * so `Dim x As [Clase Con Espacios]` emits a `references` edge to
737
- * `Clase Con Espacios` (brackets unwrapped). The unwrap is applied in
738
- * the sweep loop by picking the bracketed capture group when present.
739
- */
740
- static DIM_ALL_VARS_RE = /\b(\p{L}[\p{L}\p{N}_]*)\s+As\s+(?:New\s+)?(?:\[([^\]]+)\]|(\p{L}[\p{L}\p{N}_]*))(?:\.(?:\[([^\]]+)\]|(\p{L}[\p{L}\p{N}_]*)))?/giu;
741
- /**
742
- * Bare-declared variable capture for the `Dim|Private|Public|Global|Static`
743
- * prefix. Captures (1) the variable name. Used to register bare `Dim x`
744
- * (no `As` clause) and explicit-primitive `Dim x As Long|String|...`
745
- * declarations into `localVarTypeMap` so the type tracking is consistent
746
- * across all three Dim shapes:
747
- *
748
- * `Dim x` → outer = 'variant' (VBA default)
749
- * `Dim x As Variant` → outer = 'variant' (PRIMITIVE_TYPES member)
750
- * `Dim x As Long` → outer = 'long' (PRIMITIVE_TYPES member)
751
- * `Dim x As Foo` → outer = 'foo' (project class — non-primitive)
752
- *
753
- * Antigravity audit Task 3: the previous `DIM_ALL_VARS_RE` only matched
754
- * the `... As <Type>` form, so a bare `Dim x` was invisible to
755
- * `isLocalProjectClassVar` / `scanCallSites` and `x.Method(1)` produced
756
- * a dead-end `calls` edge to a stub named `x.Method` that no resolver
757
- * could repoint. Registering bare Dim with `outer = 'variant'` closes
758
- * the gate, so `scanCallSites` skips ONLY when the receiver is mapped
759
- * as a primitive — leaving the "undeclared receiver → stub → resolver
760
- * repoints" path intact for cross-module qualified calls like
761
- * `modUtils.Foo(1)` (`modUtils` is not in `localVarTypeMap`).
762
- */
763
- static BARE_DIM_VAR_RE = /^\s*(?:Dim|Private|Public|Global|Static)\s+(\p{L}[\p{L}\p{N}_]*)\s*(?:,|$|\b)/iu;
764
- /**
765
- * VBA primitive type names — skipped when emitted as Dim targets so
766
- * we don't pollute the graph with `As Long` / `As String` references.
767
- * Fix 4: all entries are LOWERCASE and the lookup lowercases the
768
- * captured type name so `As long`, `As LONG`, `As Long` all match.
769
- * Fix 1 (Issue #1): added `'new'` as a backstop so that if the `As New`
770
- * pattern is ever captured as a type name it is silently skipped.
771
- */
772
- static PRIMITIVE_TYPES = new Set([
773
- 'long', 'integer', 'short', 'byte', 'single', 'double', 'currency',
774
- 'string', 'boolean', 'date', 'variant', 'object', 'error',
775
- 'empty', 'null', 'longptr', 'longlong', 'new',
776
- ]);
777
- /** `WithEvents m_X As Form_Foo` — Dim/Private/Public/Global/Static prefix is optional. */
778
- static WITHEVENTS_RE = /^\s*(?:(?:Dim|Private|Public|Global|Static)\s+)?WithEvents\s+\p{L}[\p{L}\p{N}_]*\s+As\s+(\p{L}[\p{L}\p{N}_]*)/iu;
779
- sweepDimsAndWithEvents(src) {
780
- const lines = src.split('\n');
781
- let count = 0;
782
- for (let i = 0; i < lines.length; i++) {
783
- const line = lines[i] ?? '';
784
- const lineNum = i + 1;
785
- // Fix 1 + Fix 3 (Issues #1, #3): replace the old single-match
786
- // DIM_QUAL_RE / DIM_UNQUAL_RE pair with a global scan that handles
787
- // `As New <Type>`, multi-variable `Dim a As Foo, b As Bar`, and
788
- // qualified `Dim x As Foo.Bar` in a single pass.
789
- if (VbaExtractor.DIM_DECL_PREFIX_RE.test(line)) {
790
- VbaExtractor.DIM_ALL_VARS_RE.lastIndex = 0;
791
- let m;
792
- while ((m = VbaExtractor.DIM_ALL_VARS_RE.exec(line)) !== null) {
793
- const varName = m[1] ?? '';
794
- // Issue #54: DIM_ALL_VARS_RE groups (2) and (3) are alternative
795
- // captures for the same outer-type position — the bracketed
796
- // alternative wins when present, the bare one otherwise. The
797
- // captured value is already unwrapped (the `[...]` is consumed
798
- // by the regex, group (2) holds the inner content). Same shape
799
- // for the inner type at groups (4)/(5).
800
- const outerType = m[2] ?? m[3] ?? '';
801
- const innerType = m[4] ?? m[5] ?? '';
802
- // Fix 2 (Issue #2): populate the local var type map so that
803
- // `sweepCallsAndSql` can gate qualified statement-form calls.
804
- if (varName && outerType) {
805
- this.localVarTypeMap.set(varName.toLowerCase(), {
806
- outer: outerType,
807
- qualified: !!innerType,
808
- });
809
- }
810
- if (innerType) {
811
- // Qualified form (`Dim x As Foo.Bar`) — emit reference to the
812
- // outer type `Foo` (same behaviour as the old DIM_QUAL_RE path).
813
- if (outerType) {
814
- this.emitReference(outerType, lineNum, 0, 'vba-name-resolution');
815
- count++;
816
- }
817
- }
818
- else {
819
- // Unqualified form (`Dim x As SomeType`, `Dim x As New SomeType`)
820
- // — emit reference only when the type is not a primitive or keyword.
821
- if (outerType && !VbaExtractor.PRIMITIVE_TYPES.has(outerType.toLowerCase())) {
822
- this.emitReference(outerType, lineNum, 0, 'vba-name-resolution');
823
- count++;
824
- }
825
- }
826
- }
827
- // Antigravity audit Task 3: bare `Dim x` (no `As` clause) and
828
- // explicit-primitive `Dim x As <primitive>` declarations must
829
- // still register `x` in `localVarTypeMap` so the qualified-call
830
- // site scan in `scanCallSites` can gate the dead-end stub that
831
- // no resolver could ever repoint.
832
- //
833
- // Captures the FIRST variable name only. Multi-variable bare
834
- // Dim (e.g. `Dim a, b, c` without an `As` clause) is rare in
835
- // real Dysflow fixtures and is intentionally NOT tracked here —
836
- // those bare variables fall back to the undeclared-receiver
837
- // path, which is the conservative choice. Skip if the typed-form
838
- // loop already populated the entry.
839
- const bm = VbaExtractor.BARE_DIM_VAR_RE.exec(line);
840
- if (bm) {
841
- const varName = bm[1] ?? '';
842
- if (varName) {
843
- const key = varName.toLowerCase();
844
- if (!this.localVarTypeMap.has(key)) {
845
- // Look for an `As <Type>` continuation on the same line so the
846
- // outer type matches the existing typed-form behaviour. If
847
- // absent, the variable is implicit `Variant` per VBA semantics.
848
- const asRe = /\bAs\s+(\p{L}[\p{L}\p{N}_]*)/iu;
849
- const asMatch = asRe.exec(line);
850
- const outer = asMatch ? (asMatch[1] ?? '').toLowerCase() : 'variant';
851
- this.localVarTypeMap.set(key, {
852
- outer,
853
- qualified: false,
854
- });
855
- }
856
- }
857
- }
858
- }
859
- // WithEvents declarations — handled by their own regex; also populate
860
- // the local var type map for completeness.
861
- const weMatch = VbaExtractor.WITHEVENTS_RE.exec(line);
862
- if (weMatch) {
863
- const formType = weMatch[1] ?? '';
864
- if (formType) {
865
- // Extract the variable name from the WithEvents line for the map.
866
- const weVarM = /^\s*(?:(?:Dim|Private|Public|Global|Static)\s+)?WithEvents\s+(\p{L}[\p{L}\p{N}_]*)/iu.exec(line);
867
- const weVarName = weVarM?.[1] ?? '';
868
- if (weVarName) {
869
- this.localVarTypeMap.set(weVarName.toLowerCase(), {
870
- outer: formType,
871
- qualified: false,
872
- withEvents: true,
873
- variableName: weVarName,
874
- });
875
- }
876
- this.emitReference(formType, lineNum, 0, 'vba-withevents');
877
- const targetId = (0, tree_sitter_helpers_1.generateNodeId)(this.filePath, 'class', formType, 0);
878
- const subscriberEdge = {
879
- source: this.moduleOrClassNode?.id ?? '',
880
- target: targetId,
881
- kind: 'subscribes-event',
882
- provenance: 'heuristic',
883
- metadata: {
884
- synthesizedBy: 'vba-withevents',
885
- variableName: weVarName || undefined,
886
- },
887
- line: lineNum,
888
- column: 0,
889
- };
890
- this.edges.push(subscriberEdge);
891
- if (!this.moduleOrClassNode) {
892
- this.pendingModuleOrClassSource.push(subscriberEdge);
893
- }
894
- count++;
895
- }
896
- }
897
- }
898
- return count;
899
- }
900
- /** `[visibility] Enum <Name>` — opens an enum block. */
901
- static ENUM_START_RE = /^\s*(?:(Public|Private|Friend|Global)\s+)?Enum\s+(\p{L}[\p{L}\p{N}_]*)/iu;
902
- /** `End Enum` — closes the current enum block. */
903
- static ENUM_END_RE = /^\s*End\s+Enum\b/i;
904
- /**
905
- * An enum member line: a leading identifier optionally followed by `=
906
- * <value>`. Runs only inside an open Enum block, on the (already
907
- * comment-stripped) source, so a trailing `'comment` never reaches here.
908
- * `\p{L}` covers accented member names (e.g. `Sí`).
909
- */
910
- static ENUM_MEMBER_RE = /^\s*(\p{L}[\p{L}\p{N}_]*)\s*(?:=|$)/u;
911
- /** `[visibility] Const <decls>` — captures visibility (1) and the rest (2). */
912
- static CONST_DECL_RE = /^\s*(?:(Public|Private|Friend|Global)\s+)?Const\s+(.+)$/i;
913
- /**
914
- * Issue #52: shared `End Sub` / `End Function` / `End Property` marker.
915
- * Promoted from a local regex in `sweepCallsAndSql` so `sweepEnumsAndConsts`
916
- * can walk the same proc boundaries and decide Const scope per line.
917
- * The `(?:^|:\s*)` prefix tolerates colon-separated single-line procs
918
- * (`Public Sub X(): ... : End Sub`) so the proc stack pops on the same
919
- * physical line.
920
- */
921
- static PROCEDURE_END_RE = /(?:^|:\s*)End\s+(?:Sub|Function|Property)\b/i;
922
- /**
923
- * Fold a VBA visibility keyword to the canonical lowercase enum, matching
924
- * the procedure convention: `Private` → 'private'; `Public`, `Global`,
925
- * `Friend`, or none → 'public' (VBA's default module-level `Const`/`Enum`
926
- * is Private, but we follow the same broader-than-private fold the proc
927
- * sweep uses so visibility is consistent across symbol kinds).
928
- */
929
- static foldVisibility(raw) {
930
- return raw.trim().toLowerCase() === 'private' ? 'private' : 'public';
931
- }
932
- /**
933
- * Walk the (uncommented, line-joined) source and emit:
934
- * - one `enum` node per `Enum <Name>` block, with one `enum_member` node
935
- * per member and a `contains` edge enum→member;
936
- * - one `constant` node per name declared on a `Const` line (multi-name
937
- * lines emit one node per name);
938
- * - a `contains` edge from the module/class node to each enum and constant
939
- * (held in `pendingModuleOrClassSource` until the module node exists).
940
- *
941
- * Returns the number of top-level symbols (enums + constants) emitted so
942
- * the caller can flip `hasAnySymbols`.
943
- */
944
- sweepEnumsAndConsts(src) {
945
- const lines = src.split('\n');
946
- let count = 0;
947
- let currentEnum = null;
948
- // Issue #52: reset the shared scope stack + lookup key so leftover
949
- // state from a previous extract() (impossible in production but
950
- // possible in unit tests that construct a fresh extractor and run
951
- // twice) never leaks across sweeps. The walk below updates both
952
- // every iteration; `sweepCallsAndSql` resets again at its own
953
- // start, before any OpenForm/OpenQuery reader consults them.
954
- this.procStack.length = 0;
955
- this.currentProcKey = 'module';
956
- for (let i = 0; i < lines.length; i++) {
957
- const line = lines[i] ?? '';
958
- const lineNum = i + 1;
959
- // Issue #52: track proc scope so Const declarations on this line
960
- // can decide whether they belong to the module (currentProcKey
961
- // === 'module') or to the top-most procedure (write the per-proc
962
- // bucket, skip the module-level `constant` node emission).
963
- //
964
- // PROC_RE cannot overlap with CONST_DECL_RE on the same physical
965
- // line (different leading keywords), so it is safe to advance the
966
- // stack here and then fall through to the rest of the body.
967
- const procStart = VbaExtractor.PROC_RE.exec(line);
968
- if (procStart) {
969
- this.procStack.push(lineNum);
970
- this.currentProcKey = String(lineNum);
971
- }
972
- else if (VbaExtractor.PROCEDURE_END_RE.test(line) && this.procStack.length > 0) {
973
- this.procStack.pop();
974
- this.currentProcKey =
975
- this.procStack.length > 0
976
- ? String(this.procStack[this.procStack.length - 1])
977
- : 'module';
978
- }
979
- if (currentEnum) {
980
- if (VbaExtractor.ENUM_END_RE.test(line)) {
981
- currentEnum = null;
982
- continue;
983
- }
984
- const mm = VbaExtractor.ENUM_MEMBER_RE.exec(line);
985
- if (mm) {
986
- const memberName = mm[1] ?? '';
987
- if (memberName) {
988
- const memberId = (0, tree_sitter_helpers_1.generateNodeId)(this.filePath, 'enum_member', memberName, lineNum);
989
- this.nodes.push({
990
- id: memberId,
991
- kind: 'enum_member',
992
- name: memberName,
993
- qualifiedName: `${currentEnum.name}.${memberName}`,
994
- filePath: this.filePath,
995
- language: 'vba',
996
- startLine: lineNum,
997
- endLine: lineNum,
998
- startColumn: 0,
999
- endColumn: line.length,
1000
- updatedAt: Date.now(),
1001
- });
1002
- // enum → member: source is known, emit directly (not pending).
1003
- this.edges.push({
1004
- source: currentEnum.id,
1005
- target: memberId,
1006
- kind: 'contains',
1007
- });
1008
- }
1009
- }
1010
- continue;
1011
- }
1012
- const enumStart = VbaExtractor.ENUM_START_RE.exec(line);
1013
- if (enumStart) {
1014
- const visibility = VbaExtractor.foldVisibility(enumStart[1] ?? '');
1015
- const name = enumStart[2] ?? '';
1016
- if (!name)
1017
- continue;
1018
- const enumId = (0, tree_sitter_helpers_1.generateNodeId)(this.filePath, 'enum', name, lineNum);
1019
- this.nodes.push({
1020
- id: enumId,
1021
- kind: 'enum',
1022
- name,
1023
- qualifiedName: name,
1024
- filePath: this.filePath,
1025
- language: 'vba',
1026
- startLine: lineNum,
1027
- endLine: lineNum,
1028
- startColumn: 0,
1029
- endColumn: line.length,
1030
- visibility,
1031
- updatedAt: Date.now(),
1032
- });
1033
- this.pushContainsFromModule(enumId);
1034
- currentEnum = { id: enumId, name };
1035
- count++;
1036
- continue;
1037
- }
1038
- const constDecl = VbaExtractor.CONST_DECL_RE.exec(line);
1039
- if (constDecl) {
1040
- const visibility = VbaExtractor.foldVisibility(constDecl[1] ?? '');
1041
- const body = constDecl[2] ?? '';
1042
- const declarations = parseConstDeclarations(body);
1043
- for (const declaration of declarations) {
1044
- const constName = declaration.name;
1045
- if (!constName)
1046
- continue;
1047
- // Issue #52: every Const line (module-level or proc-local)
1048
- // writes into a per-scope resolution bucket so
1049
- // `DoCmd.OpenForm FORM_X` later resolves correctly; module-level
1050
- // Consts additionally emit a `constant` graph node + the
1051
- // module→constant `contains` edge. Proc-local Consts skip both
1052
- // (the const is not a module symbol, so the wrong-containment
1053
- // node + edge the pre-fix code emitted are gone), but the
1054
- // per-proc bucket keeps OpenForm/OpenQuery argument
1055
- // resolution working exactly as before.
1056
- if (declaration.value !== null) {
1057
- this.setLocalConstInScope(this.currentProcKey, constName, declaration.value);
1058
- }
1059
- if (this.procStack.length > 0)
1060
- continue;
1061
- const constId = (0, tree_sitter_helpers_1.generateNodeId)(this.filePath, 'constant', constName, lineNum);
1062
- this.nodes.push({
1063
- id: constId,
1064
- kind: 'constant',
1065
- name: constName,
1066
- qualifiedName: constName,
1067
- filePath: this.filePath,
1068
- language: 'vba',
1069
- startLine: lineNum,
1070
- endLine: lineNum,
1071
- startColumn: 0,
1072
- endColumn: line.length,
1073
- visibility,
1074
- metadata: declaration.value !== null ? { value: declaration.value } : undefined,
1075
- updatedAt: Date.now(),
1076
- });
1077
- this.pushContainsFromModule(constId);
1078
- count++;
1079
- }
1080
- }
1081
- }
1082
- return count;
1083
- }
1084
- /**
1085
- * Emit a `contains` edge from the (lazily-created) module/class node to
1086
- * `targetId`. Mirrors the pending-source pattern the procedure and
1087
- * implements sweeps use: the module node doesn't exist yet, so the edge's
1088
- * source is rewritten once `extract()` creates it.
1089
- */
1090
- pushContainsFromModule(targetId) {
1091
- if (this.moduleOrClassNode) {
1092
- this.edges.push({
1093
- source: this.moduleOrClassNode.id,
1094
- target: targetId,
1095
- kind: 'contains',
1096
- });
1097
- return;
1098
- }
1099
- const edge = { source: '', target: targetId, kind: 'contains' };
1100
- this.edges.push(edge);
1101
- this.pendingModuleOrClassSource.push(edge);
1102
- }
1103
- /**
1104
- * Walk the (uncommented) source per line. For each line:
1105
- * - If it starts a `Sub`/`Function`/`Property`, push the proc onto a stack.
1106
- * - If it ends one (`End Sub` / `End Function` / `End Property`), pop.
1107
- * - While inside a procedure, scan the line for call-site patterns and
1108
- * SQL-wrapper patterns.
1109
- *
1110
- * Call-site regex captures either `Name(...)` (same-file candidate) or
1111
- * `Receiver.Member(...)` (qualified — emit a synthetic node + heuristic
1112
- * edge). The receiver AND member alternatives accept BOTH the bare form
1113
- * (`Foo`) and the VBA bracketed form (`[Foo Bar]`) — bracketed captures
1114
- * win when present. Only one of (1)/(2) and one of (3)/(4) is ever
1115
- * populated per match. Brackets are stripped by the regex itself (the
1116
- * capture groups hold the inner content), so callers receive unwrapped
1117
- * identifiers and the `${name}.${proc}` stub shape stays canonical.
1118
- *
1119
- * Issue #54: the bracketed alternative was previously absent, so
1120
- * `[FUNCIONES UTILES].FormatearFecha(fecha)` (a real Dysflow-exported
1121
- * idiom for modules with spaces in their names) was silently dropped.
1122
- */
1123
- static CALL_RE = /(?<![\w.])(?:\[([^\]]+)\]|(\p{L}[\p{L}\p{N}_]*))(?:\.(?:\[([^\]]+)\]|(\p{L}[\p{L}\p{N}_]*)))?\s*\(/gu;
1124
- /** SQL wrapper helpers — order matters because `db.Execute` is a suffix of others. */
1125
- static SQL_WRAPPERS = [
1126
- { name: 'DoCmd.RunSQL', re: /\bDoCmd\.RunSQL\s+"((?:[^"]|"")*)"/g },
1127
- { name: '*db.OpenRecordset', re: /\b(?:\p{L}[\p{L}\p{N}_]*)?db\b(?:\(\))?\.OpenRecordset\s+"((?:[^"]|"")*)"/giu },
1128
- { name: '*db.Execute', re: /\b(?:\p{L}[\p{L}\p{N}_]*)?db\b(?:\(\))?\.Execute\s+"((?:[^"]|"")*)"/giu },
1129
- // Fix 4 (Issue #4): inline-literal forms `getdb().Execute "..."` and
1130
- // `getdb().OpenRecordset "..."` — the variable form is covered by
1131
- // SQL_VAR_EXEC_RE but the direct-literal form was missing.
1132
- { name: 'getdb().Execute', re: /\bgetdb\(\)\.Execute\s+"((?:[^"]|"")*)"/g },
1133
- { name: 'getdb().OpenRecordset', re: /\bgetdb\(\)\.OpenRecordset\s+"((?:[^"]|"")*)"/g },
1134
- ];
1135
- /**
1136
- * `DoCmd.OpenForm "<FormName>"` modelling regex — B4 (hueco 6).
1137
- *
1138
- * Real VBA idiom (matches literal and bare-identifier forms):
1139
- * `DoCmd.OpenForm "MyForm"`
1140
- * `DoCmd.OpenForm FORM_MY_FORM`
1141
- * `DoCmd.OpenForm "MyForm", acNormal, , , acFormEdit`
1142
- *
1143
- * Captures the first argument (group 1). String literals are unwrapped;
1144
- * bare identifiers resolve against local Const declarations, falling back
1145
- * to the identifier name when unknown. The trailing positional args
1146
- * (`acNormal`, `acFormEdit`, etc.) are intentionally NOT captured.
1147
- *
1148
- * Why a separate dispatch: `DoCmd` is in `RUNTIME_RECEIVER_BLACKLIST`
1149
- * (R4 invariant), so `DoCmd.OpenForm` is intentionally SKIPPED by the
1150
- * generic `CALL_RE` path that would otherwise emit a junk `calls` edge
1151
- * to a synthetic `function` node for `DoCmd.OpenForm`. This regex
1152
- * matches BEFORE the call-site scan and uses its own dispatch to emit
1153
- * the `opens-form` edge instead — sharing no logic with CALL_RE.
1154
- */
1155
- static OPEN_FORM_ARG_RE = /\bDoCmd\.OpenForm\s+("(?:(?:[^"]|"")*)"|\p{L}[\p{L}\p{N}_]*)/gu;
1156
- /**
1157
- * Issue #48: `DoCmd.OpenReport "<ReportName>"` modelling regex — sibling
1158
- * of `OPEN_FORM_ARG_RE` (hueco 6 expanded). Same literal-or-bare-id argument
1159
- * capture (group 1) and same trailing positional-args drop. The dispatch
1160
- * table `DOCMD_OPEN_DISPATCH` (below) carries the per-method metadata so
1161
- * OpenForm and OpenReport share the same scan/emit pipeline while their
1162
- * edge kinds (`opens-form` vs `opens-report`), stub node kinds
1163
- * (`form-layout` vs `report-layout`), synthetic file-path prefixes, and
1164
- * qualifiedName prefixes (`Form_<Name>` vs `Report_<Name>`) stay
1165
- * distinct.
1166
- */
1167
- static OPEN_REPORT_ARG_RE = /\bDoCmd\.OpenReport\s+("(?:(?:[^"]|"")*)"|\p{L}[\p{L}\p{N}_]*)/gu;
1168
- /**
1169
- * Issue #48 dispatch table — shared literal-or-Const argument resolution
1170
- * for `DoCmd.OpenForm` and `DoCmd.OpenReport`. Each entry is everything
1171
- * `scanDoCmdOpenCalls` + `emitOpensStubEdge` need to share the pipeline
1172
- * between methods while keeping the per-method names distinct.
1173
- *
1174
- * OpenQuery is intentionally NOT in this dispatch — it emits an
1175
- * `UnresolvedReference` (not a stub + edge), resolution to the REAL
1176
- * `query` node emitted by `SqlQueryExtractor`. See
1177
- * `OPEN_QUERY_ARG_RE` + `scanDoCmdOpenQuery`.
1178
- *
1179
- * Why a separate dispatch: `DoCmd` is in `RUNTIME_RECEIVER_BLACKLIST`
1180
- * (R4 invariant), so ALL of these methods are intentionally SKIPPED by
1181
- * the generic `CALL_RE` path that would otherwise emit a junk `calls`
1182
- * edge to a synthetic `function` node for `DoCmd.OpenX`. This dispatch
1183
- * matches BEFORE the call-site scan and uses its own emission path.
1184
- */
1185
- static DOCMD_OPEN_DISPATCH = [
1186
- {
1187
- method: 'OpenForm',
1188
- re: VbaExtractor.OPEN_FORM_ARG_RE,
1189
- edgeKind: 'opens-form',
1190
- stubKind: 'form-layout',
1191
- syntheticPrefix: 'synthetic:opensFormStub',
1192
- syntheticExtension: '.form.txt',
1193
- moduleNamePrefix: 'Form_',
1194
- cacheKey: 'OpenForm',
1195
- metadataTargetKey: 'targetFormName',
1196
- synthesizedBy: 'vba-opens-form',
1197
- },
1198
- {
1199
- method: 'OpenReport',
1200
- re: VbaExtractor.OPEN_REPORT_ARG_RE,
1201
- edgeKind: 'opens-report',
1202
- stubKind: 'report-layout',
1203
- syntheticPrefix: 'synthetic:opensReportStub',
1204
- syntheticExtension: '.report.txt',
1205
- moduleNamePrefix: 'Report_',
1206
- cacheKey: 'OpenReport',
1207
- metadataTargetKey: 'targetReportName',
1208
- synthesizedBy: 'vba-opens-report',
1209
- },
1210
- ];
1211
- /**
1212
- * Issue #48: `DoCmd.OpenQuery "<QueryName>"` modelling regex. Emits an
1213
- * `UnresolvedReference` (NOT a stub + edge) so the resolver binds to the
1214
- * REAL `query` node `SqlQueryExtractor` emits for `queries/<Name>.sql`
1215
- * — the same shape as `vba-me-control` and `vba-forms-bang`. The query
1216
- * may not yet exist in the index when the .bas is parsed; the resolver
1217
- * does the binding when the .sql is later indexed.
1218
- *
1219
- * Argument shape: identical to OpenForm/OpenReport — literal `"..."` or
1220
- * bare identifier resolved against local `Const` declarations, falling
1221
- * back to the bare identifier when unknown.
1222
- */
1223
- static OPEN_QUERY_ARG_RE = /\bDoCmd\.OpenQuery\s+("(?:(?:[^"]|"")*)"|\p{L}[\p{L}\p{N}_]*)/gu;
1224
- /** SQL assigned to a local variable, e.g. `m_SQL = "SELECT ..." & ...`. */
1225
- static SQL_VAR_ASSIGN_RE = /^\s*(\p{L}[\p{L}\p{N}_]*)\s*=\s*(.*)$/iu;
1226
- /** SQL wrapper called with a variable, e.g. `getdb().Execute m_SQL`. */
1227
- static SQL_VAR_EXEC_RE = /\b(?:\p{L}[\p{L}\p{N}_]*)?db\b(?:\(\))?\.(?:OpenRecordset|Execute)\s*\(?\s*(\p{L}[\p{L}\p{N}_]*)\s*\)?/giu;
1228
- /**
1229
- * Issue #42: `DoCmd.RunSQL <identifier>` (variable form) — the dominant
1230
- * Access idiom for executing a dynamically-built SQL string. Today only
1231
- * the literal form `DoCmd.RunSQL "DELETE FROM X"` is tracked via the
1232
- * `SQL_WRAPPERS` regex at line 1108; the variable form silently dropped
1233
- * table impact for every procedure that builds SQL in a string and runs
1234
- * it through `DoCmd.RunSQL`.
1235
- *
1236
- * This regex is the DoCmd.RunSQL analogue of `SQL_VAR_EXEC_RE` above and
1237
- * is iterated by `scanSqlInLine` (lines 2051+). When a match is found,
1238
- * the captured identifier is resolved against `sqlVariables` (populated
1239
- * by `trackSqlVariableAssignment` with `&`-accumulate semantics — Issue
1240
- * #13) and the resulting SQL string drives `emitSqlTableReferences`.
1241
- *
1242
- * The optional `(?:\(\))?` + `\s*\(?` shape lets the regex match both
1243
- * the parenthesised form `DoCmd.RunSQL(strSQL)` and the no-paren form
1244
- * `DoCmd.RunSQL strSQL` that the existing SQL_WRAPPERS literal regex
1245
- * does not cover. The captured identifier is the only thing we need —
1246
- * we DO NOT try to parse what the variable points at; that's the
1247
- * existing `sqlVariables` map's job.
1248
- */
1249
- static SQL_VAR_DOCMD_RUNSQL_RE = /\bDoCmd\.RunSQL\s*\(?\s*(\p{L}[\p{L}\p{N}_]*)\s*\)?/giu;
1250
- /**
1251
- * SQL table-name regex scoped to the clauses that introduce a table
1252
- * reference: `FROM <t>`, `JOIN <t>`, `INTO <t>`, `UPDATE <t>`. Adding
1253
- * `JOIN` lets the scanner pick up tables from joined fragments that
1254
- * arrive via `&`-concatenated wrapper literals (e.g.
1255
- * `db.Execute "FROM A" & " JOIN B"`); without it the second literal's
1256
- * table was silently dropped even though the wrapper regex now matches
1257
- * the chain.
1258
- *
1259
- * The captured table name is an optional bracketed/unbracketed schema
1260
- * prefix followed by a `.`, then a bracketed-or-bare identifier — so
1261
- * `FROM dbo.tblCustomers` and `FROM [My Schema].[My Table]` come
1262
- * through as one composite reference. Without the prefix the regex
1263
- * still matches a single identifier byte-identical to the old shape.
1264
- * Brackets in the captured composite are stripped by
1265
- * `emitSqlTableReferences` (`replace(/[\[\]]/g, '')`), so the public
1266
- * node name is the unwrapped form `dbo.tblCustomers` /
1267
- * `My Schema.My Table` — matching how plain `[Order Details]` is also
1268
- * unwrapped to `Order Details`. The identifier class
1269
- * `\[[^\]]+\]|\p{L}[\p{L}\p{N}_]*` (same as the saved-queries
1270
- * `TABLE_RE` in `sql-query-extractor.ts`) ensures bracketed names
1271
- * with spaces — `[Order Details]`, `[My Schema]`, `[My Table]` —
1272
- * are captured whole.
1273
- */
1274
- static SQL_TABLE_RE = /\b(?:FROM|JOIN|INTO|UPDATE)\s+((?:(?:\[[^\]]+\]|\p{L}[\p{L}\p{N}_]*)\.)?(?:\[[^\]]+\]|\p{L}[\p{L}\p{N}_]*))/giu;
1275
- /**
1276
- * Issue #50: TempVars — Access's global key-value store for cross-form
1277
- * state. We model each STATIC-LITERAL key as a synthetic `class` placeholder
1278
- * (same NodeKind as SQL table refs, see `emitReference`) and emit one
1279
- * `references` edge per reading/writing procedure with
1280
- * `metadata.synthesizedBy: 'vba-tempvar'` and `metadata.access` ∈
1281
- * `{'read','write'}` (the user-facing enum; lowercase single-tokens).
1282
- *
1283
- * Three scanner surfaces cover the four real idioms:
1284
- * - `TEMP_VAR_BANG_RE` — `TempVars!clave` (no parens; write or read)
1285
- * - `TEMP_VAR_PAREN_RE` — `TempVars("clave")` (parens; write or read)
1286
- * - `TEMP_VAR_ADD_RE` — `TempVars.Add "clave", v` (always a write)
1287
- *
1288
- * Bang vs paren split by line-source: the bang form has no string literals
1289
- * in scope, so we scan the MASKED line (`maskStringContent` replaces
1290
- * `"…"` content with spaces — same line source the call-site / With
1291
- * event scanners consume). The paren and Add forms have their key INSIDE
1292
- * a `"…"` literal that gets blanked by the masker, so those regexes scan
1293
- * the ORIGINAL (unmasked) line — same split the SQL_TABLE_RE/OpenForm
1294
- * literal scanners already use.
1295
- *
1296
- * Dynamic-key forms — `TempVars(strNombre)`,
1297
- * `TempVars("clave" & suffix)` — are silently unmatched by all three
1298
- * regexes (none of them tolerate a function-call or `&` arg). REQ-CODE-4
1299
- * "unresolvable is silent" applies: these stay silent by design to
1300
- * prevent placeholder-node explosion.
1301
- */
1302
- static TEMP_VAR_BANG_RE = /\bTempVars!\s*(\p{L}[\p{L}\p{N}_]*)/gu;
1303
- /**
1304
- * Issue #50 (cont.):
1305
- * `TempVars("clave")` / `TempVars( "clave" )` capture. Scanned
1306
- * over the original (unmasked) line — the literal lives INSIDE a
1307
- * string and would be stripped by `maskStringContent`.
1308
- */
1309
- static TEMP_VAR_PAREN_RE = /\bTempVars\s*\(\s*"([^"]+)"\s*\)/gu;
1310
- /**
1311
- * Issue #50 (cont.):
1312
- * `TempVars.Add "clave", value` capture. Always a write (the
1313
- * `.Add` method inserts/updates the entry). Scanned over the
1314
- * original (unmasked) line for the same string-literal reason as
1315
- * `TEMP_VAR_PAREN_RE`.
1316
- */
1317
- static TEMP_VAR_ADD_RE = /\bTempVars\.Add\s+"([^"]+)"\s*,/gi;
1318
- /**
1319
- * `Me.<ControlName>` / `Me!<ControlName>` reference capture — hole 1
1320
- * of VBA control-modeling. Extended by Issue #44 to accept the bang
1321
- * form (the default-collection shortcut Access VBA inherits from VB).
1322
- *
1323
- * Real VBA idioms:
1324
- * `Me.lblTitulo.Caption = "Hello"` ← dot form, property assignment
1325
- * `Me.txtDescripcion.Value = "World"` ← dot form, property assignment
1326
- * `Me.ComandoGrabar.Enabled = True` ← dot form, property assignment
1327
- * `Me!txtNombre = "Hello"` ← BANG form (default collection)
1328
- * `Me!txtEstado.Value = 1` ← BANG form, then property
1329
- * `If Nz(Me.MotivoBorrado, "") = "" Then` ← read in expression
1330
- *
1331
- * The existing call-site scanner (CALL_RE) only fires on `Name(`
1332
- * (paren form) and `Me` is in its keyword blacklist anyway, so
1333
- * `Me.<Control>` references are silently invisible. This regex matches
1334
- * the FIRST identifier after `Me.` or `Me!` regardless of what follows
1335
- * (a property, an index, an assignment, a call argument, etc.) so the
1336
- * form → control binding is surfaced as an UnresolvedReference for the
1337
- * resolver to pick up later. Subsequent segments (`.Caption`, `.Value`,
1338
- * `.Enabled`) are intentionally NOT captured — they are properties of
1339
- * the control, not new symbols. The bang (`!`) is the default-collection
1340
- * shortcut: `Me!txtFoo` is semantically identical to `Me.txtFoo` and
1341
- * produces a byte-identical UnresolvedReference (same `referenceName`,
1342
- * `referenceKind`, and `metadata.synthesizedBy`) — the regression test
1343
- * in `__tests__/extraction-vba.test.ts` pins this parity.
1344
- *
1345
- * Provenance: `metadata.synthesizedBy = 'vba-me-control'`. Mirrors the
1346
- * `vba-form-binding` (form→sibling-`.cls`) and `vba-name-resolution`
1347
- * (qualified Dim) patterns already documented on `UnresolvedReference.metadata`
1348
- * in `src/types.ts`.
1349
- */
1350
- static ME_CONTROL_RE = /\bMe[.!](\p{L}[\p{L}\p{N}_]*)/gu;
1351
- /**
1352
- * Issue #44: `Forms!<FormName>` / `Forms("<FormName>")!<Ctl>` cross-form
1353
- * reference capture — companion to `ME_CONTROL_RE` above. Access VBA's
1354
- * default-collection shortcut for cross-form control access, captured
1355
- * BEFORE the generic call-site scan and emitted as its own
1356
- * `UnresolvedReference` family.
1357
- *
1358
- * Real VBA idioms:
1359
- * `Forms!FormX!txtY.Value = 1` — bang form, with control segment
1360
- * `Forms!FormX.Recordsource = "..."` — bang form, trailing property access
1361
- * `Set f = Forms!FormX` — bang form, no control segment
1362
- * `Forms("FormX")!txtY.Value = 1` — paren form, with control
1363
- * `Forms![Mi Formulario]!txtY` — bracketed form name (#54 mirror)
1364
- *
1365
- * Why a dedicated scanner (mirroring `OPEN_FORM_ARG_RE` / B4 / DoCmd.OpenForm):
1366
- * `Forms` is in `RUNTIME_RECEIVER_BLACKLIST`, so the generic CALL_RE
1367
- * path silently skips both `Forms!X` and `Forms("X")!Y`. Without this
1368
- * scanner, cross-form UI traffic from `Forms!FormX!txtY.Value` would
1369
- * never surface the form → control binding that the resolver needs to
1370
- * glue form `form-layout` nodes to control `form-instance-control`
1371
- * nodes. The dedicated dispatch fires BEFORE the call-site scan and
1372
- * uses its own emission path; the runtime-blacklist constraint is
1373
- * preserved (we intentionally do not rewrite CALL_RE).
1374
- *
1375
- * What this scanner emits per match:
1376
- * - ONE `UnresolvedReference` whose `referenceName` is the form's
1377
- * identifier (stripped of any surrounding quote or bracket
1378
- * decoration), tagged `metadata.synthesizedBy = 'vba-forms-bang'`
1379
- * and `referenceKind = 'references'`. The control segment (if
1380
- * present) is consumed by the regex so `Forms!FormX!txtY.Value` is
1381
- * captured as a single match, but the control name is NOT emitted
1382
- * as its own reference — control emission is the form's
1383
- * responsibility downstream.
1384
- * - NO synthetic `function` node (W4 graph-pollution invariant — the
1385
- * form is a real `.cls` / `.form.txt` pair the resolver already
1386
- * picks up via the `vba-form-binding` path). Without this guard the
1387
- * bang form would synthesize one stub per cross-form reference
1388
- * site, identical to what `bumpShapes` from #43 audited out.
1389
- *
1390
- * What this scanner does NOT match (intentional, pinned by tests):
1391
- * - `Forms!FormX.Foo` — the trailing `.Foo` IS a property
1392
- * access on the form (e.g. `Recordsource`), NOT a control access.
1393
- * The bang alternative carries a `(?![.\w])` negative-lookahead
1394
- * after the form identifier to drop this shape. The "W4
1395
- * no-synthetic-fn" test pins both halves of this contract.
1396
- * - `rs!Campo` — recordset field access (DAO/ADO
1397
- * default-member field read) is explicitly out of scope for the
1398
- * current change. The runtime-receiver blacklist plus the absence
1399
- * of any `Forms`/`Me` prefix means the existing scanners already
1400
- * skip it cleanly; the "STRETCH SCOPE" test pins that behaviour
1401
- * so a future change to bring recordset bangs in is reviewed
1402
- * explicitly against the bang-form scope decision.
1403
- *
1404
- * Operates on the ORIGINAL (unmasked) line — the paren form
1405
- * `Forms("FormX")!txtY` has the form name INSIDE a string literal, so
1406
- * masking string content would destroy the form identifier. Same
1407
- * unmasked-line constraint as `scanOpenFormCalls`.
1408
- */
1409
- static FORMS_BANG_RE = /\b(?:Forms!(\p{L}[\p{L}\p{N}_]*|\[[^\]]+\])(?![.\w])|Forms\(\s*(?:"((?:[^"]|"")*)"|(\p{L}[\p{L}\p{N}_]*|\[[^\]]+\]))\s*\)\s*!\s*(?:\p{L}[\p{L}\p{N}_]*|\[[^\]]+\]))/gu;
1410
- /**
1411
- * Issue #46: scan `Set <var> = New <Type>[.<Inner>]` lines — the dominant
1412
- * VBA late-instantiation idiom. Run inside `sweepCallsAndSql`'s proc-stack
1413
- * loop so the surrounding procedure is known. For each match:
1414
- * - register `<var>` in `localVarTypeMap` with `outer=<Type>`,
1415
- * `qualified=<hasInner>`, `assignedWithSet=true` so the PR #61 refined
1416
- * gate lets subsequent `<var>.Member ...` qualified calls resolve via
1417
- * the resolved class name;
1418
- * - emit a `references` edge from the module/class node to a synthetic
1419
- * node named `<Type>`, tagged `synthesizedBy: 'vba-set-new'`.
1420
- *
1421
- * Groups: (1) variable name, (2) outer type, (3) optional inner type.
1422
- * Operates on the MASKED line (string-literal content already replaced
1423
- * with spaces) so `Set x = New Foo` inside a string literal never matches.
1424
- */
1425
- static SET_NEW_RE = /\bSet\s+(\p{L}[\p{L}\p{N}_]*)\s*=\s*New\s+(\p{L}[\p{L}\p{N}_]*)(?:\.(\p{L}[\p{L}\p{N}_]*))?/iu;
1426
- /** Issue #43: track the receiver for `With <expr>` / `End With` blocks. */
1427
- static WITH_START_RE = /^\s*With\b\s+(.+?)\s*$/iu;
1428
- static WITH_END_RE = /^\s*End\s+With\b/iu;
1429
- /** Keywords we never want to match as call receivers. */
1430
- static CALL_KEYWORD_BLACKLIST = new Set([
1431
- 'If',
1432
- 'For',
1433
- 'While',
1434
- 'Do',
1435
- 'Select',
1436
- 'Case',
1437
- 'Then',
1438
- 'Else',
1439
- 'ElseIf',
1440
- 'With',
1441
- 'Loop',
1442
- 'Wend',
1443
- 'End',
1444
- 'Return',
1445
- 'Dim',
1446
- 'Set',
1447
- 'Let',
1448
- 'Const',
1449
- 'ReDim',
1450
- 'Static',
1451
- 'Public',
1452
- 'Private',
1453
- 'Friend',
1454
- 'Sub',
1455
- 'Function',
1456
- 'Property',
1457
- 'Class',
1458
- 'Module',
1459
- 'Option',
1460
- 'On',
1461
- 'Error',
1462
- 'Resume',
1463
- 'Exit',
1464
- 'New',
1465
- 'Call',
1466
- 'Rem',
1467
- 'LBound',
1468
- 'UBound',
1469
- 'Me',
1470
- 'Nothing',
1471
- ]);
1472
- /**
1473
- * Access runtime objects and singletons. Calls on these receivers are
1474
- * real VBA calls but the targets are NOT user-defined modules or
1475
- * classes — they're Access/DAO/ADO runtime types. Synthesizing a
1476
- * `function` node for each would pollute the graph with ~20+ junk
1477
- * nodes per real-world file (audit W4, June 2026). Skip synthesis
1478
- * for any receiver or member in this set.
1479
- *
1480
- * Note: `DoCmd.RunSQL`, `DoCmd.OpenForm`, etc. still get SQL/edge
1481
- * tracking via the dedicated `SQL_WRAPPERS` regex path (REQ-CODE-8),
1482
- * which fires BEFORE this scan and uses its own dispatch — so
1483
- * blacklisting DoCmd here doesn't lose the SQL-flow edges.
1484
- */
1485
- static RUNTIME_RECEIVER_BLACKLIST = new Set([
1486
- // Form / page references
1487
- 'Screen',
1488
- // Access application singletons
1489
- 'Application',
1490
- 'DoCmd',
1491
- 'SysCmd',
1492
- // VBA debugging intrinsic — Debug.Print / Debug.Assert
1493
- 'Debug',
1494
- // Access object collections
1495
- 'Forms',
1496
- 'Reports',
1497
- 'Modules',
1498
- 'References',
1499
- 'CommandBars',
1500
- // Error-handling intrinsic
1501
- 'Err',
1502
- // Late-binding factories (return IDispatch — not user code)
1503
- 'CreateObject',
1504
- 'GetObject',
1505
- // DAO/ADO recordset field collection access (e.g. rcdDatos.Fields("ID"))
1506
- 'Fields',
1507
- ]);
1508
- /**
1509
- * Fix 2 (Issue #2): replace string-literal content with spaces so that
1510
- * call-site patterns inside `"..."` spans are invisible to CALL_RE and
1511
- * the statement-form detectors. Column positions are preserved (each
1512
- * character is replaced 1-for-1) so any col-based metadata stays correct.
1513
- */
1514
- static maskStringContent(line) {
1515
- let result = '';
1516
- let i = 0;
1517
- while (i < line.length) {
1518
- const ch = line[i];
1519
- if (ch === '"') {
1520
- result += ' '; // opening quote masked
1521
- i++;
1522
- while (i < line.length) {
1523
- const c = line[i];
1524
- if (c === '"' && line[i + 1] === '"') {
1525
- result += ' '; // doubled-quote escape masked (2 chars)
1526
- i += 2;
1527
- }
1528
- else if (c === '"') {
1529
- result += ' '; // closing quote masked
1530
- i++;
1531
- break;
1532
- }
1533
- else {
1534
- result += ' '; // string content → space
1535
- i++;
1536
- }
1537
- }
1538
- }
1539
- else {
1540
- result += ch;
1541
- i++;
1542
- }
1543
- }
1544
- return result;
1545
- }
1546
- /**
1547
- * Fix 2 (Issue #2): return true iff `receiverName` is a file-local variable
1548
- * (appears in `localVarTypeMap`) whose declared type is a SIMPLE (non-
1549
- * qualified, non-primitive) identifier — a candidate project-defined class.
1550
- * Qualified types (e.g. `DAO.Recordset`) and primitives (`String`, `Long`)
1551
- * return false so runtime/DAO calls are suppressed.
1552
- *
1553
- * Issue #54 (defensive): brackets are stripped from the lookup key, so a
1554
- * caller that forgets to unwrap a bracketed name still finds the
1555
- * corresponding entry. This is a no-op when the name is already bare —
1556
- * the unwrap pattern only matches an opening `[` at the start and a
1557
- * closing `]` at the end. Today's call sites (`scanCallSites`,
1558
- * `detectQualifiedStatementCall`) already unwrap in the regex captures,
1559
- * so this defensive strip is a belt-and-braces guard for any future
1560
- * caller that forgets.
1561
- */
1562
- isLocalProjectClassVar(receiverName) {
1563
- // Issue #54: strip a single leading `[` and/or trailing `]` if present.
1564
- const key = receiverName.replace(/^\[|\]$/g, '').toLowerCase();
1565
- const entry = this.localVarTypeMap.get(key);
1566
- if (!entry)
1567
- return false; // not declared in this file → silent
1568
- if (entry.qualified)
1569
- return false; // DAO.Recordset etc. → silent
1570
- if (VbaExtractor.PRIMITIVE_TYPES.has(entry.outer.toLowerCase()))
1571
- return false;
1572
- return true;
1573
- }
1574
- /**
1575
- * Qualified-call eligibility is intentionally shared by paren-form
1576
- * (`Receiver.Member(...)`) and statement-form (`Receiver.Member args`) scans:
1577
- * project-class locals are processed after type resolution, declared
1578
- * primitive/external locals are silent, and undeclared receivers remain
1579
- * candidate module names.
1580
- */
1581
- shouldProcessQualifiedCall(receiverName) {
1582
- if (this.isLocalProjectClassVar(receiverName))
1583
- return true;
1584
- return !this.localVarTypeMap.has(receiverName.toLowerCase());
1585
- }
1586
- /**
1587
- * #12a: resolve the "receiver type" used to build a qualified call-stub's
1588
- * name/qualifiedName. When `receiverName` is a file-local variable typed
1589
- * as a candidate project class (`isLocalProjectClassVar`), returns the
1590
- * RESOLVED CLASS NAME from `localVarTypeMap` (e.g. `m_NCOp` typed
1591
- * `As NCOperaciones` → `'NCOperaciones'`) so the stub's qualifiedName
1592
- * matches the real `.cls` method's `${className}.${proc}` shape and the
1593
- * post-extraction resolver (#12b) can find it via an exact qualifiedName
1594
- * match. Otherwise returns `receiverName` unchanged — this is the case
1595
- * for `.bas`-qualified module calls (`modUtils.Foo`), where the receiver
1596
- * IS already the target module's name and no resolution is needed.
1597
- */
1598
- resolveReceiverType(receiverName) {
1599
- if (this.isLocalProjectClassVar(receiverName)) {
1600
- const key = receiverName.replace(/^\[|\]$/g, '').toLowerCase();
1601
- const entry = this.localVarTypeMap.get(key);
1602
- if (entry)
1603
- return entry.outer;
1604
- }
1605
- return receiverName;
1606
- }
1607
- normalizeWithReceiver(expr) {
1608
- let receiver = expr.trim();
1609
- if (!receiver)
1610
- return null;
1611
- if (/^Call\s/i.test(receiver))
1612
- receiver = receiver.replace(/^Call\s+/i, '').trimStart();
1613
- if (receiver.startsWith('[')) {
1614
- const m = /^\[([^\]]+)\]/u.exec(receiver);
1615
- if (!m)
1616
- return null;
1617
- receiver = m[1] ?? '';
1618
- }
1619
- else {
1620
- const m = /^(\p{L}[\p{L}\p{N}_]*)/u.exec(receiver);
1621
- if (!m)
1622
- return null;
1623
- receiver = m[1] ?? '';
1624
- }
1625
- if (!receiver)
1626
- return null;
1627
- if (VbaExtractor.CALL_KEYWORD_BLACKLIST.has(receiver))
1628
- return null;
1629
- if (VbaExtractor.RUNTIME_RECEIVER_BLACKLIST.has(receiver))
1630
- return null;
1631
- return receiver;
1632
- }
1633
- detectWithMemberCall(line) {
1634
- let trimmed = line.trimStart();
1635
- if (!trimmed)
1636
- return null;
1637
- if (/^Call\s/i.test(trimmed))
1638
- trimmed = trimmed.replace(/^Call\s+/i, '').trimStart();
1639
- if (trimmed.startsWith("'") || /^Rem(\s|$)/i.test(trimmed))
1640
- return null;
1641
- if (!trimmed.startsWith('.'))
1642
- return null;
1643
- const memberRest = trimmed.slice(1);
1644
- const memberM = /^(\p{L}[\p{L}\p{N}_]*)/u.exec(memberRest);
1645
- if (!memberM)
1646
- return null;
1647
- const member = memberM[1] ?? '';
1648
- const afterMember = memberRest.slice(member.length);
1649
- if (afterMember.length > 0) {
1650
- const ch = afterMember.charAt(0);
1651
- if (ch !== '(' && ch !== ' ' && ch !== '\t')
1652
- return null;
1653
- const argsText = afterMember.trimStart();
1654
- if (argsText.startsWith('='))
1655
- return null;
1656
- }
1657
- if (VbaExtractor.CALL_KEYWORD_BLACKLIST.has(member))
1658
- return null;
1659
- if (VbaExtractor.RUNTIME_RECEIVER_BLACKLIST.has(member))
1660
- return null;
1661
- return { member };
1662
- }
1663
- sweepCallsAndSql(src) {
1664
- const lines = src.split('\n');
1665
- const procedureStartLines = new Set();
1666
- const sqlTargetsThisFile = new Set();
1667
- // Issue #52: reset the shared scope state before the per-line walk
1668
- // begins. `sweepEnumsAndConsts` already populated `procStack` /
1669
- // `currentProcKey` during its own walk; clearing here guarantees the
1670
- // `scanDoCmdOpenCalls` / `scanDoCmdOpenQuery` reads (which consult
1671
- // `currentProcKey` per call-site) start in module scope and follow
1672
- // the same push/pop discipline as the existing `stack` array below.
1673
- this.procStack.length = 0;
1674
- this.currentProcKey = 'module';
1675
- // Walk the source once, emitting call edges and SQL edges per line and
1676
- // tracking the current procedure stack. The previous implementation
1677
- // did this in two passes; audit S1 (June 2026) flagged the first pass
1678
- // as dead code (its `procStack` was never read after the loop). One
1679
- // pass suffices.
1680
- const stack = [];
1681
- const withReceiverStack = [];
1682
- const sqlVariables = new Map();
1683
- // C2 fix: track each procedure's `endLine` (the line containing the
1684
- // matching `End Sub`/`End Function`/`End Property`) keyed by its
1685
- // `startLine`. After the loop, we update every function node's
1686
- // `endLine` so `codegraph_explore` returns the full body span —
1687
- // not just the signature line.
1688
- const procEndLines = new Map();
1689
- for (let i = 0; i < lines.length; i++) {
1690
- const line = lines[i] ?? '';
1691
- const lineNum = i + 1;
1692
- const procStart = VbaExtractor.PROC_RE.exec(line);
1693
- if (procStart) {
1694
- // Issue #52: mirror the proc push into the shared
1695
- // `procStack` + `currentProcKey` so Const reads in this same
1696
- // sweep see the same scope as the const writes did (during
1697
- // `sweepEnumsAndConsts`). Uses the same `startLine` key used
1698
- // for the `localConstants` bucket.
1699
- const procStartLine = lineNum;
1700
- this.procStack.push(procStartLine);
1701
- this.currentProcKey = String(procStartLine);
1702
- const name = procStart[3] ?? '';
1703
- const bucket = this.localProcs.get(name);
1704
- if (bucket) {
1705
- // Fix 1: select the ProcInfo whose startLine matches the current
1706
- // line, not always bucket[0]. When Property Get/Let/Set share a
1707
- // name, all three exist in the bucket; pushing bucket[0] every time
1708
- // meant Let/Set bodies erroneously attributed to the Get's ProcInfo
1709
- // (wrong endLine and wrong caller source on call edges).
1710
- const proc = bucket.find((p) => p.startLine === lineNum) ?? bucket[0];
1711
- if (proc)
1712
- stack.push(proc);
1713
- }
1714
- procedureStartLines.add(lineNum);
1715
- }
1716
- else if (VbaExtractor.PROCEDURE_END_RE.test(line) && stack.length > 0) {
1717
- const ending = stack.pop();
1718
- // Issue #52: mirror the pop into the shared scope state.
1719
- this.procStack.pop();
1720
- this.currentProcKey =
1721
- this.procStack.length > 0
1722
- ? String(this.procStack[this.procStack.length - 1])
1723
- : 'module';
1724
- procEndLines.set(ending.startLine, lineNum);
1725
- continue;
1726
- }
1727
- // Fix 2 (Issue #2): mask string-literal content before call scanning so
1728
- // patterns like `modHelper.BuildQuery(` inside a string argument are not
1729
- // mistakenly treated as call sites. SQL scanning still uses the original
1730
- // line because SQL lives INSIDE string literals.
1731
- const callScanLine = VbaExtractor.maskStringContent(line);
1732
- if (stack.length > 0 && VbaExtractor.WITH_END_RE.test(callScanLine)) {
1733
- withReceiverStack.pop();
1734
- continue;
1735
- }
1736
- if (stack.length > 0 && !procedureStartLines.has(lineNum)) {
1737
- const withStart = VbaExtractor.WITH_START_RE.exec(callScanLine);
1738
- if (withStart) {
1739
- const receiver = this.normalizeWithReceiver(withStart[1] ?? '');
1740
- if (receiver)
1741
- withReceiverStack.push(receiver);
1742
- continue;
1743
- }
1744
- }
1745
- // Don't scan call sites on the line that declares the procedure — it
1746
- // would match the proc name itself in `Sub Outer()`.
1747
- if (!procedureStartLines.has(lineNum) && stack.length > 0) {
1748
- const currentProc = stack[stack.length - 1];
1749
- this.scanRaiseEvents(callScanLine, currentProc, lineNum);
1750
- this.scanCallSites(callScanLine, currentProc, lineNum);
1751
- }
1752
- // Hueco 1: capture `Me.<Control>` references (property assignments,
1753
- // read expressions, method calls, anything after `Me.`). Only inside
1754
- // procedures because `Me` is only meaningful inside a form's class
1755
- // module. Skipped on the proc-declaration line for the same reason as
1756
- // the call-site scan above.
1757
- if (!procedureStartLines.has(lineNum) && stack.length > 0) {
1758
- this.scanMeControlReferences(callScanLine, stack[stack.length - 1], lineNum);
1759
- }
1760
- // SQL wrappers — only inside a procedure (don't pollute module scope
1761
- // with a stray string literal). Use the ORIGINAL line — SQL is inside
1762
- // string literals, so the masked line would strip the SQL content.
1763
- if (stack.length > 0) {
1764
- this.trackSqlVariableAssignment(lines, i, sqlVariables);
1765
- this.scanSqlInLine(line, lineNum, sqlTargetsThisFile, sqlVariables);
1766
- }
1767
- // H1 fix: detect statement-form Sub calls (no parens, no `Call`
1768
- // keyword). Audit H1 (June 2026): the parens-only CALL_RE made the
1769
- // dominant VBA idiom invisible — `EstablecerDatos m_Error` on the
1770
- // Form_FormNCAuditoriaMotivoEliminado.cls fixture contributed
1771
- // nothing to the call graph. Walked here so we share the proc stack
1772
- // already maintained by this loop.
1773
- //
1774
- // Issue #45: the statement-call detector used to inspect only the
1775
- // FIRST identifier of the line, so for `If x Then Foo` it returned
1776
- // the keyword `If` (which is then dropped by
1777
- // `emitStatementCallEdge`'s BLACKLIST check) — the actual `Foo` call
1778
- // after `Then` was silently invisible. Same gap existed on the
1779
- // `Else`/multi-statement (`:`) and qualified paths. The fix is to
1780
- // split the line into one or more statement clauses via
1781
- // `splitSingleLineIfClauses` (which handles `If … Then <body>`,
1782
- // `Else <body>`, and colon-separated multi-statements) and run the
1783
- // detectors per clause. Lines that don't match the single-line If
1784
- // shape pass through unchanged as `[<line>]` so block-form `If`
1785
- // (where the body lives on subsequent lines) keeps working through
1786
- // the existing per-line scan that picks up the body line.
1787
- if (stack.length > 0 && !procedureStartLines.has(lineNum)) {
1788
- // Issue #46: `Set x = New <Type>[.<Inner>]` late-instantiation.
1789
- // Run BEFORE the call-site scan so a later `<x>.Member ...` line
1790
- // finds `x` already registered in `localVarTypeMap` and the PR #61
1791
- // refined gate lets the qualified call resolve to `<Type>.Member`.
1792
- const setNew = VbaExtractor.SET_NEW_RE.exec(callScanLine);
1793
- if (setNew) {
1794
- const varName = setNew[1] ?? '';
1795
- const outerType = setNew[2] ?? '';
1796
- const innerType = setNew[3] ?? '';
1797
- if (varName && outerType) {
1798
- // Skip primitives defensively — `Set x = New Long` is nonsense
1799
- // in practice but the gate is cheap and consistent with the
1800
- // Dim sweep's PRIMITIVE_TYPES guard.
1801
- if (!VbaExtractor.PRIMITIVE_TYPES.has(outerType.toLowerCase())) {
1802
- this.localVarTypeMap.set(varName.toLowerCase(), {
1803
- outer: outerType,
1804
- // Mirror `Dim x As Foo.Bar`: qualified `Set rs = New
1805
- // DAO.Recordset` registers `qualified: true` so the PR #61
1806
- // gate keeps downstream `rs.Method` calls silent (DAO is
1807
- // a runtime / external library, not a project class).
1808
- qualified: !!innerType,
1809
- assignedWithSet: true,
1810
- variableName: varName,
1811
- });
1812
- this.emitReference(outerType, lineNum, 0, 'vba-set-new');
1813
- }
1814
- }
1815
- }
1816
- const clauseLines = this.splitSingleLineIfClauses(callScanLine);
1817
- for (const clauseLine of clauseLines) {
1818
- const stmtCall = this.detectStatementCall(clauseLine);
1819
- if (stmtCall) {
1820
- const caller = stack[stack.length - 1];
1821
- this.emitStatementCallEdge(caller, stmtCall, lineNum);
1822
- }
1823
- // Fix 7 + Fix 2 + Issue #40: qualified statement-form calls
1824
- // (`Receiver.Member args`) — the dominant cross-object call shape in
1825
- // real Dysflow fixtures. `CALL_RE` only matches the paren form; this
1826
- // path covers the no-paren statement form and emits a heuristic
1827
- // `calls` edge through the unified `shouldProcessQualifiedCall`
1828
- // gate (shared with the paren form): declared project-class locals
1829
- // process, declared primitive/external locals stay silent, and
1830
- // undeclared receivers remain module-name candidates for the
1831
- // post-extraction resolver.
1832
- const qualStmt = this.detectQualifiedStatementCall(clauseLine);
1833
- if (qualStmt) {
1834
- const caller = stack[stack.length - 1];
1835
- if (this.shouldProcessQualifiedCall(qualStmt.receiver)) {
1836
- this.emitQualifiedStatementCallEdge(caller, qualStmt.receiver, qualStmt.member, lineNum);
1837
- }
1838
- }
1839
- const withReceiver = withReceiverStack[withReceiverStack.length - 1];
1840
- if (withReceiver) {
1841
- const withCall = this.detectWithMemberCall(clauseLine);
1842
- if (withCall && this.isLocalProjectClassVar(withReceiver)) {
1843
- const caller = stack[stack.length - 1];
1844
- this.emitQualifiedStatementCallEdge(caller, withReceiver, withCall.member, lineNum);
1845
- }
1846
- }
1847
- }
1848
- // B4 (hueco 6): `DoCmd.OpenForm "FormName"` modelling.
1849
- //
1850
- // The literal form name lives INSIDE a string literal, so we
1851
- // scan the ORIGINAL (unmasked) line — `callScanLine` has its
1852
- // string content replaced with spaces. The receiver is the same
1853
- // proc-stack frame so we attribute the edge to the calling Sub,
1854
- // not the file-level module.
1855
- //
1856
- // Cross-file edge filter note: the synthesized stub node is
1857
- // emitted locally (same file's extraction result), so the
1858
- // per-file filter at `index.ts:insertedIds.has(source/target)`
1859
- // passes the edge naturally — no exemption to the filter is
1860
- // required. The real form-layout node, when VbaFormExtractor
1861
- // later processes the matching .form.txt file, gets a DIFFERENT
1862
- // node id (it uses the real .form.txt path); the stub and the
1863
- // real node coexist harmlessly — the edge from THIS file still
1864
- // references the stub. Future work can collapse the stub once
1865
- // the indexer learns to re-resolve cross-file edges to the real
1866
- // node id (the same pattern the cross-file incoming-edges
1867
- // snapshot already uses at `index.ts:getCrossFileIncomingEdges`).
1868
- const caller2 = stack[stack.length - 1];
1869
- // Issue #48: shared OpenForm/OpenReport dispatch via
1870
- // `scanDoCmdOpenCalls` (formerly `scanOpenFormCalls`, refactored to
1871
- // iterate the `DOCMD_OPEN_DISPATCH` table — OpenForm behavior is
1872
- // byte-identical to pre-#48). The OpenQuery scanner emits an
1873
- // `UnresolvedReference` and stays separate from the dispatch since
1874
- // its emission shape (no stub + edge) differs.
1875
- this.scanDoCmdOpenCalls(line, caller2, lineNum);
1876
- this.scanDoCmdOpenQuery(line, caller2, lineNum);
1877
- // Issue #44: cross-form bang references (`Forms!X` / `Forms("X")!Y`).
1878
- // Same line context as `scanDoCmdOpenCalls` — the form name lives in
1879
- // a string literal in the paren form, so we MUST scan the unmasked
1880
- // line. The scanner is independent of `scanCallSites` because
1881
- // `Forms` is in `RUNTIME_RECEIVER_BLACKLIST` and would otherwise be
1882
- // dropped by the generic path.
1883
- this.scanFormsBang(line, caller2, lineNum);
1884
- // Issue #50: cross-form TempVars key accesses.
1885
- // Sibling of `scanFormsBang` — TempVars is Access's second global
1886
- // cross-form state surface (alongside Forms/DoCmd.OpenForm). Each
1887
- // static-literal key reference emits one `references` edge to a
1888
- // synthetic `class` placeholder, tagged `synthesizedBy: 'vba-tempvar'`
1889
- // and `access: 'read' | 'write'`. We need BOTH line sources: bang
1890
- // form scans the masked line, paren + Add forms scan the original
1891
- // (literal lives inside `"…"`).
1892
- this.sweepTempVars(callScanLine, line, lineNum, caller2);
1893
- }
1894
- }
1895
- // Apply endLine to every emitted function node keyed by its startLine.
1896
- // Functions without a recorded endLine (e.g. malformed VBA without an
1897
- // `End`) keep their `endLine = startLine` from sweepProcedures —
1898
- // which is the correct "single line" representation.
1899
- for (const n of this.nodes) {
1900
- if (n.kind !== 'function')
1901
- continue;
1902
- const end = procEndLines.get(n.startLine);
1903
- if (end !== undefined)
1904
- n.endLine = end;
1905
- }
1906
- }
1907
- static RAISE_EVENT_RE = /\bRaiseEvent\s+(\p{L}[\p{L}\p{N}_]*)\b/giu;
1908
- scanRaiseEvents(line, from, lineNum) {
1909
- VbaExtractor.RAISE_EVENT_RE.lastIndex = 0;
1910
- let m;
1911
- while ((m = VbaExtractor.RAISE_EVENT_RE.exec(line)) !== null) {
1912
- const eventName = m[1] ?? '';
1913
- const eventNode = this.localEvents.get(eventName.toLowerCase());
1914
- if (!eventNode)
1915
- continue;
1916
- this.edges.push({
1917
- source: this.findOrCreateFunctionNodeId(from),
1918
- target: eventNode.id,
1919
- kind: 'raises-event',
1920
- provenance: 'parser',
1921
- metadata: { eventName },
1922
- line: lineNum,
1923
- column: m.index,
1924
- });
1925
- }
1926
- }
1927
- scanCallSites(line, from, lineNum) {
1928
- VbaExtractor.CALL_RE.lastIndex = 0;
1929
- let m;
1930
- while ((m = VbaExtractor.CALL_RE.exec(line)) !== null) {
1931
- // Issue #54: CALL_RE groups (1)/(2) are alternative captures for the
1932
- // receiver position, (3)/(4) for the member position. The bracketed
1933
- // alternative wins when present; the captured value is already
1934
- // unwrapped by the regex (it captures only the inner content, not
1935
- // the surrounding `[...]`). Result: a `[FUNCIONES UTILES]` receiver
1936
- // surfaces here as `FUNCIONES UTILES` so the downstream blacklist
1937
- // checks and `resolveReceiverType` lookup see the bare identifier.
1938
- const receiver = m[1] ?? m[2] ?? '';
1939
- const member = m[3] ?? m[4] ?? '';
1940
- if (!receiver)
1941
- continue;
1942
- // Skip VBA control-flow keywords.
1943
- if (VbaExtractor.CALL_KEYWORD_BLACKLIST.has(receiver))
1944
- continue;
1945
- if (member && VbaExtractor.CALL_KEYWORD_BLACKLIST.has(member))
1946
- continue;
1947
- // Skip Access runtime objects — `Me`, `DoCmd`, `Application`, etc.
1948
- // These calls are real but the targets are NOT user code; emitting
1949
- // synthetic function nodes for them pollutes the graph (audit W4).
1950
- if (VbaExtractor.RUNTIME_RECEIVER_BLACKLIST.has(receiver))
1951
- continue;
1952
- if (member && VbaExtractor.RUNTIME_RECEIVER_BLACKLIST.has(member))
1953
- continue;
1954
- // Skip the receiver when it equals the containing procedure (self-call).
1955
- if (receiver === from.name && !member)
1956
- continue;
1957
- const col = m.index;
1958
- if (!member) {
1959
- // Bare `Name(...)` — same-file resolution.
1960
- const localFuncNode = this.findFunctionNodeByName(receiver);
1961
- if (!localFuncNode)
1962
- continue;
1963
- this.edges.push({
1964
- source: this.findOrCreateFunctionNodeId(from),
1965
- target: localFuncNode.id,
1966
- kind: 'calls',
1967
- line: lineNum,
1968
- column: col,
1969
- });
1970
- }
1971
- else {
1972
- // Qualified `Receiver.Member(...)` — synthesize the call target only
1973
- // for project-class local variables or undeclared module candidates.
1974
- if (!this.shouldProcessQualifiedCall(receiver))
1975
- continue;
1976
- // #12a: `receiverType` resolves to the real class name when
1977
- // `receiver` is a declared project-class local var (matching a real
1978
- // `.cls` method's `${className}.${proc}` qualifiedName shape so the
1979
- // #12b resolver can find it by exact match); otherwise it's the raw
1980
- // `receiver` text unchanged (e.g. `.bas`-qualified module calls).
1981
- //
1982
- // Antigravity audit Task 3 (refined gate): if `receiver` is a
1983
- // file-local variable declared as a PRIMITIVE (Variant, Object,
1984
- // Empty, Null, LongPtr, LongLong, New, Long, String, ...), skip
1985
- // emission. The previous behaviour emitted a heuristic `calls`
1986
- // edge to a stub named `<receiver>.<member>` that no resolver
1987
- // could ever repoint (Variant can hold anything, including
1988
- // runtime singletons; the stub is dead-end graph pollution).
1989
- //
1990
- // This refined gate does NOT regress cross-module qualified
1991
- // calls like `modUtils.Foo(1)` because `modUtils` is never
1992
- // declared as a file-local variable and therefore is NOT in
1993
- // `localVarTypeMap` — `localVarTypeMap.has(receiver.toLowerCase())`
1994
- // returns false and the gate is skipped. The "stub emitted for
1995
- // undeclared receivers → resolver repoints if a real module
1996
- // exists" behaviour is preserved.
1997
- const recvEntry = this.localVarTypeMap.get(receiver.toLowerCase());
1998
- if (recvEntry && VbaExtractor.PRIMITIVE_TYPES.has(recvEntry.outer.toLowerCase())) {
1999
- continue;
2000
- }
2001
- const receiverType = this.resolveReceiverType(receiver);
2002
- const qualified = `${receiverType}.${member}`;
2003
- // Avoid emitting duplicate edges for the same call (within a line).
2004
- const dedupeKey = `${from.name}->${qualified}@${lineNum}`;
2005
- if (this.callDedupe.has(dedupeKey))
2006
- continue;
2007
- this.callDedupe.add(dedupeKey);
2008
- const synthId = (0, tree_sitter_helpers_1.generateNodeId)(this.filePath, 'function', qualified, lineNum);
2009
- // Only add the synthetic function node once per (file, qualified, line).
2010
- if (!this.synthFunctionNodeIds.has(synthId)) {
2011
- this.synthFunctionNodeIds.add(synthId);
2012
- this.nodes.push({
2013
- id: synthId,
2014
- kind: 'function',
2015
- name: qualified,
2016
- qualifiedName: qualified,
2017
- filePath: this.filePath,
2018
- language: 'vba',
2019
- startLine: lineNum,
2020
- endLine: lineNum,
2021
- startColumn: col,
2022
- endColumn: col + qualified.length,
2023
- visibility: 'public',
2024
- // #12a: tag the stub so the post-extraction resolver (#12b)
2025
- // can find and repoint it. Mirrors the DoCmd.OpenForm stub
2026
- // precedent (`emitOpensFormEdge`).
2027
- metadata: { stub: true },
2028
- updatedAt: Date.now(),
2029
- });
2030
- }
2031
- this.edges.push({
2032
- source: this.findOrCreateFunctionNodeId(from),
2033
- target: synthId,
2034
- kind: 'calls',
2035
- provenance: 'heuristic',
2036
- metadata: {
2037
- synthesizedBy: 'vba-name-resolution',
2038
- stub: true,
2039
- receiverType,
2040
- member,
2041
- },
2042
- line: lineNum,
2043
- column: col,
2044
- });
2045
- }
2046
- }
2047
- }
2048
- /** Cache so we don't re-emit the same proc function node per call site. */
2049
- procNodeIdCache = new Map();
2050
- callDedupe = new Set();
2051
- synthFunctionNodeIds = new Set();
2052
- /**
2053
- * B4 (hueco 6) extended by Issue #48: cache of stub node ids we've already
2054
- * emitted for a given (method, target name) pair in this file. Avoids
2055
- * emitting duplicate stubs when `DoCmd.OpenForm "FormTest"` or
2056
- * `DoCmd.OpenReport "InformeMensual"` shows up N times across N calls.
2057
- * Keyed by `${cacheKey}:${lowerName}` so the OpenForm and OpenReport
2058
- * de-dup buckets stay disjoint — `OpenForm:Form1` ≠ `OpenReport:Form1`.
2059
- * The name part is lowercased so `FormTest` / `formtest` collapse.
2060
- */
2061
- opensStubIdsByKey = new Map();
2062
- /**
2063
- * Hueco 1: scan a line for `Me.<ControlName>` / `Me!<ControlName>`
2064
- * patterns and emit one UnresolvedReference per occurrence, tagged
2065
- * `metadata.synthesizedBy: 'vba-me-control'`. Issue #44 extended
2066
- * `ME_CONTROL_RE` from `Me\.` to `Me[.!]` so the bang form (default-
2067
- * collection shortcut) is captured byte-identically to the dot form.
2068
- *
2069
- * Operates on the masked `callScanLine` (string-literal content already
2070
- * replaced with spaces) so `Me.X` inside a string literal is not falsely
2071
- * captured. Per-site emission — every `Me.lblTitulo` reference site
2072
- * produces its own UnresolvedReference carrying the line/column; the
2073
- * resolver fans them out into multiple `references` edges at index time
2074
- * once the matching `form-instance-control` node exists.
2075
- *
2076
- * `fromNodeId` is the current procedure's function node — that's the
2077
- * "owner" of the reference (the Sub body that wrote `Me.lblTitulo = …`).
2078
- * The +3 column offset remains correct under Issue #44's regex change
2079
- * because both `Me.` and `Me!` are 3-character prefixes.
2080
- */
2081
- scanMeControlReferences(line, from, lineNum) {
2082
- VbaExtractor.ME_CONTROL_RE.lastIndex = 0;
2083
- let m;
2084
- while ((m = VbaExtractor.ME_CONTROL_RE.exec(line)) !== null) {
2085
- const controlName = m[1] ?? '';
2086
- if (!controlName)
2087
- continue;
2088
- this.unresolvedReferences.push({
2089
- fromNodeId: this.findOrCreateFunctionNodeId(from),
2090
- referenceName: controlName,
2091
- referenceKind: 'references',
2092
- line: lineNum,
2093
- column: m.index + 3, // +3 to skip the `Me.` / `Me!` prefix
2094
- filePath: this.filePath,
2095
- language: 'vba',
2096
- metadata: { synthesizedBy: 'vba-me-control' },
2097
- });
2098
- }
2099
- }
2100
- /**
2101
- * Issue #44: scan a line for cross-form bang references
2102
- * (`Forms!<FormName>[!<Ctl>]` and `Forms("<FormName>")!<Ctl>`) and
2103
- * emit ONE UnresolvedReference per match with `metadata.synthesizedBy
2104
- * = 'vba-forms-bang'`. Companion to `scanMeControlReferences` above;
2105
- * shares the emission shape (a single `UnresolvedReference` per match,
2106
- * `referenceKind: 'references'`), keeping the W4 invariant that we
2107
- * synthesize NO `function` node for forms.
2108
- *
2109
- * Operates on the ORIGINAL (unmasked) line — the paren form
2110
- * `Forms("FormX")!txtY` carries the form name INSIDE a string literal
2111
- * and would be destroyed by `maskStringContent`. Mirrors
2112
- * `scanOpenFormCalls`'s unmasked-line constraint.
2113
- *
2114
- * Regex alternatives (see `FORMS_BANG_RE`):
2115
- * 1. `Forms!<FormName>` (bare or `[bracketed]`, NOT followed by `.X`)
2116
- * — bang form, may have a trailing `!<Ctl>[.<Prop>]` that is
2117
- * consumed but NOT emitted.
2118
- * 2. `Forms("<FormName>")!<Ctl>` (quoted or bare/bracketed form arg)
2119
- * — paren form, ALWAYS with a control segment.
2120
- *
2121
- * Stripping: the form's identifier is unwrapped of `"` quotes (string
2122
- * literals) and `[…]` brackets (Issue #54 reserved-identifier shape) so
2123
- * the public `referenceName` is the bare form name. Bracketed forms
2124
- * (`Forms![Mi Formulario]`) follow the same strip rules as the paren
2125
- * form so the resolver sees `Mi Formulario` either way.
2126
- */
2127
- scanFormsBang(line, from, lineNum) {
2128
- VbaExtractor.FORMS_BANG_RE.lastIndex = 0;
2129
- let m;
2130
- while ((m = VbaExtractor.FORMS_BANG_RE.exec(line)) !== null) {
2131
- // Group 1: bang form name (bare or `[bracketed]`); group 2: paren
2132
- // form name in `"quotes"`; group 3: paren form name bare or
2133
- // `[bracketed]`. Take whichever the regex alternative produced and
2134
- // strip the surrounding decoration so the public referenceName is
2135
- // the bare form identifier the resolver will compare against the
2136
- // form's `form-layout` node name.
2137
- const raw = m[1] ?? m[2] ?? m[3] ?? '';
2138
- if (!raw)
2139
- continue;
2140
- const formName = raw
2141
- .replace(/^"|"$/g, '') // strip surrounding string-literal quotes
2142
- .replace(/^\[|\]$/g, ''); // strip surrounding bracket decoration (#54)
2143
- if (!formName)
2144
- continue;
2145
- this.unresolvedReferences.push({
2146
- fromNodeId: this.findOrCreateFunctionNodeId(from),
2147
- referenceName: formName,
2148
- referenceKind: 'references',
2149
- line: lineNum,
2150
- column: m.index, // start of the `Forms` keyword; the form name's column can be reconstructed by the UI if it cares
2151
- filePath: this.filePath,
2152
- language: 'vba',
2153
- metadata: { synthesizedBy: 'vba-forms-bang' },
2154
- });
2155
- }
2156
- }
2157
- findOrCreateFunctionNodeId(proc) {
2158
- // Fix 1: key the cache by `name:startLine` so Property Get/Let/Set
2159
- // accessors with the same name each resolve to their own node.
2160
- const cacheKey = `${proc.name}:${proc.startLine}`;
2161
- const cached = this.procNodeIdCache.get(cacheKey);
2162
- if (cached)
2163
- return cached;
2164
- // Prefer the startLine-indexed lookup (O(1), correct for all three
2165
- // Property accessors). Fall back to the name-indexed cache for procs
2166
- // whose startLine wasn't recorded (should not happen in practice).
2167
- const fn = this.functionNodeByStartLine.get(proc.startLine) ??
2168
- this.findFunctionNodeByName(proc.name);
2169
- if (fn) {
2170
- this.procNodeIdCache.set(cacheKey, fn.id);
2171
- return fn.id;
2172
- }
2173
- // Fallback: synthesize a node id matching generateNodeId's input shape.
2174
- const id = (0, tree_sitter_helpers_1.generateNodeId)(this.filePath, 'function', proc.name, proc.startLine);
2175
- this.procNodeIdCache.set(cacheKey, id);
2176
- return id;
2177
- }
2178
- findFunctionNodeByName(name) {
2179
- // O(1) via the cache populated as function nodes are emitted in
2180
- // `sweepProcedures`. Audit S2 (June 2026): the previous
2181
- // `this.nodes.find(...)` was an O(n) linear scan per call site —
2182
- // meaningful on real .cls files with hundreds of procedures.
2183
- return this.functionNodeByName.get(name);
2184
- }
2185
- /**
2186
- * Issue #45: split a single-line VBA `If <cond> Then <body>` into one or
2187
- * more statement-clause fragments that the existing `detectStatementCall`
2188
- * and `detectQualifiedStatementCall` detectors can process. Handles:
2189
- *
2190
- * - `If x Then Foo` → `['Foo']`
2191
- * - `If x Then Foo Else Bar` → `['Foo', 'Bar']`
2192
- * - `If x Then DoA: DoB` → `['DoA', 'DoB']` (colon-separated multi-statement)
2193
- * - `If x Then Foo Else A: B` → `['Foo', 'A', 'B']`
2194
- * - `If x Then GoTo fin` → `[]` (GoTo clause filtered out)
2195
- * - `If x Then Exit Sub` → `[]` (Exit clause filtered out)
2196
- *
2197
- * When the line does NOT match a single-line `If … Then` shape — for
2198
- * instance a block-form `If x Then` whose body lives on subsequent
2199
- * lines — the splitter returns `[<line>]` (the original input) so
2200
- * callers can use this method unconditionally and let the existing
2201
- * per-line scan pick up the body on a separate line.
2202
- *
2203
- * `GoTo`, `Exit`, and `Resume` clauses are filtered at the fragment
2204
- * level (defense in depth): even though `emitStatementCallEdge` already
2205
- * drops these via the `CALL_KEYWORD_BLACKLIST`, filtering here prevents
2206
- * any chance of `detectStatementCall`'s generic identifier extractor
2207
- * matching a substring (e.g. an identifier like `GoToFinishingTouches`)
2208
- * as a side effect of a richer clause where the keyword happens to be
2209
- * the leading token.
2210
- *
2211
- * `line` is the string-literal-masked scan line. The mask makes global
2212
- * `:` splitting safe: real VBA colons never appear inside string
2213
- * literals (already masked to spaces) and never inside expressions
2214
- * inside parens at the source level (a colon ends a statement in VBA,
2215
- * so it cannot appear inside a parenthesised argument list either).
2216
- * The `Else` keyword is a VBA statement-level separator and is
2217
- * forbidden inside parens or expressions, so splitting on
2218
- * `\s+Else\s+` does not need paren tracking either.
2219
- */
2220
- splitSingleLineIfClauses(line) {
2221
- const trimmed = line.trimStart();
2222
- if (!trimmed)
2223
- return [];
2224
- // Match `If <cond> Then <body>` with a non-greedy condition. Requiring
2225
- // at least one whitespace character after `Then` ensures the block
2226
- // form `If x Then` (with the body on subsequent lines) is left alone
2227
- // for the existing per-line call-site scan to handle on the body line.
2228
- const ifThenRe = /^If\s[\s\S]+?\bThen\b\s+/i;
2229
- const m = ifThenRe.exec(trimmed);
2230
- if (!m) {
2231
- // Not a single-line `If … Then` — preserve the original line so the
2232
- // existing detection path picks it up unchanged.
2233
- return [line];
2234
- }
2235
- const body = trimmed.slice(m[0].length);
2236
- // Split on top-level `Else` (case-insensitive; word-bounded). The
2237
- // statement-level-only nature of `Else` in VBA means the regex split
2238
- // is safe without paren tracking on the masked-line invariant.
2239
- const elseClauses = body.split(/\s+Else\s+/i);
2240
- const clauses = [];
2241
- for (const elseClause of elseClauses) {
2242
- // Split each Else-clause on `:` for multi-statement single-line
2243
- // `If` bodies. VBA expressions never contain `:`, so a global
2244
- // split is correct on a masked line.
2245
- const subStatements = elseClause.split(':');
2246
- for (const sub of subStatements) {
2247
- const t = sub.trim();
2248
- if (!t)
2249
- continue;
2250
- // Defense in depth: GoTo / Exit / Resume are VBA control-flow
2251
- // statements, not Sub calls — drop them before they reach the
2252
- // statement-call detectors. (The existing
2253
- // `CALL_KEYWORD_BLACKLIST` check in `emitStatementCallEdge`
2254
- // also covers this; the fragment-level filter keeps the two
2255
- // intent statements aligned and protects against any future
2256
- // detector refactor that might relax the BLACKLIST check.)
2257
- if (/^(?:GoTo|Exit|Resume)\b/i.test(t))
2258
- continue;
2259
- clauses.push(t);
2260
- }
2261
- }
2262
- return clauses;
2263
- }
2264
- /**
2265
- * H1: detect a statement-form Sub call.
2266
- *
2267
- * Real VBA idioms:
2268
- * `MySub` — bare no-argument statement call
2269
- * `MySub arg1, Nz(x, 0)` — bare statement call
2270
- * `Call MySub arg1, Nz(x, 0)` — Call keyword, no parens around call
2271
- * `MySub(arg1, arg2)` — parens call (handled by CALL_RE)
2272
- * `Call MySub(arg1, arg2)` — Call keyword + parens (also CALL_RE)
2273
- *
2274
- * Returns the called proc name if the line matches the statement form
2275
- * AND the proc name is in localProcs (same-file resolution). Returns
2276
- * null for declarations, assignments, comments, keyword lines, etc.
2277
- *
2278
- * The `Call` keyword form is handled by stripping `Call ` and
2279
- * running the same logic on the remainder — they're structurally the
2280
- * same call after the keyword.
2281
- */
2282
- detectStatementCall(line) {
2283
- let trimmed = line.trimStart();
2284
- if (!trimmed)
2285
- return null;
2286
- // Strip `Call ` keyword if present — same call shape after.
2287
- if (/^Call\s/i.test(trimmed)) {
2288
- trimmed = trimmed.replace(/^Call\s+/i, '');
2289
- }
2290
- // Skip comment lines.
2291
- if (trimmed.startsWith("'") || trimmed.startsWith('Rem '))
2292
- return null;
2293
- // Skip declarations: Dim/Private/Public/Static/Global/Const/ReDim.
2294
- if (/^(Dim|Private|Public|Static|Global|Const|ReDim)\s/i.test(trimmed))
2295
- return null;
2296
- // Extract the leading identifier.
2297
- const m = /^(\p{L}[\p{L}\p{N}_]*)/u.exec(trimmed);
2298
- if (!m)
2299
- return null;
2300
- const procName = m[1] ?? '';
2301
- const rest = trimmed.slice(procName.length);
2302
- // `MySub(...)` is parens-form and already handled by CALL_RE. Parentheses
2303
- // are valid later in statement-form argument expressions (`MySub Nz(x, 0)`),
2304
- // so only the character immediately after the proc name distinguishes the
2305
- // call form.
2306
- if (rest.startsWith('('))
2307
- return null;
2308
- // Bare `MySub` is a valid no-argument statement-form Sub call.
2309
- if (rest.length === 0)
2310
- return procName;
2311
- const nextCh = trimmed.charAt(procName.length);
2312
- if (nextCh !== ' ' && nextCh !== '\t')
2313
- return null;
2314
- const args = rest.trimStart();
2315
- // Skip leading-identifier assignments (`X = ...`). Do not reject `=` inside
2316
- // argument expressions because named arguments use `:=` and comparisons can
2317
- // appear in expressions.
2318
- if (args.startsWith('='))
2319
- return null;
2320
- return procName;
2321
- }
2322
- /**
2323
- * H1: emit a same-file `calls` edge for a statement-form Sub call.
2324
- * `caller` is the current procedure (from the stack); `procName` is
2325
- * the called procedure name. Emits a plain `calls` edge to the
2326
- * already-emitted function node.
2327
- */
2328
- emitStatementCallEdge(caller, procName, lineNum) {
2329
- if (procName === caller.name)
2330
- return; // skip self-call
2331
- if (VbaExtractor.CALL_KEYWORD_BLACKLIST.has(procName))
2332
- return;
2333
- if (VbaExtractor.RUNTIME_RECEIVER_BLACKLIST.has(procName))
2334
- return;
2335
- const target = this.findFunctionNodeByName(procName);
2336
- if (!target)
2337
- return;
2338
- this.edges.push({
2339
- source: this.findOrCreateFunctionNodeId(caller),
2340
- target: target.id,
2341
- kind: 'calls',
2342
- line: lineNum,
2343
- column: 0,
2344
- });
2345
- }
2346
- /**
2347
- * Fix 7: detect a qualified statement-form call — `Receiver.Member <args>`
2348
- * where `Receiver.Member` is NOT followed by `(`.
2349
- *
2350
- * Real VBA idioms this covers:
2351
- * `m_NCOp.Registrar m_ARAlInicio, p_Error`
2352
- * `m_Obj.Init`
2353
- *
2354
- * These are distinct from the paren form (`m_NCOp.Registrar(...)`) which is
2355
- * already handled by `CALL_RE`. Returns `{receiver, member}` or `null`.
2356
- *
2357
- * Property assignments (`Receiver.Prop = value`) are excluded via the `=`
2358
- * guard — consistent with `detectStatementCall`'s same-file assignment skip.
2359
- * Blacklisted receivers (runtime objects, keywords) are excluded too.
2360
- */
2361
- detectQualifiedStatementCall(line) {
2362
- let trimmed = line.trimStart();
2363
- if (!trimmed)
2364
- return null;
2365
- // Strip `Call` keyword — same call shape after it.
2366
- if (/^Call\s/i.test(trimmed))
2367
- trimmed = trimmed.replace(/^Call\s+/i, '');
2368
- // Skip comment lines.
2369
- if (trimmed.startsWith("'") || /^Rem(\s|$)/i.test(trimmed))
2370
- return null;
2371
- // Skip declarations.
2372
- if (/^(Dim|Private|Public|Static|Global|Const|ReDim)\s/i.test(trimmed))
2373
- return null;
2374
- // Issue #54: the receiver alternative accepts BOTH the bare form
2375
- // (`Foo`) and the VBA bracketed form (`[Foo Bar]`). The bracketed
2376
- // alternative wins when present; the captured value is already
2377
- // unwrapped by the regex (it captures only the inner content, not
2378
- // the surrounding `[...]`). The same shape applies to the member.
2379
- const receiverM = /^(?:\[([^\]]+)\]|(\p{L}[\p{L}\p{N}_]*))/u.exec(trimmed);
2380
- if (!receiverM)
2381
- return null;
2382
- const receiver = receiverM[1] ?? receiverM[2] ?? '';
2383
- const rest = trimmed.slice(receiverM[0].length);
2384
- // Must have a dot separator.
2385
- if (!rest.startsWith('.'))
2386
- return null;
2387
- // Extract member identifier.
2388
- const memberRest = rest.slice(1); // skip the dot
2389
- const memberM = /^(?:\[([^\]]+)\]|(\p{L}[\p{L}\p{N}_]*))/u.exec(memberRest);
2390
- if (!memberM)
2391
- return null;
2392
- const member = memberM[1] ?? memberM[2] ?? '';
2393
- const afterMember = memberRest.slice(memberM[0].length);
2394
- // Must NOT be followed by `(` — the paren form is handled by CALL_RE.
2395
- if (afterMember.startsWith('('))
2396
- return null;
2397
- // Must be followed by space/tab (args present) OR end of line (no args).
2398
- if (afterMember.length > 0) {
2399
- const ch = afterMember.charAt(0);
2400
- if (ch !== ' ' && ch !== '\t')
2401
- return null;
2402
- // Skip property assignments: `Receiver.Prop = value`.
2403
- const argsText = afterMember.trimStart();
2404
- if (argsText.startsWith('='))
2405
- return null;
2406
- }
2407
- // Respect the keyword and runtime blacklists.
2408
- if (VbaExtractor.CALL_KEYWORD_BLACKLIST.has(receiver))
2409
- return null;
2410
- if (VbaExtractor.RUNTIME_RECEIVER_BLACKLIST.has(receiver))
2411
- return null;
2412
- if (VbaExtractor.CALL_KEYWORD_BLACKLIST.has(member))
2413
- return null;
2414
- if (VbaExtractor.RUNTIME_RECEIVER_BLACKLIST.has(member))
2415
- return null;
2416
- return { receiver, member };
2417
- }
2418
- /**
2419
- * Fix 7: emit a heuristic `calls` edge for a qualified statement-form call.
2420
- * Same shape as the qualified-paren path in `scanCallSites` — reuses the
2421
- * same `callDedupe` / `synthFunctionNodeIds` sets for de-duplication so a
2422
- * paren and non-paren form on the same line don't create duplicate edges.
2423
- */
2424
- emitQualifiedStatementCallEdge(caller, receiver, member, lineNum) {
2425
- // Qualified-call eligibility is checked before this method. Project-class
2426
- // local receivers resolve to their class name (for exact `.cls` method
2427
- // matching), while undeclared receivers stay as raw module-name candidates.
2428
- const receiverType = this.resolveReceiverType(receiver);
2429
- const qualified = `${receiverType}.${member}`;
2430
- const dedupeKey = `${caller.name}->${qualified}@${lineNum}`;
2431
- if (this.callDedupe.has(dedupeKey))
2432
- return;
2433
- this.callDedupe.add(dedupeKey);
2434
- const synthId = (0, tree_sitter_helpers_1.generateNodeId)(this.filePath, 'function', qualified, lineNum);
2435
- if (!this.synthFunctionNodeIds.has(synthId)) {
2436
- this.synthFunctionNodeIds.add(synthId);
2437
- this.nodes.push({
2438
- id: synthId,
2439
- kind: 'function',
2440
- name: qualified,
2441
- qualifiedName: qualified,
2442
- filePath: this.filePath,
2443
- language: 'vba',
2444
- startLine: lineNum,
2445
- endLine: lineNum,
2446
- startColumn: 0,
2447
- endColumn: qualified.length,
2448
- visibility: 'public',
2449
- metadata: { stub: true },
2450
- updatedAt: Date.now(),
2451
- });
2452
- }
2453
- this.edges.push({
2454
- source: this.findOrCreateFunctionNodeId(caller),
2455
- target: synthId,
2456
- kind: 'calls',
2457
- provenance: 'heuristic',
2458
- metadata: {
2459
- synthesizedBy: 'vba-name-resolution',
2460
- stub: true,
2461
- receiverType,
2462
- member,
2463
- },
2464
- line: lineNum,
2465
- column: 0,
2466
- });
2467
- }
2468
- /**
2469
- * B4 (hueco 6) extended by Issue #48: scan one line of VBA source for
2470
- * `DoCmd.OpenX "Target"` calls where X ∈ {Form, Report} (see the
2471
- * `DOCMD_OPEN_DISPATCH` table). For each match, emit:
2472
- * - a stub node (form-layout / report-layout) for the target, cached
2473
- * per-(method, name) so the same target referenced from N sites
2474
- * emits exactly ONE stub,
2475
- * - an `opens-form` / `opens-report` heuristic edge from the calling
2476
- * Sub to that stub.
2477
- *
2478
- * Both endpoints are pushed into `this.nodes` / `this.edges`, so the
2479
- * per-file edge filter at `index.ts:insertedIds.has(source) &&
2480
- * insertedIds.has(target)` passes the edge naturally without any
2481
- * exemption to the filter.
2482
- *
2483
- * Why a stub and not a direct lookup: the target form/report lives in a
2484
- * DIFFERENT file (its own `.form.txt` / `.report.txt`), and the extractor
2485
- * doesn't have DB access at parse time. The stub's synthetic file path
2486
- * (`synthetic:opensFormStub/<Name>.form.txt` /
2487
- * `synthetic:opensReportStub/<Name>.form.txt`) guarantees a deterministic
2488
- * node id so re-indexes collapse to the same stub. When the consumer's
2489
- * `.form.txt` / `.report.txt` is later indexed, the real
2490
- * `form-layout` / `report-layout` node carries a different id (it uses
2491
- * the real file path); the stub and the real coexist harmlessly.
2492
- *
2493
- * Why a separate dispatch from CALL_RE: `DoCmd` is in
2494
- * `RUNTIME_RECEIVER_BLACKLIST` (R4 invariant), so `DoCmd.OpenForm` /
2495
- * `DoCmd.OpenReport` are intentionally SKIPPED by the generic CALL_RE
2496
- * path that would otherwise emit a junk `calls` edge to a synthetic
2497
- * `function` node for `DoCmd.OpenX`. The dispatch below matches BEFORE
2498
- * the call-site scan and uses its own emission path.
2499
- *
2500
- * Scope note: literal-string and bare-identifier argument forms are
2501
- * supported. Bare identifiers resolve only through local `Const`
2502
- * declarations; arbitrary variable data-flow remains intentionally
2503
- * out of scope.
2504
- */
2505
- scanDoCmdOpenCalls(line, caller, lineNum) {
2506
- for (const dispatch of VbaExtractor.DOCMD_OPEN_DISPATCH) {
2507
- // Each regex has /g so we MUST reset `lastIndex` before use; cloning
2508
- // the regex is the simplest way to avoid leaking state across lines
2509
- // AND across dispatch iterations.
2510
- const localRe = new RegExp(dispatch.re.source, dispatch.re.flags);
2511
- let m;
2512
- while ((m = localRe.exec(line)) !== null) {
2513
- const rawArg = (m[1] ?? '').trim();
2514
- // Issue #52: const lookup is now per-proc-bucket with module
2515
- // fallback (see `resolveLocalConst`). Two procs declaring the
2516
- // same Const name with different values no longer collide —
2517
- // each call site uses the value visible at its own scope.
2518
- const targetName = rawArg.startsWith('"')
2519
- ? unwrapVbaStringLiteral(rawArg)
2520
- : (this.resolveLocalConst(rawArg) ?? rawArg);
2521
- if (!targetName)
2522
- continue;
2523
- this.emitOpensStubEdge(dispatch, caller, targetName, lineNum, m.index);
2524
- }
2525
- }
2526
- }
2527
- /**
2528
- * B4 (hueco 6) extended by Issue #48: emit a stub `form-layout` /
2529
- * `report-layout` node for `targetName` (cached per dispatch entry so
2530
- * duplicates collapse and OpenForm/OpenReport de-dup buckets stay
2531
- * disjoint) and a single `opens-form` / `opens-report` heuristic edge
2532
- * from `caller` to that stub.
2533
- *
2534
- * The edge carries:
2535
- * - `kind` — dispatch-specific (`opens-form` / `opens-report`)
2536
- * - `provenance: 'heuristic'` — synthesized, not parsed
2537
- * - `metadata.<dispatchTargetKey>` (e.g. `targetFormName`) — the resolved name
2538
- * - `metadata.synthesizedBy` — dispatch-specific (`vba-opens-form` /
2539
- * `vba-opens-report`); distinguishes this synthesis from the
2540
- * dim/sql/event-handler families
2541
- *
2542
- * The stub's `metadata.stub: true` flag lets downstream UI render
2543
- * stubs distinctly (e.g. with a dashed border) and gives later
2544
- * re-resolution pass a hook for collapse. The stub is
2545
- * line-independent (`line = 0`) so re-indexes produce identical ids.
2546
- */
2547
- emitOpensStubEdge(dispatch, caller, targetName, lineNum, column) {
2548
- const key = `${dispatch.cacheKey}:${targetName.toLowerCase()}`;
2549
- let stubId = this.opensStubIdsByKey.get(key);
2550
- if (!stubId) {
2551
- // Synthetic file path keeps the stub's id deterministic AND
2552
- // disambiguates it from any real `.form.txt` / `.report.txt`
2553
- // indexed later. The directory prefix (`synthetic:opensFormStub/`
2554
- // or `synthetic:opensReportStub/`) is intentionally not a real
2555
- // filesystem path — it just namespaces the id space. The file
2556
- // extension (`syntheticExtension`) DOES mirror the real form/report
2557
- // file extension so a reader of the synthetic path can tell the
2558
- // stub's intent at a glance.
2559
- const syntheticFilePath = `${dispatch.syntheticPrefix}/${targetName}${dispatch.syntheticExtension}`;
2560
- stubId = (0, tree_sitter_helpers_1.generateNodeId)(syntheticFilePath, dispatch.stubKind, targetName, 0);
2561
- this.opensStubIdsByKey.set(key, stubId);
2562
- this.nodes.push({
2563
- id: stubId,
2564
- kind: dispatch.stubKind,
2565
- name: targetName,
2566
- // Convention: form module names in Access are `Form_<Name>` and
2567
- // report module names are `Report_<Name>`. We follow the same
2568
- // convention in the synthetic stub's qualifiedName so cross-file
2569
- // lookups can find it consistently.
2570
- qualifiedName: `${dispatch.moduleNamePrefix}${targetName}`,
2571
- filePath: syntheticFilePath,
2572
- language: 'vba',
2573
- startLine: lineNum,
2574
- endLine: lineNum,
2575
- startColumn: 0,
2576
- endColumn: 0,
2577
- metadata: { stub: true },
2578
- updatedAt: Date.now(),
2579
- });
2580
- }
2581
- this.edges.push({
2582
- source: this.findOrCreateFunctionNodeId(caller),
2583
- target: stubId,
2584
- kind: dispatch.edgeKind,
2585
- provenance: 'heuristic',
2586
- metadata: {
2587
- synthesizedBy: dispatch.synthesizedBy,
2588
- [dispatch.metadataTargetKey]: targetName,
2589
- },
2590
- line: lineNum,
2591
- column,
2592
- });
2593
- }
2594
- /**
2595
- * Issue #48: scan one line of VBA source for `DoCmd.OpenQuery "X"` calls.
2596
- * Each match emits ONE `UnresolvedReference` (NOT a stub + edge) so the
2597
- * resolver binds to the REAL `query` node that `SqlQueryExtractor`
2598
- * produces for `queries/<Name>.sql` (dysflow exports every saved QueryDef
2599
- * + `queries.json` manifest). Falls back to silent when the .sql is not
2600
- * yet in the index — the resolver does the binding when it's later
2601
- * indexed, exactly like `vba-me-control` and `vba-forms-bang`.
2602
- *
2603
- * Companion to `scanDoCmdOpenCalls` but intentionally NOT in the
2604
- * dispatch table — OpenQuery's emission shape (`UnresolvedReference`)
2605
- * is structurally different from OpenForm/OpenReport's (synthetic node
2606
- * + heuristic edge). The two pipelines share the literal-vs-Const
2607
- * argument resolution pattern via `localConstants.get(...)` but emit
2608
- * via two different branches of `unresolvedReferences` vs
2609
- * `nodes`/`edges`.
2610
- *
2611
- * UnresolvedReference shape (per Issue #48 spec — must match
2612
- * SqlQueryExtractor's query node name exactly):
2613
- * - `referenceName` = resolved query name
2614
- * - `referenceKind: 'references'` = same kind the resolver binds
2615
- * - `metadata.synthesizedBy: 'vba-opens-query'`
2616
- * - NO synthetic `function` node (W4 graph-pollution invariant — the
2617
- * real `query` node already exists in the index once `.sql` is
2618
- * processed, and creating stubs would compete with the binding).
2619
- */
2620
- scanDoCmdOpenQuery(line, caller, lineNum) {
2621
- const localRe = new RegExp(VbaExtractor.OPEN_QUERY_ARG_RE.source, VbaExtractor.OPEN_QUERY_ARG_RE.flags);
2622
- let m;
2623
- while ((m = localRe.exec(line)) !== null) {
2624
- const rawArg = (m[1] ?? '').trim();
2625
- // Issue #52: same per-proc-with-module-fallback lookup as
2626
- // `scanDoCmdOpenCalls` — proc-local consts resolve to their own
2627
- // values, so a `DoCmd.OpenQuery LOCAL_QUERY` inside `Sub X()`
2628
- // points at the correct query even when another proc declares a
2629
- // different `LOCAL_QUERY` (pre-fix this was whichever wrote
2630
- // the file-wide map last).
2631
- const targetName = rawArg.startsWith('"')
2632
- ? unwrapVbaStringLiteral(rawArg)
2633
- : (this.resolveLocalConst(rawArg) ?? rawArg);
2634
- if (!targetName)
2635
- continue;
2636
- this.unresolvedReferences.push({
2637
- fromNodeId: this.findOrCreateFunctionNodeId(caller),
2638
- referenceName: targetName,
2639
- referenceKind: 'references',
2640
- line: lineNum,
2641
- column: m.index,
2642
- filePath: this.filePath,
2643
- language: 'vba',
2644
- metadata: { synthesizedBy: 'vba-opens-query' },
2645
- });
2646
- }
2647
- }
2648
- /**
2649
- * Regex matching the chained `& "..."` literals that may follow a
2650
- * wrapper's first literal on the same physical line. Captures the
2651
- * literal CONTENT (group 1); the surrounding `&` and quotes are
2652
- * structural, not data. VBA allows whitespace around `&` and around
2653
- * the inner quotes — handled with `\s*`. The `((?:[^"]|"")*)` body
2654
- * mirrors the wrapper regex so a `""` inside a chained literal still
2655
- * decodes to a single `"`.
2656
- *
2657
- * Cross-physical-line concat via `_` continuation is OUT OF SCOPE for
2658
- * v1 (deferred; see commit message).
2659
- */
2660
- static SQL_WRAPPER_CHAIN_RE = /&\s*"((?:[^"]|"")*)"/g;
2661
- /**
2662
- * Given the text that follows a SQL wrapper's first literal on the same
2663
- * physical line, return the contents of every `& "..."` chained literal
2664
- * in source order. Operates per-physical-line only — VBA `_` line
2665
- * continuation across physical lines is handled separately by
2666
- * `collectStringLiteralText` for the variable-assignment path.
2667
- */
2668
- collectSqlWrapperChain(rest) {
2669
- const out = [];
2670
- const re = new RegExp(VbaExtractor.SQL_WRAPPER_CHAIN_RE.source, VbaExtractor.SQL_WRAPPER_CHAIN_RE.flags);
2671
- let m;
2672
- while ((m = re.exec(rest)) !== null) {
2673
- out.push(m[1] ?? '');
2674
- }
2675
- return out;
2676
- }
2677
- scanSqlInLine(line, lineNum, dedupe, sqlVariables) {
2678
- for (const { re } of VbaExtractor.SQL_WRAPPERS) {
2679
- // Each wrapper regex is stateful (has /g); reset before use.
2680
- const localRe = new RegExp(re.source, re.flags);
2681
- let m;
2682
- while ((m = localRe.exec(line)) !== null) {
2683
- const firstLiteral = m[1] ?? '';
2684
- // After the wrapper regex consumes up to and including the closing
2685
- // `"` of the first literal, walk the rest of the line for any
2686
- // `& "..."` chains and concatenate every literal's content. Joining
2687
- // with a space (mirrors `collectStringLiteralText`) keeps adjacent
2688
- // `FROM tblA` & `FROM tblB` separated so `SQL_TABLE_RE` finds both.
2689
- const rest = line.slice(m.index + m[0].length);
2690
- const chain = this.collectSqlWrapperChain(rest);
2691
- const joined = [firstLiteral, ...chain].join(' ');
2692
- this.emitSqlTableReferences(joined, lineNum, dedupe);
2693
- }
2694
- }
2695
- const localRe = new RegExp(VbaExtractor.SQL_VAR_EXEC_RE.source, VbaExtractor.SQL_VAR_EXEC_RE.flags);
2696
- let vm;
2697
- while ((vm = localRe.exec(line)) !== null) {
2698
- const varName = (vm[1] ?? '').toLowerCase();
2699
- const sqlString = sqlVariables.get(varName);
2700
- if (!sqlString)
2701
- continue;
2702
- this.emitSqlTableReferences(sqlString, lineNum, dedupe);
2703
- }
2704
- // Issue #42: `DoCmd.RunSQL <identifier>` (variable form). Mirrors the
2705
- // SQL_VAR_EXEC_RE path above but for the Access-style `DoCmd.RunSQL`
2706
- // idiom — the dominant pattern in real-world VBA modules. Resolve the
2707
- // captured identifier against `sqlVariables` (populated by
2708
- // `trackSqlVariableAssignment` with `&`-accumulate semantics, Issue
2709
- // #13) and feed the resolved SQL string into `emitSqlTableReferences`.
2710
- // Unresolved identifiers (no row in the map) are silently skipped —
2711
- // same graceful-no-op contract as SQL_VAR_EXEC_RE.
2712
- const docmdLocalRe = new RegExp(VbaExtractor.SQL_VAR_DOCMD_RUNSQL_RE.source, VbaExtractor.SQL_VAR_DOCMD_RUNSQL_RE.flags);
2713
- let dm;
2714
- while ((dm = docmdLocalRe.exec(line)) !== null) {
2715
- const varName = (dm[1] ?? '').toLowerCase();
2716
- const sqlString = sqlVariables.get(varName);
2717
- if (!sqlString)
2718
- continue;
2719
- this.emitSqlTableReferences(sqlString, lineNum, dedupe);
2720
- }
2721
- }
2722
- /**
2723
- * #13 fix: `sql = sql & "..."` (self-referential concatenation) must
2724
- * ACCUMULATE the new fragment onto whatever was already tracked for
2725
- * `varName`, not overwrite it. Overwriting silently dropped earlier
2726
- * fragments' tables — typically the initial `FROM <table>` in
2727
- * `sql = "SELECT * FROM tblA"` followed by `sql = sql & " WHERE x=1"`.
2728
- *
2729
- * Detection: the RHS (`m[2]`, trimmed) starts with `<varName> &`,
2730
- * case-insensitively — matching VBA's case-insensitive identifiers (`Sql`
2731
- * and `sql` are the same variable). A genuine fresh assignment (RHS does
2732
- * NOT start with the self-reference) still RESETS tracking — that
2733
- * behavior is unchanged.
2734
- */
2735
- trackSqlVariableAssignment(lines, lineIndex, sqlVariables) {
2736
- const line = lines[lineIndex] ?? '';
2737
- const m = VbaExtractor.SQL_VAR_ASSIGN_RE.exec(line);
2738
- if (!m)
2739
- return;
2740
- const rawVarName = m[1] ?? '';
2741
- const varName = rawVarName.toLowerCase();
2742
- const rhs = (m[2] ?? '').trim();
2743
- const newFragment = this.collectStringLiteralText(lines, lineIndex);
2744
- if (!newFragment)
2745
- return;
2746
- const selfRefRe = new RegExp(`^${escapeRegExpLiteral(rawVarName)}\\s*&`, 'i');
2747
- const existing = sqlVariables.get(varName);
2748
- if (existing !== undefined && selfRefRe.test(rhs)) {
2749
- sqlVariables.set(varName, `${existing} ${newFragment}`);
2750
- }
2751
- else {
2752
- sqlVariables.set(varName, newFragment);
2753
- }
2754
- }
2755
- collectStringLiteralText(lines, startIndex) {
2756
- const fragments = [];
2757
- for (let i = startIndex; i < lines.length; i++) {
2758
- const line = lines[i] ?? '';
2759
- for (const lit of (0, vba_preprocess_1.extractStringLiterals)(line)) {
2760
- fragments.push(lit.text);
2761
- }
2762
- if (!line.trimEnd().endsWith('&'))
2763
- break;
2764
- }
2765
- return fragments.join(' ');
2766
- }
2767
- emitSqlTableReferences(sqlString, lineNum, dedupe) {
2768
- // Scan the SQL string for FROM/INTO/UPDATE <table>.
2769
- // Preserve the source regex's `/u` flag (Unicode property classes)
2770
- // — hardcoding `'gi'` here would silently break non-ASCII identifiers.
2771
- const tableRe = new RegExp(VbaExtractor.SQL_TABLE_RE.source, VbaExtractor.SQL_TABLE_RE.flags);
2772
- let tm;
2773
- while ((tm = tableRe.exec(sqlString)) !== null) {
2774
- let table = (tm[1] ?? '').replace(/[\[\]]/g, '');
2775
- if (!table)
2776
- continue;
2777
- const key = `${lineNum}:${table}`;
2778
- if (dedupe.has(key))
2779
- continue;
2780
- dedupe.add(key);
2781
- this.emitReference(table, lineNum, 0, 'vba-sql-table');
2782
- }
2783
- }
2784
- /**
2785
- * Emit a `references` edge from the file's module/class node to a synthetic
2786
- * node named `targetName`. Used by Dim, WithEvents, Implements, and SQL.
2787
- */
2788
- emitReference(targetName, lineNum, column, synthesizedBy) {
2789
- if (!targetName)
2790
- return;
2791
- // Fix 5: key the synthetic node id on (filePath, kind, name) WITHOUT
2792
- // lineNum so the same type/table referenced on N different lines produces
2793
- // exactly ONE node. The edge carries the per-site `line`/`column`.
2794
- const targetId = (0, tree_sitter_helpers_1.generateNodeId)(this.filePath, 'class', // placeholder kind; cross-file resolution will re-type at lookup
2795
- targetName, 0);
2796
- if (this.synthClassNodeIds.has(targetId)) {
2797
- // Node already emitted for this name — only add the edge below.
2798
- }
2799
- else {
2800
- this.synthClassNodeIds.add(targetId);
2801
- this.nodes.push({
2802
- id: targetId,
2803
- kind: 'class',
2804
- name: targetName,
2805
- qualifiedName: targetName,
2806
- filePath: this.filePath,
2807
- language: 'vba',
2808
- startLine: lineNum,
2809
- endLine: lineNum,
2810
- startColumn: column,
2811
- endColumn: column + targetName.length,
2812
- updatedAt: Date.now(),
2813
- });
2814
- }
2815
- const edge = {
2816
- source: '', // rewritten after module node exists
2817
- target: targetId,
2818
- kind: 'references',
2819
- provenance: 'heuristic',
2820
- metadata: { synthesizedBy },
2821
- line: lineNum,
2822
- column,
2823
- };
2824
- this.edges.push(edge);
2825
- this.pendingModuleOrClassSource.push(edge);
2826
- }
2827
- synthClassNodeIds = new Set();
2828
- /**
2829
- * Issue #50: TempVars placeholder-node de-dup cache. Keys are the
2830
- * deterministic node ids produced by `emitTempVarReference` — those
2831
- * ids intentionally ignore `this.filePath` (use a synthetic
2832
- * `synthetic:tempvar/<key>` path instead) so the SAME key referenced
2833
- * from Form_A.cls AND Form_B.cls collapses to ONE placeholder node.
2834
- * Cross-file id stability is the cross-form state premise that makes
2835
- * `codegraph_explore` connect producer ⇄ consumer in one hop.
2836
- */
2837
- synthTempVarNodeIds = new Set();
2838
- /**
2839
- * Issue #50: emit one TempVar reading/writing site. Per call:
2840
- * - placeholder `class` node keyed on the synthetic
2841
- * `synthetic:tempvar/<key>` file path so cross-file extraction
2842
- * calls collapse to one node per key,
2843
- * - `references` edge from the calling `function` node
2844
- * (via `findOrCreateFunctionNodeId` — same access pattern as
2845
- * `scanDoCmdOpenCalls` / `scanDoCmdOpenQuery`) carrying
2846
- * `metadata.synthesizedBy: 'vba-tempvar'` AND
2847
- * `metadata.access: 'read' | 'write'`.
2848
- *
2849
- * Reuses the synthetic-`class`-placeholder shape established by
2850
- * `emitReference` for SQL tables, events, Dim types, etc. — so every
2851
- * synthesized `references` edge in the file already carries the same
2852
- * `kind: 'class'` target, and downstream consumers (UI, resolvers,
2853
- * search queries) filter on `metadata.synthesizedBy` rather than
2854
- * NodeKind anyway. The `metadata.synthesizedBy` tag cleanly distinguishes
2855
- * TempVars refs from SQL-table refs inside the same NodeKind bucket.
2856
- *
2857
- * Skips when stack is empty (the writer/reader is module-level code
2858
- * — REQ-CODE-4 "unresolvable/runtime reference is silent"). The
2859
- * spec wires this per-proc only; the module-level shape would need
2860
- * a different emission (no procedure source) and is out of scope.
2861
- */
2862
- emitTempVarReference(caller, key, lineNum, column, access) {
2863
- if (!caller)
2864
- return;
2865
- if (!key)
2866
- return;
2867
- const syntheticFilePath = `synthetic:tempvar/${key}`;
2868
- const targetId = (0, tree_sitter_helpers_1.generateNodeId)(syntheticFilePath, 'class', // placeholder kind — see SQL_TABLE_RE emitReference for precedent
2869
- key, 0);
2870
- if (!this.synthTempVarNodeIds.has(targetId)) {
2871
- this.synthTempVarNodeIds.add(targetId);
2872
- this.nodes.push({
2873
- id: targetId,
2874
- kind: 'class',
2875
- name: key,
2876
- qualifiedName: key,
2877
- filePath: syntheticFilePath,
2878
- language: 'vba',
2879
- startLine: lineNum,
2880
- endLine: lineNum,
2881
- startColumn: column,
2882
- endColumn: column + key.length,
2883
- updatedAt: Date.now(),
2884
- });
2885
- }
2886
- this.edges.push({
2887
- source: this.findOrCreateFunctionNodeId(caller),
2888
- target: targetId,
2889
- kind: 'references',
2890
- provenance: 'heuristic',
2891
- metadata: {
2892
- synthesizedBy: 'vba-tempvar',
2893
- // User-facing enum: lowercase single-token string values.
2894
- access,
2895
- },
2896
- line: lineNum,
2897
- column,
2898
- });
2899
- }
2900
- /**
2901
- * Issue #50: scan one line of VBA source for TempVars access sites and
2902
- * emit one `references` edge per site. Three regex runs:
2903
- * - bang form `TempVars!x` over the MASKED line (`!` itself never
2904
- * lives inside a string literal, so masked == unmasked here — using
2905
- * the masked line is conservative against false positives in
2906
- * concatenated string content),
2907
- * - paren form `TempVars("x")` over the ORIGINAL line (the literal is
2908
- * inside a string — masked would strip the key),
2909
- * - Add form `TempVars.Add "x", v` over the ORIGINAL line (same reason),
2910
- * always classified as a write.
2911
- *
2912
- * Each match classifies access by looking at the LINE SUFFIX (after the
2913
- * closing paren / bang-key) on the SAME line: if `=` (and not the
2914
- * nonexistent `==`) is the next non-whitespace character, it's a write.
2915
- * VBA has no `==` so a bare `=` suffix check is safe.
2916
- *
2917
- * The caller parameter comes from `stack[stack.length - 1]` in
2918
- * `sweepCallsAndSql`. Pass `undefined` for module-level access — we
2919
- * drop the edge (per REQ-CODE-4 spirit, runtime references have no
2920
- * static source to anchor against).
2921
- */
2922
- sweepTempVars(maskedLine, originalLine, lineNum, caller) {
2923
- // 1) Bang form — masked line. No string-literal interaction.
2924
- const bangRe = new RegExp(VbaExtractor.TEMP_VAR_BANG_RE.source, VbaExtractor.TEMP_VAR_BANG_RE.flags);
2925
- let bm;
2926
- while ((bm = bangRe.exec(maskedLine)) !== null) {
2927
- const key = bm[1] ?? '';
2928
- if (!key)
2929
- continue;
2930
- const access = detectAssignmentSuffix(maskedLine, bm.index + bm[0].length)
2931
- ? 'write'
2932
- : 'read';
2933
- this.emitTempVarReference(caller, key, lineNum, bm.index, access);
2934
- }
2935
- // 2) Paren form — original line. The literal survives only here.
2936
- const parenRe = new RegExp(VbaExtractor.TEMP_VAR_PAREN_RE.source, VbaExtractor.TEMP_VAR_PAREN_RE.flags);
2937
- let pm;
2938
- while ((pm = parenRe.exec(originalLine)) !== null) {
2939
- const key = pm[1] ?? '';
2940
- if (!key)
2941
- continue;
2942
- const access = detectAssignmentSuffix(originalLine, pm.index + pm[0].length)
2943
- ? 'write'
2944
- : 'read';
2945
- this.emitTempVarReference(caller, key, lineNum, pm.index, access);
2946
- }
2947
- // 3) Add form — original line. Always a write.
2948
- const addRe = new RegExp(VbaExtractor.TEMP_VAR_ADD_RE.source, VbaExtractor.TEMP_VAR_ADD_RE.flags);
2949
- let am;
2950
- while ((am = addRe.exec(originalLine)) !== null) {
2951
- const key = am[1] ?? '';
2952
- if (!key)
2953
- continue;
2954
- this.emitTempVarReference(caller, key, lineNum, am.index, 'write');
2955
- }
2956
- }
2957
- /**
2958
- * Maps `variableName.toLowerCase()` → declared type info.
2959
- * Built by `sweepDimsAndWithEvents`; consulted by the unified qualified-call
2960
- * gate (`shouldProcessQualifiedCall`) so declared project-class locals emit
2961
- * edges, declared primitive/external locals stay silent, and undeclared
2962
- * receivers remain module-name candidates for the resolver (Fix 2 / Issue #2).
2963
- */
2964
- localVarTypeMap = new Map();
2965
- /**
2966
- * Issue #52: Const resolution buckets, scoped per procedure. Key is
2967
- * `'module'` for module-level Consts, or the procedure's `startLine`
2968
- * (stringified) for proc-local Consts. Each bucket maps the lowercase
2969
- * constant name to its simple-literal value (used by `DoCmd.OpenForm` /
2970
- * `OpenReport` / `OpenQuery` argument resolution via `resolveLocalConst`).
2971
- *
2972
- * Two procs declaring the same Const name with different values stay
2973
- * isolated (each in its own bucket); reads look up the current proc's
2974
- * bucket first and fall back to the module bucket. The bucket-per-proc
2975
- * model was chosen over a single file-wide Map (the pre-fix shape) so
2976
- * `DoCmd.OpenForm FORM_DESTINO` resolves to the proc-local value, not
2977
- * whichever was written last.
2978
- */
2979
- localConstants = new Map();
2980
- /**
2981
- * Issue #52: shared lookup helper for `scanDoCmdOpenCalls` and
2982
- * `scanDoCmdOpenQuery`. The current scope is the procedure whose
2983
- * `startLine` is on top of `procStack` (or `'module'` when the stack is
2984
- * empty). Per-proc bucket first; module bucket is the fallback.
2985
- */
2986
- resolveLocalConst(name) {
2987
- const lower = name.toLowerCase();
2988
- const procBucket = this.localConstants.get(this.currentProcKey);
2989
- if (procBucket) {
2990
- const v = procBucket.get(lower);
2991
- if (v !== undefined)
2992
- return v;
2993
- }
2994
- const moduleBucket = this.localConstants.get('module');
2995
- return moduleBucket?.get(lower);
2996
- }
2997
- /**
2998
- * Issue #52: shared writer. `scopeKey` is `'module'` or the procedure's
2999
- * startLine-as-string. Creates the bucket lazily so callers do not have
3000
- * to pre-allocate per proc. Returns the bucket the value was written to
3001
- * (mostly useful for tests; production code ignores it).
3002
- */
3003
- setLocalConstInScope(scopeKey, name, value) {
3004
- let bucket = this.localConstants.get(scopeKey);
3005
- if (!bucket) {
3006
- bucket = new Map();
3007
- this.localConstants.set(scopeKey, bucket);
3008
- }
3009
- bucket.set(name.toLowerCase(), value);
3010
- return bucket;
3011
- }
3012
- /**
3013
- * Issue #52: the current Const-lookup scope. `'module'` when no procedure
3014
- * is open, otherwise the top-of-stack proc's `startLine` as a string.
3015
- * Both `sweepEnumsAndConsts` (to decide whether to emit a `constant`
3016
- * node) and `sweepCallsAndSql` (to drive OpenForm/OpenQuery resolution)
3017
- * keep this in sync with their per-line stack walk by pushing/popping
3018
- * `procStack` and writing the new top's key here.
3019
- */
3020
- currentProcKey = 'module';
3021
- /**
3022
- * Issue #52: per-extraction proc-stack shared between `sweepEnumsAndConsts`
3023
- * and `sweepCallsAndSql`. Each sweep clears it at the start so the file's
3024
- * mid-proc structural state never leaks across sweeps. Holds the
3025
- * `startLine` (1-based, matches `ProcInfo.startLine`) of every procedure
3026
- * whose body the sweep has not yet emitted `End Sub`/`End Function`/
3027
- * `End Property` for.
3028
- */
3029
- procStack = [];
3030
- /** Local event name (lowercase) → event node for `RaiseEvent` edge emission. */
3031
- localEvents = new Map();
3032
274
  }
3033
275
  exports.VbaExtractor = VbaExtractor;
3034
- /**
3035
- * #13 helper: escape a variable name for safe interpolation into the
3036
- * self-reference RegExp built by `trackSqlVariableAssignment`. VBA
3037
- * identifiers are alphanumeric+underscore only, so in practice nothing here
3038
- * ever needs escaping — this guards against regex metacharacters anyway
3039
- * rather than assume the input is always well-formed.
3040
- */
3041
- function escapeRegExpLiteral(value) {
3042
- return value.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
3043
- }
3044
- function parseConstDeclarations(body) {
3045
- const declarations = [];
3046
- for (const part of splitOutsideVbaStrings(body, ',')) {
3047
- const m = /^\s*(\p{L}[\p{L}\p{N}_]*)\s*(?:As\s+[^=]+?)?\s*=\s*(.+?)\s*$/iu.exec(part);
3048
- if (!m)
3049
- continue;
3050
- const name = m[1] ?? '';
3051
- const rawValue = (m[2] ?? '').trim();
3052
- declarations.push({
3053
- name,
3054
- value: rawValue.startsWith('"') ? unwrapVbaStringLiteral(rawValue) : rawValue || null,
3055
- });
3056
- }
3057
- return declarations;
3058
- }
3059
- function splitOutsideVbaStrings(value, separator) {
3060
- const parts = [];
3061
- let current = '';
3062
- let inString = false;
3063
- for (let i = 0; i < value.length; i++) {
3064
- const ch = value[i];
3065
- const next = value[i + 1];
3066
- if (ch === '"') {
3067
- current += ch;
3068
- if (inString && next === '"') {
3069
- current += next;
3070
- i++;
3071
- continue;
3072
- }
3073
- inString = !inString;
3074
- continue;
3075
- }
3076
- if (!inString && ch === separator) {
3077
- parts.push(current);
3078
- current = '';
3079
- continue;
3080
- }
3081
- current += ch;
3082
- }
3083
- parts.push(current);
3084
- return parts;
3085
- }
3086
- /**
3087
- * Issue #50 helper: return true iff `line.charAt(fromIndex..)` (after the
3088
- * matched TempVars site, including any trailing whitespace) holds an `=`
3089
- * (write-assignment) and not a `==` (VBA has no `==` operator, so a bare
3090
- * check for `=` is safe). Used by `sweepTempVars` to classify a `TempVars`
3091
- * access as read vs write from the local line context.
3092
- *
3093
- * We skip over trailing whitespace only — `.`, `(`, `<`, `>`, `*`, `+`
3094
- * etc. all mean "this isn't an assignment target", and bare `=`
3095
- * is the only access-suffix shape VBA syntax allows here. A line like
3096
- * `TempVars!x = 1` (write) or `Debug.Print TempVars!x` (read) land in
3097
- * the right bucket without further analysis.
3098
- *
3099
- * Strings should already be masked out of `line` (`maskStringContent`)
3100
- * when this is called for the bang form (no string in scope anyway);
3101
- * the paren form passes the ORIGINAL line — but for THAT form the
3102
- * captured match ends at the closing `"` of the literal, so any `=`
3103
- * that follows is unambiguously outside the string.
3104
- */
3105
- function detectAssignmentSuffix(line, fromIndex) {
3106
- let i = fromIndex;
3107
- while (i < line.length) {
3108
- const ch = line[i];
3109
- if (ch === ' ' || ch === '\t') {
3110
- i++;
3111
- continue;
3112
- }
3113
- return ch === '=';
3114
- }
3115
- return false;
3116
- }
3117
- function unwrapVbaStringLiteral(raw) {
3118
- const trimmed = raw.trim();
3119
- if (!trimmed.startsWith('"'))
3120
- return trimmed;
3121
- let text = '';
3122
- for (let i = 1; i < trimmed.length; i++) {
3123
- const ch = trimmed[i];
3124
- const next = trimmed[i + 1];
3125
- if (ch === '"' && next === '"') {
3126
- text += '"';
3127
- i++;
3128
- continue;
3129
- }
3130
- if (ch === '"')
3131
- break;
3132
- text += ch;
3133
- }
3134
- return text;
3135
- }
3136
- /**
3137
- * Hueco 3 helper: parse an Access event-handler Sub name into its
3138
- * `<ControlName>_<EventName>` components. Returns null when the name does
3139
- * not match the convention, when the split yields an empty segment, or
3140
- * when the control name is `Form` (form-level events are NOT control
3141
- * handlers — `Form_Load` is the form's own lifecycle event, not a
3142
- * command-button click).
3143
- *
3144
- * Splitting on the LAST underscore (rather than the first) lets
3145
- * multi-word event names parse correctly: `ComandoAltaPM_BeforeDelConfirm`
3146
- * yields control=`ComandoAltaPM`, event=`BeforeDelConfirm`, not
3147
- * control=`ComandoAlta`, event=`PM_BeforeDelConfirm`.
3148
- */
3149
- function parseEventHandlerName(name) {
3150
- if (!name)
3151
- return null;
3152
- const lastUnderscore = name.lastIndexOf('_');
3153
- if (lastUnderscore <= 0)
3154
- return null; // no underscore OR starts with underscore
3155
- const controlName = name.slice(0, lastUnderscore);
3156
- const eventName = name.slice(lastUnderscore + 1);
3157
- if (!controlName || !eventName)
3158
- return null;
3159
- // Form-level events live on the form, not on a control.
3160
- if (controlName.toLowerCase() === 'form')
3161
- return null;
3162
- return { controlName, eventName };
3163
- }
3164
276
  //# sourceMappingURL=vba-extractor.js.map